Hi,
You should be using the standard config file with fnirt (i.e. $FSLDIR/etc/flirtsch/T1_2_MNI152_2mm.cnf) as without it the defaults are unlikely to do a good job, and hence the warning that the registration is not good. You can add a mask when running fnirt by using the --inmask option. Note that this mask is such that it excludes areas where the value in the mask image is zero. So add the --inmask and --config options, and also use the whole-head images, and not the brain-extracted versions (as FNIRT wants to use the signal outside of the brain too, unlike FLIRT).
All the best,
Mark
On 11 Oct 2013, at 16:55, Derek Archer <[log in to unmask]> wrote:
> Hi Mark,
>
> Thanks for the response -- I used the costweightmask in flirt and it works great. However, I'm not sure what option to use in fnirt with this mask. I am using the affine matrix outputted from flirt, but I seem to be getting an error when I do this.
>
> The code I'm using is:
> fnirt --ref=MNI152_T1_2mm_brain.nii.gz --in=skullstripped_T1.nii --aff=standard.mat --iout=fnirt_output.nii
>
> The error I'm getting is:
> Warning, Jacobian not within prescribed range. Prescription is 0.01 -- 100 and obtained range is -5.94365 -- 26.3309.
>
> I tried the same fnirt code with a subject without a lesion, and it worked great. So what option do I use in fnirt to use the costweightmask? Could my issue be something else?
>
> Thanks,
> Derek
|