MIRA: Medical Image Reflection for Agentic Diagnosis

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Shengzhi Wang, Jun Yang, Kai Wu, Xiaozhong Ji, Yiwen Ye, Ziyang Chen, Mingliang Xiong, Wen Fang, Mingqing Liu, Mengyuan Xu, Miaoxuan Shan, Caiyan Liu, Bin He, Qingwen Liu
• Venue: arXiv: cs.CV
• Published: 2026-08-11
• Collected: 2026-08-12
• URL: http://arxiv.org/abs/2608.10827v1
• PDF: https://arxiv.org/pdf/2608.10827v1
• DOI: UNKNOWN
• Registry group: academic_apis_and_preprints
• Evidence role: canonical_academic_index
• Language: en
• Country / region: UNKNOWN

Plain-Language Brief

The paper introduces MIRA (Medical Image Reflection for Agentic Diagnosis), a medical visual diagnostic framework designed to autonomously search for evidence and verify its relevance and consistency in image diagnosis. The system uses various image-processing operations and web searches, with the goal of improving the accuracy and usefulness of tool use in medical diagnoses. MIRA was developed using a two-stage training strategy involving a tool-augmented Monte Carlo Tree Search data engine and reinforcement learning.

What the Document Says

• MIRA is a medical visual diagnostic framework for autonomous evidence search and reflective verification.
• It dynamically invokes image-processing operations, web searches, and evaluates the relevance and consistency of the acquired evidence.
• The training strategy involves a tool-augmented Monte Carlo Tree Search data engine and reinforcement learning.
• Across nine medical visual reasoning benchmarks, MIRA improved its Qwen3-VL-8B backbone by 7.44 points.
• It increased useful tool-use judgments from 56.2% to 73.8% and reduced harmful judgments from 8.9% to 1.6%.

Key Concepts Mentioned

• Medical Image Reflection for Agentic Diagnosis (MIRA)
• image-processing operations
• web search
• reliable diagnosis
• tool-augmented Monte Carlo Tree Search
• reinforcement learning
• diagnostic hypotheses
• visual grounding accuracy
• semantic consistency

Claims Made by the Document

• MIRA improves the accuracy and usefulness of tool use in medical diagnoses.
• The two-stage training strategy improves decision-making.

Evidence Used

• Across nine medical visual reasoning benchmarks, MIRA improved its Qwen3-VL-8B backbone by 7.44 points.
• It increased useful tool-use judgments from 56.2% to 73.8% and reduced harmful judgments from 8.9% to 1.6%.

Methods / Approach

• tool-augmented Monte Carlo Tree Search
• reinforcement learning

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

The paper does not explicitly mention any limitations.

Open Questions

• What specific image-processing operations are used by MIRA?
• How is the 'reflective principle evolution' in reinforcement learning implemented?

Notes for Future Q&A

May include interpretation. Labeled where it does.

The paper does not provide detailed information about the specific image-processing operations used by MIRA or how the reflective principle evolution in reinforcement learning is implemented.

Raw Abstract / Excerpt

Medical visual agents can use tools to inspect images and retrieve external knowledge, but indiscriminate tool use may introduce noisy or misleading evidence. Reliable diagnosis therefore requires not only acquiring additional observations, but also verifying whether tool actions are necessary and whether the resulting evidence supports the current hypothesis. We introduce MIRA (Medical Image Reflection for Agentic Diagnosis), a medical visual diagnostic framework for autonomous evidence search and reflective verification. MIRA dynamically invokes image-processing operations, including zooming, grounding, pointing, rotation, and measurement, as well as web search, while evaluating the relevance and consistency of the acquired evidence. We develop MIRA through a two-stage training strategy. First, a tool-augmented Monte Carlo Tree Search data engine explores diverse diagnostic hypotheses and jointly verifies visual grounding accuracy and semantic consistency to construct supervised fine-tuning trajectories. Second, reinforcement learning further improves decision-making through online reflective principle evolution: failure cases are distilled into candidate principles, and only principles that improve held-out rollout rewards are retained. Across nine medical visual reasoning benchmarks, MIRA achieves an average score of 64.73, improving its Qwen3-VL-8B backbone by 7.44 points. It also increases useful tool-use judgments from 56.2% to 73.8% and reduces harmful judgments from 8.9% to 1.6%. Qualitative analyses show that MIRA can re-examine evidence, correct premature conclusions, and adapt its tool-use strategy. Project page: https://MIRA-VL.github.io/