INTERACT-CMIL: Multi-Task Shared Learning and Inter-Task Consistency for Conjunctival Melanocytic Intraepithelial Lesion Grading

TL;DR

This paper presents INTERACT-CMIL, a multi-task deep learning framework that improves grading accuracy of conjunctival melanocytic intraepithelial lesions by jointly predicting multiple diagnostic axes with cross-task consistency, trained on a multi-center dataset.

Contribution

It introduces a novel multi-head deep learning model with shared feature learning and inter-dependence loss for multi-criteria pathology grading, advancing automated ocular pathology diagnosis.

Findings

Achieved up to 55.1% relative macro F1 improvement over baselines.

Provided coherent and interpretable multi-criteria predictions.

Established a reproducible benchmark for CMIL diagnosis.

Abstract

Accurate grading of Conjunctival Melanocytic Intraepithelial Lesions (CMIL) is essential for treatment and melanoma prediction but remains difficult due to subtle morphological cues and interrelated diagnostic criteria. We introduce INTERACT-CMIL, a multi-head deep learning framework that jointly predicts five histopathological axes; WHO4, WHO5, horizontal spread, vertical spread, and cytologic atypia, through Shared Feature Learning with Combinatorial Partial Supervision and an Inter-Dependence Loss enforcing cross-task consistency. Trained and evaluated on a newly curated, multi-center dataset of 486 expert-annotated conjunctival biopsy patches from three university hospitals, INTERACT-CMIL achieves consistent improvements over CNN and foundation-model (FM) baselines, with relative macro F1 gains up to 55.1% (WHO4) and 25.0% (vertical spread). The framework provides coherent, interpretable multi-criteria predictions aligned with expert grading, offering a reproducible computational benchmark for CMIL diagnosis and a step toward standardized digital ocular pathology.