Dear Jim,
This test inside invwarp for BET is very crude and does go wrong.
Simply ignore it and use the --force option to make invwarp continue
to run despite the warning.
All the best,
Mark
On 6 Dec 2011, at 02:55, James Porter wrote:
> Hello FSLers-
>
> I've been using FNIRT to do alignments on a study, and all is great except for attempting to transform standard-space ROIs back to native space for a few subjects. When running invfeatreg I get the following error:
>
> "WARNING!! Reference image has large non-zero ROI. Check that BET has run successfully on this."
>
> If I check the BET on these subjects, I can see that there is a bit of eyeball and/or parietal midline extraparenchymal tissue left over. However, none of the failed subjects seem to have worse BETs than successful subjets, and neither stricter (perhaps overzealous) BETting nor very meticulous hand cleaning of the images does any good. The inversion still fails on the adjusted registrations.
>
> Any ideas on why these large non-zero ROIs pop up for a small set of subjects?
>
> Here's a link to some images from two example subjects that may be helpful to look at:
>
> https://netfiles.umn.edu/xythoswfs/webui/_xy-20539033_1-t_FEUrXq22
>
> Thanks,
>
> Jim Porter, M.A.
> Ph.D. Candidate
> Clinical Science and Psychopathology Research
> University of Minnesota
>
|