Deep learning models to detect microsatellite instability in colorectal cancer from histological images

Zenodo (CERN European Organization for Nuclear Research)

Abstract

This repository contains trained deep learning models (based on shufflenet) for detecting microsatellite instability (MSI) in histological images of colorectal cancer. Expected input is 512x512x3, output is categorical (MSI vs. not MSI). The models are provided as MAT (Matlab R2019a) files. All models are described in detail in the publication "Clinical-grade detection of microsatellite instability in colorectal cancer by deep learning: an international multicenter study" More information on the original model: https://www.mathworks.com/help/deeplearning/ref/shufflenet.html

Authors 1

1. Jakob Nikolas Kather corresponding Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH University Aachen