CurveAlign V1.3 Beta2 User`s Manual
Contents
1. Step 4 check the results All the results are automatically saved to a new subfolder called CA_Out within the folder where the image exists In this case the path for this subfolder is C Users youmap Desktop Box Sync collagen analysis github curvelets CurveAlignTestImages input_fire_tiffbound CA_Out Eight output files are shown Compared to the output of the tutorial 1 the CT reconstructed image is not available here Bors Organize ix Open UY Favorites Ml Desktop Recent Places mi Downloads m Google Drive m Box Sync u Libraries 2 Documents am Git A aa Step 5 reset the GUI Share with a m Mew folder Mame E 1B alb_1_hist csv Gik 1B alb_1_stats csw ENI 1B alb 1 values csy Gii 1B al b_fibFeatNames csv fai 1B alb_fibFeatures csv 5 1B alb_fibFeatures mat a 16 alb_overlay titf T 1B alb_procmap titt Click Reset to start over 15 Box Sync collagen analysis github curvelets CurveAlignTestlmages input_fire_tiffhound CA Out Type Microsoft Office E Microsoft Office E Microsoft Office E Microsoft Office E Microsoft Office E MATLAB Data TIFF image TIFF image 1 KB 1 KB 1 KB 1 KB 9 KB 10 KB 193 KB 210 KB CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison 12 3 Tutorial 3 feature ranking This tutorial shows how to rank features for a training data set that includes 8 TACS 3 negative images an2. File Edit View Insert Tools Desktop Window Help ogUHU s ki AAVDL A l al 0al nana Figure 3 Map image of the overlay image showing the fibers that are within the specified distance CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison When a boundary is not selected the raw map file translates 0 to 180 degrees into 0 to 255 in gray scale In this case red indicates areas of highly aligned structures while black and green indicate areas of more randomly aligned structures The map image is overlaid on the original image with transparency set to 0 5 as shown in Error Reference source not found E CurveAlign Fiber Overlay l E CurveAlign Angle Mop E File Edit View Insert Tools Desktop Window Help File Edit View Insert Tools Desktop Window Help DOGAMs Fk AAVVEA B 08 ao DGmdes k AAaVoed a 08 ao 7 ey Figure 2 Overlay and map images when no boundary is selected Boundary Analysis The user is able to analyze the fibers that fall only within a certain distance from a boundary To enter this distance first a boundary must be drawn or opened from file Then the distance in pixels should be entered into the edit box Enter distance from boundary to evaluate in pixels Boundaries are created by alt clicking on the original image that is opened in the CurveAlign GUI When alt is released the boundary is ended and no additional points may be added to the boundary If the bou
3. Select Output Options Run Reset CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison In this manual depending on the selected Fiber analysis method in the drop down box at the top of the GUI fiber may refer to curvelets that represent a fiber segment CT fiber segments generated by CT FIRE the extracted fiber generated by CT FIRE or extracted fiber endpoints generated by CT FIRE 2 QuickStart Instructions 1 2 3 4 5 6 T 8 9 Select the fiber processing mode from among the following options CT CT Fire Segments CT FIRE endpoints and CT FIRE Fibers Select the boundary processing mode from among the following options No Boundary CSV Boundary Draw Boundary and TIFF Boundary Click Get Images and browse to an image Images may be single or stacks If you would like to process more than one image in a batch just select more than one image in the dialog using the technique appropriate for your operating system 1 e in Windows hold CTRL while selecting multiple files If a manual boundary is required not available in batch mode hold down the Alt key and click points along the desired path on the image window Release the Alt key when finished Enter the keep threshold level if fiber extraction mode CT is selected and select the desired outputs using the checkboxes on the bottom of the main window Click the run button The output data and images wil
4. CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison hist csv List of bin values and numbers of observations in each bin Bin values correspond to relative angle in if boundaries are used and absolute angle if no boundary is used relative angles as in version 2 3 using the nearest distance the defined the region of interest Statistics of the relative angles as in version 2 3 Feature ranking output including feature importance and normalized feature difference between two classes 8 Output images A few output images are saved in tiff format in the selected output directory These images are explained below 8 1 Overlay image This image shows the position of each fiber detected within the particular distance from the user defined boundary The fibers outside the specified distance are also plotted for File Edit View Insert Tools Desktop Window Help nddl RI ABRVDLA S20 a r P tiraji Wi wiry K he he 4 AL AP LMMEBEL ANDO ReRE Ka 4 siin Figure 1 Overlay image showing the fibers that are within or outside the specified distance The green lines show the fibers inside the specified distance The red lines show the fibers outside the specified distance The boundary is the line connecting the yellow mark Blue lines are the lines connecting the centers of the fibers with the nearest points on the user defined boundary CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison 8 2 Map Image The map im
5. Fiber processing modes There are four fiber processing modes To avoid errors ensure that the previously opened xls files of results of an image are closed before any other operation is performed on the same image 1 2 3 4 Curvelet Transform The curvelet transform is performed on the image s and each curvelet corresponds to an observation in the feature list Each curvelet has a unique angle and position In this case fiber length and curvature are not available since curvelets are independent from one another CT FIRE Segments The output from CT FIRE is used as the input providing information about the fibers in the image Each fiber segment in the CT FIRE output file corresponds to an observation in the feature set Each segment has a unique angle and position In this case each segment is given a fiber length and curvature value that corresponds to the length and curvature of the entire fiber this segment belongs to For this option the mat file generated by CT FIRE must be saved in the same folder as the original tif image and when prompted the tif image should be selected CT FIRE Fibers The output from CT FIRE is used as the input providing information about the fibers in the image Each fiber center point in the CT FIRE output file corresponds to an observation in the feature set Each center point has a unique angle corresponding to the average angle of the fiber The position is the midpoint between the fiber end
6. dist curvature box density 128 width box alignment 64 box alignment 32 box alignment 128 box density 64 nearest relative boundary angle total length extension point angle box density 32 extension point distance mean nearest align 0 0 2 0 4 0 6 0 8 Classification Importance 2 Feature Normalized Difference Pos Neg File Edit View Inset Tools Desktop Window Help OSGas REF r N E Sensitivity 1 00 Specificity 1 00 inside epi region std nearest dist mean nearest dist box alignment 128 box alignment 64 box alignment 32 extension point angle mean nearest align curvature earest relative boundary angle total length std nearest align width extension point distance nearest dist to bound box density 32 box density 64 box density 126 0 0 5 Normalized Difference Pos Neg The ranking results are saved in a txt file named featRank txt which is saved in the same folder as the mat feature files 13 Status Label To allow the user to keep track of what is happening in the program there is a status label on the bottom of the main window This label gives hints about what the user should do next or shows the current task that the program is working on Analyzing boundary 14 Tips 1 Output data and images are saved under a subfolder named CA_Out within the folder where the original image exists and under filenames indicating the source of the data and the type of output 2 T
7. for 1B a5 tif tif 1B al tif tif 1B a5 tif tif Select Output Options File name 1B a1b tif bd Open Run Reset Files of type tif tiff jpg jpeg hd Cancel or oo ay Step 2 select run parameters and set output options After clicking Get Images the image is displayed and new controls are enabled on the control panel as shown below and the information shown in the command window Found ctFIREout_1B alb mat Found mask for 1 B alb tif tif indicates the CT FIRE output file and Tiff boundary file were found The region of interest 1s set to the area within the 300 pixels from the boundaries All the output options are checked Le lame Som 7 CurveAlign V3 0 Beta2 o By Cinediign Figure f CT FIRE Fibers Fiber analysis method Boundary method 1B a1b tif Sel Output Options ons V Feature List Histogram v Overlay Output Val V Map Output V Bdry Assoc Run Reset i 14 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison Step 3 run fiber feature extraction click Run button Then the progress will be listed in the Command Window The overlaid and map images will also be displayed as follows D CurveAlign Fiber Overlay File Edit View Insert Tools Desktop Window Help wisi ERRER ETA O i aulliex B curveAlign Angle Map File Edit View Insert Tools Desktop Window Help URE a2 08 x U del hk boba
8. images or fibers using machine learning techniques An Support Vector Machine SVM is incorporated into CurveAlign for ranking extracted features and these features can also be used or ranked through many open source machine learning tools such as Weka and R The primary change in CurveAlign Beta version 3 0 is feature extraction from CT FIRE output Compared to the version 2 3 the major updates are 1 output up to thirty four fiber features including angle alignment density and etc these features are saved in both cvs file and mat file which can then be used to potentially classify images or fibers using machine learning techniques 2 tiff boundary can be loaded to investigate the interaction between fiber and cell boundary 3 CT FIRE fiber extraction results can be read in 4 a feature ranking using support vector machine SVM is included and 5 keep the most features of the older version 2 3 such as the statistics for the relative angles heat map to show the fiber angle alignment and the fiber boundary association etc The GUI in CurveAlign is modular so that the main user interface is in a separate window from the outputs This allows for the user to resize the output windows to their preferred size The main user interface window is shown below BJ Curvedlign V3 0 Beta ae cT Fiber analysis method No Boundary Boundary method Get Image s None Selected Feature Ranking
9. 18 19 20 21 22 23 24 25 6 8 9 14 18 23 32 Feature List Overlay Output CA Features List 1 fiber Key into CTFIRE list 13 dist to nearest 16 2 end point row 3 end point col 4 fiber abs ang 5 fiber weight 6 total length 7 end to end length 8 curvature 9 width 10 dist to nearest 2 11 dist to nearest 4 12 distto nearest 8 17 25 box alignment 32 14 mean nearest dist 26 box alignment 64 15 std nearest dist 27 box alignment 128 16 box density 32 28 nearest dist to bound 17 box density 64 29 inside epi region 18 box density 128 30 nearest relative boundary angle 19 alignment ofnearest2 31 extension point distance 20 alignment ofnearest4 32 extension point angle 21 alignment ofnearest8 33 boundary point row 22 alignment of nearest 16 34 boundary point col 23 mean nearest align 24 std nearest align CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison After clicking OK button on the feature selection window feature ranking starts and the two output windows are displayed as below including both the feature classification importance and feature normalized difference between Negative and Positive classes Feature Classification Importance File Edit View Insert Tools Desktop Window Help Aisi E AAC902H4 42 0H an Sensitivity 1 00 Specificity 1 00 std nearest align nearest dist to bound inside epi region std nearest dist mean nearest
10. A_Out and will include all of the outputs available from the CurveAlign software 10 1 Batch mode with CSV boundaries If boundaries are drawn with the CurveAlign program they should be saved in the same directory as the images The boundary files should be named like the following Image file name TACS 3a jpg Boundary file name boundary for TACS 3a jpg csv 10 2 Batch mode without boundaries 10 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison To process a directory of images without boundaries then just place the images in a directory by themselves without any boundary files then run CurveAlign with select all of the images in that directory In this case the distance from the boundary and boundary association edit boxes will be ignored 10 3 Batch mode with Tiff boundaries In batch mode boundaries may also be imported as tiff files Tiff boundary files must be 8bit binary mask images where the inside of a region has a value of 255 and everything else must be 0 Boundary tiff files must have the same number of pixels length and width as the original image and may be produced by hand or by segmentation in ImageJ FIJI Tiff boundary files should be named according to the following convention Image file name 1B a5 tif Boundary file name mask for 1B a5 tif tif 10 4 Batch mode with CT FIRE input To compare the results of the CT FIRE software to boundaries either TIFF or csv then choose one of the CT FIRE f
11. CurveAlign V1 3 Beta2 User s Manual AAO CC OM scorch sarc E acess Saree E E ace ueacd deh aaath Neuere de 2 Z Quickstart MUC Hons grseie cherie cosande ostiote cs ui doscievnte N 3 z WRelationstiip witli CR ersari an OE tadeds temnctanseccsantanceaseloededatannctemeccimtarceastere 3 A FIDE PrOCeSSIN MOCeS 22k classe sone i Sachs aac a sacs seeds dace T 4 X Boundary ProGessine MOUES cicinico cr these la arn engine dese E dee Neca satebnc ace E Ne ceaneeoeaee eee 4 O AAE chs inane merce erate rene mere etree Ore eee ree ner ree re oes Pee eee ee ere ere eee reer eres 5 A Qupu utdata ESen T 5 S OUMU MARO S E A E E 6 8 1 Overlay ma een ert ar ee me oe arse ee ner ene Pm oor oP N Re a eRe er ere rete ter ae er ae 6 52 Map 12 per ers eng E E oer rra ae ee eee 7 OY Sak pee ettcacten eet anteater a eh ote Maloy tacit AS lta Maat lect tect Asal las 9 9 1 Stacks WHE BOUIN AICS enar T A TEE T cadens cumtia chateeretentack 10 9 2 Stacks WAtHOUU DOUNGAIICS eeii enetaixte 10 10 Bate imode sse a a aie aoe a cecee ee 10 10 1 Batch mode with CSV Dondar S enron enir E EE N 10 10 2 Batch mode without DOUNGAMES occsecs nedeaceundsnadcanceaadeunssaedennecaadanncaaaddsacsaddaneddeedenadsaadanscseddaunsaade 10 10 3 Batehamode with Tiff DOUNGATIES sa cscs5ue sess test ns cer a E a tae noes ascended acetate nde eee 11 104 Baten mode with F FIRE INPI eeann EEN EAG 11 11 Pere RAKAS sc ae E E E A N ndandeee 11 12 TO oee E E E S 11 12 1 Tutorial 1 feature extr
12. TIFF image 1 255 KB Documents ii TACS 3a_procmap tiff TIFF image 1 279 KB am Git W TACS 3a_reconstructed tiff TIFF image 286 KB a Music t Pictures F Videos a Homegroup jE Computer amp Windows7_0S C a Lenovo_Recovery Q x data ftp microscopy wisc edu V G LOCI bockdrive bocklabs wisc ed J 9 items 13 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison Step 5 reset the GUI Click Reset to start over 12 2 Tutorial 2 feature extraction with CT FIRE method and Tiff boundary Step I choose the methods of fiber extraction and boundary open an image As shown in the Figure below Select the CT FIRE Fibers as the Fiber analysis method Tiff Boundary as the Boundary method and then click the Import image data button a file selection window opens allowing the user to choose the image In this tutorial 1B alb tif is selected E CurveAlign V3 0 Beta2 Seren Mgt Select image ia P J ma 4 g CT FIRE Fibers Fiber analysis method Lookin p input fie_tffbound e aam Tit Boundary Boundary method ay Recent Places F ssh q ues 5 E p Get Image s None Selected Feature Ranking Desktop ka are 3 a 1B al tif 1B alb tif 1B a5 tif mask for mask for e 1B al tif tif 1B alb tif tif z Libraries A e g xP e Computer gt F ue i ate t a t p s E default is 001 Ene Network 2 mask for orig_mask for orig_mask
13. action with fiber extraction method CT and CSV boundary 11 12 2 Tutorial 2 feature extraction with CT FIRE method and Tiff boundary eee 14 12232 Tutol teatre TANKIN E enra E N E E E E N EN O S 16 13 Staus Lapele A 18 14 TID ae E E E E T E E EE 18 15 PREMIO Ve CS ement wancanane E E EE steseieeian tieeaseecemn a eeseesieeniane 19 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison 1 Introduction The purpose of CurveAlign is to compute features that describe collagen interactions with epithelial cells It was developed in order to search for stromal changes that are correlated with disease in images of collagen and epithelial cells We have recently used CurveAlign for the feature calculations performed in a research paper by Bredfeldt et al Journal of Pathology Informatics 2014 in press Collagen images may be obtained by a number of imaging approaches however we have focused here on Second Harmonic Generation SHG images of collagen Epithelial cell information is input into CurveAlign as an 8 bit mask file that must be pre registered with the collagen image where white pixels correspond to epithelial cell regions and black pixels correspond to anything else in the image 1 e the background collagen fibers etc These mask files can be generated by any appropriate means such as manual ROI annotation in FIJI or using segmentation tools in MATLAB or FIJI The output features can then be used to potentially classify
14. age is intended to help the user identify the spatial distribution of fiber angles within the image The raw map _rawmap tiff file codes the angle of the fiber into a grey scale value The pixel in the image where the center of the fiber is located is given a value between 0 and 255 that corresponds to O to 90 degrees when a boundary is selected and 0 to 180 degrees when a boundary is not selected This file may be further processed in matlab or image according to the users preferences The processed map file _procmap tiff is a processed version of the raw map file overlaid on the original image The output is intended to show regions of aligned structures that are perpendicular to the boundary in the case of a boundary selection or regions of generally aligned structures in the case where no boundary is selected How processed map files are created When a boundary is selected the raw map file translates O to 90 degrees into 0 to 255 in gray scale The center location of each curvelet is given a gray level corresponding to its angle with respect to the boundary Then a square max filter is applied with a size of 12 pixels on a side followed by a Gaussian disc filter with a sigma of 4 pixels The color scale is then set to 0 10 degrees black 10 45 degrees green 45 60 degrees yellow and 60 90 degrees red The map image is overlaid on the original image with transparency set to 0 5 as shown in Error Reference source not found
15. cted ao E amp A Select Image Fiber analysis method Look in input_ct_csvbound e m Far t Boundary method sn Recent Places Feature Ranking Desktop Y Y N l E stack_example_s substack_exampl __TACS 3ajpg i Libraries A Computer Select Output Options p as decimal 0 005 Reconstructed Image Histogram Values Bdry Assoc Network Feature List Overlay Output File name TACS 3a jpg Map Output Files of type tif tiff jpg jpeg hI RS Lae Se I ae a eer mn Run Step 2 select run parameters and set output options After clicking Get Images the image is displayed and new controls are enabled on the control panel as shown below and information shown in the command window Found boundary for TACS 3a jpg csv indicates the CSV boundary file was found The fraction of curvelets coefficients is set to 0 005 the region of interest is set to the area within the 150 pixels distance from the boundary All the output Reset options are checked CSV Boundary Get Image s TACS 3a jpg CurveAlign Figure Fiber analysis method Boundary method Enter fraction of coefs to keep as decimal default is 001 0 005 Select Output Options Reconstructed Image ues V Bdry Assoc Feature List Overlay Output Ss 8 Map Output 12 Cur
16. d 8 TACS 3 positive images Step I prepare annotation file The file is named annotation xlsx which contains the label for each image and must be saved in the same folder as the CA extracted feature files 1 e the foder of CA_Out The annotation file for this training images is shown below x bal File At a ni Al e E a T A Faste F Font Alignment Number Styles Cells r Fal T T Clipboard Ma Editing 0 1B_A5 11B Al 0 1B AS 0 1B B7 11B C7 0 1B C9 0 1B D 1 1B D4 0 1B F1 0 1B 11 0 2B A6 1 2B F1 1 2B F3 1 2B G10 1 2B H2 1 2B H9 17 18 19 4 4b Pl Sheet1 Sheet Sheet 4 Iil ao Average 0 5 Count 32 sum 8 0 100 wherein 0 indicates the image is TACS 3 negative indicates the image is TACS 3 positive Step 2 Run feature ranking click the Feature Ranking button all other fiber feature extraction buttons are enabled off and a window is popped up to Select fiber feature directory Here TrainingSets20131113 CA_Out is selected as shown below 16 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison 2 CurveAlign V3 0 Beta2 CT Fiber analysis method No Boundary Get Image s None Selected Boundary method ep as decimal from boundary to evaluate Select Output Options Reconstructed Image His
17. iber processing modes on the main GUI The following naming convention should be strictly used Image file name 1B al tif CT FIRE results file ctFIREout_1B al mat 11 Feature Ranking Based on the importance for differentiating two class classification problem Negative and Positive using SVM SVM feature weight is used to rank the features The average feature value difference between two classes is also included An annotation file including the label for each image is needed to be prepared before running the feature ranking This file is named as annotation xIsx with two columns where the first column includes the label for each image e g 1 for positive 0 negative the second column includes the full original image name 12 Tutorials To reproduce the tutorials users need to first download the test images and boundary files at LOCI website 12 1 Tutorial 1 feature extraction with fiber extraction method CT and CSV boundary Step I choose the methods of fiber extraction and boundary open an image As shown in the Figure below Select the CT as the Fiber analysis method CSV Boundary as the Boundary method and then click the Import image data button A file selection window opens allowing the user to choose the image In this tutorial TACS 3a jpg is selected 11 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison B R20 CurveAlign Beta V3 0 1 cT CSV Boundary Get Image s None Sele
18. l automatically be placed in a folder called CA_Out in the directory with the original image s If you have drawn a boundary you will be prompted to give a file name and location for the boundary points to be saved These will be saved as a csv file This file can be used again in the future by selecting CSV boundary as the boundary processing mode Store the boundary in the same directory as the original image with the filename as image tif csv if the name of the original image is image tif The current function being performed by CurveAlign is listed in the Status label at the bottom of the main window 3 Relationship with CT FIRE CT FIRE is another software tool developed for processing images of fibers Bredfeldt J et al Journal of Biomedical Optics 2014 CT FIRE traces fibers and outputs a database of fiber objects This output database can be used as an input to CurveAlign the subject of this manual CurveAlign s main purpose is to compare fibers with boundaries and fibers to each other to measure density and alignment When CT FIRE inputs are used the fibers are compared to region boundaries either drawn manually or created by any segmentation tool CT FIRE and CurveAlign may someday be combined but the basic philosophy is that itis useful to have them separate and easy enough to just connect them by reading the CT FIRE outputs into CurveAlign when needed CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison 4
19. lignment will still be computed and output in the feature list CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison 2 Manual boundaries The user is required to manually draw boundaries with the Alt key mouse clicking technique This option is not allowed if multiple input images are selected to be processed in a batch If this option is selected and multiple files are to be processed then the No Boundaries option is automatically selected All boundary points must be selected in a succession to select a new boundary click anywhere on the image and then select a new boundary 3 CSV boundaries Previously stored boundary files can be used that contain the x y positions of points along the boundary a CSV files must adhere to the following naming convention If the image is named TACS 3a jpg then the CSV file must be named boundary for TACS 3a jpg csv b The boundary file must be in the same directory as the original image 4 Tiff boundaries Mask files are used to indicate where region boundaries are located The mask files should be 8 bit files where the inside of ROIs are given a value of 255 and outside a value of 0 This allows for features identifying whether a fiber is inside an ROI or not These files must be in pixel for pixel registry with the image files and therefore have the same dimensions as the image files Tiff boundary files must adhere to the following naming convention If the image is named 1B a5 tif then the
20. ndary mode CSV Boundary or Tiff Boundary is selected previously saved boundary file can be automatically checked and loaded in if it exists The boundary file should be saved in the same directory as the selected image and should be named like the following Image file name TACS 3a jpg Boundary file name boundary for TACS 3a jpg csv 8 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison or Boundary file name mask for TACS 3a jpg tif The position on the boundary that is being compared to each fiber may be visualized as well This allows the user to see where the angle comparisons are being made Blue lines are drawn on the overlay output image from each fiber to the point on the boundary that the fiber is associated with File Edit View Insert Tools USGas ki ASBATBOLKA BAB O8 imi Figure 3 An example of an overlay images where the associations between fibers and the boundary are indicated by blue lines 9 Stacks CurveAlignV3 0 allows for the user to select stacks of images When a stack is opened the slider bar is enabled allowing the user to choose which image in the stack should be displayed CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison Stack image gelected 1 When a stack is being analyzed then the output files that are images will also be in stack format To open these stacks in ImageJ the LOCI bioformats importer must be used For some reason these tiff stacks cannot be d
21. o read in the output files from CT FIRE CT FIRE output mat files should be saved in the same folder as the original images 3 To read in the boundary files in csv or tif format the boundary files should be saved in the same folder as the original images and should follow the strict naming convention e g if image 18 CurveAlign V3 0 Beta2 User s Manual LOCI UW Madison is named 1B a5 tif CSV boundary is named Boundary for 1 B a5 tif csv and tiff boundary is named mask for 1B a5 tif tif 4 Tiff boundary files must be 8bit binary mask images where the inside of a region has a value of 255 and everything else must be 0 Boundary tiff files must have the same number of pixels length and width as the original image and may be produced by hand or by segmentation in ImageJ FISI 5 Tiff boundary is recommended for more accurately extracting the features of collagen cell interaction 6 To run feature ranking an annotation xlsx file should be prepared and saved in the same folder as the mat feature files This file is named as annotation xlsx with two columns where the first column includes the label for each image e g 1 for positive 0 negative the second column includes the full original image name 15 Acknowledgement Guneet Singh Mehta has contributed to this manual 19
22. points In this case each center point is given a fiber length and curvature value that corresponds to the length and curvature of the entire fiber this center point belongs to For this option the mat file generated by CT FIRE must be in the same folder as the original tif image and when prompted the tif image should be selected CT FIRE Fiber Endpoints The output from CT FIRE is used as the input providing information about the fibers in the image Each fiber endpoint in the CT FIRE output file corresponds to an observation in the feature set Each endpoint has a unique angle and position The angle corresponds to the angle of the entire fiber In this case each endpoint is given a fiber length and curvature value that corresponds to the length and curvature of the entire fiber this segment belongs to When CT FIRE outputs are used as inputs to CurveAlign the CT FIRE output files must be named according to a strict naming convention For example if the image is named the following 1B a1 tif Then the CT FIRE output file must be named the following ctFIREout_1B al mat At the time of writing this manual the above naming convention has been adopted by the CT FIRE tool as well so the outputs from CT FIRE may be directly used as inputs to CurveAlign 5 Boundary processing modes There are four boundary processing modes 1 No boundaries Fibers will be compared to each other but not to a boundary For example features about fiber a
23. rag and dropped into ImageJ this will be investigated further in the next release of CurveAlign The other output files for example the histogram file and the curvelet angle spreadsheet are produced such that each image in the stack creates a new output file For example a stack with 4 images will generate 4 histogram files named stack_1_hist csv stack_2_hist csv stack_3_hist csv etc 9 1 Stacks with Boundaries If a stack is being processed then only a single boundary can be used for the entire stack Future versions of this tool will allow for a different boundary in each image plane in a stack 9 2 Stacks without Boundaries When no boundary file is selected then each image in the stack is analyzed as described in the section above about image analysis without a boundary 10 Batch mode Batch mode is used to process multiple images and potentially associated boundaries To use batch mode click the Get Images button and select more than one image in the file selection dialog box Each image will be processed according to the modes currently selected CurveAlign will first search for boundary files in the chosen directory If there are boundary files then it will process all images that are found to be associated with the located boundary files Images may be a mixture of both individual images and stacks If the image is a stack then the entire stack will be processed The batch mode output files will be stored in a directory named C
24. tiff boundary file must be named mask for 1B a5 tif tif The boundary file must be in the same folder as the original image When boundaries are used as part of the analysis CurveAlign computes up to 2 relative angles per observation The first 1s the nearest point angle and the second 1s the intersection point angle The nearest point angle is the relative angle between the observation point and the tangent angle of the region boundary at the point nearest to the observation point The intersection point angle is the relative angle between the observation point and the intersection point of the fiber interpolated out a user defined distance and a boundary 6 Input files CurveAlign requires up to three types of input files Image files These files contain images of fibers These can be produced by any method available however we have particularly focused on SHG images of collagen fibers CT FIRE files The output of CT FIRE having been run on the above mentioned Image files Boundary files The boundary files can be CSV files or Tiff files See the section about Boundary Modes above to learn about naming conventions for these file types mat files must be present in the same directory as the original image while using the results of the CT FIRE results 7 Output data files There are several files that are generated by the CurveAlign software for each image processed These files and their descriptions are listed in the table below
25. togram Values Bdry Assoc in pixels Feature List Overlay Output Map Output Browse For Folder Select Fiber feature Directory ME Desktop b So Libraries b B yuming b jE Computer b Ci Network a CA Beta V3 0_test J CA Out di CurveAlign Beta V3 0 Testlrages a i TestimagesforCA3 0 gt di input_ct_cswbound di input_ct_nobound di Input_ct_titfbound di input_fire_nobound b di input fire_tiffbound a TrainingSets20131113 Ji CA Out dJ CA_Outold gt TrainingSets20131113 Copy b gt di CAA b m Cole Mina stain project b gt di ctf 3B2_ sourcecode e GIT_LOCIUWmadison gt jb Mc Folder CA_Out After clicking OK two windows pop up as shown below One is the CA Features List showing the index and description of each of the 34 features the other is a input dialog window which shows the applicable features and can be used to enter the number of the features to be ranked In this tutorial enter 6 8 9 14 18 23 32 as shown below 1 e features 6 28 29 1 CurveAlign V3 0 Beta2 9 14 15 16 17 18 23 24 25 26 27 30 31 32 User can input arbitrary number of features to be ranked orp E pole X CT No Boundary Get Image s Fiber analysis method Boundary method None Selected Feature Ranking aJ Select features to be ranked Select features among 1 2 3 4 5 Select Output Options Reconstructed Image Histogram 16 17
26. veAlign V3 0 Beta2 User s Manual LOCI UW Madison Step 3 run fiber feature extraction click Run button Then the progress will be listed in the Command Window The overlaid and map images will also be displayed as follows 2 CurveAlig File Edit View Insert Tools Desktop Window Help OUSGAsSs ARV 9EL AB 0H af d SN BY Cuvedlign Angle Map E o r a Sa File Edit View Inset Tools Desktop Window Help Aima E e OE r A R E x 5 if sia eR AONE ey E S U Step 4 check the results All the results are automatically saved to a new subfolder called CA_Out within the folder where the image exists In this case the path for this subfolder is C Users youmap Desktop Box Sync collagen analysis github curvelets CurveAlignTestImages input_ct_csvbound CA_Out Nine output files are shown i GO gt BoxSync gt collagen analysis github curvelets CurveAlignTestImages input_ct_csvbound CA Out Organize v Include in library v Share with v New folder sk cae Name 5 Type Size E Desktop EN TACS 3a_1_hist csv Microsoft Office E 1 KB i Recent Places Gi TACS 3a_1_stats csv Microsoft Office E 1KB iB Downloads EN TACS 3a_1_values csw Microsoft Office E 6 KB my Google Drive Ei TACS 3a_fibFeatNames csv Microsoft Office E 1 KB Je Box Sync G TACS 3a_fibFeatures csv Microsoft Office E 79 KB TACS 3a_fibFeatures mat MATLAB Data 56 KB 3 Libraries TACS 3a_overlay tiff
Download Pdf Manuals
Related Search