Study Background
Histology is emerging as a potential therapeutic endpoint for ulcerative colitis (UC) driven by associations between histologic response and long-term outcomes 1. However, existing scoring systems are subjective and consequently have variable inter- and intra-reader variability 2. Furthermore, manual histologic assessment is semi-quantitative and limited in the ability to capture spatial relationships.Here we report the first machine learning (ML)-based prediction of the Geboes score, and GS-derived thresholds of histologic improvement and remission 3, directly from whole slide images (WSI) of hematoxylin and eosin (H&E)-stained mucosal biopsies. Together, PathAI models for characterization of the UC histology (IBD-Explore) and the PathAI algorithm for Geboes scoring (AI-measurement of histological improvement in UC (AIM-HI UC); for research use only) have the potential to identify clinically relevant histologic features, enable robust scoring, and ultimately advance precision medicine for patients with IBD.
1. Gonzalez-Partida, I. et al. Eur J Gastroenterol Hepatol. 2021; 33:e796-e802.
2. Romkens, TH. et al. J Crohns Colitis. 2018; 12:425-431.
3. Li, K. et al. Gastroenterol. 2020; 159:P2052-P2064. Carlos Gaitán 1, Zahil Shanis 1, Kathleen Sucipto 1, John Shamshoian 1, Jin Li 1, George Hu 1, Harshith Padigela 1, Andrew Walker 1, Harsha Pokkalla 1, Darren Fahy 1, Geetika Singh 1, Chinmay Surve 2, Natoria Wade 2, Archit Khosla 1, Mary Lin 1, Michael Montalto 1, Andrew Beck 1, Jimish Mehta 1, Ilan Wapinski 1, Michael Drage 1, Fedaa Najdawi 1, Christina K.B. Jayson
1PathAI, Boston, Massachusetts
2PathAI Diagnostics, Memphis, Tennessee
