Hi Mohamad - What you did with your binarization was that you lowered the threshold all the way to zero, i.e., included all the voxels that had any probability > 0 to be in the tract. This means that you included more voxels than the default, which is to threshold at 20% of the max probability. Changing the threshold to get a broader ROI would be an alternative to running more iterations to get a more realiable distribution. How either of those approaches would behave depends on the noise level in the data. Also, you averaged the FA/MD/etc over those voxels without multiplying with their probability, which corresponds to the unweigthed average that trac-all gives you. Note that trac-all calls the dmri_pathstats command to compute the stats, and you can run that command with a different threshold to get the same behavior as what you did. You can find an example of the dmri_pathstats command line in the trac-all.log file of any of your subjects.
Best, a.y On Mon, 24 Nov 2014, Alshikho, Mohamad J. wrote: > Dear Anastasia, > Thank you very much for your answer. > In my analysis I have two groups 23 patients and 27 controls. My problem > right now is that this change in the results (from the output of trac-all > -path -c) gave me different P values between the groups, sometimes the > results are significant and sometimes it is not! Till today I repeated this > analysis many times and I collected the results after 5 consecutive runs (all > the tracts for all the subjects) > > Recently and In order to find a solution for this problem I used the > following command line to generate the mean metrics and the tracts volume: > > fslstats - dti_FA.nii.gz -k path.pd_bin.nii.gz -M -V [ I binarized > "path.pd.nii.gz" (the output of trac-all -path -c for every tract) using > fslmaths path.pd.nii.gz -thr 0.999 path.pd_bin.nii.gz and I used this > treashold to make my mask conservative as possible] > > In this case I got stable numbers. i.e when I run this command line again the > results are the same. > Kindly is this valid? can I use "fslstats" instead of the third step in > Tracula "trac-all -path -c" to calculate the mean metrics and the tract's > volume? > > Also what is the difference between fslstats and trac-all -path -c > (technically) > > I highly appreciate your support > > Mohamad > ________________________________________ > From: freesurfer-boun...@nmr.mgh.harvard.edu > [freesurfer-boun...@nmr.mgh.harvard.edu] on behalf of Anastasia Yendiki > [ayend...@nmr.mgh.harvard.edu] > Sent: Friday, November 21, 2014 5:10 PM > To: Freesurfer support list > Subject: Re: [Freesurfer] Tracula_critical issue > > Hi Mohamad - As it's a probabilistic method that's sampling from a > distribution, there is a bit of randomness built in (which should decrease > as you increase the number of samples that you collect). You'll find that > the the weighted average measures are the most reliable, as they are less > affected by the tails of the distribution. > > a.y > > On Wed, 5 Nov 2014, Alshikho, Mohamad J. wrote: > >> Hi Anastasia, >> I started recently using Tracula to do tractography for my DTI data. I did >> the analysis exactly as mentioned >> in wiki. The problem is that when I am repeating the third step (trac-all >> -path -c ) for the same subjects to >> generate the statistics; Tracula is updating all the information mentioned >> in the file "pathstats.overal.txt" >> regarding the tract's volume and the tract's metrics values. >> Every time the output is different. I think we have a problem in the scripts >> or something like this? >> >> Attached are three "pathstats.overal.txt" files for the same subject after >> three consecutive runs for the >> "trac-all -path -c" (the same subject) >> >> >> Looking forward for your support >> Thanks >> >> > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.