Segmentation of CT scans from coronary arteries with deep learning
Loading...
Date
2026-03-03
Authors
Τουμανίδη, Βικτωρία
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Coronary artery disease (CAD) is defined as a group of cardiovascular diseases, including angina, myocardial infarction, and sudden cardiac death, which are caused by the limitation of blood flow to the myocardium, often due to atherosclerosis and cholesterol-rich plaque buildup in the arteries. Unfortunately, CAD is the prevalent cause of mortality worldwide, affecting both developed and developing countries alike.
Visualization of the coronary arteries has been a vital step in assessing and evaluating the severity of coronary artery disease in a patient. Segmentation of the coronary arteries is the cornerstone of this visualization. Furthermore, before the advent of machine learning, this process was conducted by experienced medical staff. This procedure has been tedious and highly detail-oriented. Therefore, mistakes were highly expected.
To limit these errors, artificial intelligence has had a significant impact in recent years, enabling more automated and reliable detection of vascular boundaries. More precisely, the development of artificial neural networks (ANNs) and, by extension, convolutional neural networks (CNNs) has been crucial to the precision and automation of this process.
Consequently, in order to facilitate both the physician's and the patient's experience, we have created an interactive model that utilizes the doctor’s directories to segment the selected artery. Moreover, with just five to ten clicks, the physician can indicate the artery where more information is needed. Hence, the artery is segmented, allowing the doctor to gather the necessary specifications.
In a nutshell, this thesis is a preliminary study into methods that could revolutionize the diagnostic process of cardiovascular diseases. It is a showcase of how technology and medicine are two sciences advancing interdependently, dedicated to improving human health.
Description
Keywords
Coronary artery disease, Artificial intelligence, Convolutional neural networks, Interactive segmentation