MONKEY Challenge: Machine-learning for Optimal detection of iNflammatory cells in KidnEY transplant biopsies.
AI successfully identifies immune cells in transplant biopsies with strong agreement to pathologist scoring, paving the way for computer-assisted kidney rejection monitoring.
The MONKEY challenge established the first multicenter PAS-based benchmark for automated inflammatory cell detection in kidney transplant biopsies, with top deep learning models achieving F1=0.75 overall detection performance on 118 biopsies annotated with over 131,000 cells. Integrating the winning model with tissue segmentation demonstrated strong correlation with consensus Banff lesion scoring across 180 independent biopsies, providing a validated foundation for AI-assisted transplant pathology in clinical workflows.
What the study was
- Study design
- multicenter_ml_challenge
- Category
- ai_ml_clinical_diagnostics
- Maturity
- Potentially Practice-Changing
- Journal
- Laboratory Investigation
Why it surfaced
First multicenter benchmark for DL-based Banff scoring in transplant pathology; strong correlation with expert consensus; directly enables standardized AI tools for clinical Banff classification.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.