MedAD-R1: Eliciting Consistent Reasoning in Interpretible Medical Anomaly Detection via Consistency-Reinforced Policy Optimization

2026-02-01

AI Summary

Existing medical anomaly detection models struggle with reliable reasoning and multimodal generalization due to their reliance on fragmented data. To address this, this work introduces MedAD-38K, the first large-scale, multimodal, multicenter benchmark for medical anomaly detection, and proposes a two-stage training framework. The framework first aligns structured reasoning with responses through cognitive injection and then refines logical coherence via a consistency reinforcement strategy. The core innovation is the Con-GRPO algorithm, which uniquely integrates a consistency-based reward mechanism into policy optimization to ensure high coherence between generated reasoning and final diagnoses. Built upon large multimodal models and leveraging supervised fine-tuning, Chain-of-Thought annotations, and structured visual question answering, the proposed method, MedAD-R1, achieves state-of-the-art performance on MedAD-38K, surpassing strong baselines by over 10% and significantly enhancing both accuracy and interpretability in medical anomaly detection.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Medical Anomaly Detection (MedAD) presents a significant opportunity to enhance diagnostic accuracy using Large Multimodal Models (LMMs) to interpret and answer questions based on medical images. However, the reliance on Supervised Fine-Tuning (SFT) on simplistic and fragmented datasets has hindered the development of models capable of plausible reasoning and robust multimodal generalization. To overcome this, we introduce MedAD-38K, the first large-scale, multi-modal, and multi-center benchmark for MedAD featuring diagnostic Chain-of-Thought (CoT) annotations alongside structured Visual Question-Answering (VQA) pairs. On this foundation, we propose a two-stage training framework. The first stage, Cognitive Injection, uses SFT to instill foundational medical knowledge and align the model with a structured think-then-answer paradigm. Given that standard policy optimization can produce reasoning that is disconnected from the final answer, the second stage incorporates Consistency Group Relative Policy Optimization (Con-GRPO). This novel algorithm incorporates a crucial consistency reward to ensure the generated reasoning process is relevant and logically coherent with the final diagnosis. Our proposed model, MedAD-R1, achieves state-of-the-art (SOTA) performance on the MedAD-38K benchmark, outperforming strong baselines by more than 10\%. This superior performance stems from its ability to generate transparent and logically consistent reasoning pathways, offering a promising approach to enhancing the trustworthiness and interpretability of AI for clinical decision support.

Problem

• Medical Anomaly Detection
• Interpretable Reasoning
• Multimodal Generalization
• Consistency
• Chain-of-Thought

Innovation

• Consistency-Reinforced Policy Optimization
• Chain-of-Thought Reasoning
• Medical Anomaly Detection
• Multimodal Benchmark
• Interpretable AI