Analyze NIRS data with NIRS-SPM

being updated for v01_r14 …

NIRS-SPM is a SPM5 and MATLAB-based software package for statistical analysis of near-infrared spectroscopy (NIRS) signals, developed at the Bio Imaging Signal Processing (BISP) lab. at KAIST in Korea.

prepare files needed by NIRS-SPM

1. NIRS data (in csv format or nii format. For Hitachi ETG4000, you need “File out” the measurement data, not the Hb data, into csv format)
2. (optional) Structural image (e.g. xu.img). I use SPM2 to do normalization and segmentation (below), but you can choose whatever tool you prefer.
   1. normalize the image (use template T1.mnc) and get two files, wxu.img and xu_sn.img. This step takes 2 min.
   2. segment the image (not the normalized image) into gray and white volumes and get 3 files (xu_seg1.img, xu_seg2.img and xu_seg3.img). They are gray, white, and CSF volumes. The first two images will be used in NIRS-SPM. This step takes ~6 min.
3. (optional) Coordinate file (a plain text file, e.g. RealCoordinate.txt), from probe positioning system
   This file contains 3 columns, with each row a 3-D coordinate of a point measured with 3D digitizer. The total number of rows is the sum of landmarker points (nasion, inion, etc)  plus the number of optodes. You will have to specify the two numbers in NIRS-SPM.
   • If you don’t have 3d digitizer at all but you know the positions of each optodes in your structural image (e.g. you used vitamin E marker), you can simply find the positions using SPM’s display functionality and put into the file.
   • If you don’t have MRI image at all, you can use a standard brain image. You need to check how well the registration is visually.
   • If you don’t have MRI image, and you didn’t measure optode positions with 3D digitizer, but you take a picture of the subject, you can use a standard brain, and “guess” the optode coordinate with the picture.
   • If you don’t have MRI image, or 3D digitizer measurement, or a picture, then you can use a standard brain, and guess, and cross your finger.
4. The event onset timing and duration (in seconds). For example, [12 34 55 67] etc.

NIRS-SPM system requirement:

1. MatLab (with graphic display)
2. SPM5 (note: not SPM2)

steps:

1. (optional) Login scuttlebutt (even you are on scuttlebutt computer)
   ssh fmri@scuttlebutt -Y
2. Start matlab with graphic support and spm5 in path
   e.g.
   ml7spm5 -jvm &

3. add NIRS-SPM folder into MatLab path
   addpath('/net/cibsrdata/Volumes/SPNLData05/quarry/cuixu/NIRS/NIRS_SPM_v01_r14');
   (You need to replace NIRS-SPM path to your own path)
4. Enter sample_data directory (optional)
   cd /net/cibsrdata/Volumes/SPNLData05/quarry/cuixu/NIRS/sample_data
5. run NIRS-SPM and the main window pops up
   nirs_spm
6. Click “Convert” button, data conversion window pops up
7. Select Hitachi ETG-4000
   Select the csv file, conversion begins automatically. It will take ~20s
   Click “Save .mat file” button, and save the file as “converted_NIRS.mat”
   Close data conversion window.
8. (optional) Display NIRS data.
   Click “Display NIRS Time Series” button in the main window. NIRS_TimeSeries_Viewer window pops up.
   Click “Specify NIRS(.mat) file” button and select “converted_NIRS.mat”. Time series of channel 1 is displayed.
   Click “Specify model parameter” button and input the vector of onsets and duration
   Close NIRS_TimeSeries_Viewer window
9. Select “With MRI” and Click “Spatial registration” button, two windows pops up.
   Select “T1_MRimage/uniform.img” as T1 image
   Select “wuniform.img” as normalized T1 image
   Then a SPM big window pops up and the subject’s T1 is shown.
   Find point (-70 34 36) and click + (you will see the point is added into “Indicator Locations” window)
   Repeat for other points (-82 -31 36), (-1 85 -15 nasion), (-8 -86 -58 inion)
   In “Indicator Locations” window, enter 4 and 16 for “Indicator #” and “Optode #”
   Click “Select Real Coordinate File” button and select file “RealCoordinate_txt_format.txt”. The locations will be displayed.
   Click “Get Optode Pos. in MRI” button and select “T1_MRimage/uniform_sn.mat” file. Also select the gray and white file (“T1_MRimage/c1uniform.img” and “T1_MRimage/c2uniform.img”)
   After ~30s the coordinates in MRI is calculated.
   Click “View Channel Pos” button you will see the positions of channels
   Click Save button and save the position as “channel_position.mat”
10. Click “Specify 1st Level” button and “NIRS_Specification” window pops up
    Select nirs data file “converted_NIRS.mat”
    Create directory “spm” and Select it
    Select “Oxy-Hb”.
    Click “Specification” and specify the following parameters
    “hrf (with time and dispersion derivatives”
    number of conditions: 1
    name for condition 1: right finger tapping
    vector of onsets: [42:51:501]*9.75
    duration: 21*9.75
    high-pass filter: 60
    low-pass filter: Gaussian
    Gaussian FWHM: 4
    Correct for serial correlations? none
    Then a big SPM window pops up with design matrix
    Close the windows except for the main window
11. Click “Estimate” button in the main window, NIRS_Estimation window pops up
    Select “Individual Analysis”
    Select “SPM_indiv_HbO.mat” as SPM.mat
    Click “Estimation” button. It will take 7 min. File “V_indiv_HbO.mat” is saved.
12. Click “Result NIRS” and NIRS_Results_Viewer window pops up
    Select SPM_indiv_HbO.mat
    Select channel_positoin.mat
    Click “Contrast”, SPM contrast manager window pops up
    Click “Define new contrast”. name: finger, type: t, contrast 1 0 0 0, click OK, click “Done”
    After ~20s T map is shown. Click “View the thresholded T-statistic” and try different p-values.
13. Click “Result fMRI” and select SPM.mat in fMRI_result folder
    (can’t get this to work now)

files required:

1. NIRS_data_finger_tapping.nir
2. RealCoordinates_txt_format.txt
3. uniform.img : subject’s structural image
4. wuniform.img : subject’s structural image after normalization
5. uniform_sn.mat : the mat file produced by SPM during structural image normalization
6. c1uniform.img : gray volume of segmentation
7. c2uniform.img : white volume of segmentation
8. SPM.mat (for fMRI and NIRS comparison)
9. spmT_0001.img (for fMRI and NIRS comparison)

inside the files:

1. .nir file is simply a text file. Row is time and column is channel.
2. .csv file is also text file exported from Hitachi ETG4000
3. RealCoordinates.txt is a text file with each row a point coordinate.
4. converted_NIRS, or nirs data after conversion is a mat file with a single variable nirs_data. nirs_data is a struct. An example is:

             oxyData: [5384x24 double]
             dxyData: [5384x24 double]
                  fs: 9.7500
          timeLength: 14
                 nch: 24
            lightNum: 2
       rawdataLength: 5384

5. channel_position.mat, or the channel’s position after calculation is a mat file with a single variable, preproc_info. An example is:

          wT1_info: [1x1 struct]
       rend_ch_pos: {1x6 cell}
   

   wT1_info is the volume information of the normalized brain (you can get this info by spm_vol). rend_ch_pos is a cellarray. Each element contains a different view of the brain. Try
   figure;imagesc(preproc_info.rend_ch_pos{4}.ren)
hold on;plot(preproc_info.rend_ch_pos{4}.rchn,preproc_info.rend_ch_pos{4}.cchn,'o')
   You will see how the channel positions are displayed.

6. SPM_indiv_HbO.mat, containing estimation parameters and info
7. V_indiv_HbO.mat, a matlab mat file containing variable V_nirs, which is a sparse matrix.

related links:

• NIRS-SPM official webpage
• NIRS-SPM publication

---

---

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