GraphMamba: Whole slide image classification meets graph-driven selective state space model

Open Webpage

Tingting Zheng, Hongxun Yao, Sicheng Zhao, Kui Jiang, Yi Xiao

Published: 2025, Last Modified: 16 Jan 2026Pattern Recognit. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•We design GraphMamba to model long-range relations and spatial structure for WSI analysis.•GraphMamba couples graph and Mamba to enhance local tissue structure and global relations learning.•Intra-group graph Mamba explores the correlations among instances within each group.•Cross-group feature sampling extracts discriminative features across groups.•Inter-group graph Mamba boosts comprehensive bag representation for better prediction.