Special Sessions

Exploring Generative AI in Medical Imaging: Applications, Challenges, and Ethical Considerations

Abstract: Medical imaging is pivotal for accurate disease diagnosis and effective treatment. Generative artificial intelligence (AI) is transforming this domain by enhancing tasks such as data augmentation, image synthesis, and image-to-image translation. These improvements not only bolster model robustness but also generate realistic annotated images for training and facilitate modality conversions. Additionally, generative AI is crucial in in automating diagnostic and radiotherapy processes such as radiology report generation and radiation treatment planning, thereby reducing medical cost and streamlining clinical workflows. Despite these advancements, significant challenges remain. These include data privacy concerns, a scarcity of high-quality datasets due to ethical issues, and the high computational demands of training sophisticated models. Synthetic medical imaging data, generated by AI from existing datasets, offers a promising solution by augmenting and anonymizing real data to enable new applications like modality translation and professional training. However, this approach also presents technical and ethical challenges, such as ensuring the authenticity and diversity of synthesized images, maintaining data privacy, evaluating the performance of models trained on synthetic data, and managing high computational costs. The current regulatory framework is inadequate to ensure the safe and ethical use of synthetic images, making it crucial for regulatory bodies, medical professionals, and AI developers to collaborate on developing and continually refining best practices. This discussion aims to review recent progress in using Generative AI for diagnosis and radiotherapy, highlighting the need for further innovation to enhance AI reliability and accuracy, and to ensure its ethical integration into clinical practice.

Lead Organizer: Xiaofeng Yang, Emory University & Georgia Institute of Technology, USA

NIH Special Session: Funding Opportunities and Grant Writing

Abstract: This NIH outreach session will benefit any investigators in all career stages who are interested in applying for research grants from the National Institutes of Health (NIH). The session will feature experienced researchers and NIH program officials, who will share their advice and insights on how to navigate NIH grant space and write effective grant proposals, with examples of successful applications. The attendees will also learn about current funding opportunities from NIH program officers who will introduce their programs and priorities for the National Institute of Biomedical Imaging and Bioengineering, as well as other trans-NIH programs and opportunities. The panel will also discuss emerging areas of research and strategies for international collaborations and funding. The session will be concluded by a Q&A segment where audiences can ask questions and interact with the panel members. A post-session networking event is planned to facilitate connections between attendees and NIH officials.

The topics would benefit trainees, researchers, and investigators in all the career stages. The spectrum of the presentation and panel discussion covers all the themes of ISBI 2025. Some of the knowledge discussed in the session may be beneficial for grant writing and application beyond NIH opportunities.

Lead Organizers: Qi Duan, NIH, USA; Behrouz Shabestari, NIH, USA

Charting New Territories: Innovations in Spatial Biology and Translational Research

Abstract: This session will spotlight the latest advancements in spatial biology, a rapidly emerging field that merges spatially resolved imaging with molecular and cellular analyses to unlock insights into tissue architecture and function. This session will address the unique analytical challenges posed by high-dimensional spatial data, exploring cutting-edge technologies like single-cell proteomics, spatial transcriptomics, imaging mass cytometry, and multiomics integration.

By bringing together experts in spatial biology and computational biology, the session aligns with the goals of the International Symposium on Biomedical Imaging by fostering discussions on translating spatial insights into clinical applications. This session not only highlights the technological and analytical innovations at the intersection of spatial biology and biomedical imaging but also seeks to foster interdisciplinary collaboration and knowledge exchange for a lasting impact on healthcare.

Lead Organizers: Muhammad Aminu, The University of Texas MD Anderson Cancer Center, USA; Lulu Shang, The University of Texas MD Anderson Cancer Center, USA; Jia Wu, MD Anderson, USA

Unlocking the Potential of Foundation Models in Biomedicine: Opportunities and Challenges

Abstract: Foundation models – the latest generation of AI models, such as ChatGPT and GPT-4o, have achieved remarkable success across various domains, particularly in natural language processing and computer vision. Their application within the biomedical domain offers significant potential to revolutionize healthcare by enhancing diagnostic accuracy, enabling personalized treatment plans, and driving innovative biomedical research. Despite these promising opportunities, several challenges must be addressed to fully harness the benefits of foundation models. These challenges include the necessity for large, diverse datasets to train these models effectively, addressing the unique characteristics of biomedical data, navigating privacy concerns, and ensuring robustness and interpretability in clinical environments. Overcoming these challenges is essential for unlocking the full potential of foundation models in biomedicine and healthcare. To facilitate this, we propose organizing a special session that serves as a platform to explore innovative solutions, share insights, and foster collaboration among researchers, clinicians, and AI experts.

This session seeks to bridge the gap between advanced AI technologies and their practical applications in the biomedical field, ultimately driving advancements in research and improving patient care. Specifically, the session will discuss the following topics: 1) Applications and challenges of foundational AI/ML models in biomedicine and healthcare; 2) Trustworthiness and interpretability of foundational AI/ML models in healthcare. These objectives and topics align seamlessly with the conference’s mission to promote biomedical imaging through the integration of advanced AI technologies.

Lead Organizers: Lu Zhang, Indiana University Indianapolis, USA; Haoteng Tang, The University of Texas Rio Grande Valley, USA; Dajiang Zhu, University of Texas at Arlington, USA; Tianming Liu, University of Georgia, USA; Xiang Li, Mass General Research Institute, USA; Shiaofen Fang, Indiana University Indianapolis, USA; Jingwen Yan, Indiana University Indianapolis, USA; Lichao Sun, Lehigh University, USA; Jinglei Lv, The University of Sydney, Australia; Alex Leow, University of Illinois at Chicago, USA; Islem Rekik, Imperial College London, UK