On 16 Feb 2018, at 16:51, stefano bovo <[log in to unmask]> wrote:

Hi Jesper,

thank you for answering. Maybe I left too many details in the explanation.

I previously recorded the ASL data to the T1 space with spmregister, then I inverted the matrix and I applied the transformation  from T1 to ASL space with flirt.

Than I used these for registration in MNI. 

These are the previous registrations:

spmregister --mov "$pcasl"_mc_mean_2mm.nii.gz --reg ASL_lin_T1.dat --fsvol "$dirBC"/"$subj"/"$subj"_orig-2mm_BC.nii.gz

convert_xfm -omat T1_lin_ASL.mat -inverse ASL_lin_T1.mat 

flirt -noresample -in "$FreeSurferDir"/"$subj"/mri/brainmask_masked-2mm.nii.gz -ref "$pcasl"_mc_mean_BET_2mm.nii.gz -applyxfm -init T1_lin_ASL.mat -out brain_lin_ASL.nii.gz  # porto immagine maschera cervello in ASL

flirt -noresample -in "$dirBC"/"$subj"/PET/SUVR_2mm-processing/"$subj"_orig-2mm_BC.nii.gz -ref "$pcasl"_mc_mean_BET_2mm.nii.gz -applyxfm -init T1_lin_ASL.mat -out T1_lin_ASL.nii.gz

I hope I have been clearer now,

thank you

---

Da: FSL - FMRIB's Software Library <[log in to unmask]> per conto di Jesper Andersson <[log in to unmask]>
Inviato: venerdì 16 febbraio 2018 16:58
A: [log in to unmask]
Oggetto: Re: [FSL] FNIRT fails with jacobian warning

 

Dear Stefano,

> 
> I'm applying the same code to different patients and for some the registration with fnirt fails. 
> I'm running these lines:  
> 
> flirt -noresample -searchrx -120 120 -searchry -120 120 -searchrz -120 120 -in brain_lin_ASL.nii.gz -ref "$mniref"/MNI152_T1_2mm_brain.nii.gz -out brain_lin_ASL_lin_MNI.nii.gz -omat brain_lin_ASL_lin_MNI.mat -dof 12 
> 
> fnirt --ref="$mniref"/MNI152_T1_2mm_brain.nii.gz --in=T1_lin_ASL.nii.gz --aff=brain_lin_ASL_lin_MNI.mat --config=T1_2_MNI152_2mm.cnf --fout=T1_lin_ASL_nl_MNI.mat --iout=T1_lin_ASL_nl_MNI.nii.gz
> 
> during the fnirt running there are warning/warnings like this: 
> 
> Warning, Jacobian not within prescribed range. Prescription is 0.01 -- 100 and obtained range is -12.5089 -- 85.4155
> 
> in the end the image obtained from the registration is totally unrecognizable, while there gistration with flirt seems to be successful.

hmm, am I right in thinking you are attempting non-linearly register ASL-EPI data directly to the MNI152_T1_2mm_brain.nii.gz template? If so, I would not expect it to ever work. FNIRT can only be used to register like-with-like. i.e. T1->T1, ASL-ASL etc.

If I am wrong in thinking that, please let me know.

Jesper


> 
> I hope you can solve my problem,
> 
> Thank you in advance.
> 
> Stefano