We are having problems with using vol2subfield when attempting to extract hippocampal subfield info from DTI images.

Our T1 images have been co-registered to their matching DTI images in SPM prior to being run through the recon-all pipeline output files - this looks good.

We have run segmentHA_T1.sh and the results look good.

We created a register.lta between the DTI image and the orig.mgz:

bbregister --s sub1/ --mov DTI.nii --reg register.lta --t2

The bbregister register.lta output looks good when checked with tkregister.

We then run vol2subfield as recommended:

vol2subfield --i DTI.nii --reg register.lta --sf sub1/mri/lh.hippoAmygLabels-T1-T2.v22.FS60.mgz --o dti.rh.FS60.mgz --stats stats.dat --avgwf avgwf.dat --avgwfvol avgwfvol.mgz

When we look at the output image overlayed on the DTI.nii/orig.mgz, it looks like everything has been shifted down (on the output image - dti.rh.FS60.mgz) and it doesn’t line up with the anatomy (e.g. subfields are overlayed over the ventricles). However, when we overlay the hippocampal subfield segmentation on our original DTI image, this still looks fine.

We can provide images of the various scans/registrations on request