Abstract
Effective colorectal cancer prevention and management rely
on accurate histopathological assessment of colorectal polyps, including
lesion classification, dysplasia grading, and detection of invasive carcinoma.
The growing volume of colorectal biopsies has increased diagnostic
workload, highlighting the need for automated triage tools capable
of prioritizing clinically significant high-risk cases. While histopathology
Foundation Models (FMs) have demonstrated strong performance across
pathology tasks for patch-level classification, existing slide-level aggregation
approaches may overlook small focal regions of high-grade dysplasia
(HGD) or invasive carcinoma that determine clinical diagnosis. In this
work, we propose an interpretable patch-to-slide aggregation framework
that combines foundation model patch predictions with Density-based
spatial clustering of applications with noise (DBSCAN). The proposed
framework identifies spatially contiguous pathological regions while enabling
simultaneous detection of serrated lesions, dysplasia, and invasive
carcinoma. Evaluating four state-of-the-art (SOTA) FMs, we demonstrate
that spatial aggregation improves high-risk lesion detection while
preserving clinically meaningful interpretability through localized prediction
maps.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/COMPAYL_078.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: https://papers.miccai.org/miccai-2026-sat/supp/COMPAYL_078_supp.pdf
Link to Open Review
BibTex
@InProceedings{MarLia_Explainable_MICCAISAT2026,
author = { Marraffino, Lianna AND Dy, Amanda AND Shafique, Abubakr AND Qin, Xiaoli AND Martel, Anne L. AND Craddock, Kenneth AND Khademi, April},
title = { { Explainable Slide-Level Analysis for High-Risk Disease in Colorectal Polyps using Foundation Models } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17251},
month = {pending},
page = {pending}
}