Bonsoir Emilie - Tracula draws a large number of path samples, which in the end are added up to produce the path probability distribution. "FA" shows the values of FA at each point along the path sample with the greatest probability, i.e., along a single 1D curve. "FA_Avg" averages the FA from all the path samples at corresponding points along their trajectory. Because it is an average of many values rather than a single value, FA_Avg will tend to be less noisy and more reliable.
Currently there is no way to average other values along the path, but you can trick the software into doing it. Find the "dmri_pathstats" command line corresponding to the tract that you want in trac-all.log. Rename "dmri/dtifit_FA.nii.gz" to something else temporarily, and rename the volume that you want to get values along the tract from as "dmri/dtifit_FA.nii.gz". Now when you run the same dmri_pathstats command, the pathstats.byvoxel.txt file will contain the values from the new volume in the FA/FA_Avg columns, instead of the values from the FA volume.
The coordinates in the stats files are in the subject's native DWI space.
Hope this helps, a.y
PS: For combining the pathstats.byvoxel.txt files across subjects, get the latest tracula update, if you haven't: http://surfer.nmr.mgh.harvard.edu/fswiki/Tracula/#Updates
On Wed, 9 Apr 2014, Emilie Poirion wrote:
Hello I'm using Tracula to analyse our DWI-images. Freesurfer reconstructions of the corresponding T1-images have been perfomed before. Respectively, the version used are 1.53 and 5.3. The Tracula process works fine and I have extracted the statistics for all tracts. About this statistics, I wanted to know what is the difference between FA and FA_Avg ?
An other question about the values extracted is about the coordinates in the file "byvoxel.txt". Are they correspind with the coordinates of the voxel in the dtifit image were the values were extracted?
The last question, can we extracted the same by tract statistics from an other image like a PET registered to the DWI with your code?
Thanks for your answer
-- Emilie Poirion
Master 1 Informatique Industrielle et Imagerie Industrielle Université Pierre et Marie Curie (Paris 6)