GAIA Workshop@ICCV25

News

Workshop website launched with Call-for-Papers and speakers announced.
Paper submission site is now open via OpenReview.
Paper submission deadline: August 20, 2025 (Extended to August 30, 2025).
Workshop scheduled for October 20, 2025 in Honolulu, Hawaii.

Introduction

Generative AI is transforming biomedical image analysis, creating new possibilities and solutions for healthcare. Although generative AI has significantly advanced medical imaging and diagnostics, developing reliable, clinically applicable systems remains challenging due to interpretability concerns, data quality issues, and regulatory compliance.

This workshop explores how generative AI is reshaping biomedical image analysis across three critical areas:

(1) Data Synthesis and Clinical Modeling: Generative models revolutionize training data creation and disease simulation by producing anatomically accurate images, addressing class imbalances, and enabling cross-modal image synthesis (e.g., MRI to CT). These models also simulate disease progression, empowering clinicians to visualize patient outcomes and evaluate treatment effectiveness. Additionally, conditional generative models enhance segmentation accuracy, while synthetic lesion generation enriches training datasets. Ensuring clinical reliability, reducing biases, and meeting regulatory standards remain essential challenges.

(2) Multimodal Learning: Integrating generative AI with large language models (LLMs) combines visual data with insights from medical reports and electronic health records, enabling systems to extract crucial information and generate informative summaries. This fusion enhances clinical communication and supports improved decision-making. However, significant challenges, such as interpretability, mitigating AI-generated inaccuracies, and aligning with clinical standards, must be addressed.

(3) Workflow Automation: Generative AI streamlines medical imaging workflows from acquisition to diagnosis. Intelligent AI agents automate tasks such as routine medical inspections, automated image analysis, and automated delineation of radiotherapy target areas. These advancements can significantly improve efficiency and consistency in clinical practices. Nevertheless, challenges related to regulatory approval, data privacy, and model reliability persist.

Our workshop brings together experts from computer vision, healthcare, and AI research to address these challenges and opportunities in applying generative AI to biomedical image analysis through interdisciplinary collaboration.

Invited Speakers


Dimitris N. Metaxas	Kun-Hsing Yu
Distinguished Professor Rutgers University	Associate Professor Harvard Medical School


Daguang Xu	Akshay Chaudhari
Principal Research Scientist NVIDIA	Assistant Professor Stanford University

Call for Papers

We invite submissions of full-length papers (up to 8 pages excluding the references, 4-6 pages recommended) for workshop proceedings. The topics covered in the workshop include but are not limited to:

Medical Image Generation & Synthesis
Vision-Language Foundation Models
Clinical Workflow Intelligence
Generative Disease Dynamics
Trustworthy Medical AI
LLM-Enhanced Clinical Reasoning
Distributed Medical Imaging Systems
Generative Surgical Simulation
Multimodal learning for medical image analysis
AI agents for healthcare applications

Submission Instructions

All submissions should follow the ICCV 2025 instructions. The papers will be subject to a double-blind review process, i.e. authors must not identify themselves on the submitted papers. The reviewing process is single-stage without rebuttals.

📝 Submit Your Paper

Submit via OpenReview

Online Submission System: OpenReview
Submission Format: official ICCV 2025 template (double column; up to 8 pages, 4-6 pages recommended, excluding references).

All authors submitting a paper are required to have an OpenReview profile. New profiles with institutional emails are automatically activated, while those without one undergo a moderation process, taking up to two weeks.

Timeline Table (11:59 PM, Pacific Time)

Paper submission open: June 15, 2025
Paper submission deadline: August 20, 2025
Extended deadline: August 30, 2025
Notification to authors: September 20, 2025
Camera-ready deadline: October 10, 2025
Workshop: October 20, 2025

Workshop Schedule

Half-day workshop schedule (Room 316 B):

Time (HST / PST / EST)	Event
09:00 – 09:10 12:00 – 12:10 / 3:00 – 3:10	Opening remarks and introduction
09:10 – 09:40 12:10 – 12:40 / 3:10 – 3:40	Invited Talk 1: Akshay Chaudhari (Stanford University) Title: Generative Image and Text Models for Biomedical Image Analysis Abstract: TBD
09:40 – 10:10 12:40 – 1:10 / 3:40 – 4:10	Invited Talk 2: Kun-Hsing Yu (Harvard Medical School) Title: Generative Artificial Intelligence for Cancer Pathology Diagnosis Abstract: Artificial intelligence (AI) is reshaping the landscape of cancer research and clinical diagnosis. Recent advances in microscopic image digitization, multi-modal machine learning, and scalable computing have enabled AI-powered pathology at an unprecedented scale. In this talk, I will highlight recent breakthroughs in pathology foundation models and generative AI techniques for interpreting high-resolution digital pathology images. In addition, I will present examples of real-time, AI-assisted pathology evaluations during cancer surgery, illustrating their adaptability to evolving diagnostic classifications. Furthermore, I will discuss recent approaches aimed at ensuring that AI-based diagnostic systems benefit patients across all demographic groups. Finally, I will outline persistent challenges in building robust medical AI systems and identify research directions to address these pressing issues.
10:10 – 10:40 1:10 – 1:40 / 4:10 – 4:40	Oral Session 1: Emmanuelle Bourigault (Oxford) Title: X-Diffusion: Generating Detailed 3D MRI Volumes from a Single Image using Cross-Sectional Diffusion Models Abstract: TBD Oral Session 2: Mainak Biswas (Indian Institute of Science) Title: Conditional Latent Diffusion Models reveal Structural Connectivity correlates of Brain Age Abstract: TBD
10:40 – 11:10 1:40 – 2:10 / 4:40 – 5:10	Coffee break
11:10 – 11:40 2:10 – 2:40 / 5:10 – 5:40	Invited Talk 3: Dimitris N. Metaxas (Rutgers University) Title: TBD Abstract: TBD
11:40 – 12:10 2:40 – 3:10 / 5:40 – 6:10	Invited Talk 4: Daguang Xu (NVIDIA) Title: Building Foundation Models for Generative AI in Healthcare Abstract: TBD
12:10 – 12:15 3:10 – 3:15 / 6:10 – 6:15	Closing remarks

Workshop Organizers


Yuanfeng Ji	Zhongying Deng	Xiangde Luo
Postdoctoral Researcher Stanford University	Postdoctoral Researcher University of Cambridge	Postdoctoral Researcher Stanford University


Jin Ye	Xiyue Wang	Dan Lin
Ph.D. Student Monash University	Postdoctoral Researcher Stanford University	Postdoctoral Researcher Cornell University


Junjun He	Jianfei Cai	Angelica I Aviles-Rivero
Researcher Shanghai AI Laboratory	Professor Monash University	Assistant Professor Tsinghua University


Carola-Bibiane Schönlieb	Shaoting Zhang	Ping Luo
Professor University of Cambridge	Principal Scientist Shanghai AI Laboratory	Associate Professor University of Hong Kong