Medical Image Analysis: Towards Translating Personalized Biomarkers to Clinical Care

Speaker: Eranga Ukwatta

Affiliation: Sunnybrook Research Institute

Medical image segmentation, a process of partitioning an image into multiple meaningful regions, is an important step in clinical workflows designed to extract quantitative information from medical images for diagnosis and treatment of diseases. In this talk, I will describe several image segmentation algorithms based on convex max-flow formulations’ that were developed for patient-specific analysis and modeling of cardiovascular structure and function. In particular, I will present a number of multi-region-based segmentation methods for generating morphological measurements of atherosclerotic plaque burden in the arteries using non-invasive imaging techniques, including 3D ultrasound and magnetic resonance (MR) imaging. My current research work aims to develop an image processing pipeline for building personalized computational models of the heart for simulation of cardiac electrophysiology. These virtual models can be non-invasively interrogated to gain mechanistic insights into electrical activity of the heart, and has potential to be utilized in the clinic for numerous applications, such as cardiac risk stratification and prediction of target locations for cardiac ablations. Finally, I will demonstrate several applications of image processing algorithms in other clinical domains, including prostate and brain imaging.