3DMedAgent: Unified Perception-to-Understanding for 3D Medical Analysis

3D CT analysis spans a continuum from low-level perception to high-level clinical understanding. Existing 3D-oriented analysis methods adopt either isolated task-specific modeling or task-agnostic end-to-end paradigms to produce one-hop outputs, impeding the systematic accumulation of perceptual evidence for downstream reasoning. In parallel, recent multimodal large language models (MLLMs) exhibit improved visual perception and can integrate visual and textual information effectively, yet their predominantly 2D-oriented designs fundamentally limit their ability to perceive and analysis volumetric medical data. To bridge this gap, we propose 3DMedAgent, an unified agent that enables 2D MLLMs to perform general 3D CT analysis without 3D-specific fine-tuning. 3DMedAgent coordinates heterogeneous visual and textual tools through a flexible MLLM agent, progressively decomposing complex 3D analysis into tractable subtasks that transition from global to regional views, from 3D volumes to informative 2D slices, and from visual evidence to structured textual representations. Central to this design, 3DMedAgent maintains a long-term structured memory that aggregates intermediate tool outputs and supports query-adaptive, evidence-driven multi-step reasoning. We further introduce the DeepChestVQA benchmark for evaluating unified perception-to-understanding capabilities in 3D thoracic imaging. Experiments across over 40 tasks demonstrate that 3DMedAgent consistently outperforms general, medical, and 3D-specific MLLMs, highlighting a scalable path toward general-purpose 3D clinical assistants. Code and data are available \href{https://github.com/jinlab-imvr/3DMedAgent}{here}.

CT scans are widely used in modern medicine, but reviewing them is time-consuming and requires doctors to inspect many image slices carefully. This paper presents 3DMedAgent, an AI agent designed to help analyze 3D CT scans more effectively. Instead of processing the whole scan at once, the agent breaks the task into smaller steps. It first gathers basic information about organs, then searches for relevant regions or lesions, and finally inspects selected slices when more evidence is needed. 3DMedAgent stores the evidence during the process dynamically and thus answering different types of medical questions, such as measuring organs, recognizing lesions, comparing visual findings, and supporting higher-level medical reasoning. We also introduce a new chest CT benchmark to better evaluate these abilities. Experiments show that 3DMedAgent performs better than existing general, medical, and 3D-focused AI models across diverse 3D medical imaging tasks. This work suggests a practical path toward AI assistants that can support doctors in complex 3D scans analysis.