Structural analysis of coronary microcirculation in normal and diseased hearts

Abstract

The coronary circulation is a network of vessels that supplies oxygenated blood to the myocardium and removes the by-products of metabolism. While this coronary perfusion is fundamental to maintaining cardiac performance, the heart's mechanical contraction interferes with blood ow. Despite several investigations over a prolonged period, understanding of how mechanical contraction in uences coronary blood ow remains limited. In part, this limitation exists because we cannot easily measure mechanical activity and coronary blood ow in vivo within the heart wall. A practical alternative to direct measurement is to model the myocardial-vessel interactions mathematically. Focused on this goal, researchers have developed a variety of mathematical models throughout the twentieth century. However, a major constraint on model development is the dearth of detailed data on the vessel anatomy and the myocyte structures within which it is embedded. While we possess excellent descriptions of structure at all levels of the coronary vascular network, we still lack systematic, quantitative data about the three-dimensional relationship between cardiac muscle cells and the microcirculatory networks that deliver blood to them. To investigate this relation, the construction of representative, large scale anatomic models of coronary microcirculation is primary. Employing recent advancements in tissue clearing and high-throughput imaging, our group simultaneously acquired myocytes and microcirculation images throughout the entire short-axis slices of the rat heart. This thesis sought to extract accurate information from these large-scale images and quantitatively assess the anatomical relationship between cardiomyocytes and coronary microcirculation. A chunk-based image segmentation pipeline was developed to process large-scale images and combined with an e cient graph generation technique developed to extract networks from the segmented images. Next, the dominant orientation of the myocytes was computed using structure tensor analysis. Finally, the orientation of the myocytes and microcirculation were compared and statistically analysed throughout the six short-axis cardiac ventricular slices from the rat, including three normal and three diseased hearts. The study identi ed that the microcirculatory vessels were closely aligned with the myocytes in both the normal and hypertensive hearts. Comparing the diseased and normal data sets revealed that the microvessels follow the pathophysiologically induced changes of cardiomyocytes in the event of cardiac remodelling. Additionally, this study showed that the capillaries which were misaligned with the myocytes were shorter than the well-aligned capillaries. From analysis on small blocks, the study also found that the hypertensive hearts had a higher variance of oxygen di usion distance compared to their normotensive counterparts. Irregular vascular patterns were also observed in diseased hearts. To our knowledge, this study is the rst time that such large-scale datasets of coronary microcirculation and cardiomyocytes have been collected, processed, and statistically analysed. To achieve this goal, we developed tools and techniques to process and analyse large-scale microscopy images and created and demonstrated novel methods to repair vessels networks and register multiple images with no regions of mutual information. The tools and techniques from this study will serve other medical image investigations, and the wealth of data from this study will pave the way for numerous related researches.