this is becoming quite urgent as more and more users are struggling with propseg.
The deep learning-based segmentation, based on @charleygros work is being implemented in branch bdl_scdeepseg. Work in progress.
The function is implemented and functional.
Here are some preliminary results on a T2- and T2*-weighted image from a DCM patient (with compression).
T2: https://www.dropbox.com/s/tk51xe28pjdmkwg/deepscseg_t2_DCM.mov?dl=0
T2*: https://www.dropbox.com/s/cnwpdtlgw8wu9kt/deepscseg_t2s_DCM.mov?dl=0
The results are great although not perfect. More to come soon.
Most helpful comment
The function is implemented and functional.
Here are some preliminary results on a T2- and T2*-weighted image from a DCM patient (with compression).
T2: https://www.dropbox.com/s/tk51xe28pjdmkwg/deepscseg_t2_DCM.mov?dl=0
T2*: https://www.dropbox.com/s/cnwpdtlgw8wu9kt/deepscseg_t2s_DCM.mov?dl=0
The results are great although not perfect. More to come soon.