TopoSlide: Topologically-Informed Histopathology Whole Slide Image Representation Learning

Shahira Abousamra, Asmita Sood, Sylvia Plevritis, TopoSlide: Topologically-Informed Histopathology Whole Slide Image Representation Learning, CVPR 2026.

TopoSlide is a novel approach for whole slide image (WSI) representation learning in computational pathology that incorporates topological data analysis (TDA) to capture global spatial features and local morphological patterns in histopathology images. It addresses the challenge of learning meaningful representations from gigapixel WSIs by leveraging persistent homology to encode topological features alongside traditional visual features.

1. Set up the environment

2. Download pretrained model.

3. Preprocessing

   3.1. Extract WSIs meta data.
   Check preprocessing/extract_meta_data/readme.md

   3.2. Tiling:
   Check preprocessing/tiling/readme.md

   3.3. Generate patch embeddings.
   Use pretrained patch encoder (we used CONCH model) to generate patch embeddings.

   3.4. Cluster WSI patch embeddings.
   Check preprocessing/cluster/readme.md

4. Generate Whole Slide Image (WSI) embeddings:
   Check eval_wsi_embeddings/readme.md

📄 Dual Licensing

This project is dual-licensed to support both research and commercial use: © The Board of Trustees of The Leeland Stanford Junior University. This code and associated models are released under the CC-BY-NC-ND 4.0 license and may only be used for non-commercial, academic research purposes with proper attribution. Any commercial use, sale, or other monetization of the TopoSlide model(s) and their derivatives, which include models trained on outputs from the TopoSlide model(s) or datasets created from the TopoSlide model(s), is prohibited and requires prior approval. If you are a commercial entity, please contact the corresponding author.

🔬 Non-Commercial Use (FREE)

• License: Creative Commons BY-NC 4.0
• Permitted: Research, education, personal projects
• Requirements: Attribution required
• Restrictions: No commercial use

💼 Commercial Use

• License: Commercial License Required
• Contact: shsamra@stanford.edu, plevriti@stanford.edu for licensing terms

Citation

@inproceedings{abousamra2026TopoSlide,
author      =  {Shahira Abousamra and Asmita Sood and Sylvia Plevritis},
title       =  {TopoSlide: Topologically-Informed Histopathology Whole Slide Image Representation Learning},
booktitle   =  {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
year        =  {2026},
pages		=  {13691-13701}}