Home

Brain Vision Analyzer User Manual Version 1.05

1. scsssseeeeeeeeeeeeeeeeeeeneeeeeeeeeeeeeeeeeeeeeees 197 Annex C Markers time markers cccccceeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeseeeseeeeeeeess 198 Annex D Keyboard shortcuts tices nicsiinseseaisatiteninceeianeteneducrscanietusnanensieutointnceintusndeae 200 Annex E Installation Network License USB ccccceseeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeees 201 Annex F Individual user profiles ccccccesssseeeeeeeeeeeeeeeeeseeeeeeeeeeeeeeeeeenseeeeneees 202 Annex G Command line parameters ccccssseeseeeeeeeeeeeeeeeeeeeneeeeeeeeeeeeeeeeeeeeeeeees 203 Annex H Links to raw Gata icccicisiceisinsninnsieccsninesnsinndunniecsssiwenusnensuusienseninsnnseusans 204 Vision Analyzer User Manual 5 1 Product declaration 1 1 Product identification Product name Brain Vision Analyzer Vendor Brain Products GmbH Stockdorfer StraBe 54 D 81475 Munich Classification in accordance with German legislation covering medical products Class I UMDNS number Analysis software for EEG and evoked potentials 16 307 This product conforms to the Medical Device Directive MDD 93 42 EEC 1 2 Area of application The Brain Vision Analyzer is used to analyze EEG signals using a personal computer The program may only be used by doctors or suitably trained personnel for research purposes only 2 Introduction The Vision Analyzer evaluates raw EEG data both for spontaneous EEG analyses and for evoked potentials Am
2. 1 0 50 0 75 2 0 50 1 22 3 0 66 1 99 4 1 08 3 23 5 1 75 5 26 6 2 85 8 56 7 4 64 13 92 8 7 99 22 66 9 12 29 36 87 10 20 00 60 00 Cancel Fig 10 65 Selection dialog of a layer Vision Analyzer User Manual 157 10 2 Secondary transforms 10 2 1 Grand Average This module enables one or more grand averages to be created for various averages The result is stored in secondary history files in the current workspace x m Input History Nodes amp Output Files Gk Insert Line Cr Average GrandAverage Remove Line Remove All m Input History Files JV Primary History Files Only Available Files Selected Files C Use Whole Workspace Select Individual History Files Selection Filter f Refresh Add gt lt lt Remove Add All gt gt Remove All IV Create a Data Set for Standard Deviation Load Parameters IV Enable Individual Channel Mode Save Parameters Fig 10 66 Grand Average dialog The following options are available in the dialog e Name of the Involved History Nodes The names of the involved data sets are separated by commas e Output File Names of the output files e Primary History Files Only The following selection can be confined to primary history files if you want to e Use Whole Workspace This includes all files in the workspace e Select Individual History Files 158 e Selection Filter With this option you can
3. Invertible discrete wavelet transform Discrete Wavelet Transformation Window Function S x Tas it C No Window Hanning Window Hamming Window Window Length val 4 Fig 10 61 Window function in the invertible discrete wavelet transform If you selected the invertible discrete wavelet transform instead of the discrete one you should select a window function and its width on the screens presented subsequently These selections influence how the wavelet transform is applied to the data This processing step is needed for the same reason that a window function is used in the normal spectral analysis As in that analysis this one works with data ranges of finite length for the wavelet analysis so that discontinuities would occur in the transform that would make it impossible to perform a complete reverse transform of the signal Vision Analyzer User Manual 151 Continuous wavelet transform Continuous Wavelet Transformation x Wavelet Morlet Complex Minimal Frequency Hz fi Time Domain Maximal Frequency Hz fi 5 S 0 55 Frequency Steps fi 0 IV Linear Steps Instead of Logarithmic Morlet Parameter c j2 8 Scales Filter Borders Frequency Domain Frequency Low High i 1 00 0 74 1 26 2 256 1 88 3 23 3 411 3 03 5 19 4 5 67 4 18 7 16 5 7 22 5 32 912 6 8 78 6 47 11 09 7 10 33 7 61 13 05 8 11 89 8 76 15 02 9 13 44 931 16 98 10 15 00 11 05 18 95 lt Z
4. A slightly different dialog appears if you chose individual channel mode Here you can also mark sections directly on the channels as artifacts with the mouse To delete a mark just click it This opens a popup menu with the Delete Artifact option The dialog contains the following elements e Display of the current segment number Segment x of y e A window with the text No Artifact or Segment with Artifacts indicating whether an artifact has been marked anywhere in the current segment e The Clear All Channels button to remove all artifact marks in the current segment and move on to the next segment e The Mark All Channels button to mark all channels as having artifacts and move on to the next segment e The lt lt button to move to the previous segment e The gt gt button to move to the next segment e The Goto button to go to a specific segment e Step Only to Clean Segments If you select this check box the program goes to the nearest previous segment that does not contain any artifact marks when the lt lt button is clicked The equivalent applies to the gt gt button for subsequent segments e Step Only to Segments with Artifacts This check box has the exact opposite effect to the previous one e Segments with Artifacts All marked artifacts are listed here You can display an artifact by double clicking it When you have made your choice click the OK button 10 1 1 2 Semiautomatic segment selection In
5. e Do Not Store Data Calculate Data on Demand e Cache Data to a Temporary File 136 e Store Data Compressed in History File with the option of specifying the resolution in nanovolts Segmentation Wizard Step 1 x m What kind of segmentation would you like to be performed Create new segments based on a marker position Divide data set in equal sized segments C Set new segments manually m How should the data be stored C Do not store data calculate data on demand Store data compressed in history file oro Fig 10 50 First page of the Segmentation dialog If you opt for marker based segmentation the next dialog page enables you to choose one or more markers from a list of available markers and to include them in the group of selected markers with the Add button Here you also see the input box for the above mentioned ABE Vision Analyzer User Manual 137 Segmentation Wizard Step 2 of 3 Which markers would you like to include Available Markers Selected Markers Stimulus 5S 1 400 Stimulus 5S 2 100 Add gt gt Advanced Boolean Expression _ TEN Fig 10 51 Second dialog page of marker based segmentation Segmentation Wizard Step 3 of 3 m Start and end of the new segments relative to the position of the selected markers Start ms j 300 End ms 300 Duration ms 1200 Based on data points rt point 0 Ef point 255 Points 2
6. n z m i ees Select Overlays E m 7 mo o y o A A U N _ oOo Cancel Fig 7 8 Grid View Settings dialog Overlays tab The Overlays tab contains settings that affect overlaid curves Here you can define the appearance and labeling of overlaid curves A table enables you to set the details separately for each overlaid curve There are the following options e Monochrome or Color This option defines whether the overlays should appear in monochrome or in color If you choose monochrome line patterns appear in the table for different overlaid curves Otherwise colors appear e Line Width Here you set the line width e Insert Line Remove Line and Remove All You can insert another line at the current position remove the current line or remove the entire table e Reference Node Overlays Here you define how the reference channel and the overlaid channel s are labeled These entries only have an effect if data sets are overlaid as described below in the Overlaying different data sets subsection When assigning names you can use placeholders which are replaced by current values when the curves are displayed One placeholder is n for example If you use this placeholder it is replaced by the name of the data set involved Here is a list of available placeholders Vision Analyzer User Manual 29 Placeholder Meaning c Channel name
7. selected node or pressing the Next gt button takes you to the second dialog in which you can select the channels for the LRP LRP Wizard Step 2 of 2 Channels x Please select the channels for LAP calculation Insert Line Average Left Hand Average Right Hand Remove Line Chn ChnQ Chn ChQ Remove all 2 P4 q B M P3 BUE S oS TERS TES E De TES TERS TES E ELE TER OES E S En CERE oo a neS SERE TES E Dae TES TEE 2ER E D TES TERS TER E D TES TERE TER E nun of nEn ERS al a ee oe TER TEN H IE CER ENO E D TER TERE TER E 15 aE ES H DS oo i lt Back Cancel Fig 10 30 Selection dialog for channels In calculating the LRP the module forms the difference between contralateral and ispsilateral electrodes in the parent node and the reverse difference in the second node and calculates the average of these two differences For this reason the two left hand columns of the channel list are assigned to the first data node parent and the two right hand columns are assigned to the second data node It is generally sufficient to fill the two left hand or two right hand columns The dialog module then searches for the corresponding channels in the other data set and fills the other columns accordingly when the associated channels have been found and the input boxes are still empty This specification can of course be changed afterwards If no suitable channels are found all columns have to be
8. Save to File Fig 10 24 Formula Evaluator dialog Formulas are input in a dialog You enter the name of the new channel containing the data from the calculation in the left column Name In the middle column Formula you input the formula that you want to use in calculating the data This is largely case insensitive i e no distinction is generally made between uppercase and lowercase In the right column Unit you can specify the unit for the new channel You can save formulas in ASCII files by clicking the Save to File button and import ASCII files containing formulas into the dialog by clicking the Load from File button If you want to create ASCII files with an editor or other program every line in the file must be in the following form Name Formula With the Keep Old Channels option you can keep old channels from the previous data set which have not been redefined without any changes In this case you can select New Channels on Top to ensure that the new channels are placed at the beginning There are the following options for formulas and they can be combined e The operators and as well as parentheses Mathematical rules of precedence are observed e Channel names these are interpreted point by point 86 e The constants e pi and i imaginary unit as well as numeric inputs e The mathematical functions sqrt square root abs In natural logarithm log logarithm to base 10 sin cos ta
9. Vision Analyzer User Manual 101 addition ranges that are marked as bad intervals are excluded from the calculation Correction is performed by subtracting the calculated average blood pulse curve from the EEG This procedure is carried out separately for each ECG episode and each channel Parameter settings A dialog takes you through the various parameters that have to be set In the first step you specify the methods to be executed The following options are available to you here e Detect and Correct Scanner Artifacts This option switches the calculation and correction of scanner artifacts on and off e Detect and Correct Pulse Artifacts This option switches the calculation and correction of blood pulse artifacts on and off e Store Corrected Data in History Node This option allows you to specify that the data resulting from the correction is to be stored in the history node This makes any subsequent analytical operations significantly faster Here you can also specify whether the data is to be compressed before it is stored and define the compression level as a voltage resolution MRI Artifact Correction General Settings 7 x V Detect and Corect Scanner Artifacts I Store Corrected Data in History Node F Use Compression Resolution m 500 Figure 10 31 Step 1 of the MR Artifact Correction Wizard If you select detection and correction of scanner artifacts in the second step you receive a dialog bo
10. Bell T P Jung D Ghahremani T J Sejnowski Blind separation of auditory event related brain responses into independent components Proc Natl Acad Sci USA 94 10979 10984 1997 Vision Analyzer User Manual 91 10 1 18 Level Trigger This transform enables you to set a marker of the Threshold type on one or more channels when the voltage exceeds or falls below a certain level These markers can be used as a basis for segmentation One application would be for example muscle activities which are to be used as a reference in averaging Another application would be the use of an analog channel as memory for stimuli in acquisition systems which do not have digital inputs Here different stimuli can be coded as different voltage levels Level Trigger Ed Insert Line 45 00 Positive Trend 4 Remove Line Remove All e 0 Time Tolerance ms fo Carosi Fig 10 27 Level Trigger dialog You have the following setting options in this dialog e Name Here you specify a name for the threshold in question e Threshold You input the threshold in uV in this box e Direction A list box enables you to specify the direction of the voltage shape which sets the marker when the threshold is reached Positive means that the marker is set when voltage is rising and the threshold is reached Negative sets the marker when voltage is falling and the threshold is reached e Channel Here you select the
11. Raw Data Choose Data Set History Node Available Files Selected Files Date Set m Message Logfile V Use Logfile Add All gt gt RemoveAll Gk Cancel Fig 8 2 History template application dialog Your first option in this dialog is to select a history template under Select History Template If you have not closed the history template window then the name of the current history template appears here Otherwise click the Select button and pick a history template In the next input group Starting Position in History File s you define whether the template is to be applied to the initial data set Root of the history files or to one of the subsequent data sets Choose Data Set If you choose the second option enter the name of the data set to which you want to apply the template Note that if there are several data sets of the same name in a history file only the first one that is found will be taken into consideration If you enable the Use Logfile option then all messages that are usually output in a dialog will be written to a log file In this case all Yes No questions are automatically set to Yes This prevents automatic processing from being interrupted while the program waits for an input The log file is displayed when processing has finished Now you can choose the history files to which you want to apply the template You have the following options here e Primary History F
12. This may be necessary if you interrupt data acquisition and want to resume it later with a new output file The raw EEGs are not actually appended as such Instead they are assigned to a single history file i e they are linked virtually The prerequisite for appending a raw EEG to another one is that the main properties of the two data sets such as channel name sampling rate etc are identical To append a raw EEG you first select the output file in the History Explorer Then click the book icon with the right mouse button A context menu appears Choose Append File from it A warning is output that any transforms you have carried out with this history file will be lost Accepting this warning takes you to a menu in which you can select the raw data EEG that you want to append When you have confirmed your choice the first history file is modified and the second is removed The book icon has changed to a book stack icon You can also append other raw EEGs to this history file If you want to know which raw EEGs have been appended select the context menu with the right mouse button again An additional menu item List Appended File s appears now Selecting this item displays a list of appended files To cancel this link you have to terminate the program and delete the associated history file xxx ehst and the history information file xxx hfinf in the history file folder The next time you start the Analyzer these files wil
13. al Standard Montage 00 00 00 Segment 1 1 P300 Fig 7 12 3D mapping view Vision Analyzer User Manual 35 7 7 Special properties of frequency views standard grid and head In addition to the properties that were explained above in the sections on standard grid and head views frequency views have some special properties which you can again access with the Set Display Features button on the tool bar Gridview Settings Ed Display Bands Ordinate Displayed Range gessenssnsceneeseny Voltage Begin Hz 15 C Power End Hz 35 Drawing m Display Style C Draw Data as Block M Show Legend Draw Data as Graph M Show Markers Scaling Tl Logarithmic Scaling Abbrechen Fig 7 13 Display tab of the Grid View Settings dialog You can set the following on the Display tab e Ordinate Voltage or Power display in uV or uV e Drawing Draw Data as Block or Draw Data as Graph e Scaling Logarithmic Scaling as an alternative to linear scaling e Displayed Range of frequencies e Show Legend e Show Markers 36 The Bands tab lets you define frequency bands Gridview Settings Ea Display Bands Name Begin Hz End Hz mee fo fos Delta 0 5 3 5 Theta 3 5 7 5 75 fi25 eta fiz5 f fo fe ee ee i ee a LEE Color Select Color Select Color e Select Color p Select Color n Select Color a Select
14. and whether those episodes which do not satisfy the above correlation criterion are to be flagged with a Low Correlation marker in order to facilitate subsequent problem analysis e Use Sliding Average Calculation This third method does not impose such strict criteria on the quality of the MR artifact episodes as the one described above Instead it addresses another problem of combined EEG and MRI measurements If the test subject changes the position or orientation of his head in the scanner even slightly then the resulting EEG artifacts can sometimes be modified significantly This would degrade a template calculated over all sections and the reduced correlation of MR episodes that have then changed compared to the original template would result in these episodes not being included in the subtraction template The calculation of the template on the basis of a sliding average over a given number of MR episodes is a good solution to this problem since it takes account of a certain level of fluctuation in artifact occurrence on the one hand and because of the averaging operation leads to a highly stable subtraction template on the other e Use Template Drift Compensation Vision Analyzer User Manual 107 This method uses several artifact templates in order to reduce interference due to drifts by fractions of a sampling interval in the scanned intervals with reference to one another template drift The underlying idea is to perform averagin
15. artifacts entirely For this reason the data in the corrected ranges can and should be post processed using the filters integrated in the module Markers can be used to define the intervals affected by scanner artifacts These can be written by the scanner directly or by an appropriate macro It is also possible to set the markers by hand although this is not recommended since the quality of the averaged artifact curve which is subsequently subtracted from the individual sections is a direct function of the precision with which the markers are set during the signal s timecourse It is also possible to mark a reference point generally the beginning of a scanned interval by means of methods implemented in the module There are two methods available for this the power method and the gradient method In the power method the average power is calculated for each data point across a selection of channels If this value exceeds a set threshold a Scan Start marker is set The scanned interval is then set relative to this marker in accordance with the set interval range In the gradient method the procedure is similar except that in this case the average gradient of the curves between two data points is calculated in the selected channels If this exceeds the set threshold a Scan Start marker is set However for both methods it should be noted that the accuracy of detection of the start of the artifact is not generally improved by ca
16. c1 Avg c1 c2 G J Avg c2 StdDev c1 StdDev c2 i 1 segment length j 1 segment length segment length 58 Avg average of all the values in the segment StdDev standard deviation of all the values in the segment The value 0 is added to the channels for the purpose of calculation outside the segment e Then an average is formed across all segments for each of these pairs When this transform is applied to the data in the frequency domain it is also possible to input a frequency band to which the calculation should be confined In the dialog select the channels between which you want to calculate the cross correlation and the frequency band if you are operating with data in the frequency domain Here you also have the option of selecting the entire frequency range To calculate the auto correlation of a channel specify its channel name both in the left and right list box of a row in the input table As the result you get the channels in which the average of the cross correlations are recorded for the time or frequency shift Vision Analyzer User Manual 59 10 1 4 Band rejection filters Band rejection filters are generally used in EEG analysis when the EEG signal is overlaid by a spurious signal of constant frequency Spurious signals on the electricity network or from electrical activities of poorly shielded monitors are typical examples Whereas these sources of interference can be eliminated
17. cnnekneetenannnetindaken eae 171 DIST EOS el AEE EE E E EEE EE EE EE TE 171 11 1 2 Genere Data EXDON romerret TAEAE EEE 172 EUs Makr EXPO basen tns Site is suse i a a EEE 175 11 2 Extended export components ssssssssseeeseeeeeeeeeeeeseereetertrrrrrrrererrerrereereereene 176 11 2 1 Area Information EXPO susancossssansosnasansoDiasarsnansbanse DsbanananadanssDedarsnannsaninns 176 11 2 2 Peak Information Export wissagzachaasagesee ashe aon aaheacha cade abaoshgashuosdpashaosneaceposenaia 178 12 Importing data positions ANd Markel ccseeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeees 180 12 1 Importing data stat ces tects atest tt stb thet tbe mst seks tt 180 12 t 1 Besa f rmase ee TR TER ER TN MNT STA STOR Sem Tee 180 124 2 AGONCHC Data Reader dcinsciansiqutiantiantiankianRiantiantiantiantiantinaRiantiunRianteaetial 181 12 2 Importing markers and channel positions cccceeeeeeeeeeeeeeeeeeeeeeeeeeeeees 188 BS PHU vi cctierasencicwteciscicamaenent A A A 189 14 Exporting graphics wuhtinastnyretndaiivtinesitatnd init etinncetutncealautasdphhutinuibubtcaphiutinain 192 15 Appending multiple raw data Sets cccccccssssesseeeeeeeeeeeeeeeeeeseeeeeeeeeeeeeneees 193 16 SOLUTIONS aieas ccs ccus nccccarandusecwendeaceceatuansweseuansvancedsnseadwessueatvencsduaaudanvencadsavendie 194 Annex A Raw data on removable Media eeecceeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeneees 196 Annex B Electrode Coordinate system
18. e Use Marker Existing markers are used to set the pulse intervals These may have been created by external programs macros or during previous runs of the module e Use Peak Detection An integrated method is used to detect R peaks e ECG EEG Channel This is the channel to be examined which is generally an ECG channel but an EEG channel with significant pulse artifacts can also be used e Derive ECG 110 This option should only be used in rare cases in which the ECG signal is recorded in integrated form in the MR system due to special physical circumstances This integrated signal can then be converted into a directly interpretable ECG signal by forming the first derivation e Pulse Rate This is the average pulse rate with the permitted deviations These specifications can be made as time values or as beats per minute BPM together with the permitted upper and lower deviations e Use Default R Peak Marker Name R This option allows you to specify the name of the marker that is to be written e Use Direct R Peak Detection Method A method based on AI198 is used to detect the R peaks e Use Coherence Method The method used to search for R peaks continuously determines both the coherence and also the mean amplitude correlation between the current data section and a template see above The threshold values for the coherence and the limit values for amplitude correlation can be specified in the dial
19. e Check Low Activity in Intervals If you select this check box the Low Activity criterion is applied e Lowest Allowed Activity Specify the minimum allowed activity here e Interval Length Specify the interval length within which activity is not allowed to fall below the minimum e Mark as Bad See the Gradient criterion When you have input all criteria and clicked the OK button the ranges are calculated If the Raw Data Inspector has marked intervals it shows you the channels most affected by the criteria You can remove individual channels from the calculation in a dialog Then a standard view appears with a dialog on the right Here you have the same options as described above for manual inspection In addition you can change criteria remove channels from calculation Disable Channel and look up statistics on the artifact intervals that were found 130 Raw Data Inspector Change Criteria Show Statistics Goto Next Artifact gt gt lt lt Gow Previous Arifact Tl Individual Channel Mode Artifact Channel Position Multiple 01 770 856 s Multiple 02 771 036 Cancel Fig 10 47 RDI dialog for semiautomatic inspection Click the OK button when you have finished inspecting the EEG 10 1 26 3 Automatic inspection Automatic inspection corresponds to semiautomatic inspection except you have no opportunity for making corrections manually Vision Analyzer User Manual 131 10 1 27 Recti
20. e Export types Area Mean Activity and Area as Raw Sum of Activity Values If the first option is selected the area is exported whereas if the second option is selected the average activity is exported The third option is only available to you for frequency data In this case the sum of the spectral line values of the defined range is exported without reference to the spectral line width e Output file The name of the output file is specified here Vision Analyzer User Manual 177 11 2 2 Peak Information Export This component exports data on the position and voltage of peak markers which have been set in selected history files or nodes The data is exported in an ASCII table These markers will normally have been set first with the Peak Detection transform Input p Options Peak name ee M Overwrite default decimal symbol Namefs of the involved data sets history nodes separated by commas Decimal symbol _ Peas J Export mean value around peak amts za IV Primary history files only Export individual latencies for each channel C Use whole workspace Select individual history files Output Selection filter f Refresh r Folder c Vision Export Output file Peaks txt Available Files Selected Files Add All gt gt x Cancel Fig 11 6 Peak Information Export dialog The following options can be set in this dialog e Peak Name As y
21. frequency that has been set i e 50 Hz 2 5 Hz or 60 Hz 2 5 Hz The edge rise is always 24 dB octave The cutoff frequency and time constant specify the frequency at which the signal is 3 dB less than the input signal i e the input signal has been reduced by about 70 The filters have been implemented as phase shift free Butterworth filters As far as high pass and low pass filters are concerned it is also possible to choose their slope This can be 12 dB oct 24 dB oct or 48 dB oct To minimize the influence of discontinuities on the filter and transient phenomena it is advisable to use this filter prior to segmentation m Low Cutoff M Enabled Frequency Hz 0 530516 Time constant s 0 3 Slope dB oct fiz 7 a ancel m High Cutoff IV Enabled Frequency Hz fro Slope dB oct fiz 7 m Notch I Enabled Frequency Hel 50 z Fill table with values from above m 0 530516 0 300000 m 70 000000 O m 0 530516 0 300000 0 m 70 000000 0 O so 0 530516 0 300000 12 m 70 000000 12 O e m 0530516 0 300000 12 m 70 000000 12 O Eo m 0 530516 0 300000 12 m 70 000000 12 O io m 0 530516 0 300000 J2 m 70 000000 j2 D E m 0530516 0 300000 12 m 70 000000 12 O a m 0530516 0 300000 12 m 70 000000 12 O E m 0 530516 0 300000 12 p 70 000000 j2 o m 0 530516 0 300000 12 m 70 000000 12 O 0 m 0 530516 0 300000 12 m 70 000000 12 O E
22. mapping view If a fixed number of maps is defined the interval between them is adjusted accordingly If this option is not set the interval between them is kept constant and the number of maps is changed accordingly Use Average Value of Interval When this check box is selected the average value of the selected interval is used to calculate the maps Otherwise only the first point of the interval is used The following algorithms are available for calculation Triangulation and Linear Interpolation This algorithm is explained below Interpolation by Spherical Splines This algorithm is also explained below Other setting options are Quick Graphics In Quick Graphics mode not every pixel of the map is calculated Instead only the values at the points of a rectangular grid with a certain resolution are calculated Then every rectangle of the grid is filled with the calculated color The result is a map with a lower resolution which can be calculated much faster Grayscaling Setting this option changes output from color to gray scales Show Electrodes Selecting this check box causes the electrodes to be shown on the map as small circles Automatic Scaling In this case the program calculates optimum scaling Manual Scaling This option is an alternative to automatic scaling Here you specify the voltage interval to be covered by the color spectrum displayed View From You can select one or more different views of the map
23. Color za Select Color E Select Color a Select Color ia Select Color Insert Line Remove Line Remove All Ke Abbrechen Fig 7 14 Bands tab of the Grid View Settings dialog Here you can input the name of a band together with its beginning and end in hertz Click the Select Color button to change the color This brings up a dialog in which you can choose the color for the band Three buttons are available to edit lines Insert Line Remove Line and Remove All All frequency ranges that are not defined here are shown in black Vision Analyzer User Manual 37 7 8 Block markers and transient transforms If you press the left mouse button somewhere between the channels in the standard view a bluish green block marker becomes visible If you hold the mouse button down and move the mouse the size of the block changes When you release the mouse button a context menu appears with the choice of transforms Zoom FFT and Map These transforms are transient i e their results are not stored anywhere They are only intended for visual inspection of the current data If you choose one of these transforms the current data window is split and the result of the transform appears on the right This relates to the current block If you change the block position by moving the mouse over it pressing the left mouse button down and moving the mouse in a horizontal direction the result on the right is
24. The one node should be segmented after the first condition e g movement with the left hand and the second node after the second condition e g movement with the right hand Then you should apply an average to these two nodes LRP Wizard Step 1 of 2 Second Dataset First dataset is p300a Raw Data Filters Segmentation Left Hand Average Left Hand Please select the second dataset for LAP calculation p300a Expand History File Raw Data Filters Expand Same Names Segmentation Left Hand Average Left Hand Expand All Segmentation Right Hand ese Average Right Hand Collapse All I Keep channels of parent Back Cancel Fig 10 29 Selection dialog for the second data set The LRP module then calculates the lateralized readiness potential and writes the result in new channels which have a name in the form LRP To run this calculation you must first select the path of the second data set in the dialog The structure of the selection tree in the dialog is like the Analyzer s history tree You select the path for the node by clicking the node You can search for the node by expanding certain parts of the tree with the buttons on the right edge You can also use the Keep Channels of Parent check box to define whether the channels of the parent node should be included in the new data set or whether the new data set should only consist of LRP data Double clicking the 96
25. are located on the Export menu In a similar way to transforms there are two categories of export components simple export components and extended export components The former always have one input data set ASCII exports are an example of this category Extended export components relate to multiple input data sets Peak Export which creates a table of peak values from a selection of history files or nodes is an example of this category Extended export components appear at the bottom of the Export menu kept apart from the simple export components by a separator Simple export components can operate with templates whereas extended export components cannot They can only be executed when a data set is being displayed and always relate to the active data set window in basically the same way as primary transforms The simple export components from Brain Products provide placeholders which stand for the name of the history file h and the name of the current data set n The advantage of using placeholders is that export components can also be used in history templates without existing exported files being overwritten constantly If one history file is named EEG1 and another EEG2 for example and the Average data set is exported for both of them and if the name h_ n is entered then the resultant file names will be EEG1_Average and EEG2_ Average If you mark a block in the current data set the simple export components allow just thi
26. atasetOneChannel Raw Data TESTOOO Raw Data Untitled Raw Data Use Relative Path for Filters z Template or Drag and Drop lt Back Finish Cancel Fig 10 57 Selecting the t test type of the second data set If you select the paired t test in the initial dialog the second dialog offers you the choice between a regular paired t test or a t test in which it is ascertained whether the values differ significantly from zero If you want to carry out your paired t test against a reference node however you can select this from the list of existing history files In addition by selecting the Use Relative Path for Template or Drag and Drop check box you can specify whether the comparison node relative to the reference node is to be determined and stored In this way for example you can carry out the same t test for each history file when executing history templates without having to specify the comparison data sets explicitly each time Please note however that this check box is only available for Vision Analyzer User Manual 143 comparison nodes that are in the same history file as the reference node You can also see this from that fact that when you select this option all the history files except your current one are removed from the list of available history files If you select the unpaired t test in the initial dialog the next dialog is very similar to the one for the pair
27. blink was detected a dialog opens in which you can inspect the blinks Blink Selection Number of blinks 2 Accepted as Blink Reject as blink Accept as blink gt Goto I Step only to accepted blinks T Step only to rejected blinks Fig 10 35 Semiautomatic blink detection dialog e The Reject as Blink button removes the displayed blink from the internal blink table and moves on to the next detected blink e The Accept as Blink button inserts a removed blink back into the internal blink table and moves on to the next blink e The lt lt button moves to the previous blink e The gt gt button moves to the next blink e With the Goto button you can go to a specific blink e Step Only to Accepted Blinks If you select this check box the program goes to the nearest previous accepted blink when the lt lt button is clicked The equivalent applies to the gt gt button for subsequent blinks e Step Only to Rejected Blinks This check box has the exact opposite effect to the previous one Algorithm Vision Analyzer User Manual 119 The Gratton amp Coles algorithm corrects ocular artifacts by subtracting the voltages of the eye channels multiplied by a channel dependent correction factor from the respective EEG channels The correction factors are calculated in several steps In the first step a blink detection procedure is applied to the vertical eye channel to calculate separate factors inside an
28. button appears at the top right of the data window Clicking this button removes the overlays again Vision Analyzer User Manual 41 7 10 Setting markers manually You can set markers manually in addition to those markers that are already in the data set A marker in the Analyzer has five different properties type description position channel number and length You will find more information on markers in the Annex under Markers In order to set markers you have to put the data window into marker edit mode You do this by pressing the following button on the tool bar M Marker Edit Mode Pressing the left mouse button now in the data window brings up a dialog which enables you to set a marker You can specify any type and a description You can also assign the marker to an individual channel or all channels If you assign the marker as the Voltage type and assign it to a special channel then the current voltage at this point and the time are displayed next to the description which you can input as an option Add Marker x Description Tesi X Cancel Iype Comment Voltage Channel Fpl x Fig 7 18 Add Marker dialog You can also shift markers as long as you are in marker edit mode To do this move the mouse pointer over the marker that you want to shift Then press the left mouse button hold it down and move the mouse A magenta motion indicator follows the mouse movement The marker is shifted when you release
29. can choose one of three methods to calculate it 1 Use Correlation Autocorrelation 2 Use Cross Spectrum Autospectrum 3 Calculate Only Cross Spectrum The first method calculates the coherence using the following formula Coh c1 2 f Cov c1 c2 f P Cov cr c1 Cov co c2 f in conjunction with 64 Cov cr c2 f X c1 i avg er 2 8 avg er f Y In the second formula totaling is carried out via the segment number i Formation of the average also relates to segments with a fixed frequency f and a fixed channel c The second method uses the cross spectrum instead of covariance The formula is as follows Coh ci c2 f CS c1 ca f P CS ci c1 f CS c2 c2 f in conjunction with CS C1 2 f E c1 02 1 8 Here too totaling is carried out via the segment number i The third method only calculates the cross spectrum which is specified in the formula with CS In methods 1 and 2 values between 0 and 1 are obtained for each frequency and each channel In method 3 complex values are obtained but only the amount is output When you use this module you have the choice of including all possible combinations of channels in the calculation or permitting just a selection of combinations If you opt for individual combinations you can input the combinations of channels in the table Alternatively select a channel on the left hand side of
30. data in the time domain The resulting wavelet vector thus represents a kind of point for point correlation of the raw data with the wavelet It becomes clear here why the selection of a suitable mother wavelet is important for the desired analysis in this approach Both of the wavelets illustrated above will show a high correlation at exactly those raw data points or segments that correspond to the time frequency Vision Analyzer User Manual 147 characteristic of the wavelet thus in this case at local steep rises with a particular time voltage characteristic But how can different frequency ranges be analyzed like this This is where it becomes evident that the CWT is closely related to the DWT After the initial analysis step of the wavelet with all the points in the raw data set the wavelet is widened by a small amount or scaled Consequently we talk about scales rather than frequencies in connection with the CWT and the mother wavelet has a scale of s 1 However widening the wavelet also means that when the data is filtered again with the wavelet other slower frequency components lead to a higher correlation of the raw data with the wavelet This process is then carried out for the entire frequency band required and for the required number of scales For each scale a vector of wavelet coefficients is calculated that contains the time frequency behavior of the raw signal for the corresponding scale Because the change to the w
31. grand averages No preliminary work is necessary on the data obtaining the differences or a priori averaging of the data for example in order to apply the t test Nor do the data set lengths of reference and comparison nodes have to match In this case the t test is only carried for the common time domain x C Unpaired t Test V Equal Variance Fig 10 56 Initial dialog of the t test In the initial dialog you can choose between the paired and unpaired t test For the unpaired t test you can also specify whether equal variance is to be assumed between the two comparison groups Since this is a common assumption in EEG research this option is selected by default However if you know that this assumption does not apply clear this check box Please note that an F test for equal variance is not carried out in the t test module If you select this check box the formula corrected for variance inequality is used 142 Second Dataset x Test Against Zero Test Against Second Dataset Raw Data Expand Reference History File Segmentation Average Expand Same Names P300b Ltnee EERE eee EEN EONS ERS See EEEEEEE SESE SELES EES EEEE SEES EEA ones Raw Data Filters Segmentation Collapse All E BaselineCorrection Average P300c Raw Data Filters P300d Raw Data P300e Raw Data P300f Raw Data P300g Raw Data 2 points ShortD
32. graphically and very clearly in the FFT module The Hanning window is calculated in accordance with the following equation Wp 0 5 1 cos 27x P 82 The Hamming window is calculated in accordance with the following equation Wp 0 54 0 46 cos 27 x P P signifies the above mentioned percentage Corresponding symmetric equations apply to the end of the segment It follows from the way in which the data window works as described above that the overall signal and thus the total variance of the EEG signal are weakened by the window particularly toward the edges Accordingly the use of a data window also results in the damping of the data produced by the Fourier transform For this reason before the Fourier transform is applied a correction factor for the data window and window width used is calculated and used to multiply the data after the Fourier transform This correction ensures that the total variance of the transformed signal matches the total variance of the original signal Vision Analyzer User Manual 83 10 1 14 Filters This transform makes it possible to filter the EEG with the aid of three different filter types e High pass filtering with selectable cutoff frequency or time constant e Low pass filtering with selectable cutoff frequency e Bandpass filtering for 50 Hz or 60 Hz notch filter to eliminate interference by the electricity network The notch filter has a bandwidth of 5 Hz symmetric around the notch
33. graphics card with a resolution of 1024 x 768 pixels and 32 768 colors The monitor used should have a screen size of at least 21 inches 53 cm measured diagonally e 100 MB of available disk space further space requirements depend on the size of the data that is processed Installation e Start Windows 98 NT 2000 or XP e Insert the program CD ROM in your CD ROM drive If your computer supports autostart of CD ROMs a menu will appear after a short time to guide you through the installation process Otherwise do the following e Choose Start gt Run from the task bar e Click the Browse button e Access your CD ROM drive choose setup exe on the CD ROM and click the Open button e Now follow the instructions that the program outputs Before launching the Analyzer insert the hardlock key dongle that comes with the package in one of the printer ports on your computer You can still operate a printer on this port simply by connecting it to the hardlock key It is also possible to connect several dongles to a port by inserting one into another If you have acquired a network license see the Annex Installing a network license for installation details of the Hardlock Network Dongle Now launch the Analyzer by double clicking the Vision Analyzer icon on the desktop An alternate way of launching the Analyzer is from the task bar Start gt Vision Analyzer 4 Getting started and handling the program Launch the Analyzer A win
34. h Name of history file n Name of associated data set o Ordinal number of curve p Full history path with all intermediate steps from the raw EEG to the current data set Fig 7 9 Table of possible placeholders 7 4 Head view As the name says the head view shows your data in the shape of a head To follow the explanations here you should display an EEG and activate a head view via Window gt New Window gt Head View A view appears in which a stylized head is drawn All channels whose head positions are known to the program are arranged accordingly on the head All others are grouped to the left of the head If you used electrode names according to the 10 10 or 10 20 system for data acquisition the program should have this information If you used other channel names however you can input the correct coordinates with the aid of the Edit Channels transform component In order to change the size of the channels move the mouse over the bottom right corner of any channel until the pointer turns into a double arrow Then press the left mouse button hold it down and move the mouse a little towards the left Now release the mouse button The channel has become smaller You could repeat that with all channels but that would be somewhat laborious Instead repeat the operation that we just ran through but press and hold down the Shift key before releasing the left mouse button Now all channels have the same new size Now
35. involved data sets are separated by commas e Primary History Files Only The following selection can be confined to primary history files if you want to 176 e Use Whole Workspace Here you define whether you want to include all files in the workspace e Select Individual History Files Here you can select individual history files e Selection Filter With this option you can filter selectable files by name criteria Wildcards can be used for multiple characters and for one character If the TestlH Test2G and Hest5 files are in the workspace then Test will filter out just Test1H and Test2G The filter est would accept all three files etc When you have set the filter press the Refresh button to refresh the selection of available files e Available Files The available files are shown here e Selected Files The selected files are shown here e Overwrite Default Decimal Symbol You can keep your computer s decimal symbol point or comma or choose one to meet your specific purposes Your computer s decimal symbol depends on Regional Settings e Use Activity Unsigned Values Use Voltage Signed Values Here you can select whether the data should be used in the calculation with or without a sign i e in the first case the signal is first rectified and then the area or activity is calculated This option only exists for data in the time domain If the data is complex all computations are based on the absolute values
36. length must not be lower than a certain value The individual criteria can also be combined With semiautomatic and manual inspections you can change or delete intervals manually New intervals can also be added Raw Data Inspection x Inspection Method C Manual Inspection Semiautomatic Inspection Automatik Inspection Select channels Cancel Fig 10 41 First dialog of the Raw Data Inspector In the first dialog after launching the Raw Data Inspector you define whether you want to inspect the data manually semiautomatically or automatically If you opt for semiautomatic or automatic inspection click Select Channels to select the channels that you want the RDI to include in the search for artifacts The following sections explain the three different inspection methods 126 10 1 26 1 Manual inspection If you choose manual inspection a standard view appears which has a dialog on the right side In this view you can mark intervals on the channels as artifacts using the mouse To delete a mark just click it A popup menu opens containing the Delete Artifact option Raw Data Inspector Goto Next Artifact gt gt lt lt Goto Previous Artifact P Individual Channel Mode Artifact Channel Position Userdefined 02 03 832 s Userdefined P8 11 768 s Userdefined FCI 185 276 Cancel Fig 10 42 RDI dialog for manual inspection You have the following options in the dial
37. more detail later The next three windows show the mouse position in red font The first window in this group contains the name of the channel that the mouse is over The second window indicates the applied voltage there and the third window states the time relative to the beginning of the displayed EEG section or relative to any Time 0 marker that has been set also explained later The final window shows the name of the active workspace in black font To carry out a simple operation at this juncture just choose Transformations gt Filters This brings up a dialog box in which you can define various filter settings The different transforms are explained later Simply press OK at this point All transforms can be undone later without any problems Note that the Analyzer never changes original EEGs After a short while another window opens and displays the new data set that has been generated as a result of this operation It is also shown in the History Explorer as a new icon named Filters Choose Transformations gt RMS Global Field Power now A dialog consisting of two windows appears showing available channels on the left and selected channels on the right If no channel names are displayed in the right hand window double click some channels in the left hand window to make them appear on the right Pressing OK opens a new window displaying the results of the operation Something has happened in the History Explorer too An RMS ic
38. press the following button to put the channels back into the right topographic position Set Display Features A dialog with the same settings as for the grid view appears see the previous section In addition you will find the Move Channels to Topographic Positions button here If you click it the channels go back to their topographic position 30 Headview Settings Ares Overs Ej Distance DEtWEER IEVENIites ii i ko m ke I lt E Fig 7 10 Head view settings You can optimize the position of channels manually by moving the mouse pointer over a channel name pressing the left mouse button holding it down and dragging the channel to the required position Drop the channel by releasing the left mouse button Note that the channel positions and sizes are assigned to the current montage The default montage does not store any settings You should therefore always define a montage if you want to save a certain channel arrangement All other options correspond to those of the grid view Vision Analyzer User Manual 31 7 5 Mapping view Here topographic maps are generated which show the voltage distribution on the head in the time or frequency domain In order to show the maps the program needs information on the position of the electrodes If you used electrode names according to the 10 10 or 10 20 system for data acquisition the program should have this information If you used other channel names h
39. problem begins The cortical activation that is actually of interest often takes place within quite a narrow period around the time of stimulation An example of this would be the examination of changes in the delta range directly after electrical pain stimulation In this case the changes of interest take place in the first 250 ms after stimulation To carry out an FFT analysis with a frequency resolution of 1 Hz 1000 ms must be included in the FFT analysis so that four times as much data is included in the analysis as is necessary and in addition the power in the delta band is also calculated for this data which is actually superfluous and this inevitably corrupts the results Another result of this is that in our example the frequency content of the frequency range e g delta that is actually of interest here obviously does not exist to the same extent over the entire FFT analysis segment The same applies to all the other signal components that occur in the EEG A signal in which the frequency components vary over time is referred to as non stationary However since the frequency resolution is directly linked to the number of data points included in the case of the FFT it is simply not possible to make a statement about the changes to the spectral composition of the signal in time segments that are shorter than specified by the frequency resolution Moreover if the signal in the analyzed segments is not stationary this variation in the
40. searching for local maxima the module looks for extreme values Vision Analyzer User Manual 121 within the interval and only considers the edge values if this search fails Since several local maxima can occur in an interval in certain circumstances it is also possible to weight them during the selection process With this method all maxima that are found are multiplied by the weighting factor 1 a t t and then the greatest value is searched for The value t is from the interval 1 1 and describes the variance of the position of the data point from the middle of the search interval The value a can be entered in the Weight input box and ranges between 0 and 1 With a weight of 0 the weighted method is identical to the unweighted method When a weight greater than 0 is specified peaks which lie closer to the middle of the interval are weighted more than peaks at the edges of it The peaks that are found are however local maxima of voltage distribution in any case The weighting function only has an impact on the selection of peaks when there are several possible local maxima not on their position or size Peak Detection Step 2 of 3 Peak Table x Insert Line Interval Name Start ms nd ms Polarity Reference channel Color ao oc a Al vy Select Color Remove Line mo fo fio fe a J B ceco remove at mel E 5 E u EEEE S E E uE EE mee S S E uE UEEn me OO E lt Zuriick Abbr
41. semiautomatic segment selection the Continue button takes you to a page on which you define the channels that are to be taken into account when searching for artifacts The next step takes you to the Criterion menu Here you can define the artifact criteria which will result in marking of channels in individual channel mode and in the exclusion of segments otherwise The following criteria are available e Gradient criterion The absolute difference between two neighboring sampling points must not exceed a certain value e Max Min criterion The difference between the maximum and the minimum within a segment must not exceed a certain value e Amplitude criterion The amplitude must not exceed a certain value or fall below another certain value 52 e Low activity The difference between the maximum and minimum in an interval of selectable length must not be lower than a certain value The individual criteria can also be combined Artifact Rejection Step 3 of 3 Criteria Please enter criteria for artifact rejection Gradient criterion fi 000 pi Maximal allowed voltage step sampling point Max Min criterion MV Check maximal difference of values in the segment Maximal allowed absolute difference fi 000 p m Amplitude criterion MV Check maximal and minimal amplitude Minimal allowed amplitude 200 pi Maximal allowed amplitude 200 p Test criteria Fig 10 4 Criteria dialog The dialog co
42. sense of the actual wavelet definition This explains why in many articles a c value of at least 5 is required for the Morlet wavelet so as to guarantee at least an approximation to that of the Wavelet definition and to ensure further regularity properties Alternatively a correction term is also sometimes added to the basic e function In practical applications the precise value of c is less important due to the numerical calculation However it can be used for making accurate statements about the frequency bands that are used Take for example a Morlet wavelet with c 5 At the filter frequency of 20 Hz this therefore produces a drop of the signal to 61 standard deviation at the limits 16 Hz and 24 Hz anda drop to 71 filter width at the limits 23 3 Hz and 16 7 Hz In particular however this does not mean that other frequencies are no longer contained in the wavelet transformed signal They are merely weakened accordingly as can easily be seen in the illustration above If the value for the parameter c is changed in the dialog then the width of the displayed wavelet frequency functions is bound to change as well In all of these cases a steeper Vision Analyzer User Manual 153 frequency function of a wavelet at the same central frequency i e an increasing parameter c is accompanied by a poorer time resolution This can easily be reproduced in the dialog by changing the parameter Equally these factors also mean that a wavelet s
43. set for the peak If you double click the peak cursor a gray area appears around the marker the so called slider Alternately you can activate the peak cursor in the table with a single or multiple click Now the gray slider can be adjusted with the mouse When you have found the ideal position you can fix the peak with another double click Alternately you can press the Fix Peak at Current Position button The Remove Slider button removes the slider leaving the peak at its original position If you opt for a separate search for peaks in different channels the dialog also has a combo box in which you can select the channel for which you want to adjust the peak Alternately you can click on the channel name on the left Note the double function of this mouse click It also serves to select certain channels for display as described in the Standard view section of the Views chapter The selected channel appears in red 10 1 25 Pooling This transform enables you to generate new channels by pooling existing channels The new channels are calculated for every point in time from the averages of the selected channels at this time Pooling Lx Number of Pools 1 Refresh Channels Pools Cancel Fig 10 40 Pooling dialog The elements in the dialog are as follows e Keep Old Channels If you check this option the old channels are included in the new data set i e the new channels are practically additional c
44. subfolders if you use the standard installations routine However you can also install the Solutions extension vaso for Vision Analyzer SOlutions manually and make your own categories by creating the corresponding subfolders It is also possible to delete individual Solutions and subfolders using Windows Explorer You can also copy your own Analyzer macros into the Solutions folder or subfolder These are also displayed in the Solutions menu Select Solutions gt Solutions Help to display the documentation for the solutions 8 Solutions Help Explorer i xj Solutions Eke EKG Markers Marker Timing 3 6 views s Stacked Plot EKG Markers Read Markers M EKG Markers Author Jor Ingmar Gutberlet Created 13 05 2004 19 17 50 Description EKG Markers Detect and mark EKG components P Q R 5 and T in an EKG channel using a slope criterion SYNOPSIS This solution requests the input of parameters for R wave detection from a graphical menu and searches for R waves in a selected EKG channel Parameters that can be set are the threshold of the gradient slope the interval to skip after a sucessful search For an R wave and the time range For the R wave search Once the R wave is found corresponding P Q 5 and T components and waves that have been specified in the graphical menu are searched for and marked accordingly x PA elect Solution For Description s So
45. the mouse button To delete a marker just click it briefly This opens a menu from which you can choose whether you want to delete the marker or position a new one When you have finished editing markers you quit marker edit mode by pressing the button on the tool bar again When you close the current data set or perform a transform the Analyzer will create a new data set beneath the current one with the name Markers Changed 42 8 Automation through history templates As described in the Getting started and handling chapter you can transfer an existing processing history from one history file to another You can also store such a history in a special kind of file named a history template You can use this template later to apply the history either to a single history file or to several files automatically Choose History Templates gt New to create a new history template A window opens in which there is a single entry named Root To generate a template open a history file in which you have already carried out one or more operations Drag a data set of the history file to the root node of the history template This data set now appears in the history template together with data sets that have been derived from it In fact only the operation instructions are transferred to the history template not the data 4 Analyzer History Example ehtp OF x a File Edit View Transformations Montage Export History Template Macro W
46. to be edited Editing comprises the following aspects e Hiding channels e Changing the channel and reference channel names e Changing position details Edit Channels x Labels Position Drg Label Enabled Chn Chnf Radius Theta Phi Cancel 1 Fpl Vv Fpl a fs2 72 Bese 2 Fp2 7 0 Be Ez _Detaut Pos 3 oF M E EE ee oe O 5 c3 2 7 sss 6 c E 7 P Po i fF fo p opm d i ffi fo a 9 1 Poff ff fx fn a ee ee ee 1 Al ME BO ho Ws Wo 12 A2 KES E E Ea p 377 a M E 14 re a E E oe DP 16 Te d D E Ce oo Fig 10 21 Edit Channels dialog The dialog shows the following fields for all channels from left to right e channel number This cannot be edited e Org Label original name of the channel This cannot be edited e Enabled Deselect this check box to hide the channel in question e Chn channel name This can be changed as required e Chn reference channel name This can also be changed as required Changing the name does not cause any rereferencing The name is for information purposes only e Radius Theta Phi position details Please refer to the Electrode coordinate system part of the Annex for more information on this Vision Analyzer User Manual 77 On the right of the dialog you will also see the Default Pos button This button sets all electrode coordinates to the 10 10 or 10 20 system depending on the channel names after asking if you really
47. updated accordingly Analyzer p300a Raw Data Mapping Z File Edit View Display Montage Transformations Export History Template Macro Configuration Window Help l x ojeli S SB gt alallala Aef l M fiz almal e Ce a 2468 ms S 0 0 pY 22 6 pV Cz on 4 1 1 S1 51 1 1 ia Ais 1s a gt Ready Standard Montage 00 00 00 Segment 1 1 Fe 7 830 4 43s P300 Fig 7 15 Example of a transient transform In the head and grid views you must move the mouse very close to the displayed signal and then press the left mouse button in order to activate the block marker A vertical tool bar appears on the right edge of the right hand window and you can use this to change the settings for the transient view You can change the size of the pane for displaying the transient transforms by using the mouse to shift the divider bar between the panes left mouse button In order to change the default setting for the page ratio between the two views choose Configuration gt Preferences from the menu and then select the Views tab 38 Here you will find the Default Width of Transient View option at the bottom Enter the width of the transient view as a percentage of the overall window width Preferences lt StandardMontage gt s oi 2 lt Standard Montage v Fig 7 16 Setting the default width of the transient view Pre
48. 56 M Allow overlapped segments M Skip bad intervals Fig 10 52 Third dialog page of marker based segmentation The third page enables you to specify the relative positions of the interval based on time or on data points You state the start and end of the interval or alternately the duration of the interval instead of the end If you select the Allow Overlapped Segments option overlapping segments can also be included Otherwise only the first segment of two overlapping segments is ever taken into consideration 138 Finally the Skip Bad Intervals option defines whether segments containing Bad Interval markers are to be excluded These markers are set by the Raw Data Inspector transform If you want to average later in individual channel mode see the Average section you must not select this option If you opt for time based segmentation you can specify the interval size Size of Segments in seconds or points on the second dialog page You can also specify the overlapping of segments in seconds or data points Here too you can choose Skip Bad Intervals to exclude any segments which have a Bad Interval marker Segmentation Wizard Step 2 of 3 x m Size of Segments Based on Time foo F C Based on Data Point Poarmts Overlap Segments Based on time fos C Based on Data Point poo Parts Seeetetecteceoceccescesccccsenccccnseesseoed lt Back Cancel Fig 10 53 S
49. 61 In the fourth step you can select the name of the output node You can also define a criterion according to which calculation of eigenvalues is discontinued The eigenvalues are normalized in such a way that their total amounts just to the number of variables The average eigenvalue thus has the magnitude 1 You can set this magnitude as the criterion with the first option button This means that all eigenvalues and associated eigenvectors which are greater than 1 are calculated With the second option button and input box you can specify another limit for the lowest eigenvalue that you want to be calculated You can use the Calculate Noise Variance button to help you select a suitable value In calculating noise variance the variance of all variables is compared with the variance of variables from the prestimulus intervals It is assumed that there can only be noise and no signals in the prestimulus interval The ratio of noise variance to signal variance is converted automatically in such a way that the resultant value corresponds to the eigenvalue limit that has to be set to calculate the factors which correspond to the signal Pressing the OK button in the output box for the resultant value will transfer this value directly to the input box of the PCA wizard Since all selected nodes are included in the calculation the calculation process may take some time The program tells you which node is being examined With the third option button a
50. 63 31 25 8 2 31 25 62 50 9 1 62 50 125 00 lt Back Cancel Fig 10 60 Entry parameters for discrete wavelet transform If you select Discrete Wavelet Transformation the Next button takes you to the dialog for a discrete wavelet transform Here you can select the type of wavelet for the discrete transform You currently have the choice between the Haar wavelet and Daubechies wavelets of various filter lengths For Start Level and End Level you can specify the levels of the transform i e the frequency ranges to be analyzed A transform of level 1 provides you with high frequency components between a half and a quarter of the sampling rate at a high resolution At higher levels you receive lower frequency components at a lower resolution The frequency is halved from step to step Like the continuous wavelet transform the discrete wavelet transform also supplies the transformed data in the form of a two dimensional data field where one dimension represents the time domain and the other dimension represents the logarithmically scaled frequency range The data values are displayed color coded in a rectangular area in the Analyzer 150 The limits specified should be viewed as guidelines here Since the discrete wavelet transform as mentioned above works with filter methods the components of a frequency of the EEG do not fall 100 into a single scale frequency step They also occur in neighboring scales in a very weakened form
51. Brain Vision Analyzer User Manual Version 1 05 CE Brain Products GmbH 1999 2006 The Brain Vision Analyzer software frequently abbreviated to Analyzer is designed exclusively for use in medical research Brain Products GmbH does not grant warranty or assume liability for the results of using the Analyzer The content of this document is the intellectual property of Brain Products GmbH and is subject to change without specific notification Brain Products GmbH does not grant warranty or assume liability for the correctness of individual statements herein Nor does Brain Products GmbH enter into any obligation with regard to this document Any trademarks mentioned in this document are the protected property of their rightful owners 4 August 2006 Contents 1 Product declaration ii deisisesisasscaudescsaaninanerandetuccansanneqandeauseatedsansaavianasnandanueiandaanenat 6 Tat PROGUCE id ntiticationn snanconimns a E E 6 1 2 Area of application eeeeeesenneneeeeeeeeesenttnnrtestertntnnnnnnseettnrnnnnnnnseerrnrnnn nennen 6 2 MMtrod cti N nissin hasssahenccaacdaacsasahancsaadsacsasahencsaasdsacsasatencsaaatsacsusatenasautens 7 Se nstallati ONE RAEAN TRAAN TAKANA E ERKAN E AKAA NEa 8 4 Getting started and handling the program cccssscssseeeeeeeeeeeeeeeeeeseeeeeeeeeeees 9 5r Segmentation a a e aaa ea Sp aee aona aNG eo eE assai 15 E eE o a TT 16 Ta MMO WS Se sara scs cassis ass scn suns asus ss s
52. Coherence Coherence Scale l 200 IZ Show Amplitude Amplitude Scale Position 48 62 50 38 67 94 71 16 91 57 112 56 117 51 125 42 149 86 153 47 157 42 160 40 180 57 213 53 216 58 229 46 234 26 254 27 259 26 273 07 275 28 280 42 282 98 285 49 288 03 297 50 Douo UN e Figure 10 38 Interactive processing of ECG episodes The three previously set markers and the template displayed in blue are easily identifiable The red and green curves present the coherence over time red and the amplitude correlation It is also clear that the maximums positive below of the two curves coincide exactly with the peak 112 On the right you can see a number of items which greatly simplify work with the potentially very large number of ECGepisodes On the one hand you can modify the scaling of the coherence and amplitude ratio curves here This is of great use in order to facilitate the use of minimums and maximums in these values when searching for the start of a pulse episode It remains the case that not all data records can be analyzed equally well with the same parameters Consequently at the right hand edge of the list you can see where certain detection problems occur in the data record You can double click to navigate to the corresponding locations If you are not satisfied with the detection values or with the parameter selection then you can click Change Settings to return to the parameter set
53. Montage Cid StandardMontage v Ctl 2 z cha fO t s S YS cla f cs f YS Ce f n cy O cls e a Ctrl 9 Fig 6 3 Keyboard shortcuts for montage selection Vision Analyzer User Manual 19 You can also choose a montage which is activated when a new data window is opened To do this choose the Configuration gt Preferences option from the menu and then the Views tab Here you can choose the default montage separately for unsegmented and segmented data sets Preferences lt Standard Montage oi 2 lt Standard Montage gt Fig 6 4 Selecting the Default Montage 20 7 Views 7 1 Overview A view is how the EEG is displayed e g how the channels are arranged in the window You have a variety of options to change the view Please note that the data displayed was originally digitized There is thus always a limit to the degree of accuracy that can be obtained and this is set by the digitization rate The Analyzer operates with standard grid head and mapping views All views are available to show data in the time and frequency domains You can choose which view you want to use when a new data window is opened To do this choose the Configuration gt Preferences menu option On the View tab you can choose the default view separately for unsegmented and segmented data sets The mapping view cannot be chosen here Preferences x Views Scaling Clipboard Transformati
54. Parseval theorem which states that the total power in the power spectrum should also be equal to the total variance in the time signal Complex output You can use the Produce Complex Output check box to specify whether the FFT module is to supply complex data You should generate complex data whenever you want to further process the data resulting from the FFT with transforms that have to process not only the values of the spectral lines but also their phase information and consequently require complex data Coherence analysis is an example of this Data Window Here you can define the FFT data window You can choose between the window types No Window no window or rectangular window Hanning Window and Hamming Window You can also input the window length as a percentage of the segment length The window function is presented graphically see below for information on the window function e Data Compression Here you can specify whether you want to store the data in compressed or uncompressed format Compressed data saves storage space in the history file but is less accurate than uncompressed data If you opt for compressed data you can choose the degree of resolution in nanovolts 1000 nV 1 uV The lower the resolution of the data higher values the greater the compression effect Consequently less storage space is required The procedure Strictly speaking a Fourier transform extends from 0 Hz to the maximum sampling rate but the dat
55. Sampling Rate This transform allows you to change the sampling rate of a data set The dialog displays the current sampling rate Current Rate and allows you to enter a new rate x Current Rate Hz 250 New Rate Hz 256 Cancel Fig 10 10 Change Sampling Rate dialog The conversion is carried out by means of a cubic spline interpolation third degree polynomial Please note that the length of time required for the EEG is generally reduced by several milliseconds when this method is used since the polynomial cannot be calculated up to the edges of the original data set You should therefore carry out the conversion in the raw EEG Vision Analyzer User Manual 63 10 1 7 Coherence The Coherence module can be called after a Fourier transform to determine the coherence between channels Note that calculation is only possible with complex frequency data Coherence j x Channels Method Calculate Coherence of All Combinations of Channels Use Correlation Autocorrelation Select Combination of Channels for Coherence Calculation C Use Cross Spectrum Autospectrum Calculate Only Cross Spectrum Insert Line Remove Line Remove All Display IV Use Paired Channels View Fig 10 11 Parameter input The coherence provides an indication of the dependence of the data between the individual channels You
56. Top Front Back Right and Left Algorithms Triangulation and linear interpolation In the course of mapping the surface of the head is first divided into individual triangles by means of a Delauney triangulation algorithm The electrodes are at the corner points of the triangles Then linear interpolation is applied to every triangle to calculate the voltage distribution within the triangles starting with the voltage levels at the corners Vision Analyzer User Manual 33 Interpolation with spherical splines A more precise mathematical presentation of interpolation with spherical splines is given in F Perrin et al 1989 Spherical splines for scalp potential and current density mapping Electroencephalography and Clinical Neurophysiology 72 184 187 together with a correction in Electroencephalography and Clinical Neurophysiology 76 1990 565 To calculate spherical splines three parameters are needed which can be input in the Settings dialog of the view the order of the splines labeled m in the article mentioned above and the maximum degree of the Legendre polynomial that is to be included in the calculation The interpolation will be flatter or wavier depending on which values are used for the order Interpolation with an increasing order of splines becomes flatter Basically the denser the electrode arrangement the smaller the order should be Since an infinite series of polynomials is included in the calculation this seri
57. a in the high frequency range results from conjugate complex values from the data in the lower frequency range is therefore not calculated explicitly The data values are scaled in such a way that a sine wave of 1 Hz and an amplitude of 100 uV generates a value of 50 uV at the 1 Hz position when the output format is Voltage calculation takes place without a data window and the Use Full Spectrum check box is not selected Together with the corresponding data value in the high frequency range i e when the Use Full Spectrum check box is selected this results in the original value of 100 uV This approach corresponds to the calculation of a Fourier transform with subsequent division by the number of data points used Brief information about interpolation Although the Fourier transform can in principle be used at any point in the process its use is recommended on segmented EEGs to save time The transform is executed separately for each segment It is not essential for the number of data points in a segment to be a power of 2 If it is not a power of 2 the segments are automatically extended to the next higher power of 2 and the corresponding number of zeros are added to the existing data This process which is also known as padding corresponds to the interpolation of the data in the frequency spectrum of the original data set As a result a frequency spectrum is obtained that has a higher resolution than the original data set No data is l
58. a set Compare Channels or between two data sets Compare Data Sets Comparison Wizard Step 2 of 3 Datasources C Compare Channels NecsonseocersencensenecTastecessnecnecsesea Fig 10 15 Second page of the Comparison dialog If you opt to compare channels the next page shows a channel menu on which you can select groups of channels which are subtracted from each other Vision Analyzer User Manual 69 Comparison Wizard Step 3 of 3 Channels Please select the channels for comparison Chn Chn O N Q0 nm A we N J JAIJ AIJA ER lt Back Cancel Fig 10 16 Third page of the Comparison dialog when channels are compared Here you can also insert a line remove a line or remove all lines If you opt to compare two data sets then the page that appears shows a menu for data sets instead of channels You see a list of all history files in the current workspace You can open them by double clicking a file 70 Comparison Wizard Step 2 of 2 Dataset i Averagel BaselineCorrection2 Artifact Rejection i Average2 P300f Raw Data Filter oe Stimulus1 BaselineCorrection1 Artifact Rejection i i Averagel E Stimulus2 BaselineCorrection2 Artifact Rejection2 i Average P300g Raw Data Filter E Stimulus l BaselineCorrection1 Artifact Rejectiont i Averagel BaselineCorrection2 E Artifact Rejection2 i A
59. a set to be characterized Principal component analysis combines covariant variables of the data set which can be interpreted as a common factor Principle component analysis works according to the following principle The variables in a data set can be chosen either as fixed time points of an EEG or as channels If you choose variables as time points they assume different values depending on the channel segment and EEG file If you select variables as channels the values depend on the time point segment and EEG file In principal component analysis the covariance matrix of all variables is computed first If n signifies the number of variables then there is an nxn matrix from which n factors can be extracted theoretically In practice though it is normal to restrict this to a number whose variance is greater than a certain limit to keep effects such as those that occur through noise out of the calculation If the number of required factors is m then the m greatest eigenvalues and the associated eigenvectors are calculated from the covariance matrix The factor loadings are the product of multiplying the root of the eigenvalue by the eigenvector From them the associated component is calculated for every value of a variable in such a way that the total of the products from multiplying the component by the factor loading optimally approximates the value of the variables This method of decomposing variables is just one of many possibilities tho
60. aceholders is given further below 190 e Alignment Here you can define whether the text is left center or right aligned e Font Clicking this button takes you to the Windows font dialog where you can choose the font to be used for the text Placeholder Meaning ct A comment that you can define for every data set is placed here To input such a comment for a data set move the mouse pointer to the data set in the History Explorer and right click it This brings up a context menu from which you choose Comment Now you can input the comment d Current date h Name of the history file n Name of the current data set nd Date of the data set at the beginning of the section being printed Note that not all EEG formats contain date information nt Time of the data set at the beginning of the section being printed Note that not all EEG formats contain time information In this case 0 00 h is assumed to be the beginning of acquisition This means you see the time offset of the section being printed p Full history path with all intermediate steps from the raw EEG to the current data set t Current time u Computer user currently logged on Fig 13 3 Table of placeholders and their meaning If the printing facilities offered by the Analyzer do not meet your requirements you can export graphics to programs such as MS Word MS PowerPoint and Corel Draw for further processing as de
61. all channels Convert to Big Endian Order If you want to process exported integer data on systems such as the Macintosh or on a SUN workstation the least and most significant bytes of each value have to be stored the other way round in big endian order the final page of the dialog you can define the channels that you want to include in the ort process You can specify that you want to export all channels or just selected ones 174 11 1 3 Markers Export When you export markers you save selected markers to an ASCH file Markers Export x Markers Which markers would you like to export Available Markers Selected Markers New Se 3 Stimulus S 1 842 ASCII Line Delimiters e Export File Name Line delimiters for the output file Placeholders h History file name PC format carriage return line feed n Name of current data set C UNIX format line feed h_ n Markers C Mac format carriage return Resulting file name P300a_Raw Data Markers Cancel Fig 11 4 Markers Export dialog You have the following input options in this dialog e Available Markers Selected Markers Here you select the marker types that you want to export e ASCII Line Delimiters You can specify the format of line delimiters for all exported ASCII files in order to process the data further on different operating systems The principal formats for most PC operating systems UNIX and the Macintosh
62. ans sass san duanian adnan daunin bd a isdans duini aaia 21 Sole OVEIVIOW stint tin esate itn Rehan R aie R eR ai R acta tia dg tal gig eae dg tala eae a i 21 7 2 Standard view Kaceacataiasasalasaiaiaiasacundeusanandeasawedads 25 Fade CONG MICW EE 26 Le HCA VIEW vic cris dan T 30 7 5 Mapping WIG W scat ca ttccs ted cct taht ant th ea st ieee iat OE canta nade 32 7 6 3D mapping V EW i825 o hee bale oc tice tt ce decedd and cee cn duice Srandnan done deneatnandeabenes ducncrangiande 35 7 7 Special properties of frequency views standard grid and head 36 7 8 Block markers and transient transforms cccceceeeeeceeeceeeeeeeeeeeeeeeeeteeeeeeees 38 7 9 Overlaying different data Sets cceeeeecssceceeeeeeeeeeeeeeaeeeeeeeeeeeeeeeesnaeeeeeeeees 40 7 10 Setting markers manually scsi cet fogs ead deosies eatin eens eevee gence season geeeeed oes 42 8 Automation through history templates ccccccesssseseeeeeeeeeeeeeeeeeeeeeneeeeeeeees 43 QO MACKS e EEEE ETETETT 46 TOs Transforms wssccrisssscseeichsasseactsacssschsash actsacsescteaseasteaceeuctaad dans teadeeasseadeanssaadeeuseass 48 TOT PRIMARY transforme essaiera aie oe eee eee 49 10 11 Artifact Rejeti sesoses a aa e S a aA a aa 49 TO Ree AVEIAOS ninn a 55 10 1 3 Averaged Cross Correlation ccccccccccsceeesnncceeeeeeeesseesesacanseeeesesseeenas 58 10 14 BANO TClOCHON TNGIS r a a cba E EE 60 101 9 Baseline Correcti r sistisssssa
63. ar as non bipolar montages are concerned the program inserts adequate names in the reference channel input boxes When you have defined your montage click the OK button You are prompted to save the montage Input a suitable name and save the file To test your new montage first make sure that a data window is active Then click the Display Montage menu The number of items on the menu has increased as the name of your new montage appears here now Choose your new montage The EEG is now displayed with the montage To revert to the default montage choose it on the Display Montage menu If you want to modify an existing montage select it under Display Montage gt Edit and edit it However you cannot change the reference type for an existing montage After editing the program asks you again which name you want to store the montage under You can input a new name in order to derive a new montage from an existing one in this way 18 You can assign keyboard shortcuts to montages so that you can switch between them faster The montages are activated when you press the specified shortcuts You can define these shortcuts under Display Montage gt Options The montages are assigned to the Ctrl 1 to Ctrl 0 key combinations Ctrl 1 is reserved for the default montage As far as the other combinations are concerned you can select from existing montages Montage Options Fa gt Shortcuts for Montage Selection Key Associated
64. are available e Export File Name Here you specify the name of the export file using placeholders if you want to The program then shows you the resultant file name Vision Analyzer User Manual 175 11 2 Extended export components 11 2 1 Area Information Export Here you can export the area dimensions uV ms or uV e Hz of an interval the average activity uV or the activity total in an interval to an ASCII table Area Information Export E x Input C Frequency Domain Area Interval Relative to Time 0 Start ms fo End ms 150 Name s of the Involved Data Sets History Nodes Separated by Commas Average JV Primary History Files Only C Use Whole Workspace Select Individual History Files Selection Filter i Refresh Available Files Selected Files Add gt Remove Add All gt gt lt Remove All Options J Overwrite Default Decimal Symbol Decimal Symbol gt Rectification Use Activity Unsigned Values Rectified C Use Voltage Signed Values Export Type C Area p ms Mean Activity p Area as Raw Sum of Activity Values uy Output Folder C Vision Export Output File Area tet Gk Cancel Fig 11 5 Area Information Export Dialog The following options can be set in this dialog e Area Interval Relative to Time O This is the interval to be exported e Name of the Involved Data Sets The names of the
65. ars either the specified raw file folder is empty or the Analyzer cannot yet read the format of the EEGs that are there In the latter case please get in touch with us to find out about the EEG readers that are currently available When you have successfully read in one or more EEG files you can open a history file To do that click the sign on the left of the book icon The entry expands and a Raw Data icon appears Double click this icon The EEG is displayed Vision Analyzer User Manual 9 Analyzer p300a Raw Data OL x Z File Edit View Display Montage Transformations Export History Template Macro Configuration Window Help lej x pelil S gt alas Elya Aef slir Mi fe amel 2 P3009 AAAA a i a cc o wo o Gs gt p Standard Montage 00 00 00 Segment 1 1 P8 10 36 3 42s P300 Fig 4 2 Vision Analyzer with a loaded EEG To navigate through the EEG use the navigation bar that is at the bottom On the left the navigation bar has four buttons with which you can move through the EEG along the time axis To the right of these buttons there is the marker window which shows all markers that have been set in this EEG Markers are all time related indicators such as stimuli responses comments segment boundaries DC corrections etc There is a slider window beneath the marker window The width of a blue s
66. avelet in the CWT is referred to as a change to the wavelet scale rather than to the wavelet frequency the representation of the time related data segment against the successively slowed frequency contents is referred to as a scalogram The scales of the wavelet calculation can of course be converted into corresponding frequency ranges again Consequently the wavelet results in the Analyzer can be displayed in a time frequency chart again On account of the scaling of the mother wavelet it is important that there is an interaction between the time and frequency resolution in the CWT just as there is in the DWT At lower scale values i e in the analysis of higher frequencies the resulting wavelet coefficients have a good time resolution but a poor frequency resolution At higher scales i e in the analysis of lower frequencies on the other hand wavelet coefficients with a good frequency resolution but a poor time resolution are obtained You will find more information on wavelet analysis for example in Louis Maa Rieder Wavelets Teubner Studienbiicher ISBN 3 519 12094 1 from which the algorithm used for the discrete wavelet transform was obtained Practical execution of the wavelet transform with the Analyzer In the initial dialog of the wavelet transform you have the choice between a discrete invertible discrete and a continuous wavelet transform You can also specify whether you want the wavelet coefficients their absolut
67. be taken into account when defining expressions The not operator has top precedence followed by and and then or Precedence can be changed by using parentheses i e expressions in parentheses are calculated first Example not R1 50 100 or R2 100 200 Here the expression R1 50 100 or R2 100 200 is calculated first and the result is negated with the not operator In this case all segments in which the patient pressed either RZ within 50 and 100 ms or R2 within 100 and 200 ms are not included The expression could be not R1 50 100 and not R2 100 200 Negative values can also be specified in the time window Then they relate to a time before the reference marker Any spaces in the expression are ignored Nor is any distinction drawn between upper and lower case This means that r 1 1 2 is identical to R1 1 2 In the ABE in contrast to the reference marker all markers with the same sequence of characters are handled in the same way without regard to spaces The marker value r 1 is equivalent to rl and R1 Now we come to the handling of the Segmentation component The first page of the dialog shows you the three segmentation types discussed above e Create New Segments Based on a Marker Position e Divide Data Set in Equal Sized Segments e Set New Segments Manually e Create New Segments Limited by Start and End Markers Here you will also find the choice of the various storage options described above
68. by improving shielding there are sources of interference which are inevitable when acquiring EEGs For example a spurious signal is generated as a result of magnetic field changes when an EEG and MRI in the EEG are acquired simultaneously These spurious signals have typical frequencies and can largely be eliminated by a combination of low pass filters and band rejection filters There are two steps in the wizard for the Band Rejection Filter module Step 1 of 2 Filters Settings a a a oe E 10 00000 0 10000 2 15 00000 0 10000 2 Remove Line 20 00000 0 10000 2 50 00000 2 00000 4 Pemwveal Cancel Fig 10 7 First step in the Band Rejection Filter wizard In the first step you can define any number of band rejection filters that you want to apply to the EEG The filter is determined by its frequency bandwidth and order In this process a signal at the threshold frequencies frequency bandwidth 2 is reduced to half its ampli tude The order determines the slope of the filter You can choose between an order of 2 or an order 4 A higher order results in greater filtering in the interval between the two threshold frequencies 60 In the second step in the wizard you can select those channels that you want to apply the filter to In this way you can for example exclude control signal channels which have not been contaminated by spurious signals from filtering It is possible to apply vari
69. cale with a lower central frequency has in absolute terms a smaller standard deviation in the frequency distribution On the other hand however the time resolution is comparatively poor This can be shown very impressively by moving the vertical red line in the lower frequency distribution diagram because as the frequency drops the wavelet shown in the window above becomes longer This representation is extremely useful for analyzing EEG data with the help of wavelets since it is of course necessary to find the optimum ratio between frequency and time resolution for the scales that are of interest This can be achieved easily using the parameter c and the two windows with the frequency and the time resolution of the resulting wavelets The natural arrangement of the central frequencies of the scales between the upper and lower ends of the frequency band for wavelet analyses using frequency ranges is logarithmic and therefore corresponds to the dyadic function of the discreted continuous wavelet function that is used here However complete coverage of the frequency band between the upper and lower ends of the band is often not at all desirable whereas an even linear sampling of the spectrum is This can be achieved using the Linear Steps instead of Logarithmic option A frequency band from 20 to 50 hertz sampled in 7 steps would therefore lead to six wavelet functions with central frequencies of 20 25 30 35 40 45 and 50 hertz
70. channel on which the threshold is to be searched for e Reset Value The value to reset the trigger in pV It is normally identical to the threshold but can be set to another value for certain purposes In a positive direction the value has to fall below 92 this value before a new trigger of this kind is found Analogously the same applies to a negative direction The dialog also incorporates three buttons One to insert a line one to remove a line and another to remove all lines Under Time Tolerance you can specify an interval within which triggers of the same direction in the same channel are not distinguished If several thresholds within this interval are exceeded only the maximum threshold is detected and only one marker is set The same applies analogously when values fall below several thresholds within the tolerance The size of the interval should be adjusted to the steepness of the edges of the trigger channels Vision Analyzer User Manual 93 10 1 19 Linear Derivation This transform enables you to generate new channels from linear combinations of existing channels The new channels are calculated from coefficients and are assigned to existing channels according to the following equation New channel Coeff1 Channell Coeff2 Channel You can input these coefficients in a matrix and also store load them xi IV Keep Old Channels Number of New Channels 10 Refresh Load from Fie Save to File I Ne
71. cted channels To select one or more channels mark it or them and then press the Select button On the second page of the dialog you select the channels that are to be referenced You can also define whether non referenced channels should be included Keep Remaining Channels Vision Analyzer User Manual 115 New Reference Step 2 of 3 Select the channels to which the new reference will be applied Available channels Al A2 Selected Channels IV Keep remaining channels lt Back Fig 10 32 Second page of the New Reference dialog On the final page of the dialog you specify a name for the new reference channel such as Ears for A1 A2 reference or Avg for averaged reference You can also reuse the old reference channel as a normal data channel An example of this would be a CZ reference which is converted to an Al A2 reference In this case the CZ channel can be used for further calculations New Reference Step 3 of 3 New Reference Channelh_ Name of new reference channel vd we Biei CO r ie mauu Reuse old reference channel negated new reference Heme ERELEAFE aia Fig 10 33 Part of the third page of the New Reference dialog When you have completed your input a new view appears showing the changed channels 116 10 1 23 Ocular Correction With this transform influences of eye movements on the EEG can be eliminated or at least reduced The Gratton amp Cole
72. ction factors calculation or not e Ifyou want the program to search for the markers you can also use the Write Only Markers option to determine whether you do not want any data correction at all with this analysis step but you simply want to have the blinks found by the program to be marked in the data set You can then process these markers if necessary using the options provided in the Analyzer and run the actual Ocular Correction later using the Based on Markers option Under Enable Disable Channels for Correction e Inthe bottom part of the window select the channels for which the correction should be performed e Ifyou also want to select the eye artifact channels for correction bear in mind that the information in these channels will be largely lost It is easy to detect whether the reference electrode of the eye channel is short circuited with the common reference Common Reference option because the eye channel no longer has a signal following the correction e Ifthe eye channel has its own reference channel Reference Channel option however then the information loss due to the correction takes the form of both the eye channel and the reference channel displaying the same signal after the correction It is 118 essential to bear this aspect in mind when the channels are used additionally as pure data channels Semiautomatic mode e If you opt for semiautomatic blink detection and at least one
73. cy Extraction dialog If you opt to output the power the average power in the specified frequency range is calculated for each data point If the phase is chosen then it is the phase of the signal in the specified frequency range that is calculated This number is only defined uniquely up to a multiple of 2x When different intervals are chosen the output values can therefore vary by multiples of this value It is possible to infer the dominant frequency in the calculated frequency range on the basis of the phase A rising phase indicates a dominant frequency beneath the middle of the chosen interval whereas a falling phase indicates that the dominant frequency lies above the middle of the interval 88 10 1 17 ICA Independent Component Analysis This module is used for calculating the ICA of EEG signals i e splitting the signals up into independent components using information theory methods Methods The object of ICA in what is referred to as blind source separation is to reconstruct source signals from a mixture of such signals In this case both the source signals and the mixture are not known However assumptions are made regarding the signals and the mixture of these signals In terms of the signals to be reconstructed it is assumed that they are statistically independent of one another More details of the theory used can be found in BS95 Car98 and MBJS96 so these will not be explained further at this point It is o
74. d you can define a grid yourself in the montage edit dialog under Arrange for Grid Views also see the Montages chapter Here you can input the required number of rows and columns in the channel grid Pressing the Refresh button updates the grid that is on the screen Now you can arrange the channels and the gaps between using the mouse Montage Arrangement for Grid View x Grid Size Rows E Columns E Refresh Cancel range channels with the mouse Source and target positions will be swapped Fig 7 5 Grid definition dialog In order to follow the descriptions below you should display an EEG and then activate a grid view via Window gt New Window gt Grid View If you want to overlay two or more channels move one channel over another one so that the upper left corners of the two channels coincide The border color of the motion indicator changes to red Drop the channel by releasing the left mouse button You will see two overlaid channels The original channel that you moved is still at its original position You can repeat this operation with different channels as often as you need to As soon as an overlay exists a button labeled Clear Overlays appears at the top right of the data window The overlays disappear when you click it 26 To zoom into a channel with or without overlaid channels double click the channel name The channel then occupies the entire data window Now you can also apply the Del
75. d by the Windows operating system you must explicitly enable port 475 for TCP and UDP The Analyzer Editions network licenses are now available to you Note that the licenses are only available if the computer has been started the dongle is attached and a network connection is active In addition make sure that no other Analyzer Editions network license either USB dongle or LPT dongle is active on the network If so make sure you deactivate them It is also possible to install the network license software on several computers Then you can simply insert the dongle in a USB port on another computer in the event of a server failure Vision Analyzer User Manual 201 Annex F Individual user profiles You can define individual user profiles if you want to Your various preferences are then stored under these profiles These include all options that you set under Configuration gt Preferences as well as the view options and the parameters that were last used in the various transforms Select Configuration gt User from the menu to enable individual user profiles User Profile Settings Ea M Use Individual Profiles Profile Name u is used as a placeholder for the current user asus c tmp gu aprof Cancel Fig F 1 User Profile Settings dialog Enable individual user profiles by checking Use Individual Profiles Now you can enter the name ofa file in which you want to store the profile If you input the u placeholde
76. d outside blinks In the next step after deduction of the mean an average is formed for each channel covering the various events i e the various segmentation markers To calculate the correction factors this average is subtracted from the respective channel in each segment in the third step in order to prevent event correlated data from being included in the calculation This step is omitted if you choose Gratton amp Coles without raw average subtraction as the method In the final step the correction factors are calculated on the basis of linear regression An exact description of the algorithm is given in Gratton G Coles M G H amp Donchin E 1983 A new method for off line removal of ocular artifact Electroencephalography and Clinical Neurophysiology 55 468 484 120 10 1 24 Peak Detection Peaks are local minima and maxima in an averaged EEG which are detected and marked by this module The module enables you to specify peaks with their names and range in a table You can also define whether a positive or negative peak is involved You can select certain channels to be included in the marking of peaks Peak detection can be automatic or semiautomatic With semiautomatic detection you see a cursor at the position at which the algorithms detected the peak You can use this cursor to change the position of the peak manually Two methods of setting marks are available e Searching for and marking peaks take place s
77. depends on the wavelet used As with filters the information in the dialog therefore applies to the range in which the effect of the frequencies is strongest For this reason overlapping at the limits in the figure above is completely normal If there are no overlaps on the other hand a loss of information can be expected In this case a larger number of scales should be selected The frequency distribution of the wavelet function is approximately Gaussian in shape which means the frequency limits in the display above are shown either as Gauss or filter limits 68 or 71 of the signal amplitude in the central frequency of the wavelet The illustration below shows examples of these limits for the Morlet and the Mexican hat wavelet 12 r r r r r 12 Fig 10 63 Filter limits of the Morlet left and Mexican hat wavelet right As can easily be seen the filter limits of the Morlet wavelet actually run symmetrically on either side of the central frequency whereas the filter limits of the Mexican hat wavelet are slightly asymmetrical and offset to the left The standard deviation of this function for the Morlet wavelet is calculated for a given parameter c using 1 c This means the ratio between frequency and standard deviation is exactly c The ratio of filter frequency to half filter width c in the Morlet wavelet is calculated at c 1 2c Unfortunately the Morlet wavelet is not a wavelet in the
78. distribution of activity on the individual spectral lines To this end the lower and upper limits of the normalization range can be specified in two dialog boxes The minimum lower limit is fixed at 0 5Hz since it is generally assumed that frequencies under 0 5Hz are not electrocortical in origin and that these frequencies are extremely unstable from segment to segment The area of this frequency range is calculated and set to the area value of 100 by multiplying it by a normalization factor that differs from segment to segment and channel to channel In the same way all of the spectral line values that are not within the normalization range are multiplied by this factor and thus also normalized Band comparisons such as the relative alpha share of the EEG or the alpha slow wave index can thus be calculated easily Normalization also makes it much easier to calculate the spectral corner frequency for example Full spectrum The Use Full Spectrum check box is used to specify whether only one or both halves of the spectrum is are to be used to calculate the spectral line values If both are used this effectively doubles the spectral line values This is of particular significance when power spectra are calculated since this check box allows both of the definitions of spectral power uV or as 1 V 2 used with EEGs to be taken into account If you select the Use Full Spectrum check box the resulting FFT power spectra behave in accordance with the
79. dow that is split into two panes appears on the left the History Explorer The first thing you have to do is set up a workspace to tell the History Explorer where your data and new history files are located Choose File gt New Workspace to do this New Workspace x Folders Raw Files Browse History Files Browse Export Files Browse Cancel Fig 4 1 New Workspace dialog The program asks you for the folders containing the raw data files history files and any export files that you may want to use to export the results of your analyses Raw data files are EEG files that you have acquired Specify the folder in which they are stored You can also look for the required folder by clicking the Browse button History files hold all processing steps transforms that you apply to the raw data It is history files that are shown graphically in the History Explorer later Define any folder for the history files Export files contain data that is intended to be processed further in other programs Once you have defined the settings press the Enter key or click the OK button Now a dialog appears asking you to specify the workspace file Give the file a meaningful name and press the Enter key or click the Save button The raw data is now analyzed A history file is created for every raw data file A book icon appears in the upper pane of the History Explorer for every history file If nothing appe
80. e Generic Data Reader GDR is used to read in EEG files of various formats for which no special reader exists e g proprietary laboratory formats The reader uses a header file which describes a single EEG This file is an ASCII file with the extension vhdr It will normally be given the same base name as the raw data EEG that is described in it The header file is stored in the raw data folder of the workspace The format of the header file is based on the Windows INI format It consists of sections of different names containing keynames and assigned values Here is an extract of a header file Brain Vision Data Exchange Header File Version 1 0 Data created from history path P300b Raw Data Filters Segmentation BaselineCorrection Average Common Infos DataFile P300b Average dat MarkerFile P300b Average vmrk DataFormat ASCII Data orientation VECTORIZED chl ptl chl pt2 MULTIPLEXED chl ptl ch2 ptl DataOrientation VECTORIZED DataType TIMEDOMAIN NumberOfChannels 32 The first line identifies the header file and is mandatory A semicolon at the beginning of a line identifies a comment which is ignored by the reader Blank lines are also ignored A section is identified by a line with a term enclosed in square brackets The header extract above for example contains the Common Infos section A header file can contain any number of sections The next lines show some keynames in this section and the values that hav
81. e after the triggering of a stimulus In more general terms ABE enables you to make segment selection dependent on the existence or non existence of one or more markers in one or more time spans relative to the reference marker The markers can be of any type segment markers reference markers DC correction etc You can input the selection criteria in a text line linking the marker names with time windows in milliseconds and with the not and and or operators Example You select a reference marker Now you only want to include segments in which the patient has pressed the R7 button within 50 and 100 ms after the reference marker You have to input the following expression for this R1 50 100 Vision Analyzer User Manual 135 If you want to include all segments in which the patient has pressed the R button within 50 and 100 ms after the reference marker and pressed the R2 button within 100 and 200 ms then the expression is R1 50 100 and R2 100 200 If you want to include segments in which the patient pressed either the R or R2 button within the above mentioned times then the expression is R1 50 100 or R2 100 200 If you only want to include segments when either only the R1 button or only the R2 button has been pressed within the above mentioned time exclusive OR then you can use the following expression R1 50 100 and not R2 100 200 or not R1 50 100 and R2 100 200 The precedence of operators must
82. e been assigned to them A keyname can only occur once in a section Its meaning depends on the section in which it occurs There must be no blank before or after the assignment operator equal sign Most predefined keynames have a predefined value which is used by the reader if a keyname is not found If you want to generate such a file it is best to export any EEG with the aid of the Generic Data Export function This creates a header which is compatible with the GDR Set the parameters in such a way that the format of the exported file is as close as possible to that of the one to be imported Now you can optimize the header to meet your specific requirements Vision Analyzer User Manual 181 The various predefined sections with keynames meaning and default values are listed below Common Infos This section contains general information on the EEG file Keyname Meaning Default value DataFile Name of the EEG file If the name does not None contain a path it is assumed that the EEG file is A value must be in the same folder as the header file The specified placeholder b can be used in the name It is replaced by the base name of the header file when the file is read in Example The entry DataFile b EEG dat is interpreted for a header file named Test vhdr as DataFile Test EEG dat MarkerFile Optional marker file containing a list of markers assigned to the EEG If no path is specified explicitly the marker fil
83. e common reference The following situation applies here in principle If your VEOG signal exists in the form of an individual channel then select Common Reference This is because from the program s perspective the signal is a unipolar signal which means it is a signal relating to the Common Reference However if your VEOG signal exists in the form of two individual channels e g VEOG Top and VEOG Bottom then activate Reference Channel here and select the second channel from the list of available channels The program will then treat these two channels as a bipolar channel pair in the blink detection HEOG The items under HEOG apply in the same way to the HEOG channel Blink Detection e Under the Blink Detection option you can specify whether the blinks should be searched for in the VEOG channel set further up By Algorithm option or whether you have already detected the blinks you are interested in outside the Ocular Correction e g using a macro with you own algorithm and you have identified each of these with a Blink Start and Blink End marker Based on Markers option e Furthermore for the By Algorithm option you can also determine whether the blink detection should be automatic or semiautomatic Semiautomatic Mode option In the latter case you can decide interactively for each potential blink found by the program whether it should be included in the Ocular Corre
84. e curve is particularly well suited for this assessment of the blood pulse markers and the artifact markers referred to under 2 4 Ifnot already done in 3 correct the blood pulse artifacts Generally speaking this subdivision into different steps has the advantage that the intermediate results at each stage can be evaluated and manual or automatic intermediate steps such as marker corrections or artifact searches with the Raw Data Inspector for example can be incorporated References A1198 P J Allen et al Identification of EEG Events in the MR Scanner The Problem of Pulse Artifact and a Method for Its Subtraction Neuroimage 8 229 239 1998 A1100 P J Allen et al A Method for Removing Imaging Artifact from Continuous EEG Recorded during Functional MRI Neuroimage 12 230 239 2000 114 10 1 22 New Reference Here the average of selected channels is used as a new reference The first page of the dialog enables you to select channels that are to be included in reference calculation As an option you can include the original reference channel Include Implicit Reference in Calculation of the New Reference New Reference Step 1 of 3 Select channels that form the new reference Available channels Selected Channels Al A2 Select gt gt lt lt Remove Cancel Fig 10 31 First page of the New Reference dialog On the left you see the available channels and on the right the sele
85. e is searched for in the folder in which the header file is located The format of the marker file is explained further below Here too it is possible to use the placeholder b DataFormat Possible values ASCII ASCII BINARY DataOrientation Possible values MULTIPLEXED VECTORIZED First the file contains all data points for the first channel followed by all data points for the second channel etc MULTIPLEXED Here all channels for every data point follow on from each other directly The data structure is multiplexed DataType Possible values TIMEDOMAIN TIMEDOMAIN The data is in the time domain FREQUENCYDOMAIN The data is in the frequency domain FREQUENCYDOMAIN_COMPLEX The data exists as complex frequency values Each real value is followed by an imaginary value TIMEFREQUENCYDOMAIN The data exists in several layers as in the case of the continuous wavelet transform for 182 example Each channel is represented by a vector of data at a point in time TIMEFREQUENCYDOMAIN_COMPLEX This type corresponds to TIMEFRQUENCYDOMAIN except that here each value exists as a complex number NumberOfChannels Number of channels in the EEG file None A value must be specified SamplingInterval The sampling interval is specified in us in the time domain and in hertz in the frequency domain None A value must be specified Averaged This indicates whether the data set which is t
86. e level baseline correction improves the calculation of the scanner artifact If the base level is high overall baseline correction prevents jumps in the corrected EEG since otherwise the base level influences the calculation and is also subtracted at correction In some circumstances it is best not to use all the scanned intervals to calculate the artifact curve It is possible that the data in some intervals cannot be used for this purpose due to either saturation of the amplifier or other artifacts which means it would falsify the averages obtained For this reason intervals that contain a marker of the type Bad Interval are excluded from the calculation This is also possible channel by channel if the markers provide the corresponding channel specific information Similarly saturation can be detected in individual channels of the intervals and be excluded from the calculation However these 100 intervals are then corrected in the same way Intervals with saturation are marked by the module with Bad Interval markers To check the quality of the averaging the standard deviations of the different channels are output in the operation information There are a number of different ways of selecting the intervals for the correction of the averaged artifact curve It is possible to use all the intervals or only those which have a minimum correlation with the previously determined artifact curve Alternatively you can calculate a slidi
87. e of the head at a fixed time Since voltage distribution is only known at the electrodes the procedure of spherical spline interpolation is used to calculate the total voltage distribution An exact mathematical presentation of this procedure is given in F Perrin et al 1989 Spherical splines for scalp potential and current density mapping Electroencephalography and Clinical Neurophysiology 72 184 187 together with a correction in Electroencephalography and Clinical Neurophysiology 76 1990 565 To calculate spherical splines three parameters are needed which are requested by the transform in a dialog the order of the splines labeled m in the article mentioned above and the maximum degree of the Legendre polynomial that is to be included in the calculation The interpolation will be flatter or wavier depending on which values are used for the order Interpolation with an increasing order of splines becomes flatter Basically the denser the electrode arrangement the smaller the order should be Since an infinite series of polynomials is included in the calculation this series must be discontinued at a certain degree The rule that applies here is the higher the spline order the lower the degree of the polynomial at which calculation is discontinued CSD x Spline Parameters Order of Splines 4 Maximum Degree of Legendre Polynomials fi 0 V Default Lambda 1e 5 Other Lambda fi e 005 T Keep Remaining C
88. e values or the power square of the absolute amounts to be output 148 x Wavelet Transformation Methods C invertible Discrete Wavelet Transformation C Continuous Wavelet Transformation Tl Calculate Absolute Values T Calculate Power Values Fig 10 59 First page of wavelet transform dialog The calculation of absolute coefficient values makes sense for example if you want to average the wavelet coefficients calculated from segmented data and are not interested in the polarity of the original signal only in its spectral composition A striking example of this involves investigating the EEG for induced versus evoked activity in a frequency range The only interesting factor in terms of the induced activity is the absolute extent of frequency activity whereas its time phase position is not significant However it is precisely these phase differences in the presence of induced activity that are eliminated by averaging the data for an evoked potential This means the wavelet transform of the evoked signal only contains the evoked wavelet activity The behavior is the same when calculating power values Vision Analyzer User Manual 149 Discrete wavelet transform Discrete Wavelet Transformation E x Wavelet Haar Daubechies 2 Start Level fi End Level E Maximum number of levels 13 Scales 1 9 0 24 0 49 2 8 0 49 0 98 3 0 98 1 95 4 6 1 95 3 91 5 5 3 91 7 81 6 4 7 81 15 63 7 3 15
89. e voltage distribution on the head in the time or frequency domain In order to represent the map the program needs information on the position of the electrodes If you used electrode names according to 10 10 or 10 20 during acquisition the program should have this information If you used other channel names though you can input the correct coordinates with the aid of the Edit Channels transform component Analyzer P3003 Raw Data Fiters Segmentation BaselmeCorrection Average 30Map an s Bie Edt yew Doly Montage Trandormations Export Hptory Temglite Moo Configuration Window Helo Dieja aalst Ella Mefe slin se mj ft SIE e eee es Eisd OROA Begner ifi I I Pm Fig 10 70 3D map Please refer to the 3D mapping view section of the Views chapter for more information on the 3D map Vision Analyzer User Manual 165 10 3 2 Current Source Density CSD The CSD is described earlier in this document under primary transforms If you use it as a transient transform the current source density is represented with the aid of a map Here order 4 and polynomial degree 10 are used implicitly for calculation In order to display the result of the transient transform in an optimum manner interpolation with spherical splines should always be used as the interpolation method for the map 166 10 3 3 Fast Fourier Transform FFT FFT represents the frequency spectrum of the current section It a
90. echen Fig 10 37 Second page of the Peak Detection dialog On the second page of the dialog you input the peaks that you want to search for There are the following fields from left to right e Name The peak name e g P300 e Start The time at which you want to start searching for the peaks e End The end of the search interval e Polarity The polarity of the peak You can choose between positive polarity and negative polarity 122 e Reference Channel The reference channel in which the peak is to be searched for if you selected a reference channel in the previous step e Color This specifies the color to be used for the peak indicator in the event of semiautomatic searching There are also buttons to insert a line remove a line and remove all lines Peak Detection Step 3 of 3 Channels Fig 10 38 Third page of the Peak Detection dialog On the third page you select all channels in which peak markers are to be set If you selected semiautomatic peak detection a dialog appears in which you can adjust the peaks that are detected Vision Analyzer User Manual 123 Peak Detection Fix Peak at Current Position Remove Slider O T 272 0 Fig 10 39 Part of the semiautomatic Peak Detection dialog You can select a peak cursor with a single mouse click This causes a green area to appear around the peak cursor representing the search interval that has been
91. econd column The corresponding channel names C3 and C4 appear in the third and fourth columns After the calculation a channel appears named LRP C4 C3 which is the result of C4 left C3 left C3 right C4 right 2 Further explanations of lateralized readiness potential are given in Sommer W Ulrich R Leuthold H 1996 Das Lateralisierte Bereitschaftspotential als psychophysiologischer Zugang bei der Untersuchung kognitiver Prozesse Psychologische Rundschau 47 1 14 98 10 1 21 MRI Artifact Correction This module permits the correction of different types of artifacts that occur during the recording of EEG signals in a magnetic resonance tomograph Some of these artifacts are formed as a result of changes to the magnetic field in the scanner scanner artifacts Pulse artifacts are also created in the EEG channels due to the heartbeat and the resulting blood pulse in the body as well as the associated micro movements of the head cardioballistic pulse artifacts The module allows both of these types of artifacts to be detected and corrected Methods The methods used are based on techniques that are described in the articles A1198 and A1100 The correction of scanner artifacts involves averaging the intervals in which the gradient change of the scanner takes place The averaged scanner artifact curve is then subtracted from the original curve in the relevant intervals This generally does not remove the scanner
92. econd dialog page of time based segmentation On the third page you are asked whether you want to generate a separate data set for every new segment This allows you to further process every segment separately If you choose to set new segments manually you are asked on the second page whether you want to generate a new data set for every segment When you have completed the dialog and have opted for manual setting of segments a new dialog appears where you can input the new segments Vision Analyzer User Manual 139 Set New Segments Manually Ed Data set length Offsets of existing segments 782 72 seconds 195680 data points Os 0 Enter new segment Start s 0 End s 0 004 Start point jo End point jo New segments Start End Fig 10 54 Dialog for manual input of segments This dialog displays the data set length in seconds and data points and the position of existing segments You can enter a new segment Press the Add button to include the segment in the New Segments list To remove a segment from the segment list mark it in the left column and then press the Remove button If you select the option for segmentation based on start and end markers then the second page shows you fields in which you can define the markers On the third page you are asked in the same way as for manual segment selection whether you want to create a separate node for each segment With this option the first start marker found is lin
93. ed channel c If you use this module you can choose whether you want to include all the possible combinations of channels in the calculation Calculate Covariance of All Combinations of Channels or whether you only want to permit a selection of combinations Select Combination If you decide on individual combinations you can enter the channel combinations in the table Alternatively select a channel from the list on the right If you click the Combine Channel with All button all the combinations of the selected channel are added to the table 68 10 1 9 Comparison With this module two data sets or two channels of a data set can be compared with each other after averaging The Comparison module currently supports the following compare operations Difference waves Cross correlation If the operation is performed on two data sets the data set that is currently displayed is regarded as the reference set This means that the new history node is appended to the current data set It is important for the comparative data set to have the same sampling rate Comparison Methods Difference C Cross Correlation Cancel Fig 10 14 First page of the Comparison module The first page of the Comparison dialog gives you a choice between calculating difference waves cross correlation and lateralized readiness potential On the second page you define whether you want to perform the comparison within the current dat
94. ed t test except that the option of testing against zero is not available Depending on whether you select a paired t test an unpaired t test or a t test against zero all the required calculation steps are carried out such as obtaining the differences across all segments in the paired t test for segmented data averaging of the differences and calculation of the t values 144 10 1 31 Wavelets Like the Fourier transform the wavelet transform is a method of analyzing the frequency of a signal The essential difference between the two methods however is that the Fourier transform uses the circular functions as basic functions i e the values are calculated by means of sine cosine functions whereas the wavelet transform uses wavelets i e functions whose properties can be adjusted to suit particular problems In contrast to the FFT it is possible with the wavelet transform to analyze local frequency properties The following example illustrates this difference The frequency resolution i e the accuracy with which you can make statements about the signal contents at specific frequencies in the case of the FFT is at 1 T T segment length in data points directly dependent on the length of the EEG segments included in the FFT analysis In order to obtain the accuracy of a frequency resolution of 1 Hz you therefore require EEG segments of exactly a second regardless of the sampling frequency This is exactly where the analytical
95. elay between the ECG and EEG channels e Number of Pulse Intervals Used for Average This is the number of pulse intervals used to calculate the average blood pulse artifact curve e Lowpass Filter for Pulse Artifact This option can be used after calculation of the artifact curve to apply a low pass filter to this curve in order to eliminate radio frequency interference Recommendations for using the module Although it is possible to carry out all correction steps simultaneously you should refrain from doing this particularly when it is not possible to evaluate the quality of the methods for a specific data record with the parameters that are set Instead the following three or four steps are suggested 1 Search for the scanned intervals and set the markers and then assess the stability over time of the located markers If external markers exist these two steps can be skipped 2 Correct the scanned intervals In some cases better results may be achieved if a low pass filter is applied before this step 3 Search for ECG episodes and if this is done in semi automatic mode correct the blood pulse artifacts If the search for blood pulse artifacts is not performed in semi automatic mode you are very strongly advised simply to write the markers in this step and then to assess their temporal structure The Marker Timing analyzer solution which is able to display the time period between two successive markers as a time voltag
96. electrodes of a channel To ascertain the neighboring electrodes the program needs information on the position of the electrodes If you used electrode names according to the 10 10 or 10 20 system for data acquisition the program should have this information If you used other channel names however then you can input the correct coordinates with the aid of the Edit Channels transform component e Bipolar Bipolar connection Differences between channels are formed Choose one of the four reference options To begin with it may be better to take the easiest one the original reference Edit Original Referenced Montage x Chn Chn laa Cancel Insert line I Remove line Insert Curent Channels Remove all Arrange for Grid Views EEEE PROSE ES N non kt ww N JAJJAJ JJJ AR Fig 6 2 Montage edit menu Vision Analyzer User Manual 17 Clicking the OK button takes you to the edit menu for electrodes You will see two columns titled Chn and Chn which indicate the channels and their reference channels The second column is not accessible if you chose bipolar reference On the right the obligatory OK and Cancel buttons are followed by others e Insert Line This button becomes accessible when you have written a text in the first field of the first channel If you click this button the program inserts a line above the current line e Remove Line Wi
97. eline correction should be carried out This is done by averaging the data in the specified time range Since the points of the interval affected by the scan very rarely permit effective baseline calculation an appropriate offset should be set in the interval range dialog box and only points that come before the data points that are affected by the scan should be used for the purpose of baseline calculation e Detect Saturation Here you can specify the saturation limit of your EEG amplifier Values that reach this limit are marked as bad intervals and not included in the calculation of the artifact curve e Common Use of all Channels for Saturation Bad Intervals and Correlation 106 This option allows you to specify that saturated periods for any given channel sections marked as Bad Intervals and sections with a low level of correlation with the template are to apply equivalently for all other channels e There are three mutually exclusive options for calculating the artifact template e Use all Scanned Intervals for Average If you choose this option all the artifact intervals defined via markers are included in the average value irrespectively of whether or not each individual interval actually reflects the occurrence of the artifact e Select Scanned Intervals for Average by Following Criteria Choosing this option significantly improves the quality of the resulting artifact template compared to the previous option since the i
98. en two data points here e Mark as Bad Here you specify the time span around the actual occurrence of the criterion that is to be marked as an artifact Max Min tab e Check Maximum Difference of Values in Intervals If you select this check box the Max Min criterion is applied e Maximum Allowed Absolute Difference Specify the maximum allowed voltage difference here e Interval Length Here you specify the time span within which the voltage difference must not exceed the specified value e Mark as Bad See the Gradient criterion 128 Raw Data Inspector Criteria Fig 10 44 RDI settings dialog for the Max Min criterion Raw Data Inspector Criteria M fao 2o Fig 10 45 RDI settings dialog for the Amplitude criterion Amplitude tab e Check Maximum and Minimum Amplitude If you select this check box the amplitude criterion is applied e Minimum Allowed Amplitude Specify the minimum allowed voltage level here e Maximum Allowed Amplitude Specify the maximum allowed voltage level here Vision Analyzer User Manual 129 e Mark as Bad See the Gradient criterion Raw Data Inspector Criteria x Gradient Max Min Amplitude Low Activity ssssssosse Lowest allowed activity Max Min 0 5 uy Interval length fi 00 ms Mark as bad before event after event 500 ms 500 ms Abbrechen Fig 10 46 RDI settings dialog for the Low Activity criterion Low Activity tab
99. eparately for every selected channel e Searching for peaks takes place in one specified channel only In all selected channels the peaks are marked at the position at which they were detected in the specified channel Detected peaks are output as markers of the Peak type Peak Detection Step 1 of 3 Methods Ea m Automation Methods Searching Methods Semiautomatic Detection Separate Search for Every Channel Automatic Detection Search Peak in a Reference Channel and Set Markers with Respect to this Peak Detection Methods Search for Local Maxima in Interval C Search for Weighted Local Maxima in Interval Cancel Fig 10 36 First page of the Peak Detection dialog The first page of the Peak Detection dialog enables you to input the degree of automation semiautomatic or automatic You also define whether you want to search for peak markers separately for each channel or search for a peak marker in a specified channel Search Peak in a Reference Channel Here you can also choose the method for searching for the peaks You have a choice between searching for the global maximum or minimum in a specified interval searching for a local maximum and for a weighted local maximum The difference between a local and global maximum is that in searching for a global maximum the edge points of the intervals are found as peaks if the value there is greater or less than all values within the interval When
100. ers simply by creating links to the various raw EEGs in the raw data folder of this workspace This also makes it possible for a raw EEG file to be analyzed in different workspaces without it having to be copied 204
101. es must be discontinued at a certain degree The rule that applies here is the higher the spline order the lower the degree of the polynomial at which calculation is discontinued In the article mentioned above degree 7 is regarded as adequate for order 4 The Lambda approximation parameter defines the accuracy with which the spherical splines are approximated to the data to be interpolated For various mathematical reasons a Lambda that is too large or too small leads to an inaccurate representation Unless there are methodical exceptions that speak against it the default value of le 5 should be retained 34 7 6 3D mapping view You can use the 3D mapping view as an alternative to the two dimensional map Here the map is projected onto a head The setting options correspond to those of the 2D mapping view In addition you can rotate the head that is shown To do this move the mouse over the head and hold down the left mouse button while moving the mouse in any direction The head will rotate accordingly Analyzer P300d Raw Data Filters S egmentation BaselineCorrection Average 3DMap _ a x Z File Edit View Display Montage Transformations Export History Template Macro Tools Configuration Window Help 18 xj olaja S A alas Elva Af sl 3 mj fe Simic 2 Raw Data S E Fites EM Segmentation E E BaselineConectic Average 76 120 ms 120 164 ms Ais _ts gt
102. ether all channels or only a subset should be used for calculating the ICA In this way for example trigger channels can be masked out from ICA since as a rule they do not satisfy the aforementioned criteria ICA is a statistical process which means there is no absolute need to use the entire data set length for calculation Often certain areas in the EEG signal can be even more meaningful in terms of the components For this reason the second step in the dialog offers you the chance to restrict the area of the EEG that is used On the one hand this is in order to restrict the calculation to statistically significant areas while improving the performance of the process on the other hand Vision Analyzer User Manual 89 Channels and Matrix Files E Patni Resulting File Mames ee 2 s 2 erWvision Export oj Disabled Ghannels Disable Fig 10 26 First page of ICA dialog 90 Enabled Ghannels DIOWSE References BS95 A J Bell T J Sejnowski An information maximation approach to blind separation and blind deconvolution Neural Computation 7 1129 1159 Car98 J F Cardoso Blind Signal Separation Statistical Principles Proceedings of the IEEE 86 10 1998 MBJS96 S Makeig A J Bell T P Jung T J Sejnowski Independent Component Analysis of Electroencephalographic Data Advances in Neural Information Processing Systems MIT Press Cambridge MA 8 1996 MBJGS97 S Makeig A J
103. filled When all columns of the required rows have been filled correctly pressing the Finish button concludes input of the channel list and the LRP is calculated The structure of the LRP channels is described in the operation info for your information Vision Analyzer User Manual 97 Here is a typical sample application to illustrate the input procedure In an experiment a warning stimulus of the test person conveys information on which hand is to respond to a subsequent imperative stimulus The warning stimulus imperative stimulus and the response left hand right hand are stored as markers in the raw EEG Using the Segmentation module in conjunction with advanced Boolean expressions two segmented data sets nodes are generated The one node corresponds to the response with the left hand and the other node to the response with the right hand You can find more details on this in the Segmentation section of the manual An average is calculated for the two nodes and the average nodes are given meaningful names e g Avg left Avg right The LRP data should be generated as subnodes of the Avg left node The LRP module is therefore applied to this node and the Avg right node is selected as the second data set Let us assume that the LRP is to be calculated for channels C3 and C4 i e C4 C3 is calculated for Avg left and C3 C4 for Avg right Then these differences are averaged Thus C4 is entered in the first column and C3 in the s
104. filter selectable files by name criteria Wildcards can be used for multiple characters and for one character If the TestlH Test2G and Hest5 files are in the workspace then Test will filter out just Test1H and Test2G The filter est would accept all three files etc When you have set the filter press the Refresh button to refresh the selection of available files e Available Files e Selected Files e Output File Name of the secondary history file that is to be created e Create a Data Set for Standard Deviation e Enable Individual Channel Mode Single channel mode is another important feature It is thus possible to include in the averaging data sets that do not have all the channels As a result the number of segments included in the averaging can be different for each channel e Load Parameters This button allows you to load previously saved settings from a parameter file and then apply or edit them e Save Parameters You can save the settings you have made in parameter files in order to reuse them This option gives you the opportunity in the save dialog to save the list of selected files in addition to the other parameters When the operation has been completed the new files appear in the lower pane of the History Explorer Vision Analyzer User Manual 159 10 2 2 Principal Component Analysis PCA Principal component analysis is used to reduce data and also to extract hypothetical quantities which permit a dat
105. fy This transform rectifies EEG data i e positive values remain the same and negative values are converted into positive values of the same magnitude In the dialog you can select those channels that you want to rectify If you select the Keep Remaining Channels option you can keep the other channels unchanged or if you deselect this option you can remove the other channels from the data set Disabled Channels Enabled Channels Al A2 Enable gt gt lt lt Disable V Keep Remaining Channels Fig 10 48 Rectify dialog 132 10 1 28 RMS Global Field Power The total activity of certain channels can be determined with this transform The root mean square of the individual values is ascertained at every time The result is written to an additional channel named RMS This transform can be used at any point in processing RMS channels Ea Disabled channels Enabled channels Enable gt gt lt lt Disable Cancel Fig 10 49 RMS dialog In the RMS dialog you can choose the channels for which you want to carry out a calculation These channels and the new RMS channel then appear in a new data set Vision Analyzer User Manual 133 10 1 29 Segmentation Segmentation by the Segmentation module is based on one of the following criteria e Around individual markers You specify a start and end position relative to a marker position The start and end position can also be before the marker po
106. g only on the basis of intervals that have similar template drift values For this to be possible the Template Drift Detection option must have been used during the detection of the scanned intervals in order to measure the template drift Every scanned interval is assigned the drift compared to the ideal position as a fraction of a sampling interval between 0 5 and 0 5 A range of such drifts is assigned to every artifact template managed by Template Drift Compensation For the purposes of averaging each template takes account of only those intervals which lie within the associated range You can select the number of managed templates The more templates you use the lower the level of interference caused by template drift However fewer intervals are then included in each template and the significance of the template is reduced A value of 3 is a sensible specification In the Correction Channels dialog below you can select the channels for which you want to perform the correction You can either select all the channels Correct All Channels option or select the Correct Only the Following Channels option to choose the required correction channels from the list of those available Figure 10 35 Selecting the channels that are to be corrected You can make the following entries in the following dialog Post Correction Settings 108 MRI Artifact Correction Post Correction Settings x Downsamping Factor od IV Enable L
107. gmentation Peak This marker is set by peak detection routines Black Response Response by the patient Blue Stimulus This indicates a stimulus Red Threshold This is set by the Level Trigger transform and Red because of its proximity to the stimulus marker it has the same color in a view Time 0 This marker only plays a part after averaging Long black It marks the boundary between prestimulus and dashed line poststimulus Voltage When this marker occurs it causes most views Black to show the voltage and time on the channel in question Fig C 1 Table of predefined marker types Vision Analyzer User Manual 199 Annex D Keyboard shortcuts The following table shows the assignment of some keys and key combinations in the Analyzer Key or Function key combination Ctrl 1 to Ctrl 0 Selection of montages Ctrl 1 is always the default montage Ctrl gt In the data window Show next interval Ctrl In the data window Show previous interval Ctrl In the data window Increase scaling more sensitive Ctrl 4 In the data window Reduce scaling less sensitive Ctrl C Copy the currently displayed graphic to the clipboard Del In the History Explorer and in a history template Delete a data set or a secondary history file F2 In the History Explorer and in a history template Rename a data set or a secondary history file 200 Annex E Installation Network License USB The Brai
108. hanged in a montage so that channels which were originally apart can be shown next to each other A channel can also occur multiple times in a montage Another important characteristic of a montage in the Analyzer is that certain display parameters such as position and size of a channel can be assigned to it in a head view A head view is a view in which the channels can be positioned freely in the window Its size can be changed to meet particular requirements The head view option is explained in more detail in the next chapter A montage is used for visualization purposes only i e the new data exists just temporarily and the original data is not changed in any way m Choose Reference Cancel C Average Laplacian Bipolar Fig 6 1 New Montage start dialog Choose Display Montage gt New from the menu in order to create a new montage This brings up a dialog in which you are asked about the type of reference to be used in the new montage There are four options e Original No new reference is calculated here This type of montage is only used to group channels or optimize their presentation as described above e Average The average reference is calculated here i e the average of all selected channels is used as the reference 16 e Laplacian Source derivation according to Hjorth This is a method derived from the Laplace transform in which the reference is calculated from multiple neighboring
109. hannels Cancel Fig 10 18 CSD dialog The Lambda approximation parameter defines the accuracy with which the spherical splines are approximated to the data to be interpolated For various mathematical reasons a Lambda that is too large or too small leads to an inaccurate representation Unless there are methodical exceptions that speak against it the default value of 1e 5 should be retained Vision Analyzer User Manual 73 Despite these rules of thumb it is not advisable to accept the results of this module without checking them In the CSD method various parameters can cause good or bad results in different data sets owing to approximations and rounding inaccuracies It is therefore advisable to proceed as follows when using this module e Before running the CSD calculation the parameters of the spherical splines should be checked in the current data set using the mapping view To do this generate a mapping view for the data set and select interpolation with spherical splines using the parameters that you want to use for CSD calculation e Check possibly by comparing with interpolation by triangulation whether the spherical splines satisfactorily approximate the voltage distribution on the surface of the head If not change the parameters until this is the case e Run the CSD calculation with these parameters and then make sure that the result is correct by generating a mapping view 74 10 1 11 DC Detrend This tran
110. hannels Otherwise the new data set consists of new channels only e Number of Pools Refresh Here you specify the number of new channels that you require The Refresh button then updates the channel matrix e Clicking a field in the table causes the color of the field to change between green and white Green means that the channel will be included in the calculation of the new channel White means that this channel will be ignored in pooling for the new channel Vision Analyzer User Manual 125 10 1 26 Raw Data Inspector With the aid of the Raw Data Inspector RDI you can inspect the raw data set for physical artifacts This inspection can be manual semiautomatic or automatic In semiautomatic and automatic inspections you can specify criteria for the artifacts and ranges before and after the artifact which are to be marked as bad ranges To remove artifacts after segmentation use the Artifact Rejection module which is described earlier in this document The following criteria are available e Gradient criterion The absolute difference between two neighboring sampling points must not exceed a certain value e Max Min criterion The difference between the maximum and the minimum within an interval must not exceed a certain value e Amplitude criterion The amplitude must not exceed a certain value or fall below another certain value e Low Activity The difference between the maximum and minimum in an interval of selectable
111. he component area is recalculated and then a new node is generated with the required name As with loadings channels are stored as graphs and time points as maps In addition to storing components as new nodes you can generate new EEG nodes from the combination of components and loadings To do this use the Create New EEG node button This brings up a dialog in which you enter the name of the node and the factors that you want to use You can use all factors or exclude certain of them Here too the components are calculated first and then the calculated EEG is generated as a subnode of Loadings Here are two applications as an example of generating new EEG nodes Vision Analyzer User Manual 163 Time space filters After eliminating factors containing noise the generated EEG contains data that is filtered on a time and space basis Ocular artifact correction By specifically excluding factors corresponding to artifacts the generated EEG contains corrected data If the data that is generated with PCA is not due to be processed further in the Analyzer you can also export loadings and components To do this click the Export Loadings or Export Components button This brings up a dialog like that for Generic Data Export You will find an explanation of the settings in the description of that export component 164 10 3 Transient transforms 10 3 1 3D Map If you select this option a three dimensional map is generated which shows th
112. he current segment number Segment x of y e A window with the text Remove or Keep indicating what is to be done with the current segment e The Remove button to include the current segment in the list of segments to be removed and move on to the next segment 50 The Keep button to take the current segment out of the list of segments to be removed and move on to the next segment The lt lt button to move to the previous segment The gt gt button to move to the next segment The Goto button to go to a specific segment Remove Segments The segments that are due to be removed are listed here You can display a segment by double clicking it Step Only to Kept Segments If you select this check box the program goes to the nearest previous segment that is not in the list of segments to be removed when the lt lt button is clicked The equivalent applies to the gt gt button for subsequent segments Step Only to Removed Segments This check box has the exact opposite effect to the previous one Show Artifacts This causes the marked artifacts to be displayed Artifact Rejection Segment 1 of 32 No rtifacts Clear all channels Mark all channels gt gt Goto I Step only to clean segments T Step only to segments with artifacts Segments with artifacts Segment Artifact Channel Fig 10 3 Dialog for manual segment selection in individual channel mode Vision Analyzer User Manual 51
113. his can impair the flow of the phase information which is actually rather steady and make the result of a subsequent coherence analysis appear questionable To conclude it is worth repeating that spectra interpolated like this are suitable above all for the visual inspection of data and should not form the basis for subsequent calculations Brief description of how the data window works In a Fourier transform it is assumed that the output signal is repeated periodically Since this condition is generally not given when EEG sections are transformed the difference between the voltage level at the beginning of the segment and the voltage level at the end of the segment is included in Fourier transform calculations as the point of jump discontinuity and causes the data to be corrupted To reduce this effect it is possible to lay a data window over the segment to be transformed This data window damps the EEG data at the ends When the data window has been applied the voltage level at the beginning and end of the segment is 0 and rises to the original measured value at the middle of the segment The range in which the data window is to be used can be specified in percent 100 would mean that only the value in the middle of the segment matches the original data value and all other values are damped by the data window The lower the percentage that is specified the smaller the range that is changed by the data window This principle is represented
114. hree storage options e No storage i e the data is generated on request e The data is cached in a temporary file e The data is stored in the history file in compressed format In most transforms it is not the results that are stored temporarily but information that describes the result of the operation As far as segmentation is concerned that is the position of the new segments in the output data set When data is requested for display or further processing transform or export then it is recalculated by the transform object The advantage of this approach is that no intermediate files are generated for the various operations while all intermediate results are retained A disadvantage of this approach is that the speed of certain steps such as Ocular Correction Baseline Correction and Average suffers under constant recalculation 134 To compensate for this disadvantage it is possible to store the results of segmentation temporarily in a cache file The transform object can access the cache file when data is requested This cache file exists as long as the history file is open If the history file is closed and then reopened the information is still available but is now calculated at request time Operations which follow segmentation should therefore be carried out without closing the history file in the meantime If you want to carry out operations after segmentation nevertheless you can recreate the cache file To do this yo
115. iesssaaaddiesaaade eabasaddaesaeddoanssatousyaabieessiaadoeesauaaiea s 62 Vision Analyzer User Manual 3 10 1 6 Ghange Sampling Fale scscesceveres a estate Sater ee eer eer eee 63 BOA TAE ONO VCS as se ge ee 64 GOB ieee Me Or ots g 214 ear irnna Pe n ren aa REAR oe A PRO PORT OAT eR Crea Per mn tin vere 67 POW OAC OMAN GON ia eS RA St AT 69 10 1 10 Current Source Density CSD cccccccccssssccneeeeeeeseeennnneeeeeeeeeesenennes 73 10 111 VICI OWCIO roen e aaa ee e aE aAa 75 10112 Edt Channels nenea aE E E EN 77 10 1 13 Fast Fourier Transform FFT cost SescsntetescinteSracindedeatinteGracinbedsatintsGoaens 79 VO S K PE E A E A E E A A E 84 10 119 Formula Evaluatot ec csosdatedeosdstaacuntadeaieeds Grsctndemediandsaeuneatenaicurdseanunteaeens 86 10 1 16 Frequency extraction spices senioe oie eke sarees oe eee ieee eabne ot eee eeeieedee 88 10 1 17 ICA Independent Component ANnAalySiS ccccccccccesstseteeeeeeeeeeeeees 89 TOTO MOVER TMOOCN i sesane enni 92 TOTS TINCAlDCNVANOG Scns ceiaa stor sath a Bish cai tees hot esha so ahs hos hn teehee at 94 10 1 20 Lateralized readiness potential LAP ccccccccceessseeeeeeeeetseeenees 96 10 1 21 MRIArtifact Corre0ton nsninineneninernonei naa Na 99 WORE New Referente nonosannniren aa a a a aa Na 115 10 1 23 OC lar ICON CCHON 2 22 siavsskissarcabs siapeshe caenaaance vache psaaaananiai debi igeaine sieve 117 10 1 24 Peak Detection i253 ra cece iucaszcaiuc
116. iew appears with the same setting options as in manual segment selection 10 1 1 3 Automatic segment selection Here you perform all operations in exactly the same way as for semiautomatic segment selection The only difference is that you do not have any opportunity to make corrections 54 10 1 2 Average The Average module is used for averaging data which has been segmented It is used after Filtering optional Segmentation Ocular Correction optional Artifact Rejection Local DC Detrend optional Baseline Correction Note that criteria such as marker types exclusion of segments with incorrect patient responses and the like are defined in the Segmentation module and not here With the Average module you can average either all segments or one chosen range You can also specify whether you only want to average segments with odd numbers segment 1 3 5 or even numbers 2 4 6 Another important option is individual channel mode Here the program no longer assumes that all channels for every segment are to be included in averaging but that each channel can be considered on its own If only one channel in a segment has been marked as bad all other channels are used in averaging nevertheless The result is that the number of segments included in averaging can be different for each channel To use individual channel mode you have to make preparations in various preprocessing steps e Ifyou use the Raw Data Ins
117. iles Only You can confine the selection to primary history files only e Use Whole Workspace e Select Individual History Files 44 e Selection Filter With this option you can filter selectable files by name criteria Wildcards can be used for multiple characters and for one character If the TestlH Test2G and Hest5 files are in the workspace then Test will filter out just Test1H and Test2G The filter est would accept all three files etc When you have set the filter press the Refresh button to refresh the selection of available files e Available Files e Selected Files Here you have a choice of Whole Workspace or Select from List The selected history files are processed with the specified operations when you press the OK button or the Enter key If a history file has already been processed with the template in question it is ignored Vision Analyzer User Manual 45 9 Macros A Basic interpreter has been built into the Analyzer so that users can program functions ranging from simple automation macros to complex applications This interpreter accesses the Analyzer via the OLE Automation interface This interface gives you access to many methods and properties of the Analyzer as well as access to every single data point in a data set history node in all history files Choose Macro gt New to write a new macro This causes the menu bar and tool bar to change In addition an edit window opens containing the follow
118. ing baseline activities can be taken out of the data before the experimental stimulation in order to obtain stimulus induced changes to the frequency content Please note that although the normalization and the baseline functions can be used at the same time they cannot be used over the same time ranges This is because by definition a baseline function sets the total of the values in the defined time domain to 0 and if a normalization function were to be performed at the same time it would attempt to define this 0 total as 100 Merely as a result of the finite accuracy of the numerical results of the baseline function this would mean that completely nonsensical values would be calculated but no errors would occur however The module therefore gives a warning when the same time domains are entered for the normalization and the baseline functions It goes without Vision Analyzer User Manual 155 saying however that the normalization and the baseline ranges are allowed to overlap and indeed this may be selected for quite proper reasons 156 10 1 32 Wavelets Layer Extraction Using this module you can extract an individual frequency range layer of a wavelet data set in order to continue processing it separately The resulting data set of this transform is exclusively a time domain data set in contrast to the time frequency domain of the wavelets The entry option is limited to selecting a layer Layer Extraction x Layer 8
119. ing two lines Sub Main End Sub You insert the actual macro code between these two lines The short macro shown below simply opens all history files outputs a message and then closes the files again Sub Main For Each hf In HistoryFiles hf Open Next sgBox All history files are open For Each hf In HistoryFiles hf Close Next End Sub Macro Options x Macros in Menu Macro Adam arker Apply T emplate 1 2 6 Emy H cn i Cancel Fig 9 1 Macro option dialog 46 When you have input your code you can test it with the F5 key You can save the macro with File gt Save and close the edit window To run an existing macro when you are not in the macro edit window choose Macro gt Run You can then choose the macro you want to run Alternately you can make macros appear as items on the macro menu bar Choose Macro gt Options to do this Here you can choose up to 10 different macros When you have made your choice and reselected the Macro menu you will find your macros on the bar You can now also access the chosen macros via keyboard shortcuts Alt M 1 2 3 Please refer to the Vision Analyzer Macro Cookbook and the Vision Analyzer Ole Automation Reference Manual for more details about writing macros You can find out more about the built in Basic during an editing session by means of Help gt Editor Help and Help gt Language Help Visio
120. instead of the central frequencies of 20 0 23 3 27 1 31 6 36 8 42 9 and 50 hertz that apply in the logarithmic arrangement Normalization and base line correction If the data exists in segmented form another dialog box appears before the calculation in which you can specify for the wavelet data whether the data is to be normalized or whether a baseline correction of the wavelet coefficients is to be carried out If the normalization function is selected the wavelet coefficients are normalized to a total of 100 in the selected range for each scale frequency step This enables relative comparisons of the activities between conditions or test groups to be carried out 154 Normalize and Baseline Correction x Normalize V Enable Normalize Based on Time Start ms fo End ms fico Duration ms jo C Based on Data Points Start point fa End point 24 Pomts 125 Baseline Correction IV Enable Baseline Correction Based on Time Start ms 200 End ms jo Duration ms 200 C Based on Data Points Start point 1 50 End point fi Paints fsa Finish Cancel Fig 10 64 Normalization and base line correction of the wavelet transform In the case of the baseline function the average value of the wavelet coefficients in the specified time domain is calculated for each scale and deducted from all the wavelet coefficients in the entire time domain In this way as with the analysis of time based data exist
121. intervals which do not contain any DC reset are included in the calculation In the fourth step the trend is subtracted from the original data taking the DC offset into account You can input the following parameters for this algorithm e The interval length in milliseconds or data points e The data type of the markers for which the prestimulus intervals are calculated and the length of these intervals Vision Analyzer User Manual 75 e The length of the intervals before DC corrections for which the voltage averages are to be calculated for the following data together with the resultant offset If this transform is called after segmentation then local DC trend correction is applied Here a linear function is subtracted from the data from every segment The slope and boundary values of this linear function are calculated for every segment from a certain interval at the beginning and at the end of the segment In the DC Detrend dialog you can specify the intervals based on time or data points To minimize the influence of DC trends on ocular artifact correction it is advisable to perform DC trend correction before ocular artifact correction DC Detrend for Segments x Intervals based on time C Intervals based on data points Interval at segment start ms fi 00 Interval at segment end ms fi 00 Cancel Fig 10 20 Local DC Detrend dialog 76 10 1 12 Edit Channels This transform enables channel information
122. irst dialog When you select Artifact Rejection from the Transformations menu the first dialog appears where you have a choice of three modes e Manual Segment Selection e Semiautomatic Segment Selection e Automatic Segment Selection This dialog also includes an item labeled Individual Channel Mode With this mode you do not need to reject entire segments but can simply mark individual channels as bad In this case the Average module will later search for as many segments as possible separately for each channel also see Average Vision Analyzer User Manual 49 The final option in this dialog is Mark Bad Segments Instead of Removing Them Here you define whether bad segments i e with artifacts should simply be marked instead of removed If you do not check this option the new data set will only contain the remaining segments The three segment selection methods are described in detail below 10 1 1 1 Manual segment selection When you press the Finish button a grid view appears which has a dialog on its right If you did not select individual channel mode you can remove individual segments here Artifact Rejection Segment 13 of 165 Keep Remove Segments Keep Change Criteria I Step only to kept segments I Step only to removed segments IV Show artifacts Cancel Fig 10 2 Dialog for manual segment selection For this purpose there is a dialog box with the following elements e Display of t
123. is also possible to implement simple filters such as sliding averages with the aid of the shift function Fp1 shift Fp1 1 Fp1 shift Fp1 1 3 or Fp1 shift Fp1 1 0 25 Fp1 0 5 shift Fp1 1 0 25 If your channel names are numeric place them in quotation marks in the formula Here is an example 1 shift 1 1 1 shift 1 1 3 Vision Analyzer User Manual 87 10 1 16 Frequency extraction Frequency extraction works on the principle of complex demodulation In this process the EEG signal is transformed continuously by mathematical methods so that the resultant signal only consists of those component frequencies that lie in the defined range In contrast to a Fourier transform only a specific frequency range can be extracted The advantage of this transform is that it also works at high speed with unsegmented EEGs and the output of each channel is a continuous channel again This transform is not to be confused with bandpass filtering The resulting signal is not the filtered input signal but it describes the power or phase of the selected frequency range at any time There are the following input options for the transform object choice between output of the power or the phase and choice of the frequency range to be extracted Frequency Extraction x Output Power Phase m Frequency Range Begin Hz 7 5 End Hz fi 25 Cancel Fig 10 25 Frequen
124. is changed not the data itself Set scaling Reset scaling to original value Set options for different views These options are described in the following sections EE Ea t Turn marker edit mode on off This mode is described in the Setting markers manually chapter W Turn History Explorer on off am Cascade all view windows Vision Analyzer User Manual 23 ju Tile view windows side by side Tile view windows one after another The navigation bar is used to move along the time axis dis tsb _ gt R Slider Marker window Fig 7 4 Navigation bar The buttons on the left 1s support forward backward navigation by one second or if a section is lt 1 second by 100 ms The buttons that follow are used to switch forward backward by the displayed interval minus one second i e the intervals shown in succession overlay each other by one second To the right of these buttons you will see the marker window and beneath that the slider window Both of these windows represent the entire EEG in their width The blue slider represents the section that is currently being shown You can grab the slider with the left mouse button and drag it to the left or right The EEG display is updated accordingly when you release the mouse button You can also left click both in the marker window and in the slider window In this case the EEG display is positioned accordingly Pressing the right mouse button in the marker wi
125. ist frequency It is thus clear that a time frequency analysis specifically for 32 Hz for example cannot be implemented Given a signal with a sampling rate of 256 Hz however accurate time frequency statements can be made for the frequency ranges 64 128 Hz 32 64 Hz 16 32 Hz 8 16 Hz 4 8 Hz 2 4 Hz and 1 2 Hz Cumulation of the result values is also possible so that for example the analysis of the delta band can be carried out in the time frequency curve A further advantage that often makes the DWT rather than the CWT appear to be the method of choice in the EEG field is the simplicity with which the wavelet coefficients can be converted back to the time domain This is possible because the filters used represent orthonormal bases and reverse transformation can thus be implemented easily The discrete wavelet transform can thus also be used for wavelet based filtering of the data Continuous wavelet transform CWT The continuous wavelet transform provides considerably more scope at those very points where the very elegant DWT is limited It permits specification of the frequency bands to be examined and of their resolution and it offers more options when it comes to selecting the underlying filters mother wavelets see below 146 Unlike the DWT the CWT does not work with semiband filters it works with mother wavelets instead These are short signal sections shaped on the basis of underlying functions in such a way that when filte
126. ith the Basic macro you can open and read ASCII files Sample macros for reading in channel positions and markers are described in the Vision Analyzer Macro Cookbook 188 13 Printing You can print out the currently displayed EEG section via File gt Print The normal Windows dialog for selecting the printer etc appears File gt Print Preview is available to give you a preview of the output However the actual output on the printer may differ from the preview owing to the quality of printer drivers upon which we unfortunately have no influence Choose Configuration gt Preferences from the menu and then the Graphics Export Output tab to set headers and footers as well as margins Preferences x Views Scaling Graphics Export Output Transformation Colors m Printer Output Default Paper Orientation C Landscape Margins cm Left 1 25 Right 1 25 Top 1 25 Bottom 1 25 Header 7 Footer Clipboard Size of the Enhanced Metafiles Width em fi 6 Height crm fi 2 C Use screen resolution Use printer resolution Abbrechen Fig 13 1 Print options dialog Here you can define whether the default paper orientation is portrait or landscape You can also specify the left right top and bottom margins in centimeters Note that portrait orientation means the height of the graphic is automatically restricted to about 2 3 of the width and that any conflicting bottom ma
127. ition to the loadings yellow area the components are displayed in the right hand blue area In the opposite way to loadings components are presented as graphs when channels were chosen as the variables and as maps when time points were chosen In both areas you have all facilities that you know from handling the grid and mapping views The settings can be changed in a dialog for both of these areas by clicking the Settings button on the Analyzer s menu bar 162 E PCA Loadings Component Control Nodes 1300a Raw Data Filters Seamental C Select all segments Select a single segment fi Write components to subnode Create new EEG node Export Export loadings Export components Fig 10 69 PCA view Since there may be a large number of components depending on the number of segments nodes and files not all components are calculated automatically and displayed Instead in the right hand dialog area you can select nodes and segments for which you want the components to be calculated and displayed Clicking the Update View button starts recalculation and output In this way you can view exactly those components that you want to The selected components can also be written as subnodes of Loadings to the history file so that you can process the data later with other modules To do this click the Write Components to Subnode button and enter the name of the new node in the input dialog that appears First t
128. ked with the first end marker found to form a segment and the second start marker is linked with the second end marker etc This means it is also possible to generate overlapping segments by setting markers correspondingly Any end marker that is found without a previous start marker is ignored Start markers for which no end marker is found by the end of the data set are also ignored 140 Segmentation Wizard Step 2 of 3 Marker which defines the start of a segment Start Comment Marker which defines the end of a segment End Comment lt Back Cancel Fig 10 55 Dialog to select limiting segment markers Vision Analyzer User Manual 141 10 1 30 The t test It often cannot be reliably estimated from the curves of two EEC curves the grand averages of two groups for example or two experimental conditions whether visible differences are also statistically significant Conversely particularly in the case of examinations of a more exploratory nature it is often not possible to clearly establish from the available averages the time section in which the conditions or groups have particularly marked differences In all these cases it can be extremely useful to calculate t values for the data The t test in the Analyzer permits the calculation of paired and unpaired t tests as well as t tests against zero It can be applied to segmented and unsegmented data as well as to averaged data and of course
129. l be regenerated automatically and the EEGs will have been separated Vision Analyzer User Manual 193 16 Solutions In the manual you have learned about various possibilities for manipulating data or controlling the Analyzer The first components for data manipulation are the transform modules that contain all the fundamental steps for calculating evoked potentials as well as many other standard processes of neurophysiological research Furthermore it is possible to write your own macros for manipulating data importing and exporting data and controlling the Analyzer The advantage of these macros is that individual solutions can be implemented in a relatively straightforward manner The disadvantage of programming your own macros is that you may first of all have to acquire the necessary expertise to do so Furthermore macros should be documented so that the process employed can be tracked If the documentation is not stored together with the macros it is always possible that it will become lost or that its version will not keep pace with the version of the macros Macros can also be changed easily as a result of their legibility which could perhaps have undesirable side effects The purpose of the Solutions is to bypass these disadvantages They are written by Brain Products and are stripped down solutions for a very wide range of problems and tasks A Solution is an individual file containing the runnable code The documentation fo
130. lculating the values over multiple channels Instead you are advised to perform detection over a single channel only This channel should however be strongly affected by the occurrence of the artifact in order to guarantee that the start of the artifact is detected as precisely as possible The module also makes it possible to use Template Drift Detection in order to identify the time shift template drift between the averaged artifact curve and the scanner artifact for each Vision Analyzer User Manual 99 interval This drift occurs if the scanner s repetition time is not an integer multiple of the sampling rate The artifacts of the individual intervals then appear to be shifted by the fraction of a sampling interval with reference to one another even if the Scan Start marker has been optimally positioned Since the artifact may contain very high frequencies this drift by the fraction of a sampling interval can result in considerable interference in the averaged artifact template Over time this destroys the quality of the template Template Drift Detection measures the drift of each interval and makes this available to the Template Drift Compensation correction method Template Drift Detection also uses the measured drift to adjust the Scan Start marker so that the drift is smaller than one sampling interval If the scanner artifacts follow one another without intermediate artifact free sections then it is difficult to identify the sta
131. lider in this window represents the currently displayed section whereas the window itself represents the entire EEG The slider can be moved with the mouse If you move the mouse to the marker window and press the right mouse button a context menu is displayed and you can hide the entire marker window or choose specific marker types Clicking any position in the marker or slider window displays the corresponding section of the EEG You can use the tool bar which is located at the top beneath the menu to define the time span shown the number of channels to be displayed simultaneously and other aspects You can obtain help on the functions of the navigation and tool bars by positioning the mouse on the buttons or various elements on them After a short time a tooltip with some brief information will appear in a small yellow window At the same time the status bar at the bottom of the program window will display some more details In addition the status bar contains seven windows which give information on montage the segment displayed mouse position and the current workspace 10 The first window shows the current montage in magenta font Montages are dealt with in detail later The second window indicates the time that corresponds to the beginning of the displayed EEG interval in blue font The third window shows the current segment number at the beginning of the displayed EEG interval also in blue font Segments are also described in
132. lowed by that of the second sampling time etc As far as ASCII export is concerned the data for every sampling time is written to a separate line If you export complex data the real part is written first followed by the imaginary part ASCII Line Delimiters You can specify the format of line delimiters for all exported ASCII files in order to process the data further on different operating systems The principal formats for most PC operating systems UNIX and the Macintosh are available Export File Name Here you specify the base name and extension of the file The program then shows you the resultant file name The second page of the dialog provides options for ASCII or binary parameters Depending on the format you specified the left ASCII or right binary half of the page is accessible Generic Data Export Settings Page 2 of 3 ESEI Date File Settings Binary Data File Settings IV Add channelinames to the date file EEE 32 Bit floating point format a EE EAS Ml fewest sca ayer Decimal symbool m M Set resolution manually Resolution pv 0 03052 Range mi 1 000 M individually optimized resolution foreach channel I Convert to big endian order Mac Sun 68k lt Back Cancel Fig 11 3 Second page of the Generic Data Export dialog Vision Analyzer User Manual 173 The ASCII options are Add Channel Names to the Data File The channel names then appear in the first line with m
133. lue Ch lt x gt Coordinates of an individual channel in the If the value is not listed x stands for the form here the Analyzer uses Vision Analyzer User Manual 185 channel number i e the keyname for the first channel is Chl for the second channel Ch2 etc lt Radius gt lt Theta gt lt Phi gt Example Ch1l 1 92 72 The coordinate system of the Analyzer is described in Annex B the electrode name of the channel searches for the coordinates in the 10 10 system and uses them If the channel name is unknown the coordinates are set internally to 0 0 0 GDR compatible marker file The marker file is based on the same principle of sections and keynames as the header file It should be given the file name extension vmrk and the same base name as the associated EEG file The first line identifies the marker file and is as follows Brain Vision Data Exchange Marker File Version 1 0 The various predefined sections with keynames meaning and default values are listed below Common Infos This section contain s general information on the marker file Keyname Meaning Default value DataFile Name of the EEG file If the name does not contain a path it is assumed that the EEG file is in the same folder as the marker file This information is not evaluated by the GDR Marker Infos The individual markers and their properties are listed in this secti
134. lution ouble click for documentation Pg Cara Right mouse click on an icon gives you a ontext menu with additional options Fig 16 2 Solutions help All Solutions and macros located in the Solutions folder or its subfolder s are displayed in a clear structure in a dialog A brief description of a Solution can be displayed by clicking it once Double clicking a Solution displays the full documentation Double clicking a macro displays the macro source code Alternatively right click a Solution to display a shortcut menu in which you are presented with several options including that of running the Solution Please contact Brain Products directly if you are interested in other individual Solutions Vision Analyzer User Manual 195 Annex A Raw data on removable media Removable media are storage media that can be removed from their drives such as CD ROM ZIP and MO drives If you use the same drive as the raw data source for different workspaces it could happen that a raw EEG occurs in several workspaces because the Analyzer normally will create all raw EEGs in the raw data folder when there is a change of workspace or will analyze them at program start time and may create history files The following rule applies to all removable media in order to simplify handling If there is at least one history file in the workspace for which there is no raw data counterpart on the removable medium and the medium contains EEGs which have not bee
135. lways uses the standard view and the currently selected montage Analyzer P300c Raw Data Filter FFT BEE File Edit View Transformations Montage Export History Template Macro Workspace Configuration Window Help la x oela S elelr afr ze mir alms 2 ry 13008 m P300b 5 QA P300 D E Raw Data B A Filter Stimulus EA Stimulus2 X todat Ais tsp a Ready Standard Montage 00 00 07 Segment 1 2 17 0Hz P300 Fig 10 71 FFT as a transient operation Vision Analyzer User Manual 167 10 3 4 Map Here a topographic two dimensional map is generated which represents the voltage distribution on the head in the time or frequency domain In order to represent a map valid head coordinates are needed here in the same way as for the 3D map Z Analyzer P300d Raw Data Filter Stimulus1 BaselineCorrection1 Artifact Rejection1 Average1 Mapping Z File Edit View Transformations Montage Export History Template Macro Workspace Configuration Window Help ll x sli S elafs el a 22 Mir Alms 2 Fp2 13008 R H P300b ZOS m P300 pet ees ae sees QQ P300d BS Raw Data a f Fite BB Stimuust BJ BaseineConect EB Artfact Reje Averag EB Stimuus2 fi BaselineConecti EB Artfact Reje E Avera BJ Segmentation A Average H rex m P300e m GrandAverage di
136. m 0 530516 0 300000 12 m 70 000000 12 O E m 0 530516 0 300000 12 m 70 000000 12 O 0530516 0 300000 12 p 70000000 i2 mo g m 0 530516 0 300000 12 m 70 000000 12 O So m 0 530516 0 300000 12 m 70 000000 12 O E Fig 10 23 Filters dialog You can set the following items in the Filters Dialog High pass filtering Low Cutoff and low pass filtering High Cutoff e Enable This turns the filter on off 84 e Frequency Time Constant This defines the cutoff frequency or as an alternative with high pass filtering the time constant e Slope This defines the slope of the filter Bandpass filter e Enable This turns the filter on off e Frequency This defines the cutoff frequency 50 or 60 Hz If you select the Enable Individual Channel Filters check box the filters can be set separately for each channel in the table below the check box Clicking the Fill Table with Values from Above button causes the values entered above to be written to the table You can then change the channels that need to be filtered differently Vision Analyzer User Manual 85 10 1 15 Formula Evaluator This module enables new channels to be calculated as functions of existing channels Extensive mathematical options are available x wame Formule SSOSOSSSCSCC Ise Line X1 Fpt Fp2 2 uv e a 5 uV 7 Remove Line 1 L a Remove All IV Keep Old Channels I New Channels on Top Load from File
137. manual Distance between tickmarks 100 Distance between Hekman A fi Tickmark labels m Tickmark labels M Show labels M Show labels Distance between labels ris 500 Distance between labels ii fc Cancel Fig 7 7 Grid View Settings dialog Axes tab The Axes tab enables you to set the X and Y axes in accordance with your requirements You have the following options which you can choose separately for the X and Y axes 28 Show Never Show if Size is Sufficient or Show Always Here you can define whether the axis should never be shown only if there is enough space or whether it should always be shown Position For the X axis you choose the Baseline position or Bottom For the Y axis you choose either Left or Time 0 Tickmarks Here you define the tick marks along the axes They can be calculated automatically or be set manually Set Automatic or Set Manual In the latter case you can input distances in ms or uV Tickmark Labels Here you define whether the tick marks are to be labeled and if so at what distance apart Gridview Settings x Display Axes Overlays m Display Style m Labels Placeholders Color c Channel name h History file teeny n History node Remove Line o Overlay number p Complete path Remove All Line Width Pattern Select aes Select Epp Select i Reference Node Select E Select D
138. mplex formats It gives you many setting options to describe your raw data These two methods are described below 12 1 1 Besa format ASCII files can be imported in Besa format This format has the following structure First line general information NPTS lt no of data points gt TSB lt time 0 in ms gt DI lt sampling interval in ms gt SB lt scaling of data points in 1 uV gt SC lt display scaling is ignored gt NCHAN lt no of channels gt Example NPTS 1024 TSB 100 DI 3 90625 SB 1 SC 1 NCHAN 32 This is a data set with 1024 data points and 100 ms prestimulus interval The digitization interval is 3 90625 ms which corresponds to a sampling rate of 256 Hz Scaling of data points is 1 which means the values are specified directly in microvolts Display scaling SC is ignored The data set has 32 channels Second line channel names The channel names are listed in the second line separated by spaces Example Fpl Fp2 F3 F4 etc Starting at the third line data Data starts in the third line in the shape of floating point numbers The decimal symbol is always a point Every line contains the data for one channel The individual data values are separated by spaces If you store the data in a raw data folder the Analyzer will read it in like a normal raw EEG You can generate a sample file by exporting part of an EEG as a Besa file as explained in the Export components chapter 180 12 1 2 Generic Data Reader Th
139. multilayer EEG of the 1 type TIMEFREQUENCYDOMAIN _COMPLEX LayerLowerLimit Lower limit in multilayer data In the case of the 0 type TIMEFREQUENCYDOMAIN _COMPLEX the unit is Hertz LayerUpperLimit Upper limit in multilayer data 0 LayerFunction Function that describes the intervals between the LINEAR layers of multilayer data Possible values LINEAR Linear function LOGARITHMIC Logarithmic function ASCII Infos This section is only relevant if DataFormat in the Common Infos section was set to ASCII Keyname Meaning Default value DecimalSymbol Decimal symbol that is used in the EEG file This Point symbol can be a point or comma In the header file the decimal symbol is always a point SkipLines Number of header lines to be skipped 0 SkipColumns Number of columns to be skipped at the 0 beginning of a line Channel Infos channel number i e the keyname for the first channel is Chl for the second channel Ch2 etc lt Channel Name gt lt Reference Channel Name gt lt Resolution in uV gt Example Chl Fp1 1 Here the first channel is named Fp1 The reference channel is assumed to be the common reference channel because no entry has been Keyname Meaning Default value Ch lt x gt Individual properties for the channel are specified lt Channel x stands for the with commas between them Number gt 1 0 1 Chi 1 1 for channel 1 for example 184 made Resolution i
140. n atan sinh cosh tanh real real part imag imaginary part arg argument of a complex number e The shift function this will be explained in more detail later The Formula Evaluator processes both real and complex data but the format of the output data must be identical to the format of the input data Real numbers are converted into complex numbers automatically if necessary The mathematical functions are currently available for real arguments only Numeric constants are interpreted as a fixed number across the entire range Channel names are interpreted point by point The shift function can be applied to channel names It has the following form shift channel shift index The shift index can be positive or negative This function is regarded as a left shift i e shift Fp1 1 shifts Fp1 in such a way that the data point which has index 2 becomes the data point with index 1 Accordingly with shift Fp1 2 for example the data point which has index 1 becomes the data point with index 3 Examples Many functions in existing modules can also be calculated with the aid of the Formula Evaluator Linear derivation can be calculated with a formula like this channel a channel 1 b channel 2 RMS of Fp1 Fp2 and Fz can be calculated with the following formula RMS sqrt Fp1 Fp1 Fp2 Fp2 Fz Fz 3 Similarly other measures can also be defined here Rectify can be defined with channel abs channel It
141. n Analyzer User Manual 47 10 Transforms This chapter alphabetically lists the transform components that currently belong to the Analyzer in the way they appear on the Transformations menu and in the context menu of a view after marking a block There are three basic types of transforms in the Analyzer These are primary transforms which store the nodes in a primary history file e g filters secondary transforms which generate secondary history files e g Grand Average and the transient transforms which were described in the View chapter and whose result is only kept temporarily Secondary transforms appear at the bottom of the Transformations menu kept apart from primary transforms by a separator It is in the nature of secondary transforms that they cannot be included in history templates Transient transforms are available when you mark a block as described in the Views chapter 48 10 1 Primary transforms 10 1 1 Artifact Rejection After segmentation the data set can be examined for physical artifacts with this transform Segments with artifacts can be removed or marked If artifacts need to be marked before segmentation please use the Raw Data Inspector which is described later Artifact Rejection Step 1 of 3 Methods Ed Methods Manual segment selection C Automatic segment selection Mode Tl Individual Channel Mode I Mark bad segments instead of removing them Fig 10 1 Artifact rejection f
142. n Vision Analyzer Professional Edition Network License referred to as the Vision Analyzer below software is protected Before you can use this license you must first install the network license service This can be done on any computer in the network Software requirement Analyzer Editions with a version number 1 05 0003 or higher and Windows 2000 XP 2003 To perform the installation you require the HASP HL net key referred to as the USB Dongle below Make sure that your computer is connected to the network before starting installation Carry out the installation as described below 1 Install the HASP HL device driver The installation program for the HASP HL device driver is in the DongleNetwork folder on the BrainVision CD Double click on HASPUserSetup exe there and follow the instructions that the installation program outputs You can navigate to the folder by selecting Browse the CD from the CD s welcome screen 2 Connect the USB dongle Installation has been successful if the LED on the USB dongle lights up 3 Install the HASP license manager The HASP license manager is in the DongleNetwork folder on the BrainVision CD Double click on Imsetup exe there and follow the instructions that the installation program outputs Select Service as the installation type and activate automatic port enabling for the firewall 4 Firewall port enabling If you are using a firewall other than the one supplie
143. n read into the current workspace you are asked whether you want to add the raw data to the current workspace Note that the program can only detect removable media on the local computer and not in a network 196 Annex B Electrode coordinate system The electrode coordinate system that is used in the Vision Analyzer is explained below This coordinate system is used wherever electrode positions are needed e g in mapping and for positioning electrodes in the head view You can change electrode positions with the aid of the Edit Channels transform The axis system has been defined in such a way that the z axis runs through the vertex The x axis points to the right and the y axis to the front Spherical coordinates are used to specify a point on the head A set of coordinates consists of a triple r O and Radius Theta and Phi The radius r specifies how far the point is away from the center of the coordinate system It is stated in millimeters The only exceptions are r 0 and r 1 In our coordinate system r 0 signifies an invalid position for instance when the position of an electrode is not known and r 1 means that a standard diameter has been chosen for the radius This can be used when the surface of the head is approximated by the surface of a sphere signifies the angle between the x axis and the projection of the line connecting the point and coordinate origin on the xy plane is gt 0 for the front right a
144. nal and noise are uncorrelated Consequently the average power of the signal is equal to the difference between the average total power and the average noise power The SNR is then calculated from the quotient of the average signal power divided by average noise power The result of SNR calculation is stored in the resultant history node as a description You can see the result and other information by moving the mouse pointer over the corresponding icon or over the view and pressing the right mouse button A context menu appears which contains the Operation Infos item among other things Choose this item A window opens showing information on the transform that has been performed Segment Range Odd Even Available Segments 147 I Enable odd even averaging Use full range Average only odd segments Specify range of segments Erom Tic m Statistical Data gosocsecocceoccccscsesocseseseccesesesococsoscesoscceccesosoesosesocsccocecossecscsesecesseseg C Average only even segments m Individual Channel Mode Enable individual channel mode Cancel J Calculate signal to noise ratio SNAR Fig 10 5 Average dialog The Average module displays a dialog box at the start You can make the following settings Segment Range Here you can specify the time range whose segments are to be included in averaging The number of segments available is displayed You have the following choice
145. ncorporation of the individual sections in the template depends on their correspondence with the average template Consequently you can for example use the Start at Interval Number option to specify whether a specified number of episodes at the start of the measurement are to be excluded from the calculation of the average This may be of use for example if your MR system inserts so called dummy volumes at the start of a measurement in order to stabilize the system These dummies often have a slightly or even very different temporal structure to the following MR volume measurements and their incorporation in the average artifact would therefore make the template less representative However to obtain a relatively stable template against which you can test the correlation of the subsequent episodes you can use the Include always Following number of Intervals from Start Interval option to define a number of volumes which must always be included in the template In the example above one volume is therefore ignored at the start and the next five must always be included All subsequent MR episodes must then have a correlation of at least 0 975 with the template formed in this way before they can be incorporated in the template The last two options of this method allow you to define whether there is to be an upper limit for the maximum number of MR episodes that are to be included in the template Set Upper Limit for Number of Intervals option
146. nd back left quadrants of the sphere and lt 0 for the front left and back right O signifies the angle between the z axis and the line connecting the point and coordinate origin In the left hemisphere is lt 0 and in the right hemisphere gt 0 Fig B 1 Coordinate system Vision Analyzer User Manual 197 Annex C Markers time markers In the Analyzer markers always indicate a time or time span in an EEG A marker can for instance be an item of stimulus information that is used to ascertain evoked potential but it can also mark a new segment or indicate that a DC correction was carried out at a certain time Markers are an aid to orientation in segmentation and other transforms All markers are characterized by five properties in the Analyzer Type This indicates the class of marker e g Stimulus New Segment etc There are various predefined types which perform special functions These are described further below The color of a marker when it appears in a view depends on its type Since the types are just normal texts you can create new types yourself for example with a macro or by setting markers manually as described in the Views chapter Description This is the description that is assigned to a marker It can be regarded as a subclass When you select markers e g in the course of segmentation you can normally do so on the basis of type and description The description too is stored as text and therefo
147. nd input box you can have a fixed number of eigenvalues calculated and thus a fixed number of factors Finally with the fourth option button and input box you can have so many eigenvalues calculated that the total of the variances of the calculated factors will just exceed a specified percentage of the total variance When all parameters have been input correctly start PCA calculation by pressing the Finish button First all selected nodes are included in covariance matrix calculation and then the eigenvalues are calculated in descending order The calculation process stops as soon as the defined criterion is reached The individual variances and their relationship to the total variance are written to the operation info for subsequent consideration and can be retrieved from there As the result a new secondary history file is generated in which the loadings are stored as nodes If you chose time points as variables the loadings are EEG curves and a new channel is created for every factor If you chose channels as variables a time point corresponds to every factor and the channel names are kept The form of the Loadings node however is only important for further processing by other modules If you open the Loadings node to look at it a special PCA view is used which automatically displays the data in a suitable form The data is output as a graph when time points were chosen as the variables and as a map when channels were chosen In add
148. nd length of the scanned intervals relative to the reference marker You can make these entries based on time or data points The values are used both for reference marker detection and artifact correction 104 MRI Artifact Correction Interval Range xi Stat and End of the Scanned Intervals Relative to the Postion of the Scan Start Start ms fi 0 End ms 140 Duration ms fiso C Based on Data Points c tart point s50 End nor 539 Points 750 Figure 10 33 Setting interval ranges If you cleared the Write Only Markers option for scanner artifact correction the following dialog boxes appear Vision Analyzer User Manual 105 MRI Artifact Correction Pre Correction Settings 7 x IV Enable Baseline Conection for Average Begin ms 100 End ms 50 T Detect Saturation hon Limit pV E Ti V Common Use of All Channels for Saturation Bad Intervals and Correlation C Use Al Scanned Intervals for Average C Select Scanned Intervals for Average by Following Criteria C Use Siding Ayerage Calculation Number of intervalet sidna 2vetTaoe Use Template Drift Compensation ji Number of Averaging Templates 3 lt Back Cancel Figure 10 34 Settings for methods to be performed before correction The following options are available to you e Enable Baseline Correction for Average If the base level in the various scanned intervals varies considerably or if the EEG has a high base level bas
149. ndow opens a context menu on which you can choose the marker types that you want to display The following sections describe the special properties of standard grid head and mapping views 24 7 2 Standard view The standard view corresponds to the EEG on paper The curves are shown one under another The standard view is the one used normally for spontaneous EEG analyses You can display a channel on its own by double clicking its channel name Another double click on the channel name takes you back to the original view If you want to show a selection of channels mark all the ones you want in the required order with a single mouse click Double clicking the last channel name selected then outputs the selection Another double click on one of the channel names takes you back to the original view During the selection process you can deselect a channel by clicking it again Bear in mind that a little time must pass before you can click the name again approx 0 5 1 second because the system would otherwise interpret the two clicks as a double click Press the following button on the tool bar to turn the scaling bars on the left hand side on or off el Set Display Features A dialog appears in which you can turn the scaling bars on or off Vision Analyzer User Manual 25 7 3 Grid view In this view the channels are arranged in a grid There is a standard grid for the default montage As far as other montages are concerne
150. ng average for the intervals If the intervals have been detected using Template Drift Detection then you can use the Template Drift Compensation method to distribute the intervals to multiple averaged artifacts depending on the drift After artifact correction by subtraction of the averaged artifact curve a high sampling rate may no longer be necessary For this reason the corrected data record may be downsampled reduction of the sampling rate at this point Downsampling is performed by using a Hanning window to calculate a weighted average for the data points Downsampling is recommended when the integrated filter mechanisms are to be used in addition to artifact correction Performance is considerably higher on a data record with a lower sampling rate After subtraction of the artifact curve there may still be residual problems in the scanned interval For this reason low pass and band rejection filters may be used that apply only to the range of the scanned intervals and are used after correction by means of the averaged artifact curve These are FIR filters The low pass filter is defined by a Hanning window in the range from 0 to the filter frequency Above the filter frequency the filter function has a constant value of 0 In the case of the band rejection filters it is possible to enter a number of frequencies and specify a frequency bandwidth In this way a number of frequency bands can be filtered out simultaneously The method for co
151. ngth are generated relative to a reference marker e g a stimulus This results in a data set of appended segments or epochs Owing to the extensive facilities provided by segmentation in the Analyzer it is also possible to calculate averages according to complex stimulus conditions e g behavior dependent conditions e In preparation for separate processing steps in different sections of an EEG for example to analyze different stages before and after medication In this case sections are chosen either manually or on the basis of a fixed time schedule and are converted into new data sets in the history file which can then be analyzed separately No matter whether a data set is viewed before or after segmentation you can input your preferences regarding the initial settings of montages and views when new data windows are opened Montages and views are explained in the following chapters You will find more information on segmentation in the Segmentation section of the Transforms chapter Vision Analyzer User Manual 15 6 Montages Montages enable channels to be reconnected on a software basis i e new voltage references are assigned to the channels They also serve to optimize the display of data e g by combining frontal electrodes in one montage and occipital electrodes in another one In this case when a montage is selected only those channels which have been assigned to it are displayed The sequence of channels can also be c
152. nly important to bear in mind here that ICA is a purely statistical i e mathematical process The process is based on the aforementioned assumptions and does not use any additional physiological information whatsoever It is the responsibility of the user of the module to check whether the assumptions are met and whether it is possible to do without physiological boundary conditions The meaningfulness of the results generated using ICA depends on this check The result of ICA is a set of components that are defined in the time domain in the same way as EEG channels Furthermore ICA produces a transform matrix by means of which the components can be calculated from the channels The weight matrix is determined using the Infomax algorithm an iterative gradient process that is described in MBJGS97 for example Parameter settings In the first step of the parameter dialog you can save the matrix used for calculating the ICA components from the original channels The matrix is saved as an ASCII file The same thing is possible for the inverse ICA matrix These matrices can then be manipulated and evaluated using macros and external programs They can also be used for processing in the Linear Derivation module The matrices can be written into the export folder of the Analyzer for evaluation into the raw data folder for importing into Linear Derivation or into any other folder In the first step of the dialog it is also possible to select wh
153. ntains the following elements Gradient Criterion e Check Gradient Low activity MV Check low activity in intervals Lowest allowed activity Max Min fi D uy Interval length fi 00 ms lt Zur ck Abbrechen If you select this check box the gradient criterion is applied e Maximum Allowed Voltage Step Sampling Point You specify the maximum allowed voltage difference between two data points here Max Min Criterion e Check Maximum Difference of Values in the Segment If you select this check box the Max Min criterion is applied e Maximum Allowed Absolute Difference Specify the maximum allowed voltage difference here Vision Analyzer User Manual 53 Amplitude Criterion e Check Maximum and Minimum Amplitude If you select this check box the amplitude criterion is applied e Minimum Allowed Amplitude Specify the minimum allowed voltage level here e Maximum Allowed Amplitude Specify the maximum allowed voltage level here Low Activity Criterion e Check Low Activity in Intervals If you select this check box the Low Activity criterion is applied e Lowest Allowed Activity Specify the minimum allowed activity here e Interval Length Specify the interval length within which activity is not allowed to fall below the minimum The Test Criteria button enables you to check the test criteria When you have completed the dialog you still have the opportunity to change the results The same v
154. o be read in has already been averaged This is particularly relevant for the enabling and disabling of transforms on the Transformations menu Possible values YES Yes the data set represents data that has been averaged NO No the data set represents data that has not been averaged NO AveragedSegments Number of segments included in the average This value is only evaluated when Averaged YES is set SegmentDataPoints If the data is segmented evenly then the number of data points per segment can be specified here SegmentationType Like Averaged this variable is relevant for the enabling and disabling of transforms on the Analyzer s Transformations menu Possible values NOTSEGMENTED The data set has not been segmented MARKERBASED The data set has been segmented on the basis of one or more marker positions All segments have the same length FIXTIME Segmentation was based on fixed times All segments have the same length NOTSEGMENTE D DataPoints Number of data points in the EEG file Vision Analyzer User Manual 183 If no predefined value has been specified the data is read to the end of the file As far as binary data is concerned the TrailerSize parameter can be set in the Binary Infos section as an alternative This section lists the individual channels and their properties Layers Number of layers in a
155. og The template against which the data in the ECG channel is tested can either be determined by setting markers in the raw data or can be selected by the module using an algorithm which is described below If the markers TSTART TEND and TPEAK for the template are set in the raw data section then a blue marking with the corresponding template dimensions is displayed in the data for the ECG channel Here the TREAK marker determines the position of the presumed R peak The blue marking can be modified and moved If you move the marking the peak marker retains its relative position within the blue template range If no TPEAK marker is set then the pulse marker is displayed in the middle of the template If no corresponding markers are set in the raw data then the module searches through the first 20 seconds of a sliding standard deviation window for ranges in which the standard deviation of the ECG rises suddenly since this indicates the start of a heartbeat All the detected episodes are thus averaged and the section which correlates best with the average value for all the sections is selected as the template e Peak Finding Mode In the coherence method the markers are set at the point which exhibits the greatest coherence with the template This means that the markers are not necessarily located at a dominant structure such as the R peak This is also not necessary for correction purposes However the clarity of the data can sometimes be increa
156. og e Goto Next Artifact gt gt e lt lt Goto Previous Artifact e Individual Channel Mode If you select this check box individual marked channels are declared as bad also after completion of inspection In this case the Average module can search for as many segments as possible for every channel on a separate basis also see Average Otherwise the entire interval in which one or more channels were marked is declared as bad when inspection finishes e You will see a list of all marked artifacts at the bottom of the dialog Double clicking an entry takes the program directly to the artifact in question Click the OK button when you have completed inspecting the EEG 10 1 26 2 _ Semiautomatic inspection The Criteria dialog is opened when you select semiautomatic inspection Here you can define four marking criteria There is a separate tab for each criterion Vision Analyzer User Manual 127 Raw Data Inspector Criteria x Gradient Max Min Amplitude Low Activity M Check gradient Maximal allowed voltage step sampling point 50 uv Mark as bad before event after event 500 ms 500 ms Abbrechen Fig 10 43 RDI settings dialog for the Gradient criterion You can make the following settings Gradient tab e Check Gradient If you select this check box the gradient criterion is applied e Maximum Allowed Voltage Step Sampling Point You specify the maximum allowed voltage difference betwe
157. oherence calculation options Most of the methods described in the literature can be implemented in conjunction with the formula interpreter More detailed information on this is available in a separate document The difference between using this and using the Coherence module is that this module always supplies data of the same type as the input being processed It is thus possible for example to calculate and further process coherence for complex valued coherence or to calculate the correlation of channels in the time domain CI x Calculate Covariance of All Combinations of Channels Select Combination of Channels for Covariance Calculation Insert Line Remove Line Remove All Nz P9 Pz Fz PS P7 T8 T7 F8 F A2 Al 02 01 JV Subtract Average OK Cancel Fig 10 13 Parameter input The two methods of covariance calculation are described by the following formulas Cov c1 c2 x 1 N Z c1 x avg ci x c2 i x avg c2 x 0 and Cov c1 C2 x 1 N 1 i x Co i x Vision Analyzer User Manual 67 You can use the Subtract Average button to specify whether the first or second method is to be used to carry out the calculation In both formulas totaling is carried out via the segment number i and the average value obtained applies to the segments at a fixed time or at a frequency of x and a fix
158. olors to Indicate Different Transformations Here you define whether you want to give different colors to different transforms e Adda Color Frame Around the Views If you enable this option views appear in a frame which has the defined color The Width of Color Frame subitem defines the width of the frame in pixels e Press Color Button to Change a Color You can assign the actual colors in this table Vision Analyzer User Manual 13 Preferences x Views Scaling Clipboard Transformation Colors V Add a color frame around the views Width of color frame Pixels 3 Press color button to change a color Transformation Coor __ lt Default Colors Jooo sRaw Data Artifact Rejection MM Average l Baseline correction Basic Macro MO Coherence E Comparison DC Detrend local Ce eooo o Fes New Retence fm ocular conection fmm Peak Detection Mm Raw Data Inspector Abbrechen Fig 4 5 Assigning colors to transforms The following chapters deal with the various options offered by the Analyzer in more detail 14 5 Segmentation At this juncture we want to outline what segmentation is because the following chapters refer to it repeatedly Segmentation means the division of an EEG into sections Segmentation can be based on different criteria We use segmentation in the following cases e As a preliminary stage in the analysis of evoked potentials In this process epochs of the same le
159. on Keyname Meaning Default value Mk lt x gt Here x stands for the marker number i e the keyname for the first marker is Mk1 for the second marker Mk2 etc Individual properties for a marker are specified with commas between them lt Type gt lt Description gt lt Position gt lt Points gt lt Channel Number gt lt Date gt Example Mk1 Time 0 26 1 0 Here the first marker has the type Time 0 no description the position is at data point 26 the length is 1 data point and the channel number is 0 which means that this marker relates to all channels 186 The date is optional It is only evaluated if the marker type is New Segment The date has the following format 4 digits Year 2 digits Month 2 digits Day 2 digits Hour 24 hour system 2 digits Minute 2 digits Second 6 digits Microsecond Consequently time is broken down to the microsecond level The following specification 19990311140312003012 means 11 March 1999 14 03 12 003012 Vision Analyzer User Manual 187 12 2 Importing markers and channel positions To read in additional data along with the raw data e g output files from your stimulator you can write a Basic macro that generates a new history node This node inherits the data from the raw EEG However you can delete and regenerate markers You can also change the names and position details of channels W
160. on Colors Unsegmented Data j C Grid view grid can be arranged in montage settings C Head view topographic channel position Default montage lt Standard Montage gt r Segmented Data Default view C Standard view EEG paper like Grid view grid can be arranged in montage settings C Head view topographic channel position Default montage lt Standard Montage gt x m Default width of Transient Views Relative to window width 20 Abbrechen Fig 7 1 Selecting the Default View You can also open a new view for the currently displayed data set To do this choose one of the following menu options Window gt New Window gt Standard View Window gt New Vision Analyzer User Manual 21 Window gt Grid View Window gt New Window gt Head View Window gt New Window gt Mapping View or Window gt New Window gt 3D Mapping View A data set can thus be displayed simultaneously in several windows On the Scaling tab under Configuration gt Preferences you can set the parameters for the views For the time domain you can choose Polarity Start with Display Baseline Correction on and Default Scaling Before After Averaging In the frequency domain you choose Default Scaling Before After Averaging here Preferences Fig 7 2 Input dialog for polarity and default scaling Under Set Individual Scaling Factors you can enter individual channels which a
161. on has attached itself to the Filters icon Further operations make the history file grow more There can also be multiple branches from a data set Let s assume you want to perform other analyses on your raw data which do not require a filter In this case you select the raw data as the current window by clicking on the open raw data window Alternately you can double click the Raw Data icon in the History Explorer Now choose Transformations gt RMS Global Field Power again for example Choose a few channels again and press OK A new RMS icon appears beneath Filters The history list has branched giving rise to a history tree Analyses can be created as branches at any point in this way Vision Analyzer User Manual 11 Fl fy p300a Faw Data Filters Pi Fms m P300b Fig 4 3 Branched history tree If you now want to transfer the same operations to another history file open the required history file For our example note that this file should contain the same channel numbers as the first file Now you could call the same transforms from the Transformations menu and answer the questions in the dialogs again but there is an easier way Move the mouse over the Filter icon for the first history file press the left mouse button and hold it down and drag the icon over the Raw Data icon for the second history file Now release the mouse button The Analyzer will automatically build a history tree It is not only possible
162. ong other things the program s features include Vision Analyzer User Manual EEGs with an infinite number of channels can be processed The maximum EEG length that can be processed exceeds 2 billion data points regardless of the number of channels EEG formats from various major makers are recognized The number of readable formats is constantly growing History trees record every single operation on EEG data Templates can be created from history trees which in turn can produce new history trees automatically The Vision Analyzer can be controlled remotely by other programs as OLE Automation has been implemented in it A built in Basic interpreter allows users to create both simple command files for analysis automation and sophisticated applications The individual parts of the program have a modular structure There are reader transform montage export and view components The program s functionality can be extended by adding new components Brain Products is constantly working on new components All interfaces are disclosed so skilled users can develop their own components or have them made to order 3 Installation It is essential to install the program with setup exe because the files on the disk are compressed and have to be unpacked in a specific way System requirements e Windows 98 Windows NT 4 0 Windows 2000 or Windows XP e Minimum configuration Intel 400 MHz Pentium II or compatible processor 64 MB RAM
163. orkspace Configuration window Help l x Mite amel 2 Root amp P300e P3008 P3009 E A OcularCorrection A BaselineCorrection A Average a segl1 OcularCorrection A BaselineCorrection A Average iS A segl2 A OcularCorrection A BaselineCorrection A Average a p300s fA New Reference m P300b A EditChannels E so SI A Filters i S A segl0 a woa Ready Z Fig 8 1 Example of a history template You can edit the history structure in the template in the same way as in a normal history file i e you can rename and delete nodes You can also drag individual history nodes from the open template back onto history files and thus trigger the corresponding operations To apply the history template to an entire set of history files save the current template to a file with File gt Save Then choose History Templates gt Apply to History Files The following dialog appears Vision Analyzer User Manual 43 Apply History Template to History File s EI m Select History Template r Select History File s lt Not Selected gt IV Primary History Files Only Use Whole Workspace Select Individual History Files Starting Position In History File ae Selection Filter f Refresh Root
164. ost here however and the data that results from this process does not contain any artifacts On account of the interpolations however it cannot be expected that using this process on data with any segment length and sampling frequency will always provide comparable results since the FFTs calculated in this way differ in terms of the total number of data points and thus also in terms of the information content of the original input data which inevitably leads to differences in the resulting spectra Interpolated FFT spectra are thus suitable above all for visual data inspection To avoid the interpolation of the FFT data at all resolutions you should thus ensure that the number of data points in the segments is a power of two and select the Maximum Resolution option in the module If you select a different resolution in the FFT module to the maximum spectral resolution determined by the segment length and data rate the resulting FFT spectra are also interpolated or even interpolated and integrated depending on the selected resolution As mentioned above interpolation like this always results in correct FFT spectra and does not create any Vision Analyzer User Manual 81 artifacts but it is important to remember that an FFT analysis like this cannot necessarily be compared with an FFT analysis using a data set that has the resolution selected here as the maximum spectral resolution To illustrate this let s take an example of an EEG
165. ou set it in the Peak Detection transform e Name of the Involved Data Sets The names of the involved data sets are separated by commas e Selection of history files Refer to Area Information Export above 178 e Overwrite Default Decimal Symbol You can keep your computer s decimal symbol point or comma or choose one to meet your specific purposes Your computer s decimal symbol depends on Regional Settings e Export Mean Value Around Peak You have the option of averaging the vicinity of the peak and exporting the mean Here you input a number which specifies how many points before and after the peak are to be included in averaging The number 2 means that 5 points would be included in averaging 2 before 2 after and the peak position itself e Export Individual Latencies for Each Channel If you used individual latencies for each channel when peaks were ascertained you have the option of exporting these too Otherwise the latency of the first marker that is found in a data set is exported e Output file The name of the output file is specified here Vision Analyzer User Manual 179 12 Importing data positions and markers 12 1 Importing data As far as data import is concerned the Analyzer not only provides standard readers for many commercial file formats but also allows you to import your own formats For simple ASCII formats you may be able to fall back on the Besa format You use the Generic Data Reader for more co
166. ous band rejection filters to different channels by using this module several times Step 2 of 2 Channels Disabled Channels Enabled Channels lt Back Finish Cancel Fig 10 8 Second step in the Band Rejection Filter wizard Vision Analyzer User Manual 61 10 1 5 Baseline Correction With the Baseline Correction module the baseline of every segment is adjusted It is used after Filtering optional Segmentation Ocular Correction optional Artifact Rejection Local DC Detrend optional Correction is generally applied before averaging The interval in a segment is defined whose average voltage level corresponds to the new zero point of the segment values In other words the average of the points of the previously defined interval is ascertained and this is subtracted from all points in the segment This operation is applied to all channels in the data set You can choose the interval to be used in forming the voltage average It is generally placed in the area of lowest activity i e best of all before stimulus or other reference markers To do this you enter the values in milliseconds for the beginning and end of the interval in the dialog box Baseline Correction Ea Range for mean value calculation Begin ms End ms fo Cancel Fig 10 9 Baseline Correction dialog Failure to perform Baseline Correction can lead to a flattening off of the average signal 62 10 1 6 Change
167. owever you can input the correct coordinates with the aid of the Edit Channels transform component You can find information on the coordinate system in Annex B By pressing the following button on the tool bar you can set the views and other parameters for the maps 2 Set Display Features Settings xl Maps Scaling Number of Maps fi C Automatic Scaling Interval between maps ms 144 M Symmetric Scaling J Fix Number of Maps Manual Scaling Maximum py fi 0 Triangulation and Linear Interpolation Minimum p A 0 V Use Average Value of Interval Order of Splines 4 View from Maximal Degree of Legendre Polynomials fio V Top MV Default Lambda 12 5 IV Front Other Lambda fie 005 IT Back T Bight Display T Left Standard Colors Il Quick Graphics J C Grayscaling M Show Electrodes l AllinOneMap May Anale degrees 190 Discrete Colors Set Details Fig 7 11 Setting options for the mapping view You have the following setting options in this dialog e Number of Maps to be shown simultaneously e Interval Between Maps in ms for time data and in hertz for frequency data 32 Fix Number of Maps There are two ways of defining the number of maps directly or indirectly via the interval between two maps If you set a size the other is calculated automatically The Fix Number of Maps setting applies when the width of the overall interval is changed manually in a
168. owing data types voltage in uV voltage density in uV Hz power in uV and power density in 1V7 Hz Density In the case of the voltage density and power density functions the spectral line values are scaled as if they were calculated with spectral line spacing of 1Hz which Vision Analyzer User Manual 79 80 permits comparisons between FFT analyses that have been carried out with different spectral resolutions To this end the spectral line values are multiplied by a constant factor b which corresponds to the reciprocal value of the spectral line spacing b 1 f Normalize Another kind of comparability of FFT data can be achieved by means of the Normalize Segments check box It is often important in EEG research to draw comparisons by changing the spectral composition of the EEG signal but since the total power of the EEG varies from segment to segment such comparisons are not normally carried out By means of the Normalize function however it is possible to normalize the total area in the entire EEG spectrum or in only a part of it and thus make it comparable If you do not select Normalize mode the FFT module returns data in the frequency range in the way it is generated by a Fourier transform Selecting Normalize causes the data to be multiplied after the Fourier transform by a factor which makes the area under every channel and in every segment identical In this case it is not the absolute data that is output but the relative
169. owpass Filter Cutoff Frequency Hz 70 Il Enable Bandrejection Filter lt Back Cancel Figure 10 36 Settings for methods to be performed after correction e Downsampling Factor Downsampling reduces the EEG sampling rate by this factor A factor of 1 means there is no downsampling e Enable Lowpass Filter A low pass filter with the properties described above is applied to the scanned intervals following correction e Enable Bandrejection Filter A band rejection filter with the properties described above is applied to the scanned intervals following correction The bandwidth and frequencies to be rejected can be specified If you selected the correction of blood pulse artifacts with peak detection in the initial dialog box the following dialog box appears Vision Analyzer User Manual 109 MRI Artifact Correction R Peak Detection Method C Use Marker A Peak Marker 7 v1 Use Peak Detection ECG EEG Channet Ekg2 7 l Derive ECG Pulse Rate ms 1000 250 Pulse Rate bpm Mijn fas Max feo IV Use Defaut R Peak Marker Name R R Peak Maker Name fi C Use Direct R Peak Detection Method Use Coherence Method Coherence Trigger Levet fos l Peak Finding Mode Amplitude Trigger Level Min 10 6 Max f 2 M Semiautomatic Mode Write Only Markers lt Back Cancel Figure 10 37 Settings for the R peak detection method The following options are available to you
170. pector make sure that you also enable individual channel mode here You can find more details on this in the Raw Data Inspector section e As far as segmentation is concerned you must not suppress bad intervals also see the Segmentation section e Ifyou use the Artifact Rejection module in place of or in addition to the Raw Data Inspector then you must also use individual channel mode here in order not to reject entire segments but to mark channels only also see the Artifact Rejection section You can output the standard deviation as an additional data set The signal to noise ratio SNR of the data to be averaged can also be calculated The SNR provides a measure of the quality of the EEG signal Since neither the signal nor the noise in the EEG is known exactly your average total powers must be estimated with statistical methods In this process the average noise power of the EEG is calculated for each channel first It is assumed that noise will be eliminated by averaging Thus average noise power is calculated from the total of the squares of the differences between the EEG value and the average value divided by the number of points minus 1 Vision Analyzer User Manual 55 In order to ascertain the average power of the signal in the EEG you first calculate the total power of a channel of the EEG This is a result of the mean of the squares for all data points of the channel before averaging It can be assumed that the sig
171. r it is replaced by the current user name In this way you can create separate profiles for different users of a computer If you do not specify a path name the profile file is stored in the main Vision folder If you store the profile files centrally on a network you can work with your personal profile on various computers in the network The new profile file is used when the program is next launched If it does not exist it is created and then contains the settings that were last used on the current computer Alternately you can also pass the name of the profile file as a command line parameter as explained in Annex G 202 Annex G Command line parameters The Analyzer supports various command line parameters Choose the MS DOS Prompt to enter them When the prompt appears go to the Vision folder with the command cd lt Folder gt Example C gt cd c Vision Now you can call the Analyzer with additional parameters Example C Vision gt Analyzer new pPrinter Alternately you can copy the link to the Analyzer that was placed on the desktop during installation Then right click on the icon of the copy After that select Properties from the context menu On the Shortcut tab append the parameter to the existing text in the Target field Example C Vision Analyzer exe Change to C Vision Analyzer exe new pPrinter In this way you can place various links for various parameters on the desktop Now we come to the actual pa
172. r the Solution is also contained in the file Solutions can be installed and uninstalled subsequently since they are only searched for and loaded when the program starts Solutions are shown in a menu with their file name In contrast to macros their source text cannot be manipulated This means you can be certain that a specific unadulterated Solution will be used The Brain Vision CD contains a collection of Solutions in the Solutions subfolder The InstallSolutions exe program in the same folder installs the Solutions You will find Solutions for use in various areas such as ECG EMG processing marker import ASCII export data manipulation etc Further solutions will be made available on the Brain Products website in the foreseeable future Select Base Folder for Solutions xj C Vision S olutions Browse Cancel Fig 16 1 Base folder for the Solutions You set the base folder for the Solutions under Configurations gt Select Base Folder for Solutions If you are working in a network with several colleagues we recommend using a shared base folder for the Solutions 194 The Solutions are displayed in the Solutions submenu of the Analyzer menu The submenu reflects the organization of the Solutions base folder and its subfolders This means each subfolder is displayed as a submenu Each menu item corresponds to one solution The Solutions are automatically installed in thematically arranged
173. rameters The m lt Macro gt parameter calls the specified macro when the program is launched Example C Vision Analyzer exe mCompress All calls the macro named Compress All when the program is launched The quotation marks are only needed if the macro name contains a space character Note that the macro name must follow the m directly without any space character The new parameter forces a new program instance of the Analyzer When the Analyzer is called more than once the existing program instance is normally used This behavior is suppressed with this parameter Example C Vision Analyzer exe new The p lt Profile File gt parameter uses the described user profile file during the session Example C Vision Analyzer exe pPrinter You could for example store all settings for optimum printing in the profile file named Printer Different parameter types can also be combined Note that p is always executed before m Vision Analyzer User Manual 203 Annex H Links to raw data Links can be created in Windows for example by moving a file from one folder to another one with the mouse in Windows Explorer while holding down the CTRL and Shift keys This generates a small file that points to the actual file The Analyzer can operate with links to raw EEG files These are treated in exactly the same way as if the original file was in the folder This enables you to set up a workspace containing raw EEGs from various fold
174. re can also be anything that is required The Analyzer or its reader components construct some texts such as for EEG formats which store stimuli as numeric values If an EEG contains a stimulus with the value 1 for instance the reader component will change this value to the text S 1 The description is generally shown when markers are displayed Position The position defines the data point at which the marker occurs in the EEG Points These are data points along which a marker extends Mostly markers have a length of one point Among the predefined markers there is only one type which extends over more than one point Bad Interval It is set by the Raw Data Inspector or Artifact Rejection ChannelNumber A marker can be assigned to one or all channels channel number 0 The New Segment marker also has the DateTime property i e the date and time of its occurrence is stored in every marker of this type if this information can be extracted from the raw EEG 198 Type Function Color in view Bad Interval This indicates a bad interval owing to artifacts Pink or gray Comment This is used for a comment Black DC Correction A DC correction occurs with EEGs which were Yellow acquired with a DC acquisition system Mostly there is a jump in the voltage level of the data at this time New Segment This marks discontinuities in the EEG including Green interruptions of acquisition but also se
175. re to be shown on an attenuated basis For instance this is desirable for ECG channels because they would otherwise extend considerably into the signal form of EEG channels In the table you input the channel names and the associated scaling factors by which you want the signals to be attenuated This attenuation only has an impact on the display and does not affect the data itself 22 When you open a view you can manipulate the output of the EEG with some elements from the tool bar ojeli 5 S alaj Fig 7 3 Tool bar SEME Alenfe Is in 2 m f Sme 2 Here are the elements that are relevant to views e Overlay different data sets Increase the time shown Reduce the time shown Set a time shown that can be selected individually Reset the interval shown to the default value Configuration gt Preferences lel amp 2 Show full segment This button causes precisely one segment to be shown It is only accessible when the segments are small enough Increase scaling sensitivity Reduce scaling sensitivity Reduce the number of channels shown Increase the number of channels shown am LH ih le Go to next group of channels This button is only accessible when a reduced number of channels is being shown gt Go to previous group of channels This button is only accessible when a reduced number of channels is being shown Turn baseline correction on off Only the baseline of the display
176. rgin specifications are ignored Vision Analyzer User Manual 189 Clicking the Header Footer button takes you to another dialog in which you can configure headers and footers Placeholders for Header Footer T ext ct comment d Current date h History file name n Name of data set nd Data set date at print position nt Data set time at print positon p Full history path t Current time u Current user m Header Lines Number of lines 0 5 fi Enter text and select the font for each line Alignment _ h n Printing date d time t N Footer Lines Number of lines 0 5 fi Enter text and select the font for each line Alignment User u center Font Cancel Fig 13 2 Dialog for defining headers and footers Here you can define up to five header and footer lines to appear on the printout The dialog consists of two sections Header Lines and Footer Lines You input the following e Number of Lines Number of header lines and footer lines For every header and footer line e Text Here you input the text that you want to appear in the lines You can also use placeholders which are replaced during printing or the print preview by the current data One placeholder is n for example If this placeholder is used anywhere in the text the program replaces it in the course of printing by the name of the data set in question A list of all available pl
177. rker file Resulting header file name PC format carriage return line feed P300b_Averagel vhdr C UNIX format line feed Resulting marker file name Mac format carriage return P300b_Averagel vmrk Cancel Fig 11 2 First page of the Generic Data Export dialog The first page of the dialog gives you the following setting options e Export Selected Block This option is only accessible if a block has been marked e Write Header File You can also generate an ASCII header file containing information on channels sampling rate data set type etc This header file has the INI file format which is common under Windows e Write Marker File A marker file also in INI format can be generated as an option It lists all existing markers together with their positions types descriptions etc e Data File Format Here you define whether the data is to be exported in ASCII or binary format For an ASCII export the data is written directly in uV or uV e Data Orientation You also have a choice of data orientation vectorized or multiplexed These terms relate to channels 172 Here vectorized means that all data points of the first channel are written first to the export file followed by those of the second channel etc As far as ASCII export is concerned every channel to be exported is written to a separate line Multiplexed means that the data of the first sampling time is written first fol
178. ropetties Fig 4 4 Context menu for a history node Alternately you can move the mouse over an open data window and right click there 12 You close a history file by clicking the sign next to the book icon This automatically closes all data set windows associated with the file Now is the time to introduce two new terms primary and secondary history files You have already encountered primary history files They populate the upper pane of the History Explorer What characterizes primary history files is that they represent a specific raw EEG and its processing steps Secondary history files generally represent operations which are applied to various nodes in various primary history files An example of this is Grand Average These secondary history files do not relate to a specific raw data EEG They are stored in the lower pane of the History Explorer You can delete and rename secondary history files in the way described above for history nodes You cannot delete or rename primary history files in the Analyzer You can use the mouse to shift the divider bar between primary and secondary history files up or down You can also assign colors to the individual transforms These will appear in the corresponding History Explorer icon and in the frame of views You access the color definition menu under Configuration gt Preferences on the Transformation Colors tab You will find the following options there e Use Different C
179. rrecting blood pulse artifacts uses the temporal redundancy of the pulse artifact It is first necessary to obtain a reliable detection of the ECG episodes to be able to determine the start of the artifacts associated with the heartbeat to a very high level of temporal precision As a result it is not absolutely essential to determine the R peak position since the aim is not to measure a specific component of the ECG but the temporal stability of the detection of any given component of the ECG This can be done by means of markers that are generated by macro or manually Two methods for the automatic detection of this type of stable point in the ECG are also implemented in the module The most important characteristics of R peak detection can be found in the method described in AP98 while the coherence method compares the shape and amplitude of the ECG curve over time for its consistency with a template curve and searches for peaks in the areas in which coherence exceeds a certain threshold value To average the blood pulse artifacts in the EEG channels the detected trigger points are transferred from the ECG channel to the EEG channels with a selectable time delay The correction is carried out in each case over an interval around the trigger point with the length of the average R R spacing Corresponding intervals of the preceding range of the EEG are used to obtain the average blood pulse curve Linear trends are calculated before averaging In
180. rs are applied to the raw data they represent particular signal characteristics in the time frequency domain as accurately as possible Here too filtering or folding thus takes place but with a specific filter characteristic for the signal of interest The two mother wavelets used most frequently in the CWT are the Mexican hat and the Morlet wavelet which are shown below Fig 10 58 The graphs of the complex Morlet wavelet left and Mexican hat wavelet right The complex Morlet wavelet also referred to as modulated Gauss function is defined by the formula y t Ae em Factor A is used for normalization in this case Parameter c is a special property of the Morlet wavelet It enables the number of oscillations of the wavelet to be determined and therefore also the filter width in the frequency range The real Morlet wavelet consists of the real part of the complex function The formula for the Mexican hat wavelet is y t Al t jea The Mexican hat wavelet looks rather like a sombrero as you can see in the figure above How do the two wavelets filter the raw data In an initial step the mother wavelet is successively translated via the raw data and for each point in the raw data a scalar product with all the points is calculated in the wavelet As a result the data is filtered with the wavelet or depending on how it is viewed a correlation function of the wavelet is calculated with the
181. rt of the next interval In such situations Template Drift Detection can often greatly improve the positioning of the Scan Start marker The accuracy of Template Drift Detection increases with the number of channels used for detection You can therefore improve the results by selecting more channels for Template Drift Detection than for simple Scan Start detection If the interval is defined by markers the scanned interval is set relative to this marker in accordance with the set interval range The beginning of the interval can also be placed before the marked reference point by using negative values for the start point when setting the interval range In other words the reference point does not have to correspond to the beginning of the interval but it must be at a constant distance to the limits of the interval Averaging is carried out on the basis of a number of different parameters It can be specified whether baseline correction is to be carried out before averaging and what period of the interval relative to the marker i e to the reference point is to be used for this The average value of the data values of this range is obtained and this value is subtracted from each data point of the interval for the purpose of baseline correction Baseline correction should be carried out if the base level in the various scanned intervals varies greatly or if the EEG has a high base level overall If there are considerable fluctuations in the bas
182. s s gt lt Ready p mna Segment F 038 P300 Fig 10 72 2D map With the following button on the right hand vertical tool bar you can open a dialog to change the display parameters of the map el Set Display Features The dialog gives you the same setting facilities as described in the Mapping view section of the Views chapter 168 10 3 5 Zoom Zooming enlarges the current section It always uses the standard view and the currently selected montage The zoom facility is only available in the time domain Analyzer P300d Raw Data Filter Stimulus1 B aselineCorrection1 Artifact Rejection1 Average1 Zoom Fie Edit View Transformations Montage Export History Template Macro Workspace Configuration Window Help laj x pojela S alas NAAT laf 22 mi fF Alms 2 HQ p300a P300b P300 QA P300d B E Raw Data S A Filter eB Stimuus1 B BaselineCorrecti EB Artifact Reje Averag EM Stimuus2 EEA BaselineCorrecti B A Artifact Reje Averag BJ Segmentation E jz Average H Peaks Eg CRS Ea SMe Grand verage s a gt Ready 00 00 00 Segment 1 1 Fa 0 21s P300 Fig 10 73 Zoom Vision Analyzer User Manual 169 11 Export components Export components enable data sets markers area information and the like to be exported to files for further processing in other programs The export components
183. s Use Full Range and Specify Range of Segments e Individual Channel Mode If you do not select this mode all segments which have a Bad Interval mark at any point are rejected completely e Odd Even Here you define whether you want to perform averaging on the basis of odd or even numbers The segments determined by the time range act as the basis for the count 56 e Statistical Data Here you define whether you want to output standard deviation as a new data set and whether you want to calculate the signal to noise ratio Vision Analyzer User Manual 57 10 1 3 Averaged Cross Correlation This transform enables you to calculate the averaged cross correlation between two channels on a segmented EEG It is a statistical measure of the dependency of two channels Time shifted and frequency shifted dependencies are also calculated If this transform is applied to complex data the complex cross correlation is calculated Averaged Cross Correlation Ed Channels Band Selection a C Take whole range Take band Begin Hz fi 0 End Hz 30 Ej Insert line Remove line Remove all mal 4 T A 4 O N mm A WwW N JULULULU sell blll J K Cancel Fig 10 6 Averaged Cross Correlation dialog The calculation procedure is as follows e First the cross correlation between the specified pairs of channels is calculated segment by segment using the following formula CrCorr c1 c2 j Xi
184. s 1 uV Resolution is the value by which the value of the data point is multiplied to convert it to pV Binary Infos This section is only relevant if DataFormat in the Common Infos section was set to BINARY Keyname Meaning Default value BinaryFormat Possible values INT_16 IEEE_FLOAT_32 IEEE floating point format single precision 4 bytes per value INT_16 16 bit signed integer UINT_16 16 bit unsigned integer ChannelOffset Channel offset at which the data starts This 0 offset is only relevant to vectorized data ChannelOffset and DataOffset can be used simultaneously DataOffset Size of the offset in the file at which the actual 0 data starts SegmentHeaderSize Ifthe data is segmented evenly the size of the 0 segment header can be input here in bytes TrailerSize Size of the trailer of the EEG file in bytes 0 This parameter can be specified as an alternative to DataPoints in Common Infos in order to stop reading in the data before the end of the EEG file is reached UseBigEndianOrder This only applies to integer formats It NO specifies whether big endian order is used i e whether the most significant byte in a number is stored first Macintosh Sun Possible values YES Yes big endian order is in use NO No little endian order is in use corresponds to the Intel specification Coordinates Coordinates are listed here Keyname Meaning Default va
185. s at time points IV Primary history files only r il Use whole workspace Select individual history files Selection filter Refresh Available files Selected files Additional Features aoa I VARIMAX rotation pae P300d P300e P300f P300g Add All gt gt Fig 10 67 First and second steps in the PCA wizard PCA Step 4 of 4 Output PCA Step 3 of 4 Channels Output file PCA Disabled channels Enabled channels AT a A2 al Eigenvalue Selection Criteria oa CP5 Take eigenvalue 1 as limit CPE Cz C Take eigenvalue limit i F3 F4 C Take a fixed number of eigenvalues F7 F8 C Take percentage of total variance e FCI i FCS lt lt Disable FCE lt Back Cancel Fig 10 68 Third and fourth steps in the PCA wizard In the second step you can select the history files and nodes that are to be included in calculation of the PCA In the upper input box you can enter the names of the nodes separated by commas In the lower area you can select the files that you want to use to calculate the PCA In the same way as calculating a grand average you have the option of restricting the selection to the primary history file and filtering out certain files by specifying wildcards In the third step you can select the channels that you want to include in PCA calculation At this juncture for example you can exclude ocular artifact channels or trigger channels if you want to Vision Analyzer User Manual 1
186. s block to be exported if appropriate We set the folder for export files in the introductory chapter under Workspace gt New or Workspace gt Edit The next two sections list the simple and extended export components that currently belong to the Analyzer in alphabetical order as they appear on the Export menu 170 11 1 Simple export components 11 1 1 Besa This component exports the data set or data set section in BESA ASCII format The exported file is given the extension raw In addition to exporting just the selected block and specifying the file name here you can also define whether you want to export the channel names in addition Besa Export SA Tey E ERDGTL Selected DIGtk M Fig 11 1 Besa Export dialog Vision Analyzer User Manual 171 11 1 2 Generic Data Export Here you can export data in the time and frequency domains including complex data in ASCII or binary format Generic Data Export Settings Page 1 of 3 Export selected block m Export File Name patel bee PE ENA Folder c ision Export Wildcards IV Write marker file h History file name n Name of current data set m Data File Format Data Orientation Base name Extension ASCII Vectorized shin Stt C is SSCSCSCSCS da Binary C Multiplexed Resulting data file name ASCII Line Delimiters P300b_Average1 dat Common line delimiter for ASCII data file 3 header file and ma
187. s een decscaceeancncaiepacecadeasaacadonieeeeinzenzcaies 121 101 254 SOONG cage carte AER Reopens Capua ke OE Mantua ee OOE E OE 125 10 1 26 Raw Data lInspectof 2 32300 sida ech sitet ocksd da tial cada tend ale dan saceeadsdce nadia 126 101 27 SACCUIV Dt orator Gianast anal an sian cada tead ania ahaat i EER 132 10 1 28 RMS Global Field Power csesccosnaccuscsasceesssecdernteentneresasteenioenteanenes ee 133 10 129 SCOMCMIAUON sze nson er n eet denied a E EE aR E 134 10 41 30 TNC THICSE ssa cin cmtinrmminniintin ietietietiaumnteniCEmmnenueiT 142 10 1317 Wavelets ene er E R eet E EN EEEE 145 10 1 32 Wavelets Layer Extraction 0 cccccccccccsssccseeeeeeseeeensseeeeseeeeeseneeees 157 10 2 Secondary transtorMs 22 26026 a scien eed ee ee inane eee 158 10 2 1 Grand AVC AGC eresian E iota ice sated tees 158 10 2 2 Principal Component Analysis PCA ccccccccccccccccccceeceeceeeeeeeseeeeeeeess 160 10 3 Transient IANSIONM Sita sccn eres ares decried ety tenehy tegen iaicay teeter teats 165 TOSS TOD AD i 5 Sie a ease a eae aaa aaa EE 165 10 3 2 Current Source Density CSD sisseeticckccveis vies eeets secession 166 10 3 3 Fast Fo rier Transtorm FIAT excuses cscs tite a a tat ante ee 167 KOREA ENI ET MRS RS RENTER ET TIRE TE TR eT OE TELL Sen Ten POET Oe nara e rere 168 102I LOONT a e na neon nr OPER ET ORION TITER Sea A A TOT TRE SRC A O 169 T1 Export components nss aaa Da aeaiia 170 11 1 Simple export components a
188. s method is used Ocular Correction x Method Gratton amp Coles VEDG HEOG V VEOG Channel Channel Name EOG 7 V HEOG Channel Channel Name HEOG Common Reference Common Reference Reference Channel 7 C Reference Channel z m Blink Detection By Algorithm Based on Markers Start Marker End Marker I Semiautomatic Mode P Write Only Markers r Enable Disable Channels for Correction Disabled Channels Enabled Channels HEOG VEOG Enable gt gt lt lt Disable CB kd Cancel Fig 10 34 Ocular Correction dialog When Ocular Correction is called a dialog appears with the following inputs e Method The method used At present only the Gratton amp Coles algorithm has been implemented as well as a slight modification of that named Gratton amp Coles without raw average subtraction The algorithms and their differences are described in more detail further below VEOG e VEOG Channel Select this check box if there is a VEOG channel in the EEG Vision Analyzer User Manual 117 Channel Name Select the name of the channel here Common Reference Reference Channel If you select Common Reference it is assumed that the reference electrode of the VEOG channel is shorted with the common reference If you select Reference Channel however the program assumes that the reference electrode of the VEOG channel is not identical to th
189. scribed in the Exporting graphics chapter Vision Analyzer User Manual 191 14 Exporting graphics You can export the currently displayed graphic as a vector graphic via the clipboard and process it further in other programs such as MS Word MS PowerPoint and Corel Draw The clipboard is a temporary storage area that is provided by the operating system When you have opened an EEG window choose Edit gt Copy Alternately you can click the following button on the tool bar The graphic is now on the clipboard Open a target application such as MS Word and choose Edit gt Paste The graphic should appear in the current window of the target application The quickest way of performing this operation however is to use the standard Windows keyboard shortcuts Ctrl C in the Analyzer to copy the graphic to the clipboard and Ctrl V in the target application to paste it in Enhanced Metafile EMF is the format that is used in this process Most Windows graphics programs support this format You can change the graphic s default size of 16x12 cm by choosing Configuration gt Preferences and then the Graphics Export Output tab You can also define here whether you want to use the screen or printer resolution The latter is generally higher Note that some programs have difficulties coping with EMF data at printer resolution 192 15 Appending multiple raw data sets You can append multiple raw EEGs to process them as one data set
190. sed by setting the marking at a dominant structure This can be achieved using the Peak Finding Mode which searches for and marks the dominant structure in the template s range If you set a TPEAK marker for the supposed R peak then a search is first performed along the ECG channel for correlation and amplitude values above the defined thresholds Coherence Trigger Level and Amplitude Trigger Level If such a point in time is found then the search continues until the first local maximum of the correlation and amplitude ratios is reached Within the time range defined by this operation a search is performed for the local maximum i e the presumed R peak Vision Analyzer User Manual 111 around the relative point in time of the TPEAK marker in the template and the marking in the ECG channel is set at this point e Semiautomatic Mode In this mode you can interactively modify the pulse markers detected by the module in a dialog e Write Only Markers Markers are written No correction is carried out If you select semiautomatic mode in this menu then the module first searches for ECG episodes and then displays these for interactive processing in the dialog which is now opened R Peak Detection ECG Channel Ekg2 Derive ECG No Pulse Rate bpm 60 15 Marker Name R Coherence Method Amplitude demeaned Coherence Trigger Level 0 70 Amplitude Trigger Level Min 0 60 Max 1 20 Peak Finding Mode Yes V Show
191. sform calculates the DC trend from the signal Two different approaches are taken depending on whether this transform is called before or after segmentation When called before segmentation global trend correction is used as described in Hennighausen et al 1993 A correction method for DC drift artifacts Electroencephalography and Clinical Neurophysiology 86 199 204 DC Detrend x Length of intervals based on Time C Length of intervals based on data points m Interval before marker Marker type for calculating trend Stimulus z Length of interval before markers ms 100 m Interval before DC Correction Length of interval before DC correction ms fi 00 Cancel Fig 10 19 Global DC Detrend dialog In accordance with this method in the first step the average voltage is calculated for each prestimulus interval of the specified marker type In the second step the average voltage is calculated for every DC reset in an interval directly before DC reset The difference between this average voltage before DC reset and the average voltage of the first prestimulus interval after this DC reset is added as an offset to all voltage levels after this DC reset Then in the third step all average values that have been calculated for the prestimulus interval and corrected by adding the offset are viewed along the time axis and the trend is ascertained by applying linear regression to this data Only
192. signal over time cannot be represented with the FFT either This is where wavelet analysis comes into play The wavelet transform calculates the correspondence of the EEG signal with the wavelet used over the entire period of the EEG signal and for different frequency ranges It is crucial here that after the wavelet transform for each time in the time domain there is also a wavelet value for each time and frequency range In the above example this would mean that after the wavelet transform there would also be a curve of the activity over time for the delta band The wavelet transform also permits changes in the spectral content of EEG signals to be examined over time and this is an essential difference between it and the FFT A distinction can be drawn here between discrete and continuous wavelet transforms Vision Analyzer User Manual 145 Discrete wavelet transform DWT The discrete wavelet transform is actually a special case of the continuous wavelet transform CWT but it is nevertheless dealt with here first since on account of its simple algorithm and associated high analysis speed it is already widely used in EEG research The discrete wavelet transform is based essentially on a method known as subband coding In this method the signal of interest is filtered by means of two semiband filters that are as perfect as possible One filter filters the frequency components above half of the available frequency band or Nyquist freq
193. sition You can choose any marker types e Division of the data set into time sections of the same length You can also specify the extent to which sections overlap e Manually marked segments Here no segments of equal length are guaranteed e Segments limited by start and end markers Here the start and end of new segments are defined by markers Data set sections which are not defined by the segmentation criterion are automatically suppressed i e the resultant data set appears as a sequence of segments complying with the criterion Existing segment boundaries are respected during segmentation i e no new segments are generated within which another segment boundary is located The result of segmentation can be a data set with multiple segments or a separate data set for every segment that has been calculated It is also possible to segment subsegment an already segmented data set again on the basis of even finer criteria This method is recommended for example if you want to average different stimuli in a data set separately but want to carry out Gratton amp Coles ocular artifact correction beforehand see also the section entitled Ocular Correction In this case select all the stimulus markers first segment the data set and then carry out ocular artifact correction You can then subsegment the resulting corrected data set again for each stimulus marker separately Storage options The Segmentation module gives you a choice of t
194. ss the ESC key when you want to close the right hand pane Vision Analyzer User Manual 39 7 9 Overlaying different data sets If you want to overlay several complete data sets then you can do it with the Window gt Overlay menu option Alternately you can press the following button on the tool bar E This option only works for data sets with the same sampling rate and same duration The number of channels must not necessarily be identical The view checks the channel names and only overlays those which are the same The easiest way of overlaying data sets is drag and drop To do this use the mouse in the History Explorer to select the data set that you want to overlay hold the left mouse button down and drag the mouse onto the view Then release the left mouse button This only works if the conditions described in the paragraph above are satisfied Select Data Set Parent From Same History File WithSameName Fr p300a Raw Data p300a Raw Data Filters p300a Raw Data Filters Aims p300a Raw Data Rms P300b Raw Data P300b Raw Data Filter P300b Raw Data Filter Stimulus1 P300b Raw Data Filter Stimulus BaselineCorrection1 P300b Raw Data Filter Stimulus BaselineCorection Artifact Rejection1 P300b Raw Data Filter Stimulus B aselineCorrection Artifact Rejection Average P300b Raw Data Filter Stimulus B aselineCorrection Artifact Rejection Average Diff We P300b Raw Data Filter S
195. ta Tool If you activate this by pressing the Delta Tool button you can measure distances with the mouse A Mapping Tool is also available with which you can represent maps at any position of the curve Click the Mapping Tool button to launch this facility Then click at any point on the curve A map appears You can move it with the mouse You get back to the original view by double clicking the channel name again You can set various parameters for the grid view by pressing the following button on the tool bar 2 Set Display Features This brings up a dialog with extensive setting options on three tabs Display Axes Overlays Style Horizontal level lines Vv Show Baseline I Show Horizontal level lines T Show Markers Distance t evel lir fc M Show Border IV Show Label Cancel Fig 7 6 Grid View Settings dialog Display tab On the Display tab you have the following toggle options e Show Baseline e Show Markers e Show Border e Show Label channel name Vision Analyzer User Manual 27 e Show Horizontal Level Lines The distance between lines can also be input in uV here Gridview Settings x Display Axes Overlays X Axis Y Axis C Show never Position haare Position Show if size is sufficient C Baseline Show if size is sufficient Left C Show always 2 C Show always C Time 0 m Tickmarks m Tickmarks Set automatic Set automatic Set manual Set
196. th this button you can remove the current line providing it is not the last line e Insert Current Channels This button is accessible providing you did not choose bipolar montage a data window has been activated and the montage list is empty Clicking it causes all channels in the current data window to be copied to the montage in their original sequence Then you may be able to define the required montage faster by removing and inserting individual channels e Remove All This button becomes accessible when an entry has been completed If you click it the entire content of the montage is removed following a question checking that you really want to do so e Arrange for Grid Views This button opens another dialog box in which the channels for grid views can be arranged The Grid view section of the Views chapter gives more information on this If you opted for source derivation another input box appears at the bottom of the dialog in which you can specify the number of neighboring electrodes to be included in reference calculation You can either type in the channel names or select them from the list boxes If a data window has been activated its channels are available for selection Otherwise channels according to the 10 10 system are at your disposal However you can also type in any names which are not listed in the boxes When you have completed the first 16 channels you can access the next channels with the scroll bar As f
197. the dialog box When you click the Combine Channel with All button all combinations of the selected channel are inserted in the table If the Use Paired Channels View check box is selected the result is presented in a view that is split into two parts On the left you see graphs in the frequency domain which represent the coherence calculated for every frequency point and every channel combination On the right you see the coherence figures between the channels in the shape of lines Their coloring codes the coherence from 0 to 1 If the check box is cleared the result is displayed in one of the frequency views Vision Analyzer User Manual 65 MEI Z Analyzer p300a Raw Data Filters Segmentation FFT Coherence Z Fie Edit View Display Montage Transformations Export History Template Macio Configuration Window Help lel xl ojeli Bl alafss AEF al All isiial ME amel 2 QA p300a B E Raw Data El Fites i A Segmentation a F P300 P3000 P3004 P300e P H P3009 Coh Fp1 Fp2 Coh Fp1 F3 Coh Fp1 F4 Coh Fp1 C3 Coh Fp1 C4 Coh Fp1 P3 Coh Fp1 P4 Coh Fpt 01 Band E Sub Delta F Deita Theta Alpha Beta Ais is a gt Ready Fig 10 12 Output of coherence data 66 Standard Montage 00 00 00 Segment 1 1 10 1 8 Covariance The Covariance module can be used as a preliminary step for various c
198. timulus BaselineCorection Artifact Rejection Averagel FFT P300b Raw Data Filter Stimulus B aselineCorrection1 Artifact Rejection Average Peaks P300b Raw Data Filter Stimulus1 B aselineCorrection Artifact Rejection Average Peaks Z P300b Raw Data Filter Stimulus2 e b P300b Raw Data Filter Stimulus2 B aselineCorrection2 P300b Raw Data Filter Stimulus2 B aselineCorrection2 Artifact Rejection2 P300b Raw Data Filter Stimulus2 B aselineCorrection2 Artifact Rejection2 Average2 P300b Riaw Data Segmentation Fig 7 17 Overlay dialog This dialog enables you to select one or more data sets from the whole workspace The selection is determined by one of four possible criteria which can be set at the top of the dialog These are Parent From Same History File With Same Name and From All Datasets Parent is the data set from which the current data set was calculated This option is inaccessible if the current data set represents the raw data EEG From Same History File shows all data sets in the current history file 40 With Same Name lists all data sets in the workspace which have the same name as the current data set Finally From All Data Sets lists all data sets in the current workspace If you select one or more data sets which have the same length and sampling rate as the current data set then the channels for the selected data sets appear in overlaid mode The Clear Overlays
199. tings screen where you can make the necessary changes You can then click on OK to perform a recalculation In contrast if you do not agree with an automatically calculated template then you can use the mouse to move or resize this The template can be easily moved to distant EEG positions by right clicking at the target position and then making the appropriate entries in the context sensitive menu You can then click Recalculate to recalculate the pulse positions The same is true for located markers which are displayed in color on the screen These can also be moved set or deleted with the mouse Manually set markers are displayed in yellow while markers which are automatically detected by the program are displayed in green Once all the pulse markers have been assigned automatically or with minimum manual corrections the following dialog Pulse Artifact Correction Settings is displayed in which the parameters for the blood pulse correction itself can be selected MRI Artifact Correction Pulse Artifact Correction Settings Time Delay s Number of Pulse Intervals Used for Average 10 I Lowpass Filter for Pulse Artifact Cutoff Frequency Hz 10 lt Back Cancel Figure 10 39 Pulse artifact correction using the R peak method Vision Analyzer User Manual 113 The correction of blood pulse artifacts using the R peak method takes place on the basis of the following parameters e Time Delay This is the time d
200. to drag history information between different history files but also to store it in history template files This aspect is also described later The individual data sets also called history nodes that make up a history file can be deleted and renamed To delete a node select the one in question with the mouse and press the Del key The program asks you whether you want to delete the node and all its subnodes If you confirm this question the node will disappear To rename a node select it and either press the F2 key or click the node text again after a short while The text can now be edited and you can change it to meet your requirements This approach is identical to that in Windows Explorer If you create larger data sets e g FFT and then delete them again the history file may keep its size i e it may contain gaps To remove such gaps move the mouse pointer to the icon for the history file in question and press the right mouse button This opens a context menu where you can choose Compress History File It does not take long to compress the history file To get information on a data set move the mouse to the corresponding icon and again press the right mouse button This opens a menu containing Operation Infos among other items Select this item A window opens showing information on the transform that has been performed Hide Expand All from here Collapse Alll from here Rename Delete Operation Infos Comment P
201. u open the history file and right click a history node which represents segmentation This opens a context menu where you can select the Cache Data option This recreates the cache file You can set the directory in which the temporary files are stored by choosing Configuration gt Select Folder for Temporary Files When averaging has been carried out the cache is no longer required because the Average transform stores the result in the history file An alternative to caching is to store the data in the history file This means that the raw data will no longer be needed in later operations which access the segmented data This approach has advantages especially when the raw data of a workspace is distributed over multiple CDs e g when a PCA needs to be calculated The Segmentation module compresses the data before storing it Since the Analyzer works internally with floating point numbers only but these are unsuitable for compression they must first be converted into integers A resolution precision must be defined for this This resolution should be about 100 nV for EEG data nanovolts 1 yV 1000 nV i e 0 1 uV A lower value means a higher resolution but worse compression Advanced Boolean Expression ABE ABE is a kind of conditional segmentation which is only used for segmentation relative to a marker It can be used for example to select only those segments in which a patient has pressed a response button within a certain tim
202. uency low pass and the other symmetrically filters the frequency components under half the Nyquist frequency Since the resulting signal of the low pass filter only contains frequencies up to half the Nyquist but still retains its full segment length half of the data is inevitably redundant Nyquist s law and is eliminated by subsampling i e by simply deleting every second data point As a result of this processing step the time resolution of the signal is halved since the entire signal is now characterized by half the data points At the same time however half of the data points represent the entire frequency content of the signal and the frequency resolution is thus doubled This process of constant halving of the time resolution and doubling of the frequency resolution is referred to as subband coding This subband coding is repeated for as long as frequency components of interest are to be extracted and each step results in wavelet data that corresponds to the time related composition of the frequency components in the original data This results in a set of coefficients that contains the time frequency curve for all the required subcoding steps i e for all the desired frequency ranges An obvious limitation of this method is that it is impossible to examine the time frequency behavior of a signal at a specific frequency since in the DWT the original signal is repeatedly divided into half frequency ranges beginning at the Nyqu
203. ugh A direct result of the construction of the loadings is that they are orthogonal However physiological fundamentals by no means involves such a necessity For this reason the results of principal component analysis are often subjected to subsequent rotation in the hope of getting data that corresponds more to physiological facts At any rate the results of principal component analysis with or without rotation should not be accepted without checking them They should be subjected to a personal scientific inspection More details on the principal component analysis method and subsequent VARIMAX rotation is given for example in F R sler D Manzey Principal Components and VARIMAX Rotated Components in Event Related Potential Research Some Remarks on Their Interpretation Biological Psychology 13 1981 3 26 The PCA module stores its results as secondary history files in the current workspace When you call the module the PCA wizard guides you through the various input options In the first step you have the choice of defining the variables as time points or as channels If you define the variables as time points you can also reduce the number of variables by condensing the time domain with a fixed factor In this step you can also enable or disable VARIMAX rotation 160 PCA Step 1 of 4 Methods PCA Step 2 of 4 Nodes Variables Name s of the involved data sets history nodes separated by commas Average Variable
204. ultiplexed data and in the first column otherwise Overwrite Default Decimal Symbol You can keep your computer s decimal symbol point or comma or choose one to meet your specific purposes Your computer s decimal symbol depends on Regional Settings The options for binary export are IEEE 32 Bit Floating Point Format or 16 Bit Signed Integer Format In the first instance the data is written in uV or uV as is the case with ASCII export The following options are available for the 16 bit integer format only On exp Set Resolution Manually You can manually set the resolution that is to be used for writing the data This resolution again in uV or pV specifies the minimum difference between two values that you want to store Since the 16 bit format is restricted to a maximum of 2 values too high a resolution can result in data peaks being truncated The range in mV that is achieved with the specified resolution is shown in the dialog You can also input this range instead of the resolution Alternately you can get the program to calculate and set the optimum resolution In this case however you should always export a header file as well since that is the only place you will find the resolution that has been used Individual Optimized Resolution for Each Channel If you have the resolution calculated automatically this option lets you optimize it separately for each channel Otherwise a resolution is chosen that is common to
205. ur ck Abbrechen Fig 10 62 Dialog of the continuous wavelet transform If the continuous wavelet transform is selected you can select the Morlet wavelet real complex and the Mexican hat wavelet as basic wavelets Furthermore you can set the upper and lower limit frequencies for the calculation and define for how many frequency steps the wavelet function is to be calculated 40 frequency steps have proved to be practical for the other representations of the entire EEG spectrum The greater the number of frequency steps however the longer the processing time because the wavelet function is calculated separately for each frequency step For the continuous method there are two different functions available for selection as basic wavelets the Morlet wavelet real complex and the Mexican hat wavelet Since the frequency ranges covered by the scales frequency steps cannot always be easily worked out for a continuous wavelet transform they are listed in the bottom left box to provide you with some orientation If you change the frequency range to be analyzed or the number of frequency steps this table is immediately adjusted 152 As with the discrete wavelet transform the limits of the scales should be viewed merely as guidelines with the continuous wavelet transform The wavelet transform actually distributes a particular frequency over the frequency range in the form of one or more bell shaped curves The actual shape of the curve
206. vailable to the Template Drift Compensation procedure during the subsequent correction operation The start of the drift interval is also adjusted accordingly This is particularly useful in the case of continuous artifacts Disabled Enabled Channels Here you can select channels to which the criterion is to apply In accordance with what has already been set out above in connection with channel selection only channel Fp1 is selected in the figure below Calculate Average Power at each Data Point Reference points are detected when the average power exceeds a threshold Calculate Average Gradient between Data Points Reference points are detected when the average gradient exceeds a threshold Vision Analyzer User Manual 103 MRI Artifact Correction Scanner Artifact Detection Method C Use Maker C Use Detection for Interleaved Scans Measurements with Gaps Use Detection Criterion for Continuous Scans Measurements without Gaps TAlms fi710 Settings for Artifact Detection T Write Only Markers I Use Template Drift Detection Disabled Channels Ekg Ekg2 Eog 01 n zi C Calculate Average Power at each Data Point Power Trigger iV 100000 Histogram Calculate Average Gradient between Data Points Gradient Trigger yV ms 200 Histogram lt Beck Carcel Figure 10 32 Determining scanned intervals In the next step for correcting scanner artifacts you are asked for the position a
207. verage Fig 10 17 Third page of the Comparison dialog when data sets are compared In addition four buttons are at your disposal to get to the required comparative data set faster e Expand History File This button expands the current history file e Expand Same Names This button expands all history files which have a data set of the same name up to this data set e Expand All This button expands all history files e Collapse All This button collapses all history files again Vision Analyzer User Manual 71 If you select the Keep Remaining Channels check box then all channels which do not contain an equivalent in the comparative data set are kept anyway The Use Relative Path for Template or Drag and Drop option enables you to save the relative location of the comparative data set This means that the history templates can always search for this data set in the same history file You can select a data set by double clicking it or by clicking it once and clicking the Finish button 72 10 1 10 Current Source Density CSD This transform replaces the voltage levels at electrodes by valid head coordinates through the current source density at these points The unit is 1 V m The resultant curves do not have any reference electrodes any more They can be processed further with other transforms You obtain the current source density by applying the spherical Laplace operator to the voltage distribution on the surfac
208. w Channels on Top Old Channels o olojojejojojeloje i T c c G o Fig 10 28 Linear Derivation dialog The dialog contains the following elements e Keep Old Channels If you select this option the old channels are included in the new data set as well i e the new channels are practically additional channels Otherwise the new data set consists of new channels only e New Channels on Top If you select this option the new channels appear in front of all the other channels e Number of New Channels Refresh Here you specify the number of new channels that you want The Refresh button then updates the coefficient matrix e Load from File With this button you can load a coefficient matrix from a text file if you want to 94 e Save to File Press this button if you want to save a coefficient matrix that you have input If you want to read in a self generated matrix text file then it must have the following structure Channell Channel2 Channel3 Newl Coeff Coeff Coeff New2 Coeff Coeff Coeff The decimal places of the coefficients have to be set as a point Vision Analyzer User Manual 95 10 1 20 Lateralized readiness potential LRP This module calculates the lateralized readiness potential LRP from two data sets for example movements with the left and with the right hand To do this first generate two data nodes
209. want to do this All unrecognized channels are set to position 0 0 0 invalid position 78 10 1 13 Fast Fourier Transform FFT In the course of a Fourier transform data is transformed from the time domain to the frequency domain i e the resultant items of data indicate the extent to which the individual frequencies are present in the EEG between 0 Hz and at most half the sampling rate Fast Fourier Transformation a x m Resolution Data Window Maximum resolution 0 488 Hz TE C Other Resolution Hz Output Q Voltage pV C No Window Voltage Density v Hz Hanning Window C Power W C Hamming Window C Power Density yv7 Hz Window Length 2 fi 0 m Data Compression M Use Full Spectrum M Use Compression I Normalize Segments 5 Resolution n fio interwal Star Hz 0 5 Interval End Hz 125 Cone Fig 10 22 FFT dialog You have the following input options in the FFT dialog e Resolution Here you define the resolution of the new data set in hertz You can choose between the maximum resolution resulting from the number of points in the segments and the sampling rate and any resolution that you decide to set The maximum resolution is calculated as follows resolution sampling rate segment length e Output Here you define the type of the resulting data and the unit to be used in calculating the data You can choose between the foll
210. with a sampling rate of 1024 Hz and segments of 4096 points The maximum spectral resolution is thus 0 25 Hz 1024 4096 0 25 If you select the Other Resolution option and enter 1 0 Hz instead of 1024 data points being included in the analysis as would be the case given a maximum resolution of 1 0 Hz segments of 1024 data points and a sampling rate of 1024 Hz four times that number of data points are involved The resolution of 1 0 Hz is now established by integrating four spectral line values If the resolution ratios are not exactly divisible the overlaps of the values at the spectral line edges are interpolated However since the original segments are four times longer and thus inevitably have a different information content to the short segments in spite of the FFT resolution theoretically being identical the results can by no means be expected to be the same It should be stated clearly here that this is not a problem caused by the method used it is an inevitable result of the differences in the quantity and contents of the information included in the FFT analysis when there are different spectral resolutions This problem which is inherent in interpolation applies in particular to the calculation of complex FFT values when a resolution other than the maximum spectral resolution is selected This is because in this case not only the values of the spectral lines but also their phase information have to be interpolated and or integrated T
211. x in which you enter parameters for detecting scanned intervals If your data record 102 already contains markers for identifying reference points you can select these here There are two methods available to you for obtaining scanned intervals Use Marker You should select this option if the scanned intervals are set by existing markers You can use the list box to select the appropriate marker Use Detection Criterion for Interleaved Scans If this option is selected the scanned intervals are determined via the power or gradient criteria described above if the intervals are separated by EEG sections without scanner artifacts Use Detection Criterion for Continuous Scans If this option is selected the scanned intervals are determined via the power or gradient criteria described above if the intervals succeed one another without interruption The intervals are delimited from one another on the basis of the Time of Repetition You are strongly recommended only to use this setting in combination with the Template Drift Detection option This can correct the Scan Start position and compensate for possible inaccuracies in the Time of Repetition Write Only Markers No correction is carried out but markers are written Use Template Drift Detection Drifts compared to the previous intervals are determined during the detection of scanned intervals The information concerning drifts by fractions of a sampling interval are made a

Brain Vision Analyzer User Manual Version 1.05

Contents

Download Pdf Manuals

Related Search

Related Contents