Cropping out part of hippocampus

[akhanf]

I've noticed sometimes the hippocampus gets cropped out slightly at the postprocess_labelmap step (e.g. the subject I was testing with #11) . @jordandekraker I am guessing increasing this parameter will address it?

Hippocampal_AutoTop/tools/postprocess_labelmap.m

Line 9 in d96d0f0

cropbuffer = 10; % in voxels

I'll give it a try with 20 and see how things work..

[jordandekraker]

Hmmm that's odd since the cropping is based on the contours of hippocampal labels. But yeah your solution should help.

Another quick note: the reason I'm doing this cropping and discarding headers in this script is because the UPenn ex-vivo atlas is not aligned to the medial-temporal lobe or anything like that. I found that registration was least likely to fail when both images were fairly tightly cropped and headlerless. I now added another file to the UPenn atlas which I manually aligned to the CITI right hippocampus, so potentially we could now run this registration with these headers in place so hippocampi should roughly line up, and then we might not need this cropping anymore.

[akhanf]

Yeah, the problem also might go away if we use more deformable registration, as in the cropped out cases the original niftynet labels were still in the fov, but the postproc labels were outside as they were extending past the hippocampus.. But ya, happy to test it out without cropping/stripping headers too.

…

On Wed, Nov 25, 2020 at 5:10 PM jordandekraker ***@***.***> wrote: Hmmm that's odd since the cropping is based on the contours of hippocampal labels. But yeah your solution should help. Another quick note: the reason I'm doing this cropping and discarding headers in this script is because the UPenn ex-vivo atlas is not aligned to the medial-temporal lobe or anything like that. I found that registration was least likely to fail when both images were fairly tightly cropped and headlerless. I now added another file to the UPenn atlas which I manually aligned to the CITI right hippocampus, so potentially we could now run this registration with these headers in place so hippocampi should roughly line up, and then we might not need this cropping anymore. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#12 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACXV2XM4MC4FPB6BBGHECU3SRV6FVANCNFSM4UC6BZNA> .

-- *Ali R. Khan, PhD* Canada Research Chair in Computational Neuroimaging Assistant Professor, Medical Biophysics Scientist, Robarts Research Institute Schulich School of Medicine & Dentistry, Western University e. alik@robarts.ca t. 519.931.5777 (x24280) w. www.khanlab.ca