Hi Ruchika,
yes, your approach seems valid so long as you didn't decide to look at this particular tract *after* getting your first results, but rather *before* starting your analysis...
The best thing is probably to load the probabilistic tract of interest in FSLview (this can be done using the fslview atlas tool "Structures..." dialog to "+" into into the overlay list then "Save as..." to create your image).. Then you can threshold it at 5 or 10% with fslmaths.
Cheers,
Gwenaelle
--- En date de : Mer 29.4.09, Ruchika Wadhwa <[log in to unmask]> a écrit :
> De: Ruchika Wadhwa <[log in to unmask]>
> Objet: [FSL] creating ROI's using probabilistic atlas
> À: [log in to unmask]
> Date: Mercredi 29 Avril 2009, 8h33
> Hello:
>
> I am trying to create ROI's using the JHU white-matter
> tractography atlas, which would
> then be used to study FA changes over a one-year period. I
> had two questions: first was
> related to the validity of the method that I was thinking
> of employing. Basically I was
> thinking of using the JHU-tracts-maxprob25 to select my
> tract of interest. I was then
> going to mask the mean_FA_skeleton using this tract:
>
> fslmaths myregion_mask -mas mean_FA_skeleton_mask
> final_mask
>
> And then use the final_mask to run the randomise command.
> Would this be a valid way of
> getting the mean FA in my tract of interest? Also, I was
> wondering if I should use the
> maxprob25 to select my tract of interest, which I am
> assuming is the proportion (25%) of
> people who showed the particular tract in that location? Or
> should I go lower than this to
> account for inter-subject variation in gross WM fiber
> architecture? And if I did want to go
> lower, say (5%), which image would I be using?
>
> Thanks for the help!
> -Ruchika.
|