search

dazhangyu123 / ACMIL

Attention-Challenging Multiple Instance Learning for Whole Slide Image Classification (ECCV2024)
MIT License
66 stars 8 forks source link

questions about comparison experiments #7

Closed boomboomboooo closed 7 months ago

boomboomboooo commented 7 months ago

Thank you for your excellent work! You have provided the codes and operation modes of ABMIL, CLAM, DAMIL, TransMIL and DTFD-MIL, but I noticed that the comparison table of your paper also provided the comparison results of IBMIL and MHIM-MIL. Could you please provide the operation modes of IBMIL and MHIM-MIL?

dazhangyu123 commented 7 months ago

"We've updated the implementation of IBMIL and MHIM-MIL with three files: Step3_WSI_classification_IBMIL.py, IBMIL_clustering.py, and Step3_WSI_classification_MHIM.py.

For IBMIL, follow these steps:

Initially, execute Step3_WSI_classification_IBMIL.py with args.c_path=None for the first-stage training. Next, run IBMIL_clustering.py to build the dictionary. Finally, run Step3_WSI_classification_IBMIL.py with args.c_path="path of generated clustering centroids in the clustering phase" for the second-stage training. For MHIM-MIL, follow these steps:

Start by executing Step3_WSI_classification_MHIM.py with args.model=pure and args.teacher_init="" for the first-stage training. Then, execute Step3_WSI_classification_MHIM.py with args.model=mhim and args.teacher_init="path of trained model in the first stage" for the second-stage training."

boomboomboooo commented 7 months ago

Thank you very much for your reply