Documentation‎ > ‎

Using image data from human connectome project

Download HCP data

The HCP data can be downloaded/ordered from the HCP website. This website also provides HCP data in the Sample Images page. You need to get the diffusion data of the 80 unrelated subjects. We recommend using the "processed" data (in nifti format), which has been corrected for distortion due to different phase encoding directions. The processed diffusion data are located under ID/T1w/Diffusion folder. There should be five files in the folder. 

bvals -- the bvalue text file
bvecs -- the b vector text file
data.nii.gz -- the DWI data
grad_dev.nii.gz -- the gradient deviation data
nodif_brain_mask.nii.gz -- the mask file

Parse NIFTI file and include gradient deviation information 

To process these data in DSI Studio, click "STEP1: Open source images" in DSI Studio and select the data.nii.gz file. DSI Studio will also read other files to include all necessary data for reconstruction (e.g. gradient deviation, b-table...). Once the file is loaded, there will be a popup windows showing the b-values and b-vector. 

Then click on the "Ok" button, and DSI Studio will save a "data.src.gz" file in the assigned folder. This src file can be reconstructed using different algorithms by clicking "STEP2: Reconstruction". 

To run a batch processing on all 80 subjects data. You may use the following batch file (for windows system) for command line processing. This file assumes that the DSI Studio program is located at c:\dsi_studio_64, and the HCP data are under F:\HCP\....

cd F:\HCP\
for /f "delims=" %%x in ('dir /b /ad /s') do (
call dsi_studio.exe --action=src --source=%%x\T1w\Diffusion\data.nii.gz --output=%%x.src.gz > %%x.src.txt

GUI also provides a function to batch the conversion of the nii file to src.gz. The function is under the "Tool: batch processing" tab of the main window. Click on the button labeled "Create SRC Files", and select a root directory containing all the HCP diffusion data.

Image reconstruction

The created *.src.gz files can be used to reconstruct fiber orientations. Click "STEP2: Reconstruction" and select a src.gz file to do the reconstruction.

There are two methods available for HCP data. One is GQI, which can make use of the gradient deviation field and reconstruct data in subject space. Another method is QSDR, which can also use gradient deviation, and the reconstructed data are in MNI space that allows further group wise processing.

The GQI and QSDR should use a length ratio of 1.25. If r^2 weighting is used, the length ratio should be decreased to 1.0. You may enable ODF sharpening to increase the angular resolution power.

Fiber tracking

After the reconstruction, a fib file will be saved, and it allows further fiber tracking. An example of tractography is shown in the following figure.