Exploring Vanilla U-Net for Lesion Segmentation from Whole-body FDG-PET/CT Scans

Our team name is Blackbean. We explore the ablity of U-Net for lesion segmentation on the autoPET 2022 challenge (MICCAI 2022).

How to use

• Prepare the checkpoints

  Download the checkpoints from Baidu Cloud (The password is gp3d), or Google Drive. Then extract it under test/.

• Make the docker

  Run gen_build.sh under test/.