hku-medai / conslide

[ICCV'23] ConSlide: Asynchronous Hierarchical Interaction Transformer with Breakup-Reorganize Rehearsal for Continual Whole Slide Image Analysis

HKU-MedAI.github.io

conslide's Issues

Inquiry for Reproducibility

Dear Authors,

thank you for sharing your work. I am trying to replicate your results, but I am missing important details that prevent me from reaching the results you reported in the published paper.

I have noticed that you have split folders such as:

split_dirs = [
        '../HIT/10fold_splits/NSCLC_100',
        '../HIT/10fold_splits/BRCA_100',
        '../HIT/10fold_splits/RCC_100',
        '../HIT/10fold_splits/ESCA_100'
]

Could you please release the IDs of the images they contain?

Thank you in advance,
Federico

Request for ID Slides for Test

Hi!

I'm currently using the datasets within this repository, which is based on the well-known TCGA dataset. However, the lack of unique ID slides for the dataset makes it challenging to ensure reproducibility due to frequent updates.

I kindly request the inclusion of unique ID slides or guidance on obtaining them within this repository to improve the reproducibility of analyses. Thank you for your attention to this matter.

CLAM hyperparameters

Dear authors,
First of all, I would like to thank you for your fantastic work. Since I'm very interested in your work, could you release also the CLAM configuration or instructions to easily reproduce your results? Thank you!