 |
 |
That isn't the right procedure for what you want, it is for something else I think. You should provide probtrackx a transformation matrix or warpfield from standard space to diffusion space for each subject. This will cause it to output the results from your standard space seed in standard space as well. You could then just average the tracking results with fslmaths, or do something more complicated (like divide by the waytotal, threshold at some level, and binarize on each subject before averaging). Peace, Matt. _____ From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Crystal Zhou Sent: Monday, June 25, 2012 11:05 AM To: [log in to unmask] Subject: [FSL] Averaging tracks across participants Hello there, I was wondering if anyone had any ideas about how to average ProbtrackX results across participants? Basically, what I'm interested in is if there's a way to average the tracks found using one seed in MNI space for all participants. I found this thread below, but it seems to stop at one participant, and (barring a general idea) I'm also not entirely sure what all the steps does and what it is each step is producing. Any help would be greatly appreciated! Crystal On 10/19/09 2:49 PM, "Matt Glasser" <[log in to unmask] <https:[log in to unmask]> > wrote: This thread is getting difficult to follow without all of the previous messages below. Here is the procedure you should be using: Create Transforms: Create affine transform between FA and Template FA using FLIRT (lets call this diff2standard.mat), also output the linearly transformed FA map in the template space (lets call this data_FA2standard_affine.nii.gz) Create nonlinear transform between data_FA2standard_affine.nii.gz and the template FA (lets call this diff2standardwarp.nii.gz) Invert Transforms: Use convert_xfm -omat standard2diff.mat -inverse diff2standard.mat Use invwarp -w diff2standardwarp.nii.gz -o standard2diffwarp.nii.gz -r <native diffusion space file, e.g. nodif_brain_mask> Move ROIs from Standard Space to Structural Space: applywarp -i <ROI in standard space> -r <native diffusion space file, e.g. data_FA> -o <ROI in native space> -w standard2diffwarp.nii.gz --postmat=standard2diff.mat --interp=nn To test, you can do the following test to verify that you did everything correctly: applywarp -i <data_FA in diffusion space> -r <template FA> -o <data_FA in standard space> --premat=diff2standard.mat -w diff2standardwarp.nii.gz --interp=trilinear applywarp -i < data_FA in standard space > -r <native diffusion space file, e.g. data_FA> -o <data_FA in diffusion space 2> -w standard2diffwarp.nii.gz --postmat=standard2diff.mat --interp=trilinear The output image <data_FA in diffusion space 2> should line up exactly with the original image <data_FA in diffusion space> as you will have transformed the FA into standard space and then back again into diffusion space. Note that I use trilinear interpolation instead of nearest neighbour because the FA is an image with continuous values. Peace, Matt.