Please refer to Patch-GCN

Please download the official TCGA datasets of BRCA, BLCA, GBMLGG, LUAD, and UCEC. For more details on pre-processing, please refer to CLAM and Patch-GCN.

before training and testing, please update the configs. Generally, we train the model with one 24 GB memory GPU. You can adjust the 'num_instances_maximum' to sample the number of instances in accordance with your GPU power.
Testing will be performed after each training epoch, and the last model will be employed for the final evaluation.

  e.g., python main.py

If you have any questions, please don't hesitate to contact us. E-mail: zhikang.wang@monash.edu

This work was built upon the Patch-GCN and CLAM.