Dear everyone,
I'm working on some 3T T1-weighted structural images with a spatial resolution of 0.8x0.8x0.8 mm^3, with some of the blood vessels being quite apparent within the white matter in the raw data. I've tried both SPM12 segmentation (default except no. of Gaussians, which is 2/2/2/3/4/2) and the latest CAT12 version (default except voxel resolution for normalised files being 1 mm, 1 for LAS/Skull stripping/Final clean-up), leading to quite similar segmentations, see http://we.tl/XYXL1GLaIu (Wetransder link to a screenshot for one subject), with the vessels being classified as GM. As I'm mainly interested in GM changes I thought of manually cleaning up the segmentations, but this would also be subjective of course. I vaguely recall some VBM papers briefly stating somewhere that they did do some manual corrections without providing any further details on what they tried to correct for and on how they proceeded.
I doubt this to be a relevant confound in the current data set, but nonetheless, is anyone aware of related papers, any additional (semi-)automatic cleaning procedures or some sort of protocol for manual routines?
Best
Helmut
|