Hi Jasper,
Thanks for looking into it. Yes, I'm running on non-betted data.
flirt -in highres -ref standard -out highres2standard -omat highres2standard.mat -cost corratio -dof 12 -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -interp trilinear
fnirt --iout=highres2standard_head --in=highres_head --aff=highres2standard.mat --cout=highres2standard_warp --iout=highres2standard --jout=highres2highres_jac --config=T1_2_MNI152_2mm --ref=standard_head --refmask=standard_mask
I've tried a non-dilated mask. I've also tried this:
fnirt --iout=highres2standard_head --in=highres_head --aff=highres2standard.mat --cout=highres2standard_warp --iout=highres2standard --jout=highres2highres_jac --config=T1_2_MNI152_2mm --ref=standard_head --refmask=standard_mask --warpres=5,5,5 --lambda=400,200,150,75,60,45 —subsamp=8,4,4,2,2,1 —infwhm=16,8,5,4,3,2
Thank you for your help!
-Marina
On Tue, 6 Feb 2018 19:58:22 +0000, Jesper Andersson <[log in to unmask]> wrote:
>Dear Marina,
>
>can you please let me know exactly how you ran fnirt on your data, i.e. the command lines for flirt and fnirt? Also, are you running it on non-betted data?
>
>Jesper
>
>> On 1 Feb 2018, at 16:37, Marina Shpaner <[log in to unmask]> wrote:
>>
>> Hello experts,
>>
>> Over the years, I have had a number of non-linear (FNIRT) registration fails, which I couldn't fix. It usually happens in brains with relatively large parietal sulci, so it's more common in older folks. I have had the luxury of just dropping those participants. However, I'm running an analysis on a very tight sample right now, so I don't want to drop anybody because of the registration. I'm attaching an image with the problematic registration (I circled the region where I see a problem). It may not look like much on an image here, but it's pretty substantial in FSLView. What are some possible solutions for this?
>>
>> Thank you in advance.
>> -Marina
>> <highres2standard.png><highres2standard2.png>
|