Source codes for FedWSIDD

We provide bash script for training on CAMELYON16 and CAMELYON17 dataset Before running the code, ensure you have downloaded these dataset and preprocess following CLAM

The list of useful parameters is as follows:

• data_root_dir: this is the path where you put your '.pt' files for patch features.
• feature_type: feature extractor, e.g., ResNet50, ViT, UNI, PLIP, CONCH.
• mil_method: MIL framework, e.g., CLAM, ABMIL, TransMIL.
• fed_method: Federated Learning frameworks, e.g., FedAvg, FedProto, MOON.
• heter_model: set this to True to allow Heterogeneous Local Model setup.
• ipc: number of synthetic slides per class.
• nps: number of synthetic patches per slide.
• syn_size: size of synthetic patches.
• init_real: set this to True to allow initialising synthetic patches with real patches.

Run (CAMELYON16 Example)

bash run.sh