Federated Modality-specific Encoders and Multimodal Anchors for Personalized Brain Tumor Segmentation

Introduction

This repository is for our AAAI2024 paper '[Federated Modality-specific Encoders and Multimodal Anchors for Personalized Brain Tumor Segmentation]'.

cl_train.py - training code for clients

cl_train_glb.py - training code for server

fl_train_clsPasData_async.py - training code for federated learning (our method)

Data Preparation

Get Brats2020 dataset here and refer to RFNet.

Usage

1. Clone the repository:

   git clone https://github.com/QDaiing/FedMEMA.git
   
   cd FedMEMA

2. Set hyperparameters via options.py and Train the model:

   bash run.sh