Multi-Granularity Cross-modal Alignment for Generalized Medical Visual Representation Learning

Usage

pip install -e .
pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
pip install -r requirments.txt

Training

cd mgca/models/mgca
CUDA_VISIBLE_DEVICES=0,1 python mgca_module.py --gpus 2 --strategy ddp

Finetune

cd mgca/models/mgca
CUDA_VISIBLE_DEVICES=1 python mgca_finetuner.py --gpus 1 --dataset chexpert --data_pct 0.01

More downstream tasks and pre-trained model will come soon!