For a analysis in MS patients I would like to segment the Corpus Callosum in T1w MRI images and use the resulting mask for further analyzes e.g. overlaying with lesion mask.
Unfortunately I was not able to find a good solution that works on both: healthy controls as well as patients with heavily atrophic CC. My attempts so far were:
- Register to MNI and use atlas:
This works sometimes well with controls but not with patients that have quite huge deformations - CCSeg:
This software is not maintained anymore - the binaries you can download wont run on current systems (linux64) and the Makefile isnt working anymore as well. Maybe I could use docker with an old linux-image but I’m not sure if it is worth - Freesurfer: This is the best solution I found so far. It adopts quite well to controls and even in most patients the results where useable. But its still far from perfect and a couple of patients got a very wrong CC-Segmentation.
- Smacc: I was able to find a number of papers that present Ki-based solutions, but only for smacc did I actually find a viable solution. Unfortunately, this approach is 2D-based and the resulting nifti image segments the CC in only one layer.
Does anyone here know a good solution for segmenting the corpus callosum in both controls and patients?