Dr. FangXiang Wu

Bio: Dr. FangXiang Wu is currently a full professor in the Departments of Computer Science, Division of Biomedical Engineering, and the Department of Mechanical Engineering at the University of Saskatchewan. He is a Fellow of the Engineering Institute of Canada (EIC) (for his exceptional contributions to artificial intelligence, computational biology, medical image analysis and complex bionetwork analytics), a Fellow of the Institute of Electrical and Electronics Engineers (IEEE) (for contributions to computational intelligence techniques for biomedical data analytics), a Fellow of the Institution of Engineering and Technology (IET), and a Fellow of the Asia-Pacific Artificial Intelligence Association (AAIA). He is a recipient of the University of Saskatchewan Distinguished Researcher Award (the institution’s highest research honor).

Dr Wu’s research interests include Artificial Intelligence, Machine/Deep Learning, Computational Biology, Health Informatics, Medical Image Analytics, and Complex Network Analytics. He has published over 500 journal/conference papers. His total Google scholar citations are over 20,200 and h-index is 77. Dr Wu is serving as the editorial board member of several international journals (including IEEE TCBB, Neurocomputing, etc.) and as the guest editor of numerous international journals, and as the program committee chair or member of many international conferences.

Title: Artificial Intelligence Empowered Fast PET Imaging: From physics-guided reconstruction to adaptive image-domain denoising

Abstract: Positron emission tomography (PET), as an advanced molecular imaging technique in nuclear medicine, can provide information on human metabolism and physiological function, and plays an irreplaceable role in the diagnosis of cancer, neurological disorders, and cardiovascular diseases. However, conventional PET scans are often time-consuming, relatively inefficient, and associated with limited patient comfort. Therefore, the development of fast PET imaging is of great clinical significance. Nevertheless, shortened scan time also leads to several challenges, including reduced photon counts, increased Poisson noise, blurred structural boundaries, and unstable SUV quantification. This talk focuses on the key challenges in low-count PET imaging and systematically introduces our group’s recent research progress in PET image reconstruction and post-processing algorithms. It will demonstrate how deep learning promotes the transition of fundamental nuclear medicine imaging algorithms from traditional mathematical modeling toward the integration of physical constraints and deep neural networks. The talk will also discuss how artificial intelligence can advance fast PET imaging toward higher image quality, improved efficiency, and greater intelligence while preserving PET physical interpretability and clinical quantitative reliability.