What question did this study set out to answer?

To enhance cancer prediction and subtype classification in whole slide images using a multi-directional approach.

January 17, 2026

Multi‐Directional Context Modeling With HCSMIL : Enhancing Cancer Prediction and Subtype Classification From Whole Slide Images

Key Points

To enhance cancer prediction and subtype classification in whole slide images using a multi-directional approach.
Developed HCSMIL, a framework based on the Mamba model for analyzing whole slide images.
Implemented multi-directional contextual modeling to capture spatial dependencies.
Utilized a multi-scale pyramid structure to extract features of global lesion distribution.
Validated the model using authoritative datasets such as Camelyon16 and TCGA.
HCSMIL achieved notable improvements in accuracy, F1 score, and AUC compared to existing methods.
On TCGA-LUNG, HCSMIL improved accuracy by 0.66% and F1 score by 1.42%.
On TCGA-Kidney, improvements of 1.47% in accuracy and 1.00% in AUC were noted.
The model recognized small lesions at an 84% rate, surpassing both TransMIL and MambaMIL.

Abstract

ABSTRACT The Mamba model performs excellently in natural image processing but faces limitations in analyzing whole slide images (WSIs) for cancer prediction and subtype classification in digital pathology—pathological images feature highly irregular lesion spatial distributions (especially complex small‐lesion associations), while Mamba's inherent unidirectional/limited‐direction scanning cannot effectively model such multi‐dimensional spatial dependencies, failing to capture key pathological structural features. To address this, we propose HCSMIL, a Mamba‐based optimized framework tailored to pathological image clinical analysis. It comprehensively captures local lesion spatial topology via multi‐directional contextual modeling and integrates a multi‐scale pyramid structure to extract global lesion distribution features, jointly enhancing diagnostic accuracy. Validation on authoritative datasets (Camelyon16, TCGA‐LUNG, TCGA‐Kidney) shows HCSMIL significantly outperforms existing mainstream methods: on TCGA‐LUNG, accuracy (ACC), F1 score, and AUC are 0.66%, 1.42%, and 1.25% higher than the second‐best method; on TCGA‐Kidney, these metrics increase by 1.47%, 0.09%, and 1.00%; on Camelyon16, ACC is 0.77% higher. Notably, HCSMIL achieves an 84% small‐lesion recognition rate, substantially exceeding TransMIL (70.59%) and MambaMIL (64.71%), fully demonstrating its strength in capturing complexly distributed lesions and providing reliable technical support for cancer diagnosis.

Mark Helpful

Bookmark

Relay