The MSIDETECT consortium
Microsatellite instability (MSI) is a genetic feature affecting 10-15% of all patients with colorectal cancer. These tumors require a different treatment than other colorectal tumors - especially, they are known to respond to cancer immunotherapy. Deep learning can detect MSI directly from routine histology images. Our consortium partners collaborate to improve and validate this technology. We are a non-for-profit academic consortium based in Germany, the Netherlands, the United Kingdom and the United States. Our consortium is focused on MSI detection in colorectal cancer and gastric cancer.
All of our deep learning models for MSI detection in colorectal cancer will be made publicly available for re-use under the MIT license. The most up-to-date model (Pytorch) are available here: https://zenodo.org/record/5151502 Our current backend to train new prediction models is DeepMed, which is freely available at https://github.com/KatherLab/deepmed