Hi Sneha
it looks like it is grabbing a bunch of dura. Do you happen to have a hires T2 or flair image? If so, you can use that in recon-all postprocessing to remove the dura. If not you can try using the graph cuts skull stripping to get rid of some of it. I can't tell the accuracy of the surfaces in other locations in the images. Other than dura is it reasonably acurate? cheers Bruce
On Fri, 10 Oct 2014, Sneha Pandya wrote:
Dear all,
I am running recon-all on few SPGR T1s and few MPRAGE T1s. However most of my dataset, may it be SPGR or MPRAGE has some artifacts and poor wm/gm contrast. Have attached a screen shot of bad pial and wm error (going beyond dura). Even gcut did not do a good job of cutting of all the dura, and I had to manually edit voxels in all the slices. I feel even after that if I run recon-all, it will not give me any reliable cortical thickness measures.
Can anyone please suggest me if I should still manually edit all other subjects for all the slices throughout the volume or exclude them from analysis?
PS: attached image FOV is 260 and I used -cw256 flag to restrict FOV to 256. For other images FOV is within 256, but still there is lot of dura in the brainmask.
Thanks, Sneha