Dear Dan,
Bedpostx and all other FSL tools do not handle NaN values well and are
likely to crash if the input images contain them. You should use the -nan
and -nanm options in fslmaths to remove them and make an appropriate
mask that you can combine with the nodif_brain_mask to get the desired
behaviour.
All the best,
Mark
On 2 Mar 2012, at 19:20, Daniel Peterson wrote:
> I'm interested in using bedpostx - and I have a question for the FSL community.
>
> How does bedpost handle NAN or otherwise "strange" values in the DW volumes? The reason I ask, is we have an automatic outlier rejection scheme we like to use (RESTORE) that really helps cut down on motion-related artefacts.
>
> Since bedpost runs on the original DW images, I'd like to go back and "mask out" arefactual voxels on individual DW images (perhaps be replacing them with a nonsense value). So ideally, bedpost would just ignore those data pts when building the fiber pdfs. Is this reasonable/feasable?
>
> Thanks so much,
> -Dan
>
|