Retinotopy analysis with MrVista

[under working]

This protocol is based on Rory Sayres‘ teaching.

Download and install MrVista
svn checkout https://white.stanford.edu/repos/vistasoft
If you already downloaded an older version, use
svn update https://white.stanford.edu/repos/vistasoft

Or you can download the zip file from
http://white.stanford.edu/software/

Setup data directory hierarchy
Inside a subject’s directory, say 13902, create directories
Raw
Raw/Pfiles
Raw/Anatomy/Inplane

Pay attention to the upper/lower cases.

Hint: To create nested folder, you may use -p option. e.g. mkdir abc/def/gh -p

Put inplane images (32 dicome files) into Raw/Anatomy/Inplane; put all retinotopy functional images into Raw/Pfiles. You should keep a track of which run is eccentricity or meridian (e.g. P18944 is ecc).

Functional analysis

1. Start Matlab
2. cd to MrVista root directory (/fs/quarry/cuixu/mrvista/) and add MrVista to MatLab path
   addpath(genpath(pwd))
   This step may take a long time (~1min) as there are a lot of folders.
3. cd to your subject’s directory (e.g. /fs/quarry/cuixu/retinotopy/data/13902/). You should stay in this directory when using mrvista.
4. Initialize
   mrInit2
   Initialize window pops up (below).
   
   Select all 5 “Initialization Options” and click OK.
   
   Session Description: retinotopy
   Subject: <your subject’s name> e.g. 13902
   For each run, fill either “ecc” or “mm”. You can add more info such as “ecc out” or “mm clockwise”. Click OK.
   
   Skip: 6 scans. Keep 96 scans for all runs.
   Crop Inplanes window:
   
   Select a slice where brain is the biggest
   Click the upper-left point, and then lower-right point, the rectangle between will be kept. You should keep the skull but not much neck.
   
   In “Assign analysis parameters”, select all scans, click “Assign coherence params” and input 8 as “Stimulus frequency cycles/scan”.
   
   Click OK.
   In “Preprocessing Options” window, select all scans and click OK. Progress will be shown. You may encounter warning “loadtSeries: unexpected number of tSeries frames in file.”. It’s a bug. (For now, you can set the nfrmaes to 96 by  mrSESSION.functionals(1).totalFrames=96;)
   
5. Run mrvista
   mrvista
   MrVista main window pops up.
   Resize and click menu File | Save preferences to save preferences (e.g. window size and positoin).
   You may also want to change 1st_Slice and #_slices to have different views (e.g. display multiple slices at once)
   
6. Check data quality by showing movie
   Click Menu Plots | tSeries Movie | Movie UI (set params)
   Scans: 1:5 (the number of runs)
   Slices: 1:32 (the number of slices in inplane)
   Click OK. After some computation, movie will be shown. With movie you can check visually if subject’s move their head a lot within run.
   
7. Average all ecc scans, and mm scans.
   click Analysis | Time Series | Average tSeries. “Average Time Series” window pops up. Select all ecc runs, and give annotation “avg ecc”. Progress bar shows.
   Repeat for mm.
   
8. Select dataType as “Averages” (previously it is “original”) on the upper-right corner of mrvista main interface.
9. Do analysis on averaged scans.
   click menu Analysis | Traveling Wave Analysis | compute corAnal | computer corAnal (all scans)
10. View coherence map: View | Coherence map. You see the back of brain (visual cortex) has higher coherence with stimuli (a lot of red). Coherence is kind of correlation regardless lag.
11. View amplitude map: View | amplitude map. Amplitude is usually low so we need to change colorbar. Right click the color bar in the bottom and select context menu “Clip mode”. A window pops up. Input 0 4 to see differential colors for amplitude.
12. View phase map: View | phase map. This is most important map.
    For ecc, you should see that color changes from blue to red when moving from front to back of the head.
    
    For mm, you first need to set retinotopy parameters. Click Edit | Data Type | Set retinotopy parameter | current scan. Start:165, clockwise, width 0. Click OK. You will find the color bar change to polar.
    You should see the upper left of brain is redish, upper right of brain is yellowish.
    
    
13. ROI
    First you may need to display only 1 slice to zoom ot.
    Create ROI: ROI | Create | Create Rectangle ROI. Click two points.
    Restrict ROI (to super-threshold voxels): ROI | restrict | restrict selected ROI
14. Plot ROI time series or FFT
    Click Plots | Current scan | mean time series to view time series. You should see 8 cycles of up and downs.
    
    Click Plots | Current scan | mean fft to see frequency.
    

Anatomy processing

1. Convert DICOM files to nifti file.
   cd to the directory where the structural T1 image is. They are DICOM files (124 of them).
   niftiFromDicom(pwd)
   You will find a new file created (.nii.gz file). Rename it to t1_raw.nii.gz
   Can’t make it work under scuttlebutt (mex file error), nor on rumor (stack error). But it works when I ssh scuttlebutt under my Mac and start as “ml7spm2 -jvm”.
2. Use mrViewer to identify if image directions are correct.
   mrViewer t1_raw.nii.gz
   Click “3-Axis View”
   Select menu Coordinates > I|P|R
   Select menu View > Label > Label directions
   Check if directions LR, AP, IS are correct. If not, you will have to flip.
   
3. Flip direction if necessary. If you find one direction is flipped (e.g. AP is flipped in the figure above), you need to correct it: (run these command on scuttlebutt, not Mac)
   mr = mrLoad(‘t1_raw.nii.gz’);
   mr.data = flipdim(mr.data, 2); % 2 means AP, 1 means LR, 3 means IS
   mrSave(mr, ‘t1_raw_flipped.nii.gz’);
   Confirm if flip take effect using mrViewer.
4. Reslice and ACPC the image
   mrAnatAverageAcpcNifti({‘t1_raw_flipped.nii.gz’}, ‘t1.nii.gz’, [], [1,1,1]);
   
   If you don’t know where is AC or PC, you need to refer my earlier post (how to find ACPC).
5. Now you should have t1.nii.gz which is already PCACed, isotropic (1mm).  You can view this image with mrViewer.
6. Run FIRST, initial segmentation. Note this command is run from shell, not within MatLab. This step takes ~2hr but you don’t have to wait to continue the next step.
   run_first_all t1.nii.gz 4 t1_sgm
7. Run BET to strip skull
   bet t1.nii.gz t1_bet.nii.gz -m -f .2
   You may find you get t1_bet.img instead of t1_bet.nii.gz. If so, use mri_convert of freesurfer to convert image format.
8. Confirm skull stripping using mrViewer (or fslview).
   Run mrViewer t1.nii.gz, then select menu File > Load > MR data map, select t1_bet.nii.gz file. You will see a red image. Adjust threshold (Threshold 1) slightly, you should see the skull-stripped image on top the original t1 image. Check if visual cortex is cut. If yes, you can re-run BET with different threshold.
   
9. Run FAST, segmentation on skull-stripped brain
   fast t1_bet.nii.gz
10. Swap x dimension using fslswapdim
    (After swapping I find left-right is not right in itkgray, so maybe should not swap?)
    Can’t get it work in scuttlebutt. But we can achieve the same effect by flipdim in MatLab.
    mr = mrLoad(‘t1_bet_seg.nii.gz’);
    mr.data = flipdim(mr.data, 1); % 2 means AP, 1 means LR, 3 means IS
    mrSave(mr, ‘t1_bet_seg_swap.nii.gz’);
11. Convert FIRST/FAST file to class file (takes ~1 min)
    mrGrayConvertFirstToClass(‘t1_sgm_all_th4_first.nii.gz’,’t1_bet_seg_swap.nii.gz’,’t1_class.nii.gz’)
12. Use ITKGray for manual segmentation. ITKGray runs on linux. It has been already downloaded and installed at /quarry/cuixu/itkgray/itkGray_install/bin/itksnap
    (You can download ITKGray from http://white.stanford.edu/software/downloads/itkGray/ITKGray_080128_fc4.tgz
    unzip (tar -xzvf). Then run itksnap.)
    Detailed ITKGray usage can be found at http://white.stanford.edu/newlm/index.php/ITKGray-Using
    
13. In ITKGray, load original t1.nii.gz
    File> Load Data> Greyscale Image (load t1.nii.gz)
    Load segmentation data
    File> Load Data> Segmentation Data (load t1_class.nii.gz)
14. Use Paint Brush tools to manually segment the brain. Overall label opacity is very useful – you need to change it constantly to identify where is gray/white matter.

Useful links: http://white.stanford.edu/newlm/index.php/Segementation-FSL-Pipeline

Related: stimulus parameters

---

---

写作助手，把中式英语变成专业英文

---

Want to receive new post notification? 有新文章通知我