PathMem: Toward Cognition-Aligned Memory Transformation for Pathology MLLMs

TL;DR

PathMem introduces a memory-augmented multimodal framework for pathology large language models, enabling structured knowledge integration and improved diagnostic reasoning aligned with human cognition, leading to state-of-the-art performance.

Contribution

The paper presents PathMem, a novel memory-centric architecture that models hierarchical knowledge organization and dynamic memory transition for pathology MLLMs.

Findings

Achieves state-of-the-art results on WSI-Bench report generation.

Improves open-ended diagnosis accuracy by nearly 9%.

Enhances reasoning capabilities with structured knowledge grounding.

Abstract

Computational pathology demands both visual pattern recognition and dynamic integration of structured domain knowledge, including taxonomy, grading criteria, and clinical evidence. In practice, diagnostic reasoning requires linking morphological evidence with formal diagnostic and grading criteria. Although multimodal large language models (MLLMs) demonstrate strong vision language reasoning capabilities, they lack explicit mechanisms for structured knowledge integration and interpretable memory control. As a result, existing models struggle to consistently incorporate pathology-specific diagnostic standards during reasoning. Inspired by the hierarchical memory process of human pathologists, we propose PathMem, a memory-centric multimodal framework for pathology MLLMs. PathMem organizes structured pathology knowledge as a long-term memory (LTM) and introduces a Memory Transformer that models the dynamic transition from LTM to working memory (WM) through multimodal memory activation and context-aware knowledge grounding, enabling context-aware memory refinement for downstream reasoning. PathMem achieves SOTA performance across benchmarks, improving WSI-Bench report generation (12.8% WSI-Precision, 10.1% WSI-Relevance) and open-ended diagnosis by 9.7% and 8.9% over prior WSI-based models.