Graph Laplacian Transformer with Progressive Sampling for Prostate Cancer Grading

2025-12-11

AI Summary

Prostate cancer whole-slide image (WSI) grading faces challenges including extreme image scale, high tissue heterogeneity, and difficulty in localizing diagnostically critical regions; existing methods suffer from redundant information due to static patch sampling, limiting performance. We propose a Graph Laplacian-constrained Transformer architecture with a progressive iterative sampling mechanism—the first to jointly integrate graph filtering, learnable spatial regularization, and importance scoring from a frozen foundation model—enabling dynamic selection of diagnosis-relevant regions and spatially consistent modeling. Our method combines ResNet50-based local feature extraction, convex combination aggregation, and Graph Laplacian regularization. Evaluated on five public and one private dataset, it consistently outperforms state-of-the-art approaches, achieving significant improvements in grading accuracy and spatial structural fidelity while maintaining efficient inference.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Computer Vision: Other Foundations of Computer Vision
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Prostate cancer grading from whole-slide images (WSIs) remains a challenging task due to the large-scale nature of WSIs, the presence of heterogeneous tissue structures, and difficulty of selecting diagnostically relevant regions. Existing approaches often rely on random or static patch selection, leading to the inclusion of redundant or non-informative regions that degrade performance. To address this, we propose a Graph Laplacian Attention-Based Transformer (GLAT) integrated with an Iterative Refinement Module (IRM) to enhance both feature learning and spatial consistency. The IRM iteratively refines patch selection by leveraging a pretrained ResNet50 for local feature extraction and a foundation model in no-gradient mode for importance scoring, ensuring only the most relevant tissue regions are preserved. The GLAT models tissue-level connectivity by constructing a graph where patches serve as nodes, ensuring spatial consistency through graph Laplacian constraints and refining feature representations via a learnable filtering mechanism that enhances discriminative histological structures. Additionally, a convex aggregation mechanism dynamically adjusts patch importance to generate a robust WSI-level representation. Extensive experiments on five public and one private dataset demonstrate that our model outperforms state-of-the-art methods, achieving higher performance and spatial consistency while maintaining computational efficiency.

Problem

• Prostate cancer grading from whole-slide images is challenging due to large-scale data and heterogeneous tissue structures.
• Existing methods suffer from redundant or non-informative patch selection, degrading performance.
• The paper aims to enhance feature learning and spatial consistency by selecting relevant tissue regions.

Innovation

• Graph Laplacian Attention Transformer for tissue connectivity
• Iterative refinement module for relevant patch selection
• Convex aggregation for robust whole-slide representation