Finding diffusion directions using Bvec and Bval Files


I need help finding out how to find diffusion directions using Bvec and Bval files for dwi data. My first theory is to count the number of bval types (e.g., if there are 20 1000s/mm2 bvals then that would equal 20 diffusion directions). Is there a way to find out how many diffusion direction per type of bval? I know mine is a multshell acquisition based on information from Andrew Jahn's website (Andy's brain book) on MRTrix3.

Diffusion: For each subject, we collected 2 runs of diffusion weighted imaging per session using monopoly scheme with multiple shells of diffusion values (b = 100, 200, 250, 350, 450, 550, 650, 900, and 1000 s/mm2.

Example from Khalilian et al., (2021): For data acquisition, four shells were selected at b = 1000,3000, 5000 and 10,000 s/mm2 with 64, 64, 128, and 256 diffusion directions.


This is just saying that there were 64 volumes associated with b=1000 and 3000, 128 volumes at b=5000, and 256 at b=10000. You can look at your bval text file and count how many images are associated with each of your shells to get the numbers of directions tested.

The BVEC information by itself only tells you what direction gradient was used in the scanner, not anything about how water is diffusing in the brain. You will need to preprocess your DWI data, and use a model to find the primary diffusion directions. The most simple model is the diffusion tensor, where the longest eigenvector (V1) will point in the primary diffusion direction. More complex models, such as constrained spherical deconvolution, can model primary diffusion directions of multiple fiber populations even if they are in the same voxel.