From Multi-Resolution Cells to Gigapixel Whole Slide Images Foundation Model for Computational Pathology

2026-08-04

AI Summary

This work addresses the limited generalizability of existing computational pathology models, which are typically trained at a single resolution and thus struggle to process whole-slide images (WSIs) of arbitrary scales or effectively integrate diagnostic information across cellular, tissue, and slide-level contexts. To overcome this, the authors propose the Multi-Resolution Pyramid Transformer (MRPT)—the first foundation model for computational pathology capable of accepting inputs at any resolution. Built upon the Vision Transformer architecture, MRPT introduces a novel continuous cross-resolution attention mechanism and a multi-resolution semantic alignment strategy. By hierarchically aggregating multi-scale features and modeling scale-invariant interactions, MRPT substantially enhances generalization performance. Evaluated across 34 diverse datasets, it consistently outperforms current foundation models and multimodal large language models in tasks including cancer subtyping, histological phenotyping, and WSI-based visual question answering.

Abstract

Vision Transformers (ViTs) and their hierarchical variants have achieved strong performance in Computational Pathology (CPath). However, most are pre-trained on single-resolution Whole Slide Images (WSIs), limiting their generalization across arbitrary resolutions. Gigapixel WSIs inherently contain diagnostic patterns at multiple scales, including cellular morphologies, tissue architectures, and global context, mirroring how expert pathologists examine WSIs. We introduce Multi-Resolution Pyramid Transformer (MRPT), a model that hierarchically aggregates multi-resolution information from cellular to tissue and WSI levels. MRPT employs a biologically meaningful Consecutive Cross-Resolution Attention (CCRA) mechanism to capture scale-independent interactions and enforces multi-resolution semantic consistency by aligning embeddings across resolutions, yielding robust and generalizable WSI representations. Pre-trained in a multi-resolution self-supervised manner on 624M patches, 2.4M regions, and 36K WSIs, MRPT learns rich coarse-to-fine histopathology features. Extensive experiments on 34 diverse datasets show that MRPT surpasses recent foundation models and Multimodal Large Language Models (MLLMs) in cancer subtype classification, tissue phenotyping, and Visual Question Answering (VQA) for WSI understanding.

Problem

• Computational Pathology
• Whole Slide Images
• Multi-Resolution
• Gigapixel
• Scale Generalization

Innovation

• Multi-Resolution Pyramid Transformer
• Consecutive Cross-Resolution Attention
• Computational Pathology
• Whole Slide Image
• Self-Supervised Pretraining