UC San Diego

The left atrium (LA) of the heart is implicated in many functional cardiovascular diseases. Global LA function is often measured through noninvasive imaging; however, the local nature of some diseases requires higher-resolution regional measurements to fully characterize changes in function. Atrial fibrillation (AF) produces complex local changes in the LA tissue which increase the risk of stroke, exacerbate heart failure, and limit quality of life. In this thesis, high-resolution measures of LA strain are developed from retrospective 4DCT images. A novel curvilinear coordinate system is devised to compare these metrics across the cardiac cycle, across scans, and across patients. In a pilot study using pre-procedure and post-procedure 4DCT, we demonstrate that the methods can measure the deleterious impact of catheter ablation as local regions of reduced strain. In a separate study, we create a clinical-imaging database from a single-center retrospective study of 385 patients who received a 4DCT scan before AF ablation. Regional strain had the strongest correlation with ablation outcome and added modest predictive value over global LA function and clinical history in machine learning models. Statistical atlases of regional LA strain revealed that regional strain is reduced in persistent vs. paroxysmal AF and in patients who have received prior ablations. The algorithms and database produced by this work can facilitate progress in the understanding of regional LA mechanics as a prognostic indicator and surrogate endpoint for examining the impact of therapy.