A dynamic CT-based morphometric investigation of the aorta

Morphometric evaluation is an important indicator to understand the mechanisms of cardiovascular function. This investigation plays a key role if images are acquired in a non-invasive way under the concept of personalized treatment planning and ongoing disease monitoring: the extraction of morphometric features can facilitate the comprehension of several cardiovascular diseases in terms of progression and onset. This thesis presents an advanced image-based morphometric framework designed to quantify the geometric features of the thoracic aorta and its coupling with the left ventricle. The entire analysis is performed by processing high-resolution ECG-gated CT imaging during the cardiac cycle. In fact, in this work the thoracic aorta and left ventricle district across 20 distinct phases of the cardiac cycle were evaluated, allowing a comprehensive and dynamic view of morphological changes over time. By using an automatic in-house deep learning algorithm to process the CT images, the 3D models of both regions of interest were extracted. The morphometric analysis was based on the vessel centreline and on its associated features such as the relative angles between specific tracts. In particular, the tilt angle, considered as the angle between two consecutive tangent vectors, and the twist angle, defined as the angle between the normal vectors were calculated. In conclusion, by combining classic geometric parameters with temporal angular analysis, this work offers a robust tool for assessing the morphometric properties and dynamic behaviour of the left ventricular and thoracic aortic region. This framework can be applied to various research settings where detailed geometric characterizations are necessary and may help in detecting early structural alterations that predispose to pathology, supporting preventative strategies and enhancing patient-specific care.