Jun Li - Ph.D. Student | Vision-Language Models

Jun Li

Ph.D. Student
Technical University of Munich, Munich Center for Machine Learning
Hobbies: 🛹🎹🚴‍♀️🎧🏋️👩‍💻
🌟 Keep your eyes on the stars, and your feet on the ground.🤞🏻

News

[10.2026] Started a new position in the AI-BAY-HEALTH project on medical foundation models, co-supervised by Prof. Daniel Rueckert, Prof. Julia Schnabel, and Prof. Nassir Navab.

[09.2026] Two papers, DDX-TRACE and RealICU, accepted by NeurIPS 2026 Evaluation Track!

[09.2026] Nominated twice for the MICCAI 2026 Outstanding Reviewer Awards (awards).

[09.2026] Started serving as a reviewer for Medical Image Analysis (MedIA) and IEEE Transactions on Medical Imaging (TMI).

[05.2026] Our paper Dynamic Decision Learning (DDL) was accepted by ICML 2026!

[04.2026] Joined the MCML UK delegation visit to Imperial College London, University of Oxford, University of Cambridge, King's College London, and University College London.

[04.2026] Research visit at Imperial College London, hosted by Prof. Wenjia Bai's group.

[04.2026] Invited visit to Google DeepMind for a research discussion with Rishabh Kabra.

[04.2026] As part of the delegation visit, gave invited talks at Imperial College London, King's College London, and University College London.

[03.2026] Organizing MultiTab 2026, the MICCAI Workshop on Multimodal Learning with Medical Tabular Data.

[12.2025] Selected as a Junior Member Representative of the Munich Center for Machine Learning (MCML).

[09.2025] Our paper NOVA Benchmark accepted as Oral paper by NeurIPS 2025 Datasets and Benchmarks Track!

[09.2025] Our paper K2Sight received Early Accept by WACV 2026 (Top 6.4%)!

[09.2025] Invited by ACM Computing Surveys as reviewer (Impact Factor:28.0).

[08.2025] Honorable Mention: MICCAI 2025 Outstanding Reviewer Award! See MICCAI 2025 Outstanding Reviewer Awards.

[04.2025] My new seminar 📚 AI for Vision-Language Models in Medical Imaging (IN2107, IN45069) is now open! For more details, please visit the course GitHub repository.

[09.2023] My seminar 📘 AI for Vision-Language Pre-training in Medical Imaging (IN2107) is now open! For more details, please visit the course GitHub repository.

[03.2024] Started PhD studies at Technical University of Munich under the supervision of Prof. Julia Schnabel.

[07.2023] Completed Master's degree at University of Chinese Academy of Sciences.

↕️ Scroll

Short Bio

I am currently a Ph.D. student in the School of Computation, Information and Technology, Technical University of Munich, supervised by Prof. Julia Schnabel. I am funded by the Munich Center for Machine Learning (MCML) and work on medical foundation models in the AI-BAY-HEALTH project. Previously, I received the M. Eng. degree from University of the Chinese Academy of Sciences , under the supervison from Prof. Ying Hu.

My research focused on the intersection of deep learning and healthcare, particularly in the analysis of medical images. My passion lies in improving the practicality of deep learning algorithms, with a primary focus on Vision and Language models, Cross-Modality Generation, and Multi-Modality Learning. Through my work in these areas, I aim to advance deep learning techniques and their transformative impact on healthcare.

Publications

DDX-TRACE: A Benchmark for Medical Diagnostic Trajectories in VLMs

Accepted by NeurIPS 2026 Evaluation Track.

Jiazhen Pan, Wenjun Shen, Jun Li, Julian Canisius, Felix Bitzer, Paula Roßmüller, Jiarui Yang, et al.

[paper]

RealICU: Do LLM Agents Understand Long-Context ICU Data? A Benchmark Beyond Behavior Imitation

Accepted by NeurIPS 2026 Evaluation Track.

Chenguang Shen, Wenjun Shen, Tobias Susetzky, Jun Li, Yixin Liu, Xi Zhang, Zeyu Gong, Daniel Rueckert, et al.

[paper]

Dynamic Decision Learning: Test-Time Evolution for Abnormality Grounding in Rare Diseases

Accepted by ICML 2026.

Jun Li, Mingxuan Liu, Jiazhen Pan, Che Liu, Wenjia Bai, Cosmin I. Bercea, Julia A. Schnabel

[paper] [homepage]

ReEvalMed: Rethinking Medical Report Evaluation by Aligning Metrics with Real-World Clinical Judgment

Accepted by EMNLP 2025.

Ruochen Li*, Jun Li*, Bailiang Jian, Kun Yuan, Youxiang Zhu

[paper]

Does DINOv3 Set a New Medical Vision Standard?

Che Liu, Yinda Chen, Haoyuan Shi, Jinpeng Lu, Bailiang Jian, Jiazhen Pan, Linghan Cai, Jiayi Wang, Yundi Zhang, Jun Li, Cosmin I. Bercea, Cheng Ouyang, Chen Chen, Zhiwei Xiong, Benedikt Wiestler, Christian Wachinger, Daniel Rueckert, Wenjia Bai, Rossella Arcucci

[paper]

Knowledge to Sight: Reasoning over Visual Attributes via Knowledge Decomposition for Abnormality Grounding

Accepted by WACV 2026 (Early Accept, Top 6.4%).

Jun Li, Che Liu, Wenjia Bai, Mingxuan Liu, Rossella Arcucci, Cosmin I. Bercea, Julia A. Schnabel

[paper] [homepage]

NOVA: A Benchmark for Anomaly Localization and Clinical Reasoning in Brain MRI

Accepted as Oral paper by NeurIPS 2025 Datasets and Benchmarks Track.

Cosmin I. Bercea, Jun Li, Philipp Raffler, Evamaria O. Riedel, Lena Schmitzer, Angela Kurz, Felix Bitzer, Paula Roßmüller, Julian Canisius, Mirjam L. Beyrle, Che Liu, Wenjia Bai, Bernhard Kainz, Julia A. Schnabel, Benedikt Wiestler.

[paper] [huggingface]

Enhancing Abnormality Grounding for Vision-Language Models with Knowledge Descriptions

Jun Li, Che Liu, Wenjia Bai, Rossella Arcucci, Cosmin I. Bercea, Julia A. Schnabel.

[paper] [project] [huggingface]

Organizing Unstructured Image Collections using Natural Language

Accepted by CVPR 2026 Findings.

Mingxuan Liu, Zhun Zhong, Jun Li, Gianni Franchi, Subhankar Roy, Elisa Ricci.

[paper]

Fmbench: Benchmarking fairness in multimodal large language models on medical tasks

Peiran Wu, Che Liu, Canyu Chen, Jun Li, Cosmin I Bercea, Rossella Arcucci.

[paper]

Language Models Meet Anomaly Detection for Better Interpretability and Generalizability

Accepted by MMMI 2024.

Jun Li, Su Hwan Kim, Philip Mller, Lina Felsner, Daniel Rueckert, Benedikt Wiestler, Julia A. Schnabel, Cosmin I. Bercea.

[paper] [project]

Design as Desired: Utilizing Visual Question Answering for Multimodal Pre-training

Tongkun Su*, Jun Li*, Xi Zhang, Haibo Jin, Hao Chen, Qiong Wang, Faqin Lv, Baoliang Zhao, Yin Hu

[paper] [Code]

Ultrasound Report Generation with Cross-Modality Feature Alignment via Unsupervised Guidance
IEEE Transactions on Medical Imaging (IF:10.6).

Jun Li, Tongkun Su, Baoliang Zhao, Faqin Lv, Qiong Wang, Nassir Navab, Ying Hu, Zhongliang Jiang.

[paper] [project]

A Self-guided Framework for Radiology Report Generation
Accepted by MICCAI 2022. (Early Accept)
(Student Travel Award, Top 5%)

Jun Li, Shibo Li, Ying Hu, Huiren Tao.

[paper] [project]

XctNet: Reconstruction network of volumetric images from a single X-ray image
Computerized Medical Imaging and Graphics (CMIG), 2022.

Zhiqiang Tan, Jun Li, Huiren Tao, Shibo Li, Ying Hu.

[paper]

Teaching

AI for Vision-Language Models in Medical Imaging

(S 25) Master seminar, Technical University of Munich.

Course ID: IN2107, IN45069.

Course Page

AI for Vision-Language Pre-training in Medical Imaging

(S 24/25) Master seminar, Technical University of Munich.

Course ID: IN2107.

Course Page

Records

K2Sight: Knowledge-Guided Abnormality Grounding

WACV 2026 presentation

How to Read a Paper? How to Review a Paper?

Master seminar lecture record

Lecture 1: Introduction to VLMs

AI for VLMs in Medical Imaging

More Records on YouTube

Rio channel

Projects

Awesome Nano Banana for Medical Imaging

A comprehensive collection of resources for Nano Banana framework applied to medical imaging tasks.

Includes implementations, benchmarks, and best practices for medical image analysis.

Github Repo

Reconstruct 3D CT from X-rays

This project focuses on exploring techniques for reconstructing 3D CT images from 2D X-rays.

Project worked at Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences.

Github Repo

Spine Cobb Angle Measurement

Automatic and semi-automatic Cobb angle measurement tool for spinal X-ray images (coronal & sagittal planes), implemented with OpenCV + NumPy.

Project worked at Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences.

Github Repo

Text to Image Generation

Using GAN to generate ultrasound scan from ultrasound repots.

Project worked at Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences.

Github Repo

Services

Teaching:

(S 25) AI for Vision-Language Models in Medical Imaging (IN2107, IN45069)

(S 24/25) AI for Vision-Language Pre-training in Medical Imaging (IN2107)

Reviewer:

Medical Image Analysis (MedIA), Reviewer, 2026

IEEE Transactions on Medical Imaging (TMI), Reviewer, 2026

ACM Computing Surveys (ACM CSUR), Invited Reviewer, 2025. (Impact Factor:28.0)

AAAI Conference on Artificial Intelligence (AAAI), Reviewer

Information Processing in Medical Imaging (IPMI), Reviewer

International Conference on Machine Learning (ICML) 2026

International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2024, 2025, 2026

Workshop Organizer:

MultiTab 2026: MICCAI Workshop on Multimodal Learning with Medical Tabular Data

Roles:

Junior Member Representative, Munich Center for Machine Learning (MCML)

Academic Visits:

MCML UK delegation visit to Imperial College London, University of Oxford, University of Cambridge, King's College London, and University College London, 2026

Research visit at Imperial College London, hosted by Prof. Wenjia Bai's group, 2026

Invited visit to Google DeepMind for a research discussion, 2026

Talks and Presentations:

Research talks during the MCML UK delegation visit at Imperial College London, King's College London, and University College London, 2026

Activities:

International Summer School in Computer Vision (ICVSS), Sicily, 2024