AI4Rad Lab
Medical imaging AI research at Shanghai Sixth People's Hospital, Shanghai Jiao Tong University School of Medicine.
News
- 2026.08 AbdomenNet — a foundation model for acute abdomen diagnosis stratification and triage on non-contrast CT — published in Nature Communications (IF 18.1). Paper · Code
- 2026.07 Imaging-Grounded Stroke MLLM — a multi-agent multimodal LLM framework for imaging-grounded treatment recommendation in acute ischemic stroke — published in npj Digital Medicine. Paper
- 2026.07 BoneCoT — a whole-body skeleton foundation model for bone metastases guided by clinician-derived chain of thought — accepted by Nature Biomedical Engineering (IF 26.3). Paper · Demo
- 2026.07 HR-LLM-Stroke — a multi-agent LLM framework for emergency stroke treatment recommendation — published in Journal of Medical Internet Research. Paper · Demo
- 2026.06 LumbarSR — paired clinical CT & micro-CT dataset for lumbar vertebra super-resolution — published in Scientific Data. Paper · Challenge
- 2026.04 BrainMIND — benchmarking LLMs for brain MRI diagnostic impression generation — published in npj Digital Medicine. Paper · Code
Research
Our research is built on four core AI capabilities, applied across the lab's seven clinical imaging directions.
- Vision foundation models — visual representation and structured understanding of medical images: diagnosis, segmentation, detection, and quantification.
- Multimodal large language models integrating images, reports, clinical history, and knowledge for diagnostic reasoning and structured reporting.
- Medical image generation — clinically useful synthesis, restoration, denoising, and enhancement, including cross-modal generation.
- Spatial intelligence & physical simulation — 3D modeling, registration, blood-flow and musculoskeletal mechanics, and interpretable quantitative analysis.
Selected Projects
Imaging-Grounded Stroke MLLM — A multi-agent multimodal LLM framework that coordinates multimodal imaging interpretation, contraindication screening, and treatment recommendation for acute ischemic stroke. npj Digital Medicine, 2026. Paper
HR-LLM-Stroke — A multi-agent large language model framework for emergency treatment recommendation in acute ischemic stroke. Journal of Medical Internet Research, 2026. Paper · Code · Demo
BrainMIND — A multicenter benchmark and reader study evaluating 10 LLMs on diagnostic impression generation from brain MRI reports. npj Digital Medicine, 2026. Paper · Code
Cerebrovascular & Stroke AI — Aneurysm detection on CTA, automatic ASPECTS scoring, and thrombus characterization in acute stroke. Radiology / European Radiology.
Join Us
We recruit PhD and Master students in medical imaging AI and machine learning. Send your CV to zhangrp@sjtu.edu.cn.