search

ivadomed / model-gm-contrast-region-agnostic

MIT License
0 stars 0 forks source link

Try butterfly-seg model on T2* images (dcm-brno) #1

Open KaterinaKrejci231054 opened 5 months ago

KaterinaKrejci231054 commented 5 months ago

This issue is about testing @Nilser3's butterfly-seg model for grey matter segmentation on test data from the dcm-brno dataset (data will be shared soon).

I've tested the model on several subjects - here is a result of gray matter segmentation in C-spine:

image

Nilser3 commented 5 months ago

Thanks you for sharing @KaterinaKrejci231054

For context: butterfly-seg is a 2D nnUnet model trained for BrainHack School 2024 (details here) using invivo and exvivo open data.

I think the results are good in T2star contrast on cervical data, It would be good to see results in other contrasts and also regions.

May be you can also try sct_deepseg_gm for this data?

Thanks you!