Pytorch implementation of our method AIVA (Submitted to MICCAI workshop).

• Install wandb using https://docs.wandb.ai/quickstart
• git clone https://github.com/yishayahu/AIVA.git
• cd AIVA
• pip3 install -r requirements.txt
• cd ..
• git clone https://github.com/deepmind/surface-distance.git
• pip install surface-distance/

• Download CC359 from: https://www.ccdataset.com/download
• Download MultiSiteMri (msm) from: https://liuquande.github.io/SAML/
• point paths to the downloaded directories at paths.py
• run python3 -m dataset create_all_images_pickle

• the results will be visible at https://wandb.ai/
• source can be any number between 0 and 5.

python3 trainer.py --source {source} --target {source} --mode pretrain --gpu {device}

python3 trainer.py --source {source} --target {source} --mode pretrain --gpu {device} --msm

• the results will be visible at https://wandb.ai/

• source and be target can be any number between 0 and 5.
• source and target should not be the same

python3 trainer.py --source {source} --target {target} --mode clustering_finetune --gpu {device}

• target can be any number between 0 and 5.

python3 trainer.py  --target {target} --mode clustering_finetune --gpu {device} --msm