Spatiotemporal Representation Learning for Cardiac Function Estimation in Medical Imaging

PhD Qualifying Examination


Title: "Spatiotemporal Representation Learning for Cardiac Function Estimation 
in Medical Imaging"

by

Mr. Leong Chit Jeff KWAN


Abstract:

Left ventricular ejection fraction (LVEF) is a central clinical index for 
assessing cardiac systolic function. It is measured from medical images that 
depend on ventricular anatomy, cardiac phase, imaging modality, acquisition 
quality, and observer convention. LVEF estimation is a useful lens for 
studying spatiotemporal representation learning in cardiac AI: the task is 
narrow enough to support benchmarking, yet rich enough to expose how deep 
learning models capture anatomy, motion, and function in dynamic cardiac data. 
This survey reviews the cardiac imaging literature through spatial 
segmentation, temporal phase modelling, integrated LVEF estimation, toward 
broader cardiac modelling. Across these areas, the literature suggests that 
effective cardiac AI requires both scale and structure. Supervised learning 
utilizes task-specific learning signals, foundation models improve transfer 
across heterogeneous data, and inductive priors help preserve clinically 
meaningful anatomical and temporal organization. Using LVEF estimation as a 
methodological probe, the proposed research direction asks how the above 
factors inform an integrated paradigm of accurate, robust, and clinically 
interpretable spatiotemporal representation learning.


Date:                   Friday, 17 July 2026

Time:                   2:00pm - 4:00pm

Venue:                  Room 5501
                        Lift 25/26

Committee Members:      Prof. Albert Chung (Supervisor)
                        Dr. Dan Xu (Chairperson)
                        Dr. Ouyang Xiaomin