Hi FSL group,
I conducted an exploratory probabilistic tractrography analysis in 16 ROIs for all pairwise connections. I then created a probabilistic tractography atlas for each track by combining all the subjects' tracks together. I want to threshold this atlas to exclude noise and voxels with low connectivity values. What is a good value to use? I see in the FSL JHU ICBM tracts atlas thresholding at 0, 25% and 50%...
My tracks connect from one ROI to another ROI, but I'm worried about the fact that there seem to be random paths that the tracks take to get there and that getting the FA value from these tracks might not be valid. In comparing the SLF in the JHU-ICBM-tracts-maxprob-thr0-2mm.nii.gz atlas vs. JHU-ICBM-labels-2mm.nii.gz atlas, it's obvious that tractography is a lot noisier. Is this the nature of tractography (which I can fix by thresholding at 25-50%) or do I potentially have a serious problem?
Best,
Dana
|