Federal Grant Award Summary

The University of Michigan received a $777,956 Project Grant award dated March 1, 2026, from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) under the Diabetes, Digestive, and Kidney Diseases Extramural Research program (CFDA 93.847). The project, extending through December 31, 2029, will develop and validate computer vision and machine learning models to improve the detection of dysplasia (pre-cancerous lesions) in patients with inflammatory bowel disease (IBD) during colonoscopy procedures and histologic evaluation. The research addresses a critical clinical need, as IBD-related dysplasia is difficult to detect using conventional colonoscopic visualization due to flat, irregular lesions obscured by inflammation, and existing enhancement techniques are resource-intensive without proven efficacy improvements.

The deliverables include advanced computer vision models trained to match or exceed expert diagnostic capabilities in identifying dysplasia on both colonoscopic imagery and histologic samples. By adapting the research team's prior machine learning methodology developed for endoscopic disease activity assessments, the project will create tools to reduce pathologist diagnostic ambiguity and eliminate the need for expert second-opinion referrals for dysplasia grading. This research directly supports NIDDK's programmatic emphasis on improving prevention, diagnosis, and treatment of digestive diseases through fundamental and clinical investigation.