ClinFusion: A Vision-Centric Multimodal LLM System for Holistic Medical Understanding

2026-07-27

AI Summary

This work addresses key vision-centric challenges in deploying multimodal large models for clinical settings, particularly insufficient understanding of heterogeneous 2D/3D medical images and the absence of fine-grained, clinically aligned evaluation protocols. The authors propose a cascaded spatial-aware local fusion encoder to unify native 2D and 3D medical image modeling and introduce a Vision-Grounded evaluation framework—comprising MedIF-Bench and region-of-interest (RoI)-grounded metrics—that enables, for the first time, automatic, fact-based, and clinically aligned assessment grounded in anatomical regions. Integrated with retrieval augmentation and agent-based tool invocation, the model outperforms existing open-source counterparts on 20 out of 24 benchmarks and surpasses GPT-5.2 and Gemini-3-Flash on 13 out of 16 tasks. Blind evaluations by radiologists confirm its superior report quality, with RoI-grounded metrics showing the strongest correlation with expert judgments.

Abstract

Multimodal large language models (MLLMs) hold immense potential to revolutionize clinical practice, yet deploying them in the medical domain is fundamentally a vision-centric challenge: models must absorb knowledge from heterogeneous 2D and 3D medical images, and evaluation protocols must align with radiologists' clinical practice and provide an accurate, fine-grained and factualness-driven assessment. In this paper, we introduce ClinFusion, a vision-centric MLLM designed for holistic medical understanding that systematically addresses these limitations. We propose a compositional and cascaded vision encoder architecture featuring a Cascade Spatial-Aware Locality Fusion operator that unifies diverse 2D and native 3D medical image understanding within a fused encoder. We further introduce a vision-grounded evaluation framework, including MedIF-Bench for instruction-following assessment and a region-of-interest-grounded method for clinically aligned and factualness-driven report generation evaluation. We show that ClinFusion sets a new state-of-the-art across a comprehensive suite of 2D and 3D multimodal medical benchmarks---spanning visual question answering, report generation, and instruction following---as well as textual medical tasks, outperforming leading open-source medical MLLMs (\textit{e.g.}, Hulu-Med, Lingshu) on 20 out of 24 benchmarks and demonstrating multimodal capabilities better than powerful proprietary models such as GPT-5.2 and Gemini-3-Flash on 13 out of 16 benchmarks, and can be further augmented with agentic tool use for retrieval-augmented and tool-assisted clinical workflows. A blinded evaluation by board-certified radiologists confirms that ClinFusion produces the highest-ranked reports, and validates our RoI-grounded metric as achieving the strongest correlation with expert judgment among all automatic evaluation metrics examined.

Problem

• vision-centric
• multimodal large language models
• medical image understanding
• clinical evaluation
• factualness-driven assessment

Innovation

• vision-centric MLLM
• Cascade Spatial-Aware Locality Fusion
• 3D medical image understanding
• RoI-grounded evaluation
• multimodal medical benchmarking