Talk Summary:
Molecular imaging provides quantitative, biologically specific measurements that are central to preclinical and translational research, in vivo therapy monitoring, biomarker development, and therapeutic discovery. AI methodologies are revolutionizing many areas of medical image analysis; however, effective use of AI models requires understanding of both its promise and challenges in particular applications. In molecular imaging, challenges to AI methods include smaller datasets, limited labeled data, heterogeneous acquisition protocols, and substantial domain shift across scanners, tracers, and clinical settings. This educational lecture will review data-efficient and generalizable AI approaches that can be leveraged for molecular imaging, with emphasis on segmentation of PET images and batch effects in cellular imaging. Topics will include weak supervision, semi-supervised learning, self-supervised learning, foundation models, domain adaptation and validation strategies.

Presenter Biography:
Dr. Tasdizen is a Professor and Associate Chair of Electrical and Computer Engineering and a faculty member of the Scientific Computing and Imaging (SCI) Institute at the University of Utah. A Senior Member of IEEE, he has served as Senior Area Editor (2019–2024) and Associate Editor (2016–2019) for IEEE Transactions on Image Processing, and as Associate Editor for IEEE Signal Processing Letters. He was named an Outstanding Editorial Board Member by IEEE Transactions on Image Processing in 2024 and received the NSF CAREER Award in 2012. Dr. Tasdizen’s research focuses on AI and machine learning for image analysis, with applications in biomedical imaging (radiology, pathology, cardiology), public health, and materials science. His contributions span self- and semi-supervised learning, domain adaptation, and model interpretability. He has led multiple federally funded research projects from NIH, NSF, DOE, and DOD, and has authored over 100 peer-reviewed publications. A recognized educator and mentor, Dr. Tasdizen has received multiple awards for teaching excellence and advised numerous graduate students and postdocs. He has extensive experience in professional service, including program chair and area chair roles at major conferences (MICCAI, MIDL, ISBI), and as General Chair for MIDL 2025.

Author
Tolga Tasdizen
University of Utah