RACR-MIL: Rank-aware contextual reasoning for weakly supervised grading of squamous cell carcinoma using whole slide images

TL;DR

RACR-MIL is a novel weakly-supervised learning framework that improves grading accuracy and tumor localization in squamous cell carcinoma whole slide images by leveraging attention mechanisms, graph modeling, and rank-ordering constraints.

Contribution

It introduces a hybrid graph and rank-aware attention mechanism for SCC grading, enhancing robustness and interpretability over previous methods.

Findings

Achieves 3-9% higher grading accuracy

Improves tumor localization by up to 16%

Enhances grading efficiency in clinical pilot study

Abstract

Squamous cell carcinoma (SCC) is the most common cancer subtype, with an increasing incidence and a significant impact on cancer-related mortality. SCC grading using whole slide images is inherently challenging due to the lack of a reliable protocol and substantial tissue heterogeneity. We propose RACR-MIL, the first weakly-supervised SCC grading approach achieving robust generalization across multiple anatomies (skin, head and neck, lung). RACR-MIL is an attention-based multiple-instance learning framework that enhances grade-relevant contextual representation learning and addresses tumor heterogeneity through two key innovations: (1) a hybrid WSI graph that captures both local tissue context and non-local phenotypical dependencies between tumor regions, and (2) a rank-ordering constraint in the attention mechanism that consistently prioritizes higher-grade tumor regions, aligning with pathologists diagnostic process. Our model achieves state-of-the-art performance across multiple SCC datasets, achieving 3-9% higher grading accuracy, resilience to class imbalance, and up to 16% improved tumor localization. In a pilot study, pathologists reported that RACR-MIL improved grading efficiency in 60% of cases, underscoring its potential as a clinically viable cancer diagnosis and grading assistant.