Computational Pathology Research Framework

A PyTorch-based framework for whole slide image analysis and deep learning in digital pathology

Matthew Vaishnav

Abstract

This framework provides tested implementations for computational pathology research, enabling reproducible experiments on whole slide image (WSI) analysis. Built on PyTorch 2.0+, it includes working pipelines for PatchCamelyon and CAMELYON16 benchmarks, achieving 94% accuracy on patch-level classification and functional slide-level aggregation with multiple pooling strategies.

The codebase emphasizes research reproducibility with comprehensive unit testing (55% coverage, 972 tests), modular architecture, clinical workflow integration, and extensive documentation.

Research Use Only: This framework is designed for research purposes and has not been validated for clinical diagnostic use.

Key Contributions

Benchmark Implementations

Complete pipelines for PatchCamelyon (94% accuracy) and CAMELYON16 slide-level classification with aggregation strategies.

Pretrained Models

Integration with 1000+ models from torchvision and timm, featuring automatic extraction and dimension detection.

Analysis Tools

Model profiling, ONNX export, prediction CSV generation, and visualization utilities for publication.

Tested and Documented

55% code coverage with 972 tests across Ubuntu/macOS/Windows, comprehensive documentation, and reproducible configurations.

Quick Links

Getting Started RTX 4070 Guide Run Experiments API Reference Documentation

Installation

git clone https://github.com/matthewvaishnav/computational-pathology-research.git
cd computational-pathology-research
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt
pip install -e .

See the Getting Started Guide for detailed instructions.

Citation

If you use this framework in your research, please cite:

@software{vaishnav2026sentinel,
  title = {Computational Pathology Research Framework},
  author = {Vaishnav, Matthew},
  year = {2026},
  url = {https://github.com/matthewvaishnav/computational-pathology-research},
  note = {A PyTorch framework for whole slide image analysis}
}