MedVLM-R1: Incentivizing Medical Reasoning Capability of Vision-Language Models (VLMs) via Reinforcement Learning
read the original abstract
Reasoning is a critical frontier for advancing medical image analysis, where transparency and trustworthiness play a central role in both clinician trust and regulatory approval. Although Medical Visual Language Models (VLMs) show promise for radiological tasks, most existing VLMs merely produce final answers without revealing the underlying reasoning. To address this gap, we introduce MedVLM-R1, a medical VLM that explicitly generates natural language reasoning to enhance transparency and trustworthiness. Instead of relying on supervised fine-tuning (SFT), which often suffers from overfitting to training distributions and fails to foster genuine reasoning, MedVLM-R1 employs a reinforcement learning framework that incentivizes the model to discover human-interpretable reasoning paths without using any reasoning references. Despite limited training data (600 visual question answering samples) and model parameters (2B), MedVLM-R1 boosts accuracy from 55.11% to 78.22% across MRI, CT, and X-ray benchmarks, outperforming larger models trained on over a million samples. It also demonstrates robust domain generalization under out-of-distribution tasks. By unifying medical image analysis with explicit reasoning, MedVLM-R1 marks a pivotal step toward trustworthy and interpretable AI in clinical practice. Inference model is available at: https://huggingface.co/JZPeterPan/MedVLM-R1.
This paper has not been read by Pith yet.
Forward citations
Cited by 23 Pith papers
-
FETAL-GAUGE: A Benchmark for Assessing Vision-Language Models in Fetal Ultrasound
Fetal-Gauge benchmark shows state-of-the-art vision-language models reach only 55% accuracy on fetal ultrasound tasks, well below clinical needs and highlighting the requirement for domain-adapted models.
-
PlanBench-V: A Spatial Planning Map Benchmark for Vision-Language Models
PlanBench-V is a new benchmark and dataset for evaluating VLMs on spatial planning map interpretation via a four-stage framework of Perception, Reasoning, Association, and Implementation.
-
Does Language Shift Break Medical Vision-Language Models? Indonesian Radiology Visual Question Answering Case Study
Introduces IndoRad-VQA dataset and reports 8-25% performance gap in medical VLMs between English and Indonesian radiology VQA prompts.
-
Weather-R1: Logically Consistent Reinforcement Fine-Tuning for Multimodal Reasoning in Meteorology
Weather-R1 is a multimodal reasoning model for meteorology that uses logical consistency rewards during reinforcement fine-tuning to cut self-contradictory outputs and raises benchmark accuracy by 9.8 points over baselines.
-
FractalMamba++: Scaling Vision Mamba Across Resolutions via Hilbert Fractal Geometry
FractalMamba++ scales Vision Mamba across resolutions by using Hilbert fractal serialization, hierarchy-based skip connections, and fractal-aware 2D rotary position encoding.
-
Look-Closer-Then-Diagnose: Confidence-Aware Ultrasound VQA via Active Zooming
A structured Zoom-then-Diagnose paradigm with uncertainty-aware GRPO rewards improves lesion localization by 39.3% on liver, breast, and thyroid ultrasound VQA datasets by encouraging caution under ambiguity.
-
Look-Closer-Then-Diagnose: Confidence-Aware Ultrasound VQA via Active Zooming
Introduces Zoom-then-Diagnose paradigm and uncertainty-aware reward in GRPO for confidence-aware ultrasound VQA, reporting 39.3% improvement in lesion localization across liver, breast, and thyroid datasets.
-
From Failure to Feedback: Group Revision Unlocks Hard Cases in Object-Level Grounding
A group-revision paradigm for GRPO-based RL fine-tuning of VLMs converts failure responses into improvement signals that refine rewards and advantages, yielding gains on referring segmentation, REC, and counting benchmarks.
-
RadThinking: A Dataset for Longitudinal Clinical Reasoning in Radiology
RadThinking releases a large longitudinal CT VQA dataset stratified into foundation perception questions, single-rule reasoning questions, and compositional multi-step chains grounded in clinical reporting standards f...
-
DIYHealth Suite: Dataset, Model, and Benchmark for Health Management at Home
DIYHealth Suite introduces a large home-care dataset, DIYHealthGPT model with Hybrid Hyper Low-Rank Adaptation, and DIYHealthBench, claiming SOTA results on 11 tasks over general and medical baselines.
-
Navigating the Clutter: Waypoint-Based Bi-Level Planning for Multi-Robot Systems
Waypoint-based bi-level planning with curriculum RLVR improves multi-robot task success rates in dense-obstacle benchmarks over motion-agnostic and VLA baselines.
-
MedRCube: A Multidimensional Framework for Fine-Grained and In-Depth Evaluation of MLLMs in Medical Imaging
MedRCube is a new fine-grained evaluation framework that benchmarks 33 MLLMs on medical imaging, ranks Lingshu-32B highest, and finds a significant positive link between shortcut behaviors and diagnostic performance.
-
Locate-Then-Examine: Grounded Region Reasoning Improves Detection of AI-Generated Images
Locate-Then-Examine improves AI-generated image detection by localizing suspicious regions first then performing region-aware re-examination, while releasing the TRACE dataset of 20k annotated images.
-
HeartcareGPT: A Unified Multimodal ECG Suite for Dual Signal-Image Modeling and Understanding
HeartcareGPT proposes Dual Stream Projection Alignment (DSPA) on a structure-aware tokenizer for unified ECG signal-image modeling, supported by Heartcare-400K dataset and Heartcare-Bench.
-
Real-World Doctor Agent with Proactive Consultation through Multi-Agent Reinforcement Learning
DoctorAgent-RL trains a Qwen2.5-7B doctor agent via multi-agent RL on the new MTMedDialog dataset to conduct dynamic, question-driven consultations, reaching 70% exact diagnostic match in real-patient trials.
-
EchoSonar-R: A Multi-View Reasoning-Enabled Model for Disease Classification and Report Generation in Echocardiography
EchoSonar-R is a multi-view VLM for echocardiography that jointly does disease classification and report generation via SFT followed by GRPO reinforcement learning, reporting accuracy gains on private and public data.
-
Mechanistically Interpreting the Role of Sample Difficulty in RLVR for LLMs
Sample difficulty in RLVR shows non-monotonic effects on LLM reasoning, with easy/medium problems strengthening computation and reasoning features while hard problems often yield weak or harmful signals.
-
RealSR-R1: Reinforcement Learning for Real-World Image Super-Resolution with Vision-Language Chain-of-Thought
RealSR-R1 introduces VLCoT-GRPO with four rewards to add understanding and reasoning to real-world image super-resolution models.
-
Confidence Calibration for Multimodal LLMs: An Empirical Study through Medical VQA
A multi-strategy interrogation method with auxiliary expert assessment reduces expected calibration error by 40% on average across three medical VQA datasets for MLLMs.
-
Vision-EKIPL: External Knowledge-Infused Policy Learning for Visual Reasoning
Vision-EKIPL injects high-quality actions from external models into RL training to expand exploration and raise the reasoning ceiling of MLLMs, reporting up to 5% gains on the Reason-RFT-CoT benchmark.
-
Ask Patients with Patience: Enabling LLMs for Human-Centric Medical Dialogue with Grounded Reasoning
APP is a multi-turn LLM framework for medical dialogue that combines empathetic questioning, Bayesian active learning, and guideline-based reasoning, outperforming baselines on a new simulated-patient benchmark in acc...
-
From System 1 to System 2: A Survey of Reasoning Large Language Models
The survey organizes the shift of LLMs toward deliberate System 2 reasoning, covering model construction techniques, performance on math and coding benchmarks, and future research directions.
-
Multimodal Chain-of-Thought Reasoning: A Comprehensive Survey
The paper provides the first comprehensive survey of multimodal chain-of-thought reasoning, including foundational concepts, a taxonomy of methodologies, application analyses, challenges, and future directions.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.