 |
 |
Hi again, Dear Jesper, I am aware there is differences in intensity for different volumes depending on b-value. That's why, in my script I compute mean and standard desviation for each volume separately. For each voxel inside a volume I check if its intensity value goes away from [mean-2*std,mean+2*std] range. Therefore, I compare the voxel intensity value with the mean of volume which voxel belongs. And results looks fine. I am not sure I understand exactly what you are doing. Are you taken the mean of an entire DWI volume and then look for voxels outside of +- 2std? If that is the case I am afraid I am still not convinced. The local differences can be very big. You would also risk selecting against white matter voxels where the preferential direction is parallel to the diffusion gradient of that volume (low signal) and against voxels where the preferential direction is perpendicular (high signal). On the other hand, if I would use RESTORE method, for example, does exist a FSL function or software to use the model approach and get outliers? There isn’t a function specifically for that, but you can certainly use the output from the --sse parameter to implement something along the lines of RESTORE. But note also that RESTORE is publicly available. Jesper Thanks very much. M del Mar