When algorithms serve the heart – a new era in coronary diagnostics

Coronary artery stenosis – when your heart needs a plumber
Our heart, which pumps blood all the time, needs the constant supply of oxygen and nutrients. This is provided by coronary arteries – small but crucial vessels that surround the heart muscle. When atheromas cause those vessels to narrow, for instance as a result of the plaque breaking up and closing the lumen of the vessel, this may cause a heart attack.
The narrowing of the arteries is a bit like when pipes get blocked – you need a plumber to clear them. The same has to be done with narrowed coronary vessels. Modern technologies allow for using transcutaneous techniques, without opening the chest.
Despite considerable success in treating coronary syndromes, one of the major questions in modern cardiology is: Which atheroma poses enough of a threat to require invasive treatment, and when is pharmacotherapy enough? With the development of computer-based technologies, we are getting closer to finding the answer and figuring out the characteristics of atheromas that are the most likely to rupture. And non-invasive diagnostics, which carries less risk, has been playing an increasing role in evaluating coronary arteries.
From anatomy to physiology – a paradigm shift
For decades, coronarography has been the main tool in assessing coronary arteries; it involved taking images of the interior of the vessels using a contrast medium delivered directly to the vessel. Coronarography is an invasive procedure and has certain limitations: it shows the shape and extent of the stenosis, but does not say anything about its functional impact. Research has shown that for moderate stenoses, not only the extent of the narrowing is important, but also how it affects the blood flow. That is why, in recent years, techniques based on physiological parameters have been gaining popularity, as they are helpful in assessing whether a given stenosis actually limits blood flow. Those measurements are routinely made to complement coronarography. However, the development of technology has allowed for obtaining results for those indicators in a less invasive way.
Those indicators may be estimated with high accuracy through the use of new non-invasive techniques based on the principles of fluid biomechanics and digital reconstruction of vessels, often supported by artificial intelligence. This helps both in choosing the treatment path and in evaluating the risk of pain exacerbation or the occurrence of clinical events. Evaluating physiology with this method could potentially also help in diagnosing diseases of the microvessels in the heart, which are an important part of coronary circulation but are not visible in diagnostic imaging.
A non-invasive future for the diagnostics of atherosclerotic disease
Modern medicine aims at limiting the use of invasive techniques to the necessary minimum. Increasingly precise, computed tomography has been a comprehensive diagnostic tool for patients with coronary syndromes. What up until recently required inserting a pressure sensor into the artery, can be achieved today using totally non-invasive computed tomography. Moreover, the procedure also allows for evaluating the atheroma for high-risk features. Tomography allows for conducting early diagnostics, choosing the right preventive strategy, planning interventions with precision, and avoiding unnecessary invasive procedures.
Can biomechanics help? New concepts
The ever broader access to comprehensive software and algorithms has made it possible for researchers to look at the role of the biomechanics of coronary blood flow, i.e. the impact of shear force on how fast atherosclerosis progresses and the increased vulnerability of atheromas to rupture.
We can explain how shear force works on the example of our hands rubbing the tabletop. As the surface of the table resists, our hands feel friction. This is shear force, which works parallel to the surface and makes the layers “stretch out” alongside one another. Similarly, when blood flows along the walls of the vessel, there is friction between them, which influences many biological processes such as the activation of endothelial cells or the regulation of vessel tension.
The unfavorable characteristics of shear forces in narrowed arteries may impair the function of the endothelium, which in turn facilitates the progression of atherosclerotic disease and exacerbates blood flow problems. Understanding the role of shear force in this process may help researchers develop new diagnostic and therapeutic methods that will better predict the risk of heart attack. Yet there are no convincing data at present regarding the relationship between the biomechanical profile and the presence of higher-risk atheromas as identified in a non-invasive procedure. We hope that the research projects I am involved in will offer new discoveries in this area.
Artificial intelligence – support for the physician
Taking full advantage of the potential of imaging and physiological analysis requires immense computing power and fast data analysis. This is where artificial intelligence steps in. Machine learning models are already able to recognize subtle patterns in coronary artery images, predict the risk of cardiac incidents, and even suggest the best treatment strategy. Another key role of artificial intelligence is in supporting non-invasive diagnostics, e.g. in analyzing flow dynamics. This allows for simulating the result of a procedure even before the physician reaches the cath lab. Supported by the smart algorithm, the physician can quickly and precisely adjust the treatment to the patient’s needs.
Imaging, data, and AI – towards personalized diagnostics of the heart
The research I have been working on at the 1st Chair and Department of Cardiology of the MUW under the supervision of my advisor, Professor Mariusz Tomaniak, focuses on the comprehensive assessment of stenoses in coronary arteries.
I analyze the relationship between the properties of atheroma visible in computed tomography and the biomechanical profile of the flow. I also evaluate the effectiveness of software in obtaining a physiological profile directly from computed tomography in patients with severe aortic stenosis who are deemed eligible for transcutaneous valve replacement.
The goal is to be able to ensure that each patient gets personalized treatment based on multi-parameter analysis, especially when it is non-invasive. At the same time, the department has been working on developing an artificial intelligence algorithm that will allow for non-invasive assessment of the function of coronary microvasculature.
Research on the comprehensive assessment of atheroma characteristics, as well as the bold application of computer-based techniques in this area offer a real opportunity for better, less invasive and more accurate diagnostics. It may be possible to develop a new standard in cardiac care – data-driven, supported by technology, and patient-oriented.
About the author
Adrian Bednarek is a fourth-year student of medicine and a doctoral student at the 1st Chair and Department of Cardiology at the Medical University of Warsaw. Author of 16 research papers (total IF score of 48.5), he has received the Scholarship of the Minister of Health for his academic achievements, and a grant from the Ministry of Science and Higher Education for the use of AI in diagnosing microvascular dysfunction. Winner of the Great Synapse, a national competition in physiology, and prizewinner in the national Scapula Aurea anatomy competition. He has represented the MUW at the Medical School of Your Future (a camp for the best students of medicine) on two occasions, and has been the president of the Student Cardiology Association. As a student, he has done internships and has been collaborating with the CORRIB CoreLab at the University of Galway (Ireland). He has presented his papers at numerous international conferences, including in France, South Korea, and China. Apart from AI and cardiology, his research interests include research methodology and statistics.