Converting inner/outer surface files to WM and pial format in Freesurfer

[khazarAhmadi]

Hi HippUnfold Experts,

Thanks a lot for your efforts to create a great package! I am working on a 7T laminar-resolution fMRI project to extract BOLD signal across different gray matter layers of the hippocampal subfields. To this end, I need to first delineate pial/GM and WM/GM boundaries of the hippocampus and subsequently bin the GM area to several depths using this toolbox . It works with Freesurfer surfaces namely .pial and .white outputs but freesurfer does not provide surface-based tissue delineations for subcortical regions (see the attached file as an example). That's where HippUnfold comes handy as it generates inner, outer and midthickness .gii files that I believe correspond to WM, pial and GM, respectively. My questions to you: 1) is there a way to convert those inner and outer .gii files to .white and .pial format? I tried mris_convert to change the .gii files to ASCII which is readable by OpenFmriAnalysis toolbox, but the outputs lose geometry and do not match the corresponding anatomy anymore. 2) Could you please clarify what exactly the midthickness surface represent? I would greatly appreciate hearing your thoughts.

Many thanks and best wishes,
Khazar

[jordandekraker]

Hi Khazar,

2: The midthickness surfaces are very similar to those generated by Freesurfer (in the image below, cyan is similar to orange; green to yellow, and blue to red). That is, it is 50% of the distance between the inner and outer surfaces using this method to correct for shifting laminae between sulci and gyri.

1: For converting surface files, I'm surprised that mris_convert didn't work out-of-the-box, but you could you try again with the extansion .mgh in the output to specify that it is a surface?

Also note that Nibabel or Matlab's gifti toolbox work great for loading surfaces!

[khazarAhmadi]

Hi Jordan,
Thanks for the quick feedback! As recommended, I changed the output extensions and it worked. No geometry issues anymore :)
Just a follow-up question. I ran HippUnfold with singularity using a T2w modality (inplane resolution 0.43 X 0.43, slice thickness = 1.5 mm) and added the --t1-reg-template flag. The T1 image has an isotropic voxel size of 0.75 mmm. I suppose this approach is superior than running HippUnfold using the T1 modality only. Right? Is the T2 image being aligned to individual T1 image during the processing? or the use of T1 is meant to guide the template registration? It would be great if you could kindly elaborate on this. Thanks a lot already in advance!

Cheers,
Khazar

[jordandekraker]

Glad to hear the surface conversion worked!

Yes your correct that the --t1-reg-template flag is ONLY for registration to template, and this is often not even needed as the T2w can usually be registered without issue (i usually would only use this flag if the T2w registration had failed already).

T2w images are often preferred in the hippocampal subfields literature, but aren't necessarily better in HippUnfold. It sounds like your T1w and T2w images have comparable voxel volumes, so I expect they should perform equally well in HippUnfold. I personally have a preference towards T1w isotropic images since they tend to show a bit more detail in the head and tail of the hippocampus, but it will also depend on image quality (eg. SNR or artefacts like dropout). You could judge by what you can subjectively see in the two types of images, or just try both (ie. using a separate output directory for each modality) and see which looks nicer (again by subjective judgement. If you post a few screenshots I'd be happy to offer my opinion, but I suspect you'll do fine on your own too!)