DTI gradient tables (bvecs)

The bvecs, short for b vectors, are the list of diffusion sensitizing directions that correspond to the different diffusion-weighted volumes that were acquired. Typically, this information is stored in the DICOM header and then output in a simple text file by the DICOM converter (e.g. the excellent dcm2nii that is included with MRIcron) when these raw files are converted to a more compact form like NIfTI.

Correcting bvecs

If you are working with a new data set, and the starting set of tractography fibers just looks wrong, chances are there is some mismatch between what is described in the bvecs, and what is actually in the data. This can happen when you are working with

  • Files that have been preprocessed (possible orientation shifts might have been introduced).
  • A program that is making different assumptions about what orientation conventions have or have not already been applied to the data.

Thankfully, most of these mismatches are simple and recoverable. Here is a useful example to illustrate the two most common types of errors that can happen with the bvecs.

Initial tensor fitting and tracking gave the following set of fibers (filtered to only show those that pass through a certain coronal slice plane).


Although this certainly looks wrong, we can see that there is still some structure in the data. In fact, we can still see major parts of the anatomy - they just look like the wrong colors. Using the 3 most obvious landmarks, we can construct a table of apparent directions, together with what the proper directions should be for these tracts.

Tract Apparent direction Proper direction
Corticospinal tract Y Z
Corpus callosum Z X
Cingulum X Y

Now we can easily see that the order of the bvecs components are simply permuted from what they should be. To get them correct, we permute the present X-Y-Z order into Z-X-Y. Once we do that and rerun the tractography, we see the following output:


Things are definitely looking better, as at least now the colors of the tracts are correct. But there is still a problem: All the fibers are short. If we look closely at the corticospinal tract fibers, we can see that they are tilted in the wrong x direction. Instead of converging inwards towards the brainstem as they move inferiorly, all the fibers are pointing outwards. This tells us that the x component of the bvecs needs to be inverted. Once we do that and rerun the tractography, we get this:


Finally, the output looks like it should.

neuroimaging/bvecs.txt · Last modified: 2011/02/24 11:37 am PST by John Colby
Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Noncommercial-Share Alike 4.0 International
Recent changes RSS feed Donate Powered by PHP Valid XHTML 1.0 Valid CSS Driven by DokuWiki