This repository contains all files for the AJP-GI publication, DOI: https://doi.org/10.1152/ajpgi.00058.2024. Included are scripts and detailed protocols for machine learning-based detection and quantification of biliary forms within whole-slide scanned liver sections. To access and properly utilize the source material, please download the contents of this repository, and follow the appropriate protocols. The code should be functional in any murine liver tissue, as well as human liver tissue (not as well validated). Training and validation data are all from immunofluorescence sections, but we have done small pilots using IHC with positive results. Any membrane-targeted antibody for biliary epithelial cells should result in highly accurate type distinction (see protocols).
Editorial Commentary, DOI: https://doi.org/10.1152/ajpgi.00173.2024
Questions: dhellen@mit.edu