Original title: Semantic-Aware Subgraph State Space Model for WSI Classification in Histopathology
Feixing Chen, Hao Lu, Lin Luo, Yan Xu · September 2026
Doctors identify diseases by looking for key tissue patterns in huge microscope images called whole slide images. These patterns often cover irregular shapes, but standard computer methods chop the images into rigid grids of square patches. This breaks up important tissue structures and ignores how they fit together. To fix this, researchers built the Semantic-Aware Subgraph State Space Model. First, it groups connected image patches into irregular clusters called subgraphs, using visual clues to capture whole tissue structures. It treats these patches as connected network nodes to keep their spatial layout intact. Next, a graph neural network reads the shape inside each cluster, while a Mamba-based model tracks relationships across many separate clusters across the entire slide. Tests on four slide datasets showed this method consistently beat top existing models. It also stayed robust and learned efficiently when tested with small cohorts and limited training data.
Imagine sorting laundry where each piece of clothing is tangled or has irregular shapes. If you cut each piece into separate squares first, you lose how they were originally folded or connected. Instead, you group them into bundles that keep their natural form. In this paper, tissue patches are like those clothing pieces; the model groups them into semantic units that preserve their spatial layout, then uses a graph to track connections and an efficient method to see the whole laundry pile at once.
Doctors analyzing tissue slides for cancer diagnosis often lose important patterns because current methods break images into tiny pieces. This model keeps those patterns intact, which could lead to more accurate diagnoses, especially when there isn't much data available.
The abstract does not specify limitations of the model. It focuses on showing consistent advantages over existing methods and robustness under small-cohort and few-shot settings, suggesting it may not address all possible scenarios in histopathology.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Semantic-Aware Subgraph State Space Model for WSI Classification in Histopathology — Feixing Chen, Hao Lu, Lin Luo et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.