PathScale-R1: Cross-scale Reasoning for Pathological Image Analysis

2026-07-26

AI Summary

This work addresses the limitations of existing pathological visual question answering (VQA) methods, which are often confined to single-scale analysis and susceptible to textual or superficial visual shortcuts, thereby hindering clinically relevant cross-scale reasoning. To overcome these challenges, the authors introduce PathScale-VQA, the first shortcut-resistant, cross-scale pathological VQA benchmark, along with the PathScale-R1 model. They construct a high-quality dataset via adversarial text-only filtering and structure-controlled perturbation sampling, and propose a scale-aware reasoning framework incorporating a structural reward mechanism. This framework leverages difficulty-driven reasoning distillation and reinforcement learning to optimize multi-scale evidence integration. Experimental results demonstrate that PathScale-R1 achieves state-of-the-art performance on cross-scale pathological reasoning tasks and generalizes effectively to conventional single-scale VQA settings.

Abstract

Pathological diagnosis is inherently multi-scale, requiring the integration of global tissue architecture at low magnification with cellular morphology at higher magnification. However, existing pathology benchmarks and vision-language models (VLMs) are still largely developed under single-scale settings, limiting their ability to learn clinically meaningful multi-magnification reasoning. Moreover, naively constructed visual question answering (VQA) tasks may be susceptible to text-only or superficial visual shortcuts, leading to unreliable assessments of visual understanding. To address these limitations, we introduce a benchmark and training framework for shortcut-resistant cross-scale pathology reasoning. We design an Adversarial Text-only Screening strategy for semantic reasoning questions and a Structure-controlled Distractor Sampling strategy for visual grounding questions, encouraging models to rely on cross-scale visual evidence. Based on this pipeline, we construct PathScale-VQA, a high-quality cross-scale pathology VQA benchmark with 10,373 multiple-choice questions grounded in 1,368 diagnostic paths across multiple magnification levels. Building on the semantic reasoning set, PathScale-R1 is optimized through Difficulty-driven Reasoning Distillation supervised fine-tuning followed by reinforcement learning with a Scale-aware Reasoning Structure reward, which encourages the use of evidence across magnifications. Extensive experiments demonstrate state-of-the-art performance of PathScale-R1 on cross-scale reasoning tasks and effective transfer to conventional single-scale pathology VQA. Our code is available at https://github.com/iMVR-PL/PathScale-R1.

Problem

• cross-scale reasoning
• pathological image analysis
• vision-language models
• visual question answering
• shortcut learning

Innovation

• cross-scale reasoning
• vision-language models
• pathology VQA
• adversarial screening
• reinforcement learning