Comp2Comp migration to TotalSegmentatorV2 and nnUNetV2

[jmweaver-Rush]

Are there any plans to fully migrate to V2 versions of both TotalSegmentator and nnUNet and remove dependencies for non-V2 versions?

I'm currently attempting to get Comp2Comp running on a Windows machine (I know, I know...☹). Portions of the inference pipeline that use V2 versions of TotalSegmentator and nnUNet are working for me (Spine) but others (MuscleAdiposeTissue) are failing due to the reliance on nnUNet V1, which does not support Windows due to some multiprocessing issues.

[aedendavis]

Hi, did you ever find a solution to this issue?

[jmweaver-Rush]

I did not, unfortunately. I'm still doing body composition work, but I'm just using TotalSegmentator by itself (version==2.9.0, with Python 3.10.18). I ended up writing my own scripts to calculate body composition from various vertebrae levels (with some inspiration from the Comp2Comp source code).