Home

User Manual

1. 0 None Compiler Afterwards a list of all installed compilers is displayed You can choose either one but for compatibility we recommend to use the Matlab Lcc Compiler This configuration has to be performed only once For future compilations the settings will remain After configuration we can start compiling out code with the command gt gt mcc m ScannerRecon m This will produce an executable ScannerRecon exe in the local directory after completion Please make sure that all the programs you need for your reconstruction to run are either in the Matlab path or in the local directory Packing the required files for the scanner When the compilation is performed manually we have to make sure to get all the files which are required to run our own reconstruction on the scanner Therefore you should collect e The compiled executable ScannerRecon exe e The Matlab runtime environment MCRInstaller exe which is located in MATLAB rootNtoolboxNXcompilerNdeployNwin32 32bit version MATLAB rootNtoolboxNcompilerNdeployNwin64 64bit version e The license file license lic which is located in MRecon installation path license When you have collected these files copy them to a memory stick or transfer them to the scanner over the net Installation on the Scanner To install the reconstruction on the scanner execute the following steps 1 Install the Matlab runtime environment MCRInstaller exe The runtime envir
2. If data has been written to a temporary file read it again if MR Parameter Chunk NrLoops 1 r temp MRecon exported datafile r temp ReadData r temp SortData MR Data r temp Data fclose all Delete the temporary files delete exported datafile 2009 2011 GyroTools LLC 17 WWW GYROTOOLS COM ReconFrame User Manual delete exported listfile clear r temp end fprintf n Reset the Parameters for the next recon MR Parameter Reset case Cpx MR ReadData MR CombineCoils case Rec MR ReadData end end Subclasses and Function Overloading Even though the MRecon class contains all the basic reconstruction steps its functionality is far from being complete to cope for all possible reconstructions which are available or presented in the MR community If you want to implement your own reconstruction which differs from the Perform function above but still want to use the basic functionalities from MRecon then deriving a subclass is an elegant and easy way Deriving a class means creating your own reconstruction object which has access to all functions and properties of its superclass MRecon Furthermore own functions can be implemented in the new derived class extending the functionality of its superclass And last but not least existing functions can be overwritten by defining one with the same name in the derived subclass This procedure is called function overloading Such a modular design ma
3. Parameter Recon ACNrVirtualChannels The number of virtual coils used in array compression Parameter Recon ACMatrix The compression matrix used in array compression Recon Flag Parameter ReconFlags isread Location Beginning of recon Formats Raw ExportedRaw Cpx Rec Bruker Description Algorithm The ReadData function reads the MR data from the selected file and stores it into Matlab The function supports selective reading meaning that the user can specify what data should be read via r Parameter Parameter2Read After calling the ReadData function the data is stored in the order it was saved in the data file This means that for raw data we obtain a matrix containing the individual k space profiles in the order they have been measured For Rec and Cpx data however we obtain partly reconstructed images in image space Generally the data is stored in a cell array where the different rows hold different data types e g imaging data correction data noise data etc and the columns different data sizes e g kt undersampled and training data The number of rows in the cell array is predefined while the number of columns varies depending on the data in the file Generally the cell array has the following form Data size 1 Data size 2 1 Accepted imaging dat STD STD 2 Rejected imaging data REJ REJ 3 Phase correction data PHX PHX 4 Frequency correction data FRX FRX 5 Noise data NOI NOI If a
4. pn pn normal3 1ab pn 4 4ReconPars m X Parameter2ReadPars After Starting it the project dialog appears Choose a project name e g ScannerRecon prj a location where to save your project e g MRecon installation path ScannerRecon and as Target choose Console Application Afterwards click OK n A Deployment Project New Open Name Target E Console Application ScannerRecon prj Location X martin IBT general matlab programs MRecon Releases Current After the deployment tool has opened execute the following steps 2009 2011 GyroTools LLC 24 WWW GYROTOOLS COM ReconFrame User Manual A Deployment Tool ga c c mm Et c Te a perm File Edit Project Debug Desktop Window Help EJ ScannerRecon prj Build Package O Z Main File on file program s entry point Add main file Shared Resources and Helper Files H Also place here Functions that are called using eval and its variants Functions that are not on the MATLAB path I External C or C code that you want to compile and link into the application Add files directories H U i Place images data files and GUIs fig files here if they are referenced by any of your functions Dessen E ee eee 1 Click on Add main file and add the ScannerRecon m which is located in MRecon ins
5. Downloadable Installation Files M ReconFrame rxxx exe Scanner Software patch compiled ReconFrame rxxx Source zip Scanner Software patch source code MhRecon x X Xxx exe Matlab recon library Windows installer MhRecon x x xxx zip Matlab recon library zipped version for linux or OSX PacknGo exe Packn Go tool MCRinstaller exe Matlab runtime environment ReconFrame_Manual pdf This document Some components of ReconFrame are intended for installation on the scanner others on offline workstations only Follow the instruction bellow for installing the various component of ReconFrame Scanner Software Patch Compiled Version Copy the Software patch ReconFrame rxxx exe to a memory stick and plug it into the scanner Double click on ReconFrame rxxx exe If asked to continue press yes If desired change installation path and press unzip P nb 2009 2011 GyroTools LLC 6 WWW GYROTOOLS COM ReconFrame User Manual Start the patch by clicking on the Windows Start button gt GyroTools gt Start ReconFrame Start the user interface and begin scanning Please make sure that Save raw data is enabled in the postproc tab for the scans you wish to reconstruct with ReconFrame When leaving the scanner stop the patch via the Windows Start button gt GyroTools gt Stop ReconFrame Source Files 1 Copy ReconFrame rxxx Source zip to your VMware 2 Unpack the 3 source files mparc__g c mguacqrcd h mmipr mxg
6. Encoding Parameter Used to read and format the correction profiles approprately Recon Flag Parameter ReconFlags isdepicorr 2009 2011 GyroTools LLC 32 WWW GYROTOOLS COM ReconFrame User Manual Location Between FFT in readout and phase encoding direction Formats Raw ExportedRaw Bruker Description Algorithm In EPI imaging multiple k space lines are acquired after a single excitation Notes of the imaging volume This is achieved by inverting the readout gradient after every acquired profile and thus sampling the k space in a forward backward fashion Due to eddy current effects however the positive readout gradients forward sampling is not exactly the same as the negativ gradient backward sampling which results in a temporal shift between the forward and backward sampled profiles After fourier transformation this shift results in a ghost like artifact located at half of the FOV To correct for this effect EPI correction profiles are acquired prior to every scan The correction profiles consist of a full EPI shot but acquired without phase encoding blips therefore sampling the central k space line multiple times Since these profiles should be identical apart from relaxation effects the temporal shift caused by eddy current effects of the readout gradients can be calculated A temporal shift in k space corresponds to a linear phase in image space Therefore the correction profiles are first transferred into image
7. Methods Events Superclasses In the initial state the Parameter2Read struct tells us exactly what data to expect in the raw file In this case we have a dataset with three different image types standard data phase correction data noise data 8 coils and 97 phase encoding profiles Every time data is read MRecon checks the Parameter2Read struct and only loads the data specified in it Thus to read data selectively we change the struct fields before calling the read function For example let s assume that we only want to read the first 2 of the available 8 coils To do this we set the chan channels field to r Parameter Parameter2Read chan 0 1 Please note that all the values except the typ attribute start from O To actually load the data we call the reading function r ReadData This loads the data and updates the Data Property oft MRecon r has now the following form r MRecon handle Properties Parameter 1x1 MRparameter Data 5x1 cell Methods Events Superclasses We see that the Data property is now set to a cell array with 5 rows r Data ans 320x194 single 1 320x2 single 1 19840x2 single 2009 2011 GyroTools LLC 13 WWW GYROTOOLS COM ReconFrame User Manual The reason for this is that we have read different image types Parameter2Read typ Every image type is written in its own cell element as they usually differ in size and dimension In this case our dataset co
8. end end end Between the lines Place your own code here place whatever code is necessary to reconstruct the final images from the raw data In the example above an MRecon object is created and a standard reconstruction is performed If for example you have written an own reconstruction object My Recon with a Perform function then simply replace MRecon with My Recon 2 Compiling Packing and Installation on the Scanner In order to run our reconstruction on the scanner the Matlab code has to be compiled There are two ways to do that manually or using the matlab s DeployTool Both methods are described below Please note that Matlab s compiler toolbox has to be installed to execute the following steps Also for compatibility we strongly recommend using the 32bit version of Matlab for compilation Manual Installation Compiling 2009 2011 GyroTools LLC 22 WWW GYROTOOLS COM ReconFrame User Manual In a first step the Matlab compiler has to be configured To do so please execute the following command in the Matlab command line gt gt mbuild setup Please choose your compiler for building standalone MATLAB applications Would you like mbuild to locate installed compilers y n y When asked to locate the installed compilers say yes Select a compiler 1 Lecc win32 C 2 4 1 in C PROGRA 2 MATLAB R2010a sys lcc 2 Microsoft Visual C 2005 SP1 in C Program Files x86 MMicrosoft Visual Studio 8
9. lab cpx par and fills its Parameter property Displaying r yields the following r MRecon handle Properties Parameter 1x1 MRparameter Data Methods Events Superclasses We can observe the two properties of MRecon Parameter already filled and Data still empty The Data property will be filled the next time ReadData is called Clicking on the Methods link displays the functions of the reconstruction object MRecon Methods for class MRecon CombineCoils Perform ZeroFill DcOffsetCorrection RandomPhaseCorrection addlistener EPIPhaseCorrection ReadData delete ExportLabels RemoveOversampling eq GridData RingingFilter findobj GridderNormalization RotateImage findprop I2K SENSEUnfold ge K2I ScaleData gt K2IM SortData isvalid K2IP WriteCpx le MRecon WriteExportedRaw lt MeasPhaseCorrection WritePar ne PDACorrection WriteRec notify PartialFourier WriteXMLPar In the next sections we will explain how to use these functions Reading amp Sorting Data Before loading the data into Matlab we can specify what to read For that purpose we have a look at the Parameter2Read struct by typing the following command r Parameter Parameter2Read ans Parameter2ReadPars handle 2009 2011 GyroTools LLC 12 WWW GYROTOOLS COM ReconFrame User Manual Properties typ 3x1 uint8 mix 0 dyn 0 card 0 echo 0 loca 0 chan 8x1 double extrl 0 extr2 0 ky 97x1 double kz O aver 0 rtop 0
10. 1131 1139 doi 10 1002 mrm 21237 Possible values are yes no ACNrVirtualChannels Specifies how many virtual coils are created after array compression Possible values are Any integer number between 1 and the number of physical coils ACMatrix Specifies the compression Matrix for array compression The matrix is automatically calculated when array compression is switched on or the number of virtual coils is changed However it is possible to specify the matrix manually by overwriting it Possible values are any matrix with the size nr_of_virtual_coils x nr_of_physical_coils ImmediateAveraging Switches immediate averaging on or off If set to off every average is reconstructed individiually Possible values are yes no ExportRECImgTypes Specifies what image types are written to the rec and par files when the WriteRec WritePar or WriteXMLPar functions are called Magnitude Phase Real or Imaginary images can be written to the rec file Possible values are M P R I AutoUpdatelnfoPars Specifies if the Imagelnformation parameters should be updated after every reconstruction step While its desirable that these parameters are always up to date reconstruction performance might suffer for large datasets with a lot of images when set to yes For optimal performance switch this parameter to no at the beginning of your reconstruction and to yes at the end of it Possible values are yes no SENSE Switches the SENSE reconstruction on or
11. Parameter Encoding FFTShift Specifies in which dimension the image should be shifted after the fourier transform This is a 3 elements vector specifying the shift in matrix notation Parameter Encoding XRange Parameter Encoding YRange Parameter Encoding ZRange Define the shifts of the image after the fourier transform for every dimension Recon Flag Parameter ReconFlags isimspace Location k space Formats Raw ExportedRaw Cpx Rec Bruker Description Algorithm K21 first checks which dimensions should be transformed defined in Parameter Encoding FFTDims Afterwards it checks which of these dimensions are in k space and then only transforms these ones The fourier transformation along one dimension is given by img sqrt n fftshift ifft ifftshift kspace dim dim dim where n are the number of samples along the specific dimension After the fourier transform from k to image space the images are shifted in image space as defined in Parameter Encoding X Y ZRange Thereby the images are shifted by the amount which leads to a symmetric range around O E g if the range is given as 42 21 then the image is shifted by 10 pixel to the right 42 21 10 32 31 K2IM Performs a fourier transformation from k space to image space in measurement direction See Ki for more information K2IP Performs a fourier transformation from k space to image space in phase encoding directions direction See K l for
12. c to a directory of your choice 3 Copy the three files to the corresponding directory under G ppe Rxxxxxxx system and overwrite the existing files Or Merge the files with your own patch 4 Compile and build the code 5 Copythe compiled patch to the scanner and start it 6 Please make sure that Save raw data is enabled in the postproc tab for the scans you wish to reconstruct with ReconFrame Pack n Go Tool After scanning the raw data can be obtained via the Pack n Go Tool porc gr unu ES 10 11 12 13 Copy the Pack n Go Tool PacknGo exe to a memory stick and plug it into the Scanner Double click on PacknGo exe Select Setup If asked to continue press yes If desired change the installation path and press unzip After Completion the Matlab Compiler Runtime Enviroment 7 13 installer will start Install the Matlab Compiler Runtime If desired chose a different installation path After installation start the Pack n Go Tool via the Windows Start button gt GyroTools gt Pack n Go Tool Chose a Patient in the left window Set the filter in the upper right corner to RAW to display only the raw data Select a scan in the right window and Press Pack amp Go Choose a directory where to save the raw data and press Save Obtain the zip file from the selected location Matlab Library MRecon O gw Double click on MRecon x x xxx exe and install it to a directory of your choice Open Matlab Navigate to Fi
13. check if the data is already sorted Recon Flag Parameter ReconFlags isoversampled Location k space Image space Formats Raw ExportedRaw Cpx Bruker Description Algorithm Generally the receiver samples data denser in readout direction than specified in the user interface The denser sampling produces a larger field of view FOV and prevents fold over artifacts in readout direction Oversampling in readout direction does not increase the scan time as the receiver can sample at a much higher frequency than desired In 3D acquisitions the data is usually oversampled in slice encoding direction as well to correct for improper slice excitation of the outer slices Oversamlling in slice encoding direction does affect the scan time however Since the denser sampling basically results in the acquisiton of a larger FOV we have to remove the oversampling by cropping the image in image space and thereby preserving the signal to noise ratio SNR Removing the oversampling in k space would not be beneficial as we would lose SNR In MRecon the RemoveOversampling can be called at three different locations 1 Before SortData The oversampling is only removed in readout direction by fourier transforming the data in that direction and cropping it This is usually done to reduce the amount of data and save memory 2009 2011 GyroTools A WWW GYROTOOLS COM ReconFrame User Manual 2 After SortData The oversampling is removed in all directions by
14. dimension depending whether we are in k space or image space For example the dimension in y direction after zero filling in k space is given by yres r Parameter Encoding YRes cur ovs where the current oversampling cur ovs is given by cur ovs r Parameter Encoding KyOversampling if oversampling is not removed Or cur ovs 1 if oversampling is removed Whether and in which directions the data is oversampled is defined by the oversampling ReconFlag Parameters ReconFlags isoversampled Zero filling in k space is used to correct for anisotropic voxel sizes and the artificially increase image resolution Zero filling in k space isused to correct for a rectangular field of view and to produce quadratic images 2009 2011 GyroTools LLC 52 WWW GYROTOOLS COM ReconFrame User Manual MRECON PROPERTIES Parameter Parameter2Read Specifies what data to read at the next call of ReadData Upon initialization the fields display all scan parameters which occur in the dataset Please note that not all possible combination of the field values might exist in the dataset typ Defines the image types to be read The values ranges from 1 5 where each number corresponds to a certain image type which might occur in the raw file 2 Standard Data 2 Rejected Data 3 Phase Correction Data 4 Frequency Correction Data 5 Noise Data mix Defines the mixes to be read dyn Defines the dynamics to be read card Defines
15. k space after SortData Formats Raw ExportedRaw Description Algorithm The fast fourier transform FFT requires the data to be on a regular cartesian grid Therefore in order to use the FFT on non cartesian data such as radial or spiral acquistions we have to grid the k space data from the nominal trajectory to a cartesian grid There are several predefined trajectories already built in into MRecon including 2D radial 3D stack of radials 3D kooshball 2D 3D spiral EPI ramp sampling On the other side user defined trajectories can easily be defined by setting the Gridder Parameters Please see the example below on how to do that If the Gridder Parameters are not set manually MRecon automatically detects the type of the scan and decides if it should be gridded or not Notes For non cartesian scans the zero filling process in k space and oversampling removal is automatically performed during gridding by chosing the grid appropriately See example 3 The Gridder parameter can also be filled using the GridderCalculateTrajectory function Examples 1 Reconstruct a golden angle radial acquisition by changing the radial angles in the gridder parameters Golden angle scans exhibit a radial angle increment of 111 246 whilethe rest is exactly the same as a normal radial trajectory Therefore all we have to do to reconstruct a golden angle scan is to change the radial angles in the gridder parameter Create a new reconstruction objec
16. more information MeasPhaseCorrection Corrects for an offset phase which might have been added to certain profiles Syntax r MeasPhaseCorrection Parameters used Parameter Recon MeasPhaseCorrection yes no Enables disables the measurement phase correction Parameter DataFormat Used to check if raw data is passed to the function 2009 2011 GyroTools LLC 39 WWW GYROTOOLS COM ReconFrame User Manual Parameter LabelLookupTable Defines the profiles which are currently stored in the Data array Parameter Labels Index meas phase Holds the measurement phases Recon Flag Parameter ReconFlags ismeasphasecorr Location k space Formats Raw Description Algorithm Different profiles can sometimes exhibit a phase shift which originates from either the RF pulse phase or the acquistion phase specified in the AQ object Different averages for example are usually acquired with a phase difference of pi between them If we wouldn t correct for this phase then the signal would effectively cancel out when summing up the averages MeasPhaseCorrection therefore corrects for this possible phase shift using the values in the labels received by the scanner Let mi be the measurement phase of the i th profile mi Parameter Labels Index meas phase i then the phase added to that profile is given by phase mi pi 2 PDACorrection Corrects the profile dependent amplification PDA Syntax r PDACorrection Parame
17. off Please note that the coil sensitivities have to be provided by the user If the coil sensitivities are not present the SENSEUnfold function has no effect Possible values are yes no Sensitivities Specifies the coil sensitivities for the SENSEUnfold function The sensitivities have to have the same geometry and orientation as the actual data Furthermore the different coils have to be stored in the 4 th dimension of the matrix If the matrix size of the sensitivities are different than the ones from the actual scan they are interpolated to the correct size SENSEPsi Specifies the noise covariance matrix for the SENSE reconstruction If the noise covariance is not given the identity matrix is used in the reconstruction The noise covariance has to be a matrix with dimension coil x coil 2009 2011 GyroTools LLC 59 WWW GYROTOOLS COM ReconFrame User Manual Gridder Defines the gridder parameters There are certain presets e g radial or spiral which are predefined parameter sets for a certain application Beside the presets the gridder can also be configured freely Preset Specifies the Gridder preset Possible values are None Radial Spiral A Gridder presets defines all necessary gridder parameter such as k space sampling positions weights kernel size etc automatically Kpos The k space sampling positions Koos has to be an array with dimensions sampling points x 3 RadialAngles For radial scans a user defined radial angle
18. space and the phase difference between two subsequent profiles is calculated The phase difference is then linearly fitted and the linear phase is added to every odd or even profile thus realigning them in k space Optionally a nonlinear correction can be enabled via Parameter Recon EPICorrectionMethod In the nonlinear correction the calculated linear phase is subtracted from the original phase difference The remaining part is then smoothed and added to the linear phase Optionally an image based correction can be enabled via Parameter Recon EPI2DCorr This correction tries to improve the images further by minimizing the amount of ghosts in the image Thereby the offsets found by the regular EPI crrection are slighly modified and the amount of ghosts in the resulting images is quantified to find the optimal correction value Every coil has its own EPI correction parameters The x axis for the linear fit is given by x floor n 2 ceil n 2 1 where n is the number of samples in the correction profiles The linear fit is weighted by the magnitude of the correction profiles points with a high magnitude value are weighted higher in the fit The EPI correction profiles are labeled as data type 3 and can be read by setting r Parameter Parameter2Read typ 3 followed by r ReadData Examples The user has the possibility to specify own correction parameters and pass them to the function To do that Parameter EPICorrData must be
19. 011 GyroTools LLC 53 WWW GYROTOOLS COM ReconFrame User Manual This information is also used for writing the parameter file in the WritePar and WriteXMLPar function The individual fields in the Imagelnformation struct can also be updated manually by the user if desired This is useful when user defined information should be written to the par file Slice The slice number of the specified image Coil The coil whereby the specified image was acquired Dynamic The dynamic number of the specified image CardiacPhase The cardiac phase of the specified image Echo The echo number the specified image Location The location stack in 3D or slices in MS and M2D of the specified image Mix The mix number of the specified image Extr1 The extr1 number of the specified image Extr2 The extr2 number of the specified image Average The average of the specified image Resolution The resolution matrix size of the specified image Rescalelntercept Used for scaling of the images on the scanner console The displayed value on the console is given by displayed value value in rec file RescaleSlope Rescalelntercept RescaleSlope Used for scaling of the images on the scanner console The displayed value on the console is given by displayed value value in rec file RescaleSlope Rescalelntercept ScaleSlope Used to restore the original floating point value of the complex data from the integer values in the rec file Thu
20. 2 E rf 07012011 1519090 4 2 wipsagfhaV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB 7 Bibliotheken A rf 07012011 1519250 5 1 wipsagfhpV42 rec 07 01 2011 15 22 VLC media file rec 128 KB n FH rf 07012011 1519250 5 2 wipsagfhpV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB es rf 07012011 1519330 6 1 wiptraaplV42 rec 07 01 2011 15 22 VLC media file rec 128 KB Computer E f 07012011 1519330 6 2 wiptraaplV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB A rf 07012011 1519540 7 1 wiptraaprV42 rec 07 01 2011 15 22 VLC media file rec 128 KB Matzet m rf 07012011 1519540 7 2 wiptraaprV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB d a rf 07012011 1520020 8 1 wiptrarlaV42 rec 07 01 2011 15 22 VLC media file rec 128 KB FA f 07012011 1520020 8 2 wiptrarlaV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB rf 07012011 1520180 9 1 wiptrarlpV42 rec 07 01 2011 15 22 VLC media file rec 128 KB FH rf 07012011 1520180 9 2 wiptrarlpV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB rf 07012011 1520270 10 1 wipcorrifV42 rec 07 01 2011 15 22 VLC media file rec 128 KB Dateiname E Oe Dateityp All Philips MR Data sri raw cpx rec Abbrechen A EC J 2009 2011 GyroTools LLC 11 WWW GYROTOOLS COM ReconFrame User Manual Calling MRecon with a directory as input will open the same file dialog but its initial path will be the specified directory r MRecon C MyData MyRawData Upon creation MRecon reads the associated parameter file
21. E Research Wil EE 1607 0111275 ke 1 1 Panscan Georerr Al ri glaners System Ready WCSecen Compose Mor lan GE 71 30 HIE 1 Geos s 7130 nt 2 Geos a gt 1215185 A Gana amp VIF T2 15165 gt We sensors detected VOGeResp ss Zeta scan tne TS mp zen Stop Scan E sE If you now copy the just reconstructed scan your own reconstruction will automatically start after the scan and you don t have to repeat the process above 2009 2011 GyroTools LLC 28 WWW GYROTOOLS COM ReconFrame User Manual 4 yame 9 Sk Sa EE EX Alimidpanes HC Screen Compose No sensor s detected VOG Resp Ss S Remaining scan eve Aetuegen Start Sean Stop Sean g SAR 2009 2011 GyroTools LLC 29 WWW GYROTOOLS COM ReconFrame User Manual MRECON METHODS Average Averages the data in the 12th dimension of the data array Syntax r Average Parameters used Parameter Scan Diffusion Used to check if it is a diffusion scan Only the magnitude is averaged for diffusion scans since phace cancelations can occur otherwise Location k space image space Formats Raw ExportedRaw Description Algorithm This function averages the data in the 12th dimension of the data array where the different averages are stored This function only has an effect if immediate averaging is disabled r Parameter Recon ImmediateAveraging If immediate averaging is enabled then the data is averaged
22. Frame User Manual and perform the recon with r Perform After that the r Data array is filled with filtered k space data seen below Adding your own Parameter to MRecon Sometimes it may be necessary to add an own parameter to the existing parameter groups in MRecon Assume that we want to extend the example above by adding an own Recon Flag checking if we already performed the Hamming filter or not Own parameters can be added in the corresponding parameter files under MRecon rootdir userdata In our example we would add our own Recon Flag to the file ReconFlagParsUsr m as shown below classdef ReconFlagParsUsr ReconFlagPars User Defined Reconstruction Flags properties Add your own parameter here ishammingfiltered 0 end methods Add your own functions here end end We now have added our own Recon Flag which is visible under Parameter ReconFlag after creating the reconstruction object clear classes r My Recon raw file name r Parameter ReconFlags ans 2009 2011 GyroTools LLC 20 WWW GYROTOOLS COM ReconFrame User Manual ReconFlagParsUsr handle Properties ishammingfiltered isread issorted isgridded isimspace iscombined isoversampled isreadparameter israndphasecorr ispdacorr isdcoffsetcorr isdepicorr ismeasphasecorr 0 0 0 OO OO OO Or OO OO Methods Events Superclasses As you can observe our new parameter shows up on top of the existing ones Th
23. Gyrolools WWW gyrotools com G3ECON Classdef My Recon lt Properties MRecon C1 end methods funct x My Recon n MRecon MR nd MR MRecon on Pe MR E User Manual Recon Programming Environment for Scientists Document Version 2 2 April 2013 WWW GYROTOOLS COM ReconFrame User Manual INDEX OF CONTENTS PE ei net AN M MU cur A SI S 4 RUE e Ee AE TEE ID D IL IL LOL LLL ERE 4 Legal NotiCes oorpore vr adidas 4 Introduction and System OvervieW cocccccncconcninnnnnonononononnnnnnnnnnnnnnnnnnnnnnnnn nine nan nas nsns s assess essen testi isa a nas nis 5 o LIE 5 Tete E SIT E EIE DRESDEN EN 6 Mee UE 6 D wnloadabl Installation UE 6 Scanner Software Patchs eet d desire d RESET eve gua eyes Ee aeae dete pas Ea RE RETE 6 COMPIlSA Versione M 6 Source Files a 7 Packs GO TOOL aa 7 Matlab Library IM RECOM ii areniscas 7 Reconstr ction e ET 9 Matrices andiSIiz s ncc enr ci id ees 10 hie unirse 11 Online Documentation 11 Object Creation Inittalzaton noo nanonnnnnrononanonnnnnnnna ano ronnnnnnnannnonns 11 Reading amp Sorting Data 12 Performing a Complete Reconstructlon nennen nennen enne nennen ananas nsns 15 Subclasses and Function Overloading ono nnnnnnnnnnnnnnnnnnnannnnnonnnononnnnnnns 18 Adding your own Parameter to Mecon 20 Performing the Reconstruction on the Scanner 22 1 Modify ScCanner
24. MRecon To get an overview over the matrix and data sizes during reconstruction please refer to the next section Please refer to the section MRecon Usage for a description of how this reconstruction process is implemented in MRecon Sorting and combination of raw data DC offset correction l Profile dependent amplification PDA correction l Random phase correction l Measurement phase correction l Sample order reversal Gridding EPI Pech Spiral Radial Ringing filter optional partial echo half scan filter l FFT in measurement direction Frequency response and gridder amplitude correction Phase correction EPI GraSE FFT in phase encoding direction SENSE unfolding Image production Inhomogeneity shutter clipping scaling 2009 2011 GyroTools LLC SortData DcOffsetCorrection PDACorrection RandomPhaseCorrection MeasPhaseCorrection done automatically in ReadData GridData Ringing Filter PartialFourier K21M GridderNormalization EPIPhaseCorrection K2IP SENSEUnfold ZeroFill RemoveOversampling CombineCoils ScaleData Rotatelmage WWW GYROTOOLS COM MATRICES AND SIZES ReconFrame User Manual The data size and matrix dimensions are adapted several times during the reconstruction process The sizes for the individual reconstruction steps are defined by the Parameter Encoding struct For illustration a graphical representation of the reconstruction process is shown belo
25. Patch 2 Performing the Reconstruction on the Scanner 1 Acquire a scan where Save raw data is switched to yes in the Postproc tab 2 After completion of the scan switch to the view context T1 3D ffE 1 Geos Ti 30 ft 2 Geos Start Sean Stop Scan B cT 3 Right click on the acquired scan and chose PRIDE Research In the window on the right chose ScannerRecon from the drop down menu and click on Submit Job 2009 2011 GyroTools LLC 26 WWW GYROTOOLS COM ReconFrame User Manual Bask Viewing Help Y Research oo to run Scammeritacon ta PRIDE Research Dveriay Package 5 Your own reconstruction will start and a black command window appears Wait until it has disappeared again PEI Arca backageconigunti Ae see Mi bor Conta d Scirnerfecor qi de Logihaapeikate D chose recon TI at CVrogram fles 2M ME Lem 6 Wait a few seconds until the newly reconstructed scan appears or right click in the left window and chose refresh The new scan is now reconstructed with your own recon 2009 2011 GyroTools LLC 27 WWW GYROTOOLS COM ReconFrame User Manual Session Case Views Tools Analysis Help PHILIPS mouse H E 1329 B mb343456 01 01 2001 7 If you switch back to the scan environment you can load the newly reconstructed scan by clicking on the little symbol Patent System Help 1321 D Mms 01 01 2001 mb343456 FRE
26. ReconFrame patch is need for these parameters to be filled out completely ScanMode Specifies the scan mode Possible values are 2D 3D MS M2D SV 1D FastimgMode Specifies the fast imaging mode Possible values are None TSE TFE EPI GRASE TFEEPI TSI TFEfactor 2009 2011 GyroTools LLC 55 WWW GYROTOOLS COM ReconFrame User Manual The TFE factor which was used in the current scan AcqMode Specifies the acquisition mode Possible values are Cartesian Radial Spiral Propeller Dependencies ky and kz are updated for radial and spiral scans ky kz lables start from 0 Gridder Preset is updated UTE Specifies if Ultra Short TE was enabled KooshBall For 3D radial sampling this parameter specifies if the trajectory was a Kooshball TR The repetition time TE The echo time FlipAngle The flip angle which was used in the current scan FOV The field of view RecVoxelSize The voxel size in the reconstructed images AcqVoxelSize The acquired voxel size Offcentre The offcentres which were used in the current scan in mm Please note that every location and every mix has its own offcentre value Angulation The angulations which were used in the current scan in degrees Please note that every location and every mix has its own angulation value Orientation The orientations used in the current scan Possible values are TRA Transversal COR Coronal SAG Sagit
27. Res KzOversampling Location k space image space Formats Raw Rec Description Algorithm Please see the transformation manual for more information on the different coordinate systems and examples To transform a coordinate from one coordinate system to another the following parameters have to be known Patient Position defines coordinate system Patient Orientation defines coordinate system Image Orientation defines coordinate system Fold Over Direction defines coordinate system Fat Shift Direction defines coordinate system Angulation fixed used in Transform Offcentre fixed used in Transform Slice Gap fixed used in Transform Resolution variable used in Transform Resolution curFOV size Data Image Matrix variable used in Transform for ijk REC systems The first 5 parameters are used to define the different coordinate systems ijk MPS RAF xyz REC which are stored in Parameter Scan and used in the Transform function Therefore the parameters itself are not needed in the Transform function anymore Basically the parameters which are used in the Transform function can be divided into 2 groups fixed and variable parameters The fixed ones such as the image angulation do not change during the reconstruction process whereas the variable ones can change The image resolution for example changes if we perform a k space zero filling In transform the current image resolution is calcu
28. age function transforms the Data from the current coordinate system usually MPS the final orientation displayed on the scanner console Examples The Rotate Image basically performs a transformation between the system specified in r Parameter Scan ijk and r Parameter Scan REC These coordinates systems are always expressed in the patient system RL AP FH Let s assume these values are given as r Parameter Scan ijk RL FH AP r Parameter Scan REC HF RL AP To transform the ijk system to the REC system we have to switch the first and the second axis of the Data matrix permute it RL FH AP FH RL AP and then flip the first axis flipdim FH RL AP gt HF RL AP SENSEUnfold Performs a SENSE reconstruction using the coil sensitivity information provided by the user Syntax r SENSEUnfold Parameters used Parameter Recon SENSE yes no Enables disables the SENSE unfolding Parameter Recon Sensitivities The sensitivity object MRsense used in the unfolding process The sensitivity object holds the coil sensitivities the noise covariance matrix as well as regularisation images Parameter Recon SENSERegStrength The regularization strength used during the SENSE unfolding A higher regularizazion results in a stronger suppression of low signal values in the unfolded images 2009 2011 GyroTools LLC 46 WWW GYROTOOLS COM ReconFrame User Manual Parameter Scan SENSEFacto
29. al Please note that every location and every mix has its own orientation value FoldOverDir The fold over directions used in the current scan Possible values are RL Right Left AP Anterior Posterior FH Feet Head Please note that every location and every mix has its own fold over direction FatShiftDir The water fat shift directions used in the current scan Possible values are R Right L Left A Anterior P Posterior F Feet H Head Please note that every location and every mix has its own water fat shift direction SENSEFactor The SENSE factors used in the current scan Rotate Specifies the operation which has to be performed to orient the image in the same way as on the scanner screen NrDatabaselmages Specifies how many images are expected from the database 2009 2011 GyroTools LLC 56 WWW GYROTOOLS COM ReconFrame User Manual Encoding Provides information about the encoding such as matrix sizes sampling ranges oversampling factors etc For a more detailed description about the matrices and sizes in the reconstruction process refer to the previous section NrMixes The number of mixes in the current dataset NrEchoes The number of echoes in the current dataset Mix Specify the mixes that corresponds to the different rows in the following encoding parameters Echo Specify the echoes that corresponds to the different r
30. ans Only then correction parameters are provided by the scanner DcOffsetCorr Corrects for a signal offset that might occur in the acquired samples Syntax r DcOffsetCorrection Parameters used Parameter Recon DcOffsetCorrection yes no Enables disables the DC offset correction Parameter DataFormat Used to check if raw data is passed to the function Parameter LabelLookupTable 2009 2011 GyroTools LLC 31 WWW GYROTOOLS COM ReconFrame User Manual Defines the profiles which are currently stored in the Data array Parameter Encoding KxRange If no noise samples are present in the raw file this parameter specifies where the outer k space noise like samples are located Recon Flag Parameter ReconFlags isdcoffsetcorr Location k space before the partial Fourier reconstruction Formats Raw Description Algorithm Acquired raw samples can sometimes exhibit a signal offset originating from hardware inaccuracies of the receiver If not corrected this offset can cause an artifact which will appear as bright signal peak in the middle of the image Since noise in the MR signal is normally distributed with a mean of zero the offset can be calculated by calculating the mean of the noise samples Therefore if the raw data contains noise samples from a noise acquisition prior to the scan these samples are read and the mean is calculated If the data does not contain any noise the offset is calculated from the no
31. ay dimension corresponds to an image attribute r Data kx ky kz coils dynamics cadiac phases echoes locations mixes extra1 extra2 averages Transform Transforms coordinates from one scanner coordinate system to another Syntax xyz transformed A r Transform xyz from system to system stack nr Or A r Transform from system to system stack nr 2009 2011 GyroTools LLC 48 WWW GYROTOOLS COM ReconFrame User Manual Parameters used Parameter Scan ijk Parameter Scan MPS Parameter Scan xyz Parameter Scan REC The different scanner coordinates system expressed in the patient coordinate system Parameter Scan Angulation Parameter Scan Offcentre Angulation and offcentre of the scan Parameter Scan curFOV The current field of view of the scan curFOV corresponds to the FOV of the current data and does change during the reconstruction process Removing the oversampling for example reduces the FOV Parameter Scan SliceGap The slice gap of the scan Parameter Encoding XRes Parameter Encoding YRes Parameter Encoding ZRes Parameter Encoding KyOversampling Parameter Encoding KyOversampling These parameters are only used if the Data matrix is empty The default matrix size used in the Transform function when no data is present is the one which we would obtain after the SENSE unfolding which has to be used for calculating the coil sensitivities It is given by XRes YRes KyOversampling Z
32. beconstruction m eene eene nnnnnnnnn nnns nns en nennen tentia sanas sanas san sns nnn 22 2 Compiling Packing and Installation on the Scanner esses nennen 22 2 Performing the Reconstruction on the Scanner 26 MRecon Methods itio e teet tSp te aa eege 30 ncc M 30 COMIDING CONS soe dt A A A erae e ai 30 ConcomitantFieldCorrectiON coccocccnccnonnnananonononononnnnnnnnncnnnnnnnnnnnnr ono nnnnnnnnnnnnnnnnnnnnnnannnnnr nro rnnnnnnnnns 31 DCORSEEC OM iia A A di id 31 EPIPhaseCortectlOFi eie ctae A a Ei bags 32 lee Ee E 34 e Tou disque 34 Elle 1D E EE 35 GridderCalculateTrajectory ooooocococnncnnnnnncnnnnannnananononononnnnnnnnonnnnnnnnnnnnnn nn nr nnnnnnnnnnnnnnnnnnnnnnnrnnnnrnnnnnnnnns 37 GridderNormalization cccccsssscecceccececcecceeeeeeeeseaaeaeceeeeesesesecsseeseeeseeaeaaeeseeeeseseseesseesseseeusanaesaeees 37 pic DM 38 Ml A 38 Kee Caveat siesta dt denen E Eege 39 K2Ip us comte eege over de tuve Boece Gack 39 MeasPliaseCorrection it et e e ree Re E poe Rena epe ls RE AA 39 cip gelido e A A aaa 40 Partial FOUN OR certc e t esr Pv eese ia T edet e ead perve dep du o Aere NER 41 2009 2011 GyroTools LLC 2 WWW GYROTOOLS COM ReconFrame User Manual PerfOL mette ob RU n st dM 41 Randombbaseorrectlon cessisse a E ee AEE assai s sess essen setis asa a aee diea aeti 42 ReadBData ir t o eese het e EI ee Ee 42 RemoveOversatmpl
33. can be set if desired e g golden angle scans When left empty then the standard angles are used which are predefined by the scanner Weights The gridding weights which describe the sampling density in k space Weights has to be a vector with the size nr of samples x 1 KernelWidth The gridding kernel width KernelBeta The kernel beta GridOvs The gridding oversampling factor Normalize Specifies if the data is normalized after gridding needs an additional Fourier transform Cardiac Synchronization Specifies the method of cardiac synchronization Possible values are None Retro retrospective gating In retrospective gated scans there are no predefined heart phases Instead every profile is assigned the acquisition time in the RR interval as well as the length of the RR interval When the Synchronization is set to Retro then a heart phase is assigned to every profile based on the timing information Since the heart rate changes during the measurement there might still be missing profiles holes in k space after the assignment These holes are filled with the best matching profile we have acquired Dependencies Cardiac phase labels RetroPhases and PhaseWindow are updated Furthermore additional labels are created which correspond to the holes in k space RetroPhases Specifies the number of heart phases which should be created This number might be different from the number which was set in the scanner user interface or was ef
34. coils are already combines Isoversampled Specifies if the oversampling has already been removed Isreadparameter Specifies if the parameter file has already been read Israndphasecorr Specifies if the data has already been random phase corrected Ispdacorr Specifies if the data has already been PDA corrected Isdcoffsetcorr Specifies if the data has already been DC offset corrected Isdepicorr Specifies if the data has already been EPI phase corrected Ismeasphasecorr Specifies if the data has already been measurement phase corrected Isunfolded Specifies if the data has already been SENSE reconstructed Labels Contains all the information from the parameter file The Labels struct can be different for the different file formats Most of the parameter in the Labels struct are passed to other parameter groups such as the encoding or Scan parameters 2009 2011 GyroTools LLC 61 WWW GYROTOOLS COM ReconFrame User Manual In raw files Labels Index contains the label information for every individual profile in the order they have been measured 2009 2011 GyroTools LLC 62 WWW GYROTOOLS COM ReconFrame User Manual DATA Contains the data at the current state of the reconstruction process Every time a reconstruction function method is called the data is updated Data is a predefined 12 dimensional array where the single dimensions are defined as x y z coils dynamics cardiac phases echoes locations mixes
35. data type does not occur in the file the the selected cell element is left empty If only one data type and size is read cell array with one element then a ordinary Matrix is returned instead of a cell Examples Read only imaging data from the first coil and the third dynamic r MRecon rawfile raw 2009 2011 GyroTools LLC 43 WWW GYROTOOLS COM ReconFrame User Manual r Parameter Parameter2Read typ 1 Imaging data r Parameter Parameter2Read chan 0 First coil r Parameter Parameter2Read dyn 2 Third dynamic r ReadData Read the noise samples from a raw file r MRecon rawfile raw r Parameter Parameter2Read typ 5 Noise data r ReadData RemoveOversampling Removes the oversampling from the image by cropping the data in image space Syntax r RemoveOversampling Parameters used Parameter Recon RemoveMOversampling yes no Enables disables the oversampling removal in readout measurement direction Parameter Recon RemovePOversampling yes no Enables disables the oversampling removal in phase encoding direction s Parameters ReconFlags isimspace Used to check if the data is in k or image space Parameter Encoding KxOversampling Parameter Encoding KyOversampling Parameter Encoding KzOversampling The oversamping factors in x y z directions After calling the RemoveOversampling function the Data matrix is smaller by these factors Parameters ReconFlags issorted Used to
36. e clear classes command at the beginning is necessary to reinitialize all classes Without it our modifications would not have any effect The newly created Recon Flag can now be used in the HammingFilter function for example if MR Parameter ReconFlags ishammingfiltered error The data is already Hamming filtered end MR Parameter ReconFlags ishammingfiltered 1 2009 2011 GyroTools LLC 21 WWW GYROTOOLS COM ReconFrame User Manual PERFORMING THE RECONSTRUCTION ON THE SCANNER When you are finished developing your reconstruction offline you can then perform it directly on the scanner console after acquiring the data After the reconstruction is finished the images will be written back to the database seemingly integrating ReconFrame into the workflow To perform your own reconstruction on the Scanner the following steps have to be executed 1 Modify ScannerReconstruction m After you are finished developing your reconstruction open the file ScannerRecon m which is located in MRecon installation pathNScannerReconN classdef ScannerRecon properties end methods function SR ScannerRecon lab filename raw filename r MRecon lab filename raw filename r Perform r Parameter Recon ExportRECImgTypes M try r WriteRec G patch pride tempoutputseries DBIEX REC r WriteXMLPar G patch pride tempoutputseries DBIEX XML catch r WriteRec DBIEX REC r WriteXMLPar DBIEX XML end
37. e scanner are written back in the scanner patient database allowing for a seamless integration of custom reconstructions into the scanner workflow MRecon ReconFrame contains an extensive object oriented Matlab library implementing many common image and spectrum reconstruction task The object oriented design makes it easy to expand or alter the existing functionality Code examples of complete reconstructions are provided They can serve as a starting point for custom development Scanner Software Patch On the scanner ReconFrame installs a scanner software patch which for each scan will complement the standard raw data with all necessary scan parameters required for reconstruction The raw data is then the only input to ReconFrame based reconstructions Sequence parameters exported in such a way are decoded and directly available in Matlab The scanner software patch source code is provided and thus any additional scan parameter can easily be added to the ReconFrame data interface Pack n Go Tool The pack and go tool is ReconFrame s independent patient database browser Installed on the scanner it allows accessing the database and exports raw and rec data files using conveniently compressed file archives 7 NM Offline Recon Online Recon Custom Reconstruction Raw Files Acquisition Scanner Patch Compiled code Custom Reconstruction Parameter File lab Data File ra
38. eck if the data has the right form if MR Parameter ReconFlags isread error Please read the data first end if MR Parameter ReconFlags issorted error Please sort the data first end MR Data can be a cell or an array depending on the dataset To make our function work with all kinds of data we convert MR Data to a cell in any case and then let the hamming filter work on every cell element using cellfun MR Data MR Convert2Cell MR Data Filter the Data by calling the filter function on every cell element MR Data cellfun x MR filter_image x MR Data UniformOutput 0 Convert MR Data back to oroginal state MR Data MR UnconvertCell MR Data End of public methods functions are Hidden to the user Static Hidden Actual code for the Hamming filter called by cellfun function data filter_image data end end end if isempty data Obtain the number of images in MR Data data size size data nr images prod data size 3 end Define filter hfilter hamming size data 1 hamming size data 2 hfilter hfilter max hfilter Filter every single image for i 1 nr_images data i data i hfilter end end End of hidden methods End of My Recon class After implementing the code above we can create our own reconstruction object by r My Recon raw file name 2009 2011 GyroTools LLC 19 WWW GYROTOOLS COM Recon
39. econ ImmediateAveraging When enabled the averages are added immediately during SortData If you want to reconstruct the individual averages seperately then disable this parameter Parameter Labels The data labels which hold the image attributes for every profile Parameter LabelLookupTable Links the profiles which are currently stored in the Data array with the data lables Parameter Encoding KyRange Parameter Encoding KzRange The sampled k space ranges in phase encoding directions used to place the data correctly in k space e g in halfscan acquisitions Parameter Encoding KyOversampling Parameter Encoding KyOversampling The oversampling factors in phase encoding directions used to place thedata correctly in k space Recon Flag Parameter ReconFlags issorted Location After ReadData Formats Raw ExportedRaw ExportedCpx Description Algorithm After reading the data into Matlab with ReadData the acquired profiles are stored in temporal order in the order they have been measured The SortData function sorts the profiles by their image attributes phase encoding number cardiac phase dynamic scan number etc After the sorting the data the data array holds proper k space data which can be fourier transformed to generate images Please note that averages are summed up immediately during sorting if immediate averaging is enabled see above After SortData the data array has a predifined form with 12 dimensions where every arr
40. en such that the range becomes symmetric around O after shifting ZRange Specifies the range in k space image space along y phase encoding direction after k space zero filling and oversampling removal Note that images in that stage already have isotropic voxel sizes in the y x plane An asymmetric range means that the image should be shifted after Fourier transformation The amount of the shift should be chosen such that the range becomes symmetric around O after shifting XRes Specifies the number of pixels in k space image space along measurement direction after k space zero filling and oversampling removal This number should correspond to XRange Note that images in that stage already have isotropic voxel sizes in the y x plane YRes 2009 2011 GyroTools LLC 57 WWW GYROTOOLS COM ReconFrame User Manual Specifies the number of pixels in k space image space along y phase encoding direction after k space zero filling and oversampling removal This number should correspond to YRange Note that images in that stage already have isotropic voxel sizes in the y x plane ZRes Specifies the number of pixels in k space image space along z phase encoding direction after k space zero filling and oversampling removal This number should correspond to ZRange Note that images in that stage already have isotropic voxel sizes in the y x plane XReconRes The final resolution of the reconstructed image in measurement direction YReconRes The fina
41. erted kx Create the modified k space These are the coordinates we grid on By definition the coordinates must be of dimension no samples x no y profiles x no z profiles x 3 See Gridder Parameters for more information k cat 4 kx inverted ky kz Assign the new trajectory to the Gridder r Parameter Gridder Kpos k 2009 2011 GyroTools LLC 36 WWW GYROTOOLS COM ReconFrame User Manual Assign gridder weights sampling density In this example the samples are uniformly distributed on the whoe k space all weights are set to 1 r Parameter Gridder Weights ones no kx samples no ky samples no kz samples Perform the reconstruction r Perform Show the data Observe that the image is flipped in readout direction compared to the ungridded reconstruction a flip in k space corresponds to a flip in image space r ShowData GridderCalculateTrajectory Calculated the gridder trajectory for the current scan and fills the gridder parameter Syntax r GridderCalculateTrajectory Parameters used Parameter Gridder The gridder parameter Parameter Labels NusEncNrs The non uniform sample position for EPI scans In EPI acquisitions the samples in readout direction are not uniformly spread due to the sampling during gradient ramp up Therefore EPI scans are usually gridded in readout direction using the k space positions specified here Location k space after SortData Formats Raw ExportedRaw Descrip
42. extr1 extr2 averages 2009 2011 GyroTools LLC 63
43. fectively measured However note that the data is simply temporally interpolated if the number of heart phases is set to a larger number than the one which was actually measured Dependencies Cardiac phase labels are updated Furthermore additional labels are created which correspond to the holes in k space PhaseWindow Alternatively a single heart phase instead of several ones can be reconstructed The PhaseWindow is a 2 dimensional vector which specifies the start and end of the window in ms after the cardiac trigger Dependencies RetroPhases and Parameter2Read rtop are updated RNAV 2009 2011 GyroTools LLC 60 WWW GYROTOOLS COM ReconFrame User Manual The respiratory navigator positions if enabled as calculated by the scanner The positions are split between the navigator preparation phase and the acquisition phase DataFormat Specifies the data format of the current dataset Possible values are Raw ExportedRaw Cpx Rec Filename Data The filename of the data file Parameter The filename of the parameter file ReconFlags Provide informations about the current status of the reconstruction Isread Specifies if the data has already been read Issorted Specifies if the data has already been sorted Ispartialfourier Specifies if the data has already been corrected for halfscan Isgridded Specifies if the data has already been gridded Isimspace Specifies if the data is in image space Iscombined Specifies if the
44. file ZeroFill Zero pads the data either in k space or image space Syntax r ZeroFill Parameters used Parameter Recon kSpaceZeroFill yes no Enables disables the k space zero filling Parameter Recon ImageSpaceZeroFill yes no Enables disables the image space zero filling Parameters ReconFlags isimspace Used to check if the data is in k or image space Parameters ReconFlags isoversampled Used to check if and in which directions the data is oversampled Parameter Encoding XRes Parameter Encoding YRes Parameter Encoding ZRes 2009 2011 GyroTools LLC 51 WWW GYROTOOLS COM ReconFrame User Manual The data is zero filled to these dimension in k space times the remaining oversampling factor Parameter Encoding XReconRes Parameter Encoding YReconRes Parameter Encoding ZReconRes The data is zero filled to these dimension in image space times the remaining oversampling factor Parameter Encoding KxOversampling Parameter Encoding KyOversampling Parameter Encoding KzOversampling The oversamping factors in x y z directions Parameters ReconFlags issorted Used to check if the data is already sorted Recon Flag Parameter ReconFlags iszerofilled Location k space Image space Formats Raw ExportedRaw Cpx Bruker Description Algorithm The ZeroFill function zero pads the current data to the dimensions specified in the Encoding parameters see above Thereby we have a different
45. filled Every coil has its own slope and offset To identify the coil we must label them in the correction parameter r Parameter EPICorrData coil r Parameter Parameter2Read chan Assign the same slope for all the coils r Parameter EPICorrData slope 0 1 Assign a different offset for every coil 4 coils in this example 2009 2011 GyroTools LLC 33 WWW GYROTOOLS COM ReconFrame User Manual r Parameter EPICorrData offset 0 0 2 0 4 0 6 Call the EPI correction r EPIPhaseCorrection ExportLabels Exports the imaging labels which are stored in the lab file to a textfile Syntax r ExportLabels filename Parameters used Parameter Labels Index The labels to be exported Location k space image space Formats Raw ExportedRaw Bruker Description Algorithm This function exports the imaging labels which were sent from the spectrometer to the reconstructor to a textfile Every row in the file corresponds to an measured k space line in the order they have been acquired Exporting the labels is useful to get an overview over the sampling order for example Notes The format of the textfile is virtually identical to the one of the list file of the exported raw data format GeometryCorrection Corrects for gradient non linearity s over large FOV s Syntax r GeometryCorrection Parameters used Parameter Recon GeometryCorrection yes no Enables disables the geometry correctio
46. fourier transforming the data in every oversampled direction and cropping it in image space 3 In image space The oversampling is removed in all directions by cropping the data Notes The oversampling is not removed for data which has to be gridded when called before SortData radial spiral EPI RingingFilter Removes ringing artifacts by applying a hamming filter on the k space data Syntax r RingingFilter Parameters used Parameter Recon RingingFilter yes no Enables disables the ringing filter Parameter Recon RingingFilterStrength A value between 0 and 1 which specifies the strength of the ringing filter O means no filter is applied and 1 is the maximum filter strength The filter strength can be specified in every direction Measurement Phase Encoding Slice Encoding Location k space Formats Raw ExportedRaw Description Algorithm Sharp edges in k space can result in ringing artifacts in image space Such edges are introduced by zero filling or by sampling a low resultion image The ringing filter tries the soften the edges by applying a hamming filter on the k space data lowering the values on the edge The amount by which the edge values are lowered is specified by the filter strength A filter strength of 0 5 for example will bring down the edge value by 5096 A strength of brings the value down to 0 and therefore removing the edges completely However a large filter strength might cause image blurrin
47. function corrects for that by dividing the image by the fourier transform of the bessel kernel I2K Performs a fourier transformation from image space to k space Syntax r I2K Parameters used Parameter Encoding FFTDims Specifies in which dimension the fourier transformation should be executed This is a 3 elements vector specifying the dimension in matrix notation Parameter ReconFlags isimspace Used to check which dimensions are already in image space Recon Flag Parameter ReconFlags isimspace Location mage Space Formats Raw ExportedRaw Cpx Rec Bruker Description Algorithm DK first checks which dimensions should be transformed defined in Parameter Encoding FFTDims Afterwards it checks which of these dimensions are in image space and then only transforms these ones The fourier transformation along one dimension is given by kspace 1 sqrt n fftshift fft ifftshift img dim dim dim where n are the number of samples along the specific dimension K2I Performs a fourier transformation from k space to image space Syntax r K2l Parameters used Parameter Encoding FFTDims 2009 2011 GyroTools LLC 38 WWW GYROTOOLS COM ReconFrame User Manual Specifies in which dimension the fourier transformation should be executed This is a 3 elements vector specifying the dimension in matrix notation Parameter ReconFlags isimspace Used to check which dimensions are still in k space
48. g Therefore it is important to find an optimal compromise between artifact removal and image blurring The default filter strength is set to 0 25 Notes The ringing filter should be applied before k space zero filling to work effectively RotateImage Rotates the images such that they are oriented in the same way as on the scanner console Syntax r Rotatelmage Parameters used Parameter Recon Rotatelmage yes no Enables disables the image rotation Parameter Scan ijk Specifies the coordinate system of the images before the rotation current coordinate system Parameter Scan REC 2009 2011 GyroTools LLC 45 WWW GYROTOOLS COM ReconFrame User Manual Specifies the coordinate system of the images after the rotation coordinate system of the Rec images Recon Flag Parameter ReconFlags isrotated Location Image space Formats Raw Description Algorithm After reading the raw data it is oriented in The MPS system meaning that the rows in the Data matrix are aligned along the measurement direction and the columns along the phase encoding direction The MPS system however heavily depends on the chosen scan parameters such as orientation fold over direction fat shift direction etc The reconstructed images displayed on the scanner console on the other hand are oriented in a well defined fashion expressed in the patient coordinate system right left RL anterior posterior AP feet head FH The Rotatelm
49. g the Perform function the Data corresponds to the final squared image with the reconstructed dimension of 256 pixels If we visualize the image we get 2009 2011 GyroTools LLC 15 WWW GYROTOOLS COM ReconFrame User Manual The code for the Perform function can be found below As you see we pretty much follow the scheme which is shown in the reconstruction overview Note that we reconstruct the data in chunks and save the images of every reconstructed chunk in a temporary file on the disk function Perform MR switch MR Parameter DataFormat case ExportedRaw Raw The Perform function is not working on spectro files yet if strcmpi MR Parameter Scan ScanType Spectro error Error in Perform The Perform function does not work on spectroscopy data yet You can however run the individual reconstruction steps like ReadData SortData ero t s end Reconstruct only standard imaging data MR Parameter Parameter2Read typ 1 MR Parameter Parameter2Read Update Check if enough memory is available to reconstruct the whole file at once Otherwise reconstruct the data in chunks if ispc MemoryNeeded MemoryAvailable MaxDataSize MR GetMemoryInformation if MemoryNeeded MemoryAvailable MR Parameter Chunk Def kx ky kz chan end oe dp d end Switch off for performance reasons after recon it is switched back to original state again AutoUpdateStatus MR Parameter Recon AutoUpda
50. ge space Parameter Encoding KxRange Parameter Encoding KyRange Parameter Encoding KzRange Specifies the sampling ranges in k space These parameters are used to check if the scan is acquired using partial Fourier and define the fully sampled k space center Parameter Scan SENSEFactor The SENSE factor used to define the fully sampled k space center Recon Flag Parameter ReconFlags ispartialfourier Location k space image space Formats Raw ExportedRaw Bruker Description Algorithm In partial fourier imaging only a fraction of k space is acquired for improved scan or echo time If the k space is partly sampled in readout direction we speak of partial echo and halfscan for a partly sampled space in phase encoding direction In the first case the echo time is reduced while in the second one the scantime is shortened Performing a standard imaging reconstruction on the partly sampled k space however leads to image bluring and reduced SNR Therefore a partial fourier reconstruction is performed which utilizes the k space symmetry to reconstruct images of better quality In MRecon a so called homodyne reconstruction is implemented Noll D C Nishimura D G Macovski A Homodyne detection in magnetic resonance imaging Medical Imaging IEEE Transactions on vol 10 no 2 pp 154 163 Jun 1991 Notes In principle the partial fourier reconstruction can either be performed in k or image space However if executed in k space it is not
51. in the SortData function already Some data however should not be averaged in k space e g diffusion scans That is why this function is called in image space to average the remaining averages which have not been processed yet Notes For diffusion scans only the magnitude is averaged CombineCoils Combines the individual coil images into one single image Several combination methods have been implemented in MRecon see below Syntax r CombineCoils Parameters used Parameter Recon CoilCombination sos pc When set on sos a sum of squares combination is performed Note that the image phase is lost with this method pc is used in phase contrast flow reconstructions where the magnitude is a sum of squares combination and the phase is a magnitude weighted combination of the individual phase images Recon Flag Parameter ReconFlags iscombined Location mage space Formats Raw ExportedRaw Cpx Bruker Notes For a non phase contrast scan the image phase is lost after CombineCoils If you want to preserve the phase please perform a SENSE CLEAR reconstruction see the SENSEUnfold function 2009 2011 GyroTools LLC 30 WWW GYROTOOLS COM ReconFrame User Manual The CombineCoils function destroys the label lookup table Parameter LabelLookupTable since the profiles cannot be related to the labels after combining the coils The coils have to be in the 4th dimension of the Data array to be combined Conc
52. ing citi t ZS re Pe res td adan 44 RINgIN grille vetet tiae seed ics 45 R tatelmage EE 45 SENSEU nifold siete itia c bk ec i Ts 46 S aleData zt ee ema utes mu ROE BT Tee s ee 47 SHOW Data tti tiroteo oes AE uta edu eus 47 SOM Data PE 48 RRE ell ENIM CITUR E 48 WriiteExportedRaw iit ee ERE RT PIEVE E Pee VRBE TFI MEE Ie 50 WHITER al 4i eer tod es ten tete dee e d ete eee ie etn 50 O ere ee tete mieten stet tete deve et deste e ET 50 WYriteXMLP atic n 51 ZetoFill2 rr Er eee e RM D RR i 51 MRecon Propettles sc EE eei t aia Ee 53 len 53 Parameter2Rea3d E E Ug ECRANS AA A ER EUR ANE REVE S VERE ERES 53 ImagelnfortmiatlOFic A et oerte petet veta et uec dete en cen 53 Sa cmi Zeta ed ees 55 lerne T 55 ENCON mE 57 hisjeel cer LE 58 CCT il fe lo T 60 Gr go f caren 60 BET ele E 61 Elena uerg eege r ege ee ee a geed egene EE 61 Rec li 61 Label ars dodo 61 DE Le PES 63 2009 2011 GyroTools LLC 3 WWW GYROTOOLS COM ReconFrame User Manual PREFACE This manual describes the functionality and the instructions for installation and operation of the ReconFrame enviroment for Philips Achieva and Ingenia systems DISCLAIMER Attention ReconFrame is a
53. ise like outer k space positions After calculating the offset it is subtracted from the data Notes The offset is calculated separately for every channel EPIPhaseCorrection Performes the EPI correction to correct for the FOV 2 ghost originating from eddy current effects Syntax r EPIPhaseCorrection Parameters used Parameter Recon EPIPhaseCorrection yes no Enables disables the EPI correction Parameter DataFormat Used to check if the right data format is passed to the function Parameter EPICorrData The linear nonlinear phase used in the EPI correction If a linear correction is used default then the phase is described by a slope and offset value For the nonlinear phase correction the full correction profile is given If this parameter is empty these values are calculated in the function itself The parameter can also be filled with user defined slopes and offsets prior to the correction Please see the example below on how to do that Parameter ReconFlags iszerofilled Parameter ReconFlags isoversampled and Parameter ReconFlags isgridded Used to check in what reconstruction state the data currently is The correction profiles have to be brought into the same state to match the imaging data Parameter Recon EPICorrectionMethod linear nonlin Enables disables the non linear EPI correction see below Parameter Recon EPI2DCorr yes no Enables disables the image based EPI correction see below
54. kes it very easy to create new or extend an existing reconstruction without the need to rewrite or copy paste existing code For a better illustration of this concept we have a look at an example where we create a new reconstruction object My Recon and extend it by implementing an additional ringing Hamming filter Furthermore we overwrite the existing Perform function of the MRecon class with our own one The new Perform function should only deliver filtered k space data instead of performing a full reconstruction Create an own reconstruction object My Recon by deriving it from MRecon classdef My Recon MRecon properties No additional properties needed end methods function MR My Recon filename Create an MRecon object MR upon creation of My Recon MR MR MRecon filename end Overload overwrite the existing Perform function of MRecon function Perform MR sReconstruct only standard imaging data MR Parameter Parameter2Read typ 1 Produce k space Data using existing MRecon functions MR ReadData MR DcOffsetCorrection MR PDACorrection MR RandomPhaseCorrection 2009 2011 GyroTools LLC 18 WWW GYROTOOLS COM ReconFrame User Manual end MR MeasPhaseCorrection MR SortData MR GridData Filter k space data using an own function defined in My Recon MR HammingFilter Definition of the Hamming filter function function HammingFilter MR end end These methods Prerequisites Ch
55. l resolution of the reconstructed image in y phase encoding direction ZReconRes The final resolution of the reconstructed image in z phase encoding direction KxOversampling The oversampling factor in measurement direction KyOversampling The oversampling factor in y phase encoding direction KzOversampling The oversampling factor in z phase encoding direction FFTShift Specifies if the image should be shifted after the Fourier transformation or not The 3 entries correspond to x y z direction Please note that this enables disables only the check if the image is shifted or not If the image has a symmetric X Y ZRange than it is never shifted even when the FFTShift is enabed Possible values for FFTShift are 0 1 FFTDims Specifies in which directions the data is Fourier transformed The 3 entries correspond to x y z direction Possible values for FFTDims are 0 1 Recon Defines various parameters which are used during the reconstruction process These parameters are used when the Perform function of MRecon is called Entire reconstruction steps can be switched on and off and parameters for single steps can be set DcOffsetCorrection Switches the dc offset correction on or off Possible values are yes no PDACorrection Switches the profile dependent amplification correction on or off Possible values are yes no RandomPhaseCorrection Switches the random phase correction on or off Possible values are yes no MeasPhaseCorrec
56. lated by dividing the current FOV by the current matrix size Therefore if you are performing your own reconstruction and are not using the built in MRecon functions 2009 2011 GyroTools LLC 49 WWW GYROTOOLS COM ReconFrame User Manual e g RemoveOversampling always make sure that Parameter Scan curFOV corresponds to the current FOV of your data when calling the Transform function WriteExportedRaw Exports the data to Philips exported raw format Syntax datafile listfile r WriteExportedRaw Filename Parameter2Write Location k space image space Formats Raw ExportedRaw Description Algorithm Exports the images in the data array to Philips exported format data list file pair The Parameter2Write which have to be given as input is usually the current Parameter2Read struct r Parameter Parameter2Read WritePar Exports the parameter to a Philips par file Syntax r WritePar filename Parameters used Parameter Scan Holds the general parameters uper part of the par file Parameter Imagelnformation Holds the individual parameters for every image lower part of the parfile Location image space Formats Raw ExportedRaw Cpx Rec Description Algorithm Exports the parameters in to a Philips par file WriteRec Exports the data to Philips rec format Syntax r WriteRec filename Parameters used Parameter Recon ExportRECImgTypes M P R I Specifies the i
57. le Set Path see image below Add the MRecon installation directory to the Matlab path see image below Execute the following command in the command line MRecon MachinelD Send the displayed MachinelD to GyroTools to obtain a license martin buehrer gyrotools com Copy the separately provided license file license lic to the license directory MRecon installation path license and overwrite the existing file in that directory 2009 2011 GyroTools LLC 7 WWW GYROTOOLS COM ReconFrame User Manual Open Selection per nex Close ter DCH es te Med om new 20941 10 SS EE 570111 0933 Seve te 070131 0933 Senat OT D 290030 18a 040111 1429 Pb ContigurtionorSeanrutaconm o RD snae Sonata Conti NA poe pen Data 3 ege A ld m o O IS orx figure plot abs som 21 Data 1 figure plot abs sum 21 Peter shove sli image slide r Dara ind ide ri Date s 1r rend Ade r3 Data 1 214 1end 311de et Data 3 4 end Ade 73 Data 1 4 41en4 trendi Ready 2009 2011 GyroTools LLC 8 WWW GYROTOOLS COM RECONSTRUCTION OVERVIEW ReconFrame User Manual An overview over the reconstruction can be observed in the chart below The corresponding function names in MRecon are displayed on the right side in italic letters Note that all the different datatypes Raw Exported Raw Cpx Rec shown in the chart can be read and processed with
58. les instead of the 97 as shown in the Parameter2Read struct The reason for that is that Halfscan was enabled in the current scan and the sort function already places 2009 2011 GyroTools LLC 14 WWW GYROTOOLS COM ReconFrame User Manual the data correctly in the k space and fills the rest with zeros This is seen best if we visualize the data As we can see from the plot above we actually measured 97 profiles asymmetrically in k space Please note that the recframe patch has to be used for MRecon to know the correct sampling pattern Also note that the data is still unusable in this stage To reconstruct meaningful images several corrections have to be performed Most importantly the Random Phase Correction Please refer to the Reconstruction Overview section for more information about the reconstruction steps Performing a Complete Reconstruction To perform a complete reconstruction we more or less follow the reconstruction steps shown in the section Reconstruction Overview The easiest way to do so is to call the Perform function of MRecon which executes a compete reconstruction However before calling the function we Reset the Parameter property of MRecon to undo all the changes we did in the last section e g reconstruct all coils not just the first two r Parameter Reset r Perform r MRecon handle Properties Parameter 1x1 MRparameter Data 256x256 single Methods Events Superclasses After callin
59. mages which should be exported Magnitude Phase Real part Imaginary part Parameter Imagelnformation Rescalelntercept Parameter Imagelnformation RescaleSlope Defines the scaling of the images in the rec file If these values are empty default then the scaling is calculated automatically in the WriteRec function For details see below Location image space 2009 2011 GyroTools LLC 50 WWW GYROTOOLS COM ReconFrame User Manual Formats Raw ExportedRaw Cpx Rec Description Algorithm Exports the images in the data array to Philips rec format During the export the data is rescaled according to Magnitude data Real part Imagenary part Value in Rec File round Value in r Data RescaleIntercept RescaleSlope Phase data Value in Rec File round 1000 Phase in r Data RescaleIntercept RescaleSlope Notes When the exported rec data is reimported into the scanner database the values are rescaled such that they correspond the original values in r Data before exporting WriteXMLPar Exports the parameter to a Philips XML par file Syntax r WriteXMLPar filename Parameters used Parameter Scan Holds the general parameters uper part of the par file Parameter Imagelnformation Holds the individual parameters for every image lower part of the par file Location image space Formats Raw ExportedRaw Cpx Rec Description Algorithm Exports the parameters in to a Philips XML par
60. n Parameter Labels GeoCorrPars Holds the correction factors Geometry parameters The correction has to be performed in the scanner fixed xyz coordinate system Therefore all necessary geometry parameters for the coordinate transformation Transform function have to be available Recon Flag Parameter ReconFlags isgeocorrected Location mage space Formats Raw 2009 2011 GyroTools LLC 34 WWW GYROTOOLS COM ReconFrame User Manual Description Algorithm Imaging gradients are not perfectly linear over the whole possible FOV The further away we are from the isocentre the stronger the non linear effects become which will result in a slight distortion of the resulting image This effect is corrected with the correction parameters provided by the scanner Notes The correction parameters are scanner specific and vary with the gradient system GridData GridData Grids k space data on a predefined trajectory Syntax r GridData Parameters used Parameter Recon Gridding yes no Enables disables the gridding Parameter Gridder The gridder parameter Parameter Labels NusEncNrs The non uniform sample position for EPI scans In EPI acquisitions the samples in readout direction are not uniformly spread due to the sampling during gradient ramp up Therefore EPI scans are usually gridded in readout direction using the k space positions specified here Recon Flag Parameter ReconFlags isgridded Location
61. n investigational tool and is not certified for use in clinical diagnosis Operation of ReconFrame on the scanner requires a clinical science research agreement with Philips Healthcare The software comes with no warranties In particular image quality may be compromised by defects and improper use of the software Any consequences of the use of this software are the sole responsibility of the end user LEGAL NOTICES This manual is copyrighted and protected by worldwide copyright laws and treaty provisions No part of this manual may be copied reproduced modified published or distributed in any form or by any means for any purpose without prior written permission of GyroTools LLC All brand product and company names mentioned in this manual are trademarks or registered trademarks of their respective owners 2009 2011 GyroTools A WWW GYROTOOLS COM ReconFrame User Manual INTRODUCTION AND SYSTEM OVERVIEW ReconFrame is a software platform providing various tools and all the functionality required to develop and perform a complete reconstruction of MR data All available Philips MR data formats can be read reconstructed and exported Using Matlab as programming environment data reconstruction procedures can be designed and tested offline A compiled version of such reconstructions can also be installed directly on the scanner and run instead or in parallel to the standard Philips reconstruction Images reconstructed with ReconFrame on th
62. n the profile labels and the phase which is subtracted is given by Let ri be the random phase label of the i th profile ri Parameter Labels Index meas phase i then the phase subtracted to that profile is given by phase 2 pi double intmax uint16 ri ReadData Reads all supported data formats into Matlab Syntax r ReadData Parameters used All formats Parameter DataFormat Used to check which data format is passed to the reader Parameter Parameter2Read Specifies what data is read from the file Parameter Filename ijk 2009 2011 GyroTools LLC 42 WWW GYROTOOLS COM ReconFrame User Manual The filename of the data and parameter file to be read Parameter Labels The data labels which hold the image attributes for every profile Raw and ExportedRaw data Parameter Encoding KxRange The sampled k space range in readout direction used to place the data correctly in k space e g in partial echo acquisitions Please note that for data which has to be gridded e g radial spiral EPI data this parameter specifies the k space range after gridding Parameter Recon ArrayCompression Enables the array compression according to Buehrer M Pruessmann K P Boesiger P and Kozerke S 2007 Array compression for MRI with large coil arrays Magnetic Resonance in Medicine 57 1131 1139 The individual profiles are compressed to the userd defined number of virtual coils immediately after reading
63. ntains of standard data first row phase correction data third row and noise data fifth row Let s assume we only want to reconstruct standard imaging data To do so we modify the typ field and read the data again r Parameter2Read typ 1 r ReadData r MRecon handle Properties Parameter 1x1 MRparameter Data 320x194 single Methods Events Superclasses Since we now read only one image type the Data property is no longer a cell but a simple 2 dimensional matrix The first dimension corresponds to the measurement direction featuring 320 samples The second dimension corresponds to the number of loaded profiles which are sorted in the order they have been measured To sort the data into a matrix where each dimension corresponds to an imaging parameter we call the sort function r SortData r z MRecon handle Properties Parameter 1x1 MRparameter Data 4 D single Methods Events Superclasses The Data property has now been reshaped to a multidimensional array with size size r Data ans 320 156 1 2 The dimensions of the array are predefined as x y z coils dynamics cardiac phases echoes locations mixes extr1 extr2 averages This means our dataset contains 320 samples 156 phase encoding profiles and 2 coils we only loaded the first 2 coils as specified earlier in the Parameter2Read struct You might wonder why we got 156 phase encoding profi
64. omitantFieldCorrection Performs a concomitant field correction on the current data Syntax r ConcomitantFieldCorrection Parameters used Parameter Recon ConcomitantFieldCorrection yes no Enables disables the concomitant field correction Parameter Labels ConcomFactors holds the correction factors Geometry parameters The correction has to be performed in the scanner fixed xyz coordinate system Therefore all necessary geometry parameters for the coordinate transformation Transform function have to be available Recon Flag Parameter ReconFlags isconcomcorrected Location mage space Formats Raw Description Algorithm Maxwell s equations imply that imaging gradients are accompanied by higher order spatially varying fields concomitant fields that can cause artifacts in MR imaging The lowest order concomitant fields depend quadratically on the imaging gradient amplitude and inversely on the static field strength On the scanner these extra magnetic fields are expressed along the directions of the physical gradients scanner fixed xyz coordinate system Therefore the images have to be transformed into that system before the correction The correction itself adds a quadratic phase to the images defined by the correction parameters Notes The concomitant field correction will not work on data measured without the ReconFrame patch The concomitant field correction is only executed for phase contrast flow sc
65. onment only has to be installed once on the scanner 2 Reboot the scanner console 3 Copy ScannerRecon exe and license lic to G Site AvailablePatches ReconFrame pride ScannerRecon 2009 2011 GyroTools LLC 23 WWW GYROTOOLS COM 4 Startthe ReconFrame Patch Using the Matlab Deployment Tool Another way to compile and pack the necessary files is to use the Matlab Deployment Tool The Deploy Tool is a graphical user interface for compiling and packaging Matlab programs The tool can be started over the Matlab Start button Details Workspace Ta gj El y stack Name Valu 4 ans 0 H build 91 st dest path XM Hfid 3 i 2 ab ibt path XN d loop 3 H main 3 H minor 0 new MR Or ld 4 MATLAB D Toolboxes a Shortcuts D SP Desktop Tools Web D Get Product Trials Check for Updates By Preferences i Find Files Help 9 Demos Import Wizard Profiler GUIDE GUI Builder Notebook Plot Tools Time Series Tools Help Demos MATLAB Central Web Product Page Web MATLAB Builder EX MATLAB Builder JA MATLAB Builder NE D MATLAB Compiler 4 MATLAB Report Generator EI 9 Q EI EI E U iv E SystemTest 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 ReconFrame User Manual r WriteRec try copyfi copyfi catch end end Deployment Tool deploytool Help Product Page Web K IS x pon m pn normal3 lab ata
66. ows in the following encoding parameters KxRange Specifies the k space sampling range in measurement direction Usually the sampling is symmetric but can become asymmetric in case of partial echo for example KxRange is an array with dimensions NrMixes NrEchoes x 2 KyRange Specifies the k space sampling range in y phase encoding direction Usually the sampling is symmetric but can become asymmetric in case of half scan for example KyRange is an array with dimensions NrMixes NrEchoes x 2 Dependencies ky labels are updated KzRange Specifies the k space sampling range in z phase encoding direction KzRange is an array with dimensions NrMixes NrEchoes x 2 Dependencies ky labels are updated XRange Specifies the range in k space image space along measurement direction after k space zero filling and oversampling removal Note that images in that stage already have isotropic voxel sizes in the y x plane An asymmetric range means that the image should be shifted after Fourier transformation The amount of the shift should be chosen such that the range becomes symmetric around O after shifting YRange Specifies the range in k space image space along y phase encoding direction after k space zero filling and oversampling removal Note that images in that stage already have isotropic voxel sizes in the y x plane An asymmetric range means that the image should be shifted after Fourier transformation The amount of the shift should be chos
67. possible to perform a SENSE or CLEAR reconstruction afterwards since the phase information will be lost Therefore we suggest to execute it in image space after SENSEUnfold which works in any case Perform Performs a complete reconstruction for imaging Only standard data typ 1 is read and reconstructed Furthermore if not enough memory is available the data is reconstructed in chunks of 2009 2011 GyroTools LLC 41 WWW GYROTOOLS COM ReconFrame User Manual x y z coils The complete code for this function can be found earlier in this manual or in the file StandardRecon m in the example directory RandomPhaseCorrection Corrects for the random phase which is added to the measured profiles Syntax r RandomPhaseCorrection Parameters used Parameter Recon RandomPhaseCorrection yes no Enables disables the random phase correction Parameter DataFormat Used to check if raw data is passed to the function Parameter LabelLookupTable Defines the profiles which are currently stored in the Data array Parameter Labels Index random phase Holds the random phases which are subtracted from the profiles Recon Flag Parameter ReconFlags israndphasecorr Location k space Formats Raw Description Algorithm The receiver adds a random phase to every acquired profile to reduce artifacts This random phase has to be subtracted in reconstruction since the phase is used for spatial encoding The added phase is stored i
68. r The SENSE reduction factor Recon Flag Parameter ReconFlags isunfolded Location image space Formats Raw ExportedRaw Description Algorithm Please see the MRsense manual for more infomation on the implemented algorithms and examples ScaleData Clips the data such that 9896 of its histogram is preserved Syntax r ScaleData Parameters used None Location mage space Formats Raw ExportedRaw Cpx Rec Bruker Description Algorithm A histogram is calculated over all values in the Data array and the value is found which includes 9896 of the histograms integral The data is then clipped at that value This function is usefull to remove unwanted signal peaks ShowData Displays the Data array using MRecon s built in 3D image viewer Syntax r ShowData Parameters used None Location k space Image space Formats Raw ExportedRaw Cpx Rec Bruker Description Algorithm Opens the built in 3D image viewer of MRecon and displays the Data array Notes The MRecon image viewer is a standalone tool and can be used to visualize any Matlab matrix It can be called as image slide Matrix r ShowData therefore is aequivalent to image slide r Data 2009 2011 GyroTools LLC 47 WWW GYROTOOLS COM ReconFrame User Manual SortData Sorts the acquired profiles by their image attributes and generates proper k space data Syntax r SortData Parameters used All formats Parameter R
69. s chunks Every time a chunk is defined MRecon automatically calculates how many of these data blocks occur in the raw file Def Defines the chunk by specifying which parameter of the raw file should be read and reconstructed at once The chunk is defined via a cell array consisting of several strings defining the parameter Hereby the notation is the same than in the Parameter2Read struct To reconstruct blocks of x y z coils for example set the Chunk Def to ky kz chan Possible values for Def are ky kz chan dyn phase loca extr1 extr2 mix aver echo Dependencies Chunk CurLoop Chunk NrLoops Chunk Loops Parameter2Read are updated CurLoop A number referring to the chunk which is currently reconstructed CurLoop cannot be larger than Chunk NrLoops Every time CurLoop is changed the Parameter2Read struct is automatically updated such that the desired chunk is read the next time ReadData is called Dependencies Parameter2Read is updated NrLoops Specifies how many chunks defined in Def occur in the raw data Loops Is a similar structure as the Parameter2Read struct but contains only fields which are not in the chunk definition and have more than one entry The sum over all entries in this struct corresponds to the number of loops NrLoops Scan Contains basic scan parameters which can be used in the reconstruction Please Note that the
70. s if both phase and magnitude images are stored in the recfile the original k space can be restored The floating point value is thereby given as floating point value value in recfile ScaleSlope WindowCenter Used for clipping of the images on the scanner console All values of the image which are larger or smaller than WindowCenter WindowWidth 2 are clipped when displayed on the console WindowWidth Used for clipping of the images on the scanner console All values of the image which are larger or smaller than WindowCenter WindowWidth 2 are clipped when displayed on the console Angulation The angulation of the specified image Offcentre The offcentre of the specified image SliceThickness The slice thickness of the specified image 2009 2011 GyroTools LLC 54 WWW GYROTOOLS COM ReconFrame User Manual SliceGap The slice gap of the specified image Orientation The orientation of the specified image The orientation is encoded in numbers where 1 Transversal 2 Sagital 3 Coronal AcqVoxelSize The acquired voxel size of the specified image RecVoxelSize The reconstructed voxel size of the specified image EchoTime The echo time of the specified image TriggerTime The trigger time of the specified image FlipAngle The flip angle of the specified image TurboFactor The turbo factor of the specified image Chunk To be memory efficient the data can be read and reconstructed individually in portion
71. t with a golden angle scan r MRecon golden angle scan raw 2009 2011 GyroTools LLC 35 WWW GYROTOOLS COM ReconFrame User Manual Calculate the total number of phase encoding profiles nr profs r Parameter Scan Samples 2 r Parameter Scan Samples 3 Set the radial angles to an increment of 111 246 r Parameter Gridder RadialAngles 0 pi 180 111 246 nr profs 1 pi 180 111 246 Perform the recon r Perform 2 Define a completely new trajectory and weights In this example we mirror the final image in readout direction by gridding a cartesian scan to a inverted kx trajectory To do that we define a completely new trajectory and pass it to the gridder parameters Create a new reconstruction object using a cartesian scan r MRecon cartesian raw Obtain the number of samples in every direction no kx samples r Parameter Scan Samples 1 no ky samples r Parameter Scan Samples 2 no kz samples r Parameter Scan Samples 3 Create a regular cartesian grid These are the sampled k space coordinates of a cartesian scan By definition the k space coordinates of a sampled trajectory are in the range of floor n 2 ceil n 2 1 where n is the number of samples kx ky kz ndgrid floor no kx samples 2 ceil no kx samples 2 1 floor no ky samples 2 ceil no ky samples 2 1 floor no kz samples 2 ceil no kz samples 2 1 Modify the k space grid by inverting the readout direction kx inv
72. tallation path ScannerRecon 2 If any additional files are needed for the compilation c code Matlab figures etc they can be added under Add files directories Usually nothing has to be added 3 When finished click on Build When the compilation is complete click on the Package tab n 4h Deployment Tool i File Edit Project Debug Desktop Window Help EJ ScannerRecon prj Build Package Package 162 MB H G 17 license lic E ScannerRecon exe 1 Click on Add files directories and add the license file license lic located in MRecon installation path license 2 Click on Add MCR 3 When finished click on Package and chose a location for the file to be saved 2009 2011 GyroTools LLC 25 WWW GYROTOOLS COM ReconFrame User Manual After completion copy the resulting file ProjectName pkg exe e g ScannerRecon pkg exe in the selected output directory to a memory stick or copy it to the scanner over the net Installation on the Scanner To install the reconstruction on the scanner execute the following steps 1 Plugin the Memory Stick and double click on ProjectName pkg exe e g ScannerRecon pkg exe 2 Waituntil the files are unpacked and the runtime environment has been installed 3 Reboot the scanner console 4 Copy ScannerRecon exe and license lic to G Site AvailablePatches ReconFrame pride ScannerRecon 5 Start the ReconFrame
73. teInfoPars MR Parameter Recon AutoUpdateInfoPars no o o 5 Define new counter counter Counter Performing Recon Chunk d d Loop over all chunks for cur loop 1 MR Parameter Chunk NrLoops 2009 2011 GyroTools LLC 16 WWW GYROTOOLS COM ReconFrame User Manual 9 Update Counter if strcmpi MR Parameter Recon StatusMessage yes counter Update cur loop MR Parameter Chunk NrLoops end Set the chunk loop which automatically determines the Parameter2Read struct for the current chunk MR Parameter Chunk CurLoop cur loop MR ReadData MR RemoveOversampling MR PDACorrection MR DcOffsetCorrection MR RandomPhaseCorrection MR MeasPhaseCorrection MR SortData MR GridData MR RingingFilter MR ZeroFill MR K2IM MR EPIPhaseCorrection MR K2IP MR GridderNormalization MR SENSEUnfold MR PartialFourier MR DivideFlowSegments MR CombineCoils MR Average MR GeometryCorrection MR RemoveOversampling MR ConcomitantFieldCorrection MR ZeroFill MR FlowPhaseCorrection MR RotateImage Perform the Reconstruction for the Current Chunk End The current chunk is now reconstructed If the data is reconstructed in more than one chunk write the result to a temporary file on the disk if MR Parameter Chunk NrLoops gt 1 exported datafile exported listfile MR WriteExportedRaw MR Parameter Filename Data temp data MR Parameter Parameter2Read end end
74. ters used Parameter Recon PDACorrection yes no Enables disables the PDA correction Parameter DataFormat Used to check if raw data is passed to the function Parameter LabelLookupTable Defines the profiles which are currently stored in the Data array MR Parameter Labels PDAFactors Holds the PDA correction factors MR Parameter Labels Index pda index Specifies which correction factor should be taken for a certain profile Recon Flag Parameter ReconFlags ispdacorr Location k space Formats Raw Description Algorithm n some cases the receiver weights certain k space profiles differently than others In 3D imaging for example the outer k space lines are amplified compared to the central 2009 2011 GyroTools LLC 40 WWW GYROTOOLS COM ReconFrame User Manual ones to ensure a sufficient signal range Of course the amplification has to be corrected in reconstruction if not the edges in the image would be enhanced The amplification factors are provided by the scanner and this function simply divides the profiles by them PartialFourier Performs a partial fourier reconstruction homodyne method Syntax r PartialFourier Parameters used Parameter Recon PartialFourier yes no Enables disables the partial fourier reconstruction Parameter DataFormat Used to check if the right data format is passed to the function Parameter ReconFlags isimspace Used to check if the data is in k or ima
75. the cardiac phases to be read echo Defines the echoes to be read loca Defines the locations to be read Locations depict either different slices in M2D or MS scans or different stacks in 3D acquisitions chan Defines the coils channels to be read extr1 Defines the extr1 attribute to be read extr1 depicts different types of data depending on the scan In flow imaging the different flow encoding directions segments are assigned to the extr1 label extr2 Defines the extr2 attribute to be read Extr2 depicts different types of data depending on the scan ky Defines the y phase encoding profiles to be read kz Defines the z phase encoding profiles to be read aver Defines the averages to be read rtop Defines the trigger delay range to be read The trigger delay of a certain profile is defined as the time between the ECG trigger and the acquisition of that profile ImagelInformation Provides information for every image in the Data property of the MRecon object such as image angulations offcentres resolution etc The Imagelnformation property is a struct array with the same dimensions as the Data property To access information about a specific image in the Data property use indexing like Parameter Imagelnformation 5 or Parameter Imagelnformation 1 2 5 3 1 In the first case information about the 5 th image is provided and in the second case information from the first slice second coil gn dynamic etc 2009 2
76. tion Switches measurement phase correction on or off Possible values are yes no PartialFourier Switches the partial fourier reconstruction partial echo half scan on or off Possible values are yes no PartialFourierNrlterations Specifies the number of iterations for the partial fourier reconstruction A larger value results in a longer reconstruction If the value is set to 1 then the reconstruction is time dependent and does not take longer than 20 seconds Possible values are 1 any integer number lager than 0 kSpaceZeroFill 2009 2011 GyroTools LLC 58 WWW GYROTOOLS COM ReconFrame User Manual Switches the k space zero filling on or off Possible values are yes no EPIPhaseCorrection Switches the EPI phase correction on or off Possible values are yes no CoilCombination Specifies the coil combination type At the moment only a sum of squares reconstruction is implemented Possible values are sos sum of squares ImageSpacezZeroFill Switches the image space zero filling on or off Possible values are yes no Rotatelmage Switches the image rotation on or off Possible values are yes no ArrayCompression Enables Array Compression which is a method for k space coil precombination to save memory and reconstruction time A detailed description can be found here Buehrer M Pruessmann K P Boesiger P and Kozerke S 2007 Array compression for MRI with large coil arrays Magnetic Resonance in Medicine 57
77. tion Algorithm When the gridder parameter are not defined by the user they are calculated internally and will not be visible or changeable If one wants to observe or change the gridding parameter this function can be called which calculates the trajectory for the gridding process and fills the gridder parameter accordingly The function can be called for any preset scan radial spiral EPI before calling the GridData function This is useful if the user wants to modify a preset trajectory without calculating it by himself To do so one would call this function modify the gridder parameter followed by the call of GridData GridderNormalization Corrects for the signal weighting in image space originating from the convolution with a bessel kernel in the gridding process Syntax r GridderNormalization Parameters used Parameter Gridder 2009 2011 GyroTools LLC 37 WWW GYROTOOLS COM ReconFrame User Manual The gridder parameter Parameter ReconFlags isgridded Used to check if the data was gridded Parameter Encoding WorkEncoding KxOversampling The resulting oversampling after gridding needed in the normalization Location mage space Formats Raw ExportedRaw Description Algorithm Gridding is essentially a convolution with a bessel kernel in k space In image space this corresponds to a multiplication with the fourier transform of the bessel kernel leaving a signal weighing on the reconstructed image This
78. ucted You can either pass both the parameter and datafile r MRecon raw data file lab raw data file raw or you can pass just one of these files and MRecon tries to find the corresponding data parameter file automatically MRecon raw data file raw r or MRecon raw_data_file lab r Please note that when you pass both filenames the parameter file always has to be the first argument The above commands create a reconstruction object of a raw dataset Thereby MhRecon automatically detects the data format of the dataset Therefore to read and process reconstructed REC datasets the same command can be used r MRecon rec data file rec Calling MRecon without arguments will open a file dialog where the MR datafile can be chosen manually The initial path of the file dialog will be the current directory of Matlab r MRecon ia F gt gt bl Ah Pick a file y j e Tr e x Suchen in on EI ek E s Name o Anderungsdatum Typ GroBe s E era A rf 07012011 1518190 2 1 wipsagapfV42 rec 07 01 2011 15 22 VLC media file rec 128 KB E f 07012011 1518190 2 2 wipsagapfV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB Ww A rf 07012011 1519000 3 1 wipsagaphV42 rec 07 01 2011 15 22 VLC media file rec 128 KB Desktop E rf 07012011 1519000 3 2 wipsagaphV4 raw 07 01 2011 15 21 RAW Datei 1 691 KB r 07012011 1519090 4 1 wipsagfhaV42 rec 07 01 2011 15 22 VLC media file rec 128 KB
79. w Matlab Runtime Scanner Database Pack n Go Tool 2009 2011 GyroTools LLC 5 WWW GYROTOOLS COM ReconFrame User Manual INSTALLATION Prerequisites Scanner Access to the gyrotest account is required If the account is not available on your scanner ask your local Philips representative to enable it Download the Scanner Software Patch version that matches the scanner software release Example For the scanner software release R3 2 1 the installation file ReconFrame r321 exe is required For running compiled custom reconstructions on the scanner the Matlab runtime libraries must be installed Make sure to install the library version that matches the Matlab version the reconstruction was compiled with Offline Workstation On the offline workstation a full Matlab installation is required The MRecon library is compatible with Windows Linux and MacOS systems both 32bit and 64bit Recommended is Matlab 64bit along with a 64bit operating system and 8GB or more of installed RAM Required Matlab Toolboxes Image Processing Toolbox To be able to compile custom reconstructions and run them on the scanner a Windows Matlab installation is required along with the Matlab Compiler Toolbox To extend the ReconFrame parameter interface with additional scan parameter or to integrate ReconFrame into custom scanner software patches the full Philips Pulse Programing environment is required
80. w Here a 2 dimensional cartesian dataset with the following parameters is reconstructed FOV Acquired Voxel Size Reconstructed Voxel Size HalfScan Partial Echo KxRange KyRange XRange YRange XRes YRes XReconRes YReconRes KxOversampling KyOversampling 320 x 250 mm 2x1 6 mm 1 25 x 1 25 mm Yes No 160 159 19 77 128 27 200 1 256 200 256 256 2 1 Matrices are displayed as rectangles shaded grey in areas with values unequal zero acquired data and white in zero valued areas The reconstruction steps in italic correspond to the function names in MRecon and the ranges to the Encoding struct fieldnames The cross marks the center of k space SortData asueyxy KyRange 2009 2011 GyroTools LLC ZeroFill sayx adueyx SuiduesiSAQXy SIYX sayx adueyx BUIN WESIZAOXA y SIYX K21 li EE nt YRes KyOversampling ar YRes KyOversampling 6ujjdwnsianganoway ZeroFill SOYUOIIYX SOYX e8ueux el YReconRes M YRange YRes 10 WWW GYROTOOLS COM ReconFrame User Manual MRECON USAGE Online Documentation The online documentation is built into the Matlab help browser To display the MRecon Documentation from Matlab type the following into the command promt doc MRecon Object Creation Initialization The first step is to create a reconstruction object by calling MRecon and passing the filename s of the data to be reconstr

User Manual

Contents

Download Pdf Manuals

Related Search

Related Contents