Periodic Reporting for period 1 - CARDIOTRIALS (Using CARDIac simulations to run in-silicO clinical TRIALS)
Période du rapport: 2022-10-01 au 2025-03-31
Clinical trials are a key tool for advancing medical knowledge, but they consist of a long and costly process entailing the recruitment of a representative cohort of participants to properly account for the population statistical variability. Computational engineering is a promising approach to gain more insight into patients’ cardiac pathologies and to test innovative medical devices before running conclusive in–vivo experiments on animals or medical trials on humans. This technological breakthrough, however, is limited by some technical and epistemic challenges: (i) the reliability of cardiovascular computational models depends on the accurate solution of the hemodynamics coupled with the deforming biologic tissues; (ii) the resulting multi-physics solver requires immense computational power and long time–to–results; (iii) a great variability among individuals exists, thus calling for a statistical approach.
The ERC project CARDIOTRIALS is accomplishing and employing a computational platform for determining the outcome of pathologies or device implantation by combining my GPU–GPU-accelerated multi-physics solver for the accurate solution of cardiac dynamics with an uncertainty quantification analysis to account for the individual's variability. Simulation campaigns will then be run to obtain the synthetic population response in the case of selected pathologies like myocardial infarction and the optimal stimulation pattern for cardiac resynchronization therapy. Our approach removes the main barrier that keeps up a systematic use of computational engineering to run in–silico clinical trials.
The ERC project CARDIOTRIALS is accomplishing and employing a computational platform for determining the outcome of pathologies or device implantation by combining my GPU–GPU-accelerated multi-physics solver for the accurate solution of cardiac dynamics with an uncertainty quantification analysis to account for the individual's variability. Simulation campaigns will then be run to obtain the synthetic population response in the case of selected pathologies like myocardial infarction and the optimal stimulation pattern for cardiac resynchronization therapy. Our approach removes the main barrier that keeps up a systematic use of computational engineering to run in–silico clinical trials.
Several scientific and technical activities are run simultaneously by the CARDIOTRIALS research group.
We have created a semi-automatic procedure for segmenting CT scans to get cardiac anatomies timely. The procedure combines automatic algorithms (running in a few seconds) and human operations for better accuracy and assessment of the results.
The resulting cardiac anatomies are the input computational domain of our multiphyics solver. The latter is a groundbreaking computational model of the whole human heart, including all four chambers, valves, and the initial sections of major arteries and veins. The model simulates the complex mechanics of heart dynamics, including blood flow, muscle motion, and electrical signals, using high-performance-computing. To ensure high accuracy, it handles up to one billion spatial degrees of freedom and half a million-time steps per heartbeat.
First, we have used the model to replicate the cardiac dynamics of a healthy heart. Then, to simulate the left bundle branch block condition by disrupting the heart’s electrical signals between the atrioventricular node and the left bundle branch. This causes a decline in cardiovascular performance, reflecting real-life clinical symptoms. Pacemaker therapy is then simulated to resynchronize the heart. By testing different positions for the pacemaker lead in the left ventricle, the study generates a small-scale clinical trial to analyze therapy outcomes.
The model is also applied to study the effects of aortic stenosis (AS), a condition where the aortic valve narrows, restricting blood flow. The simulations reveal how AS increases blood velocity and pressure differences across the valve, known as transvalvular pressure drop (TPD). Severe cases result in peak jet velocities of 4.9 m/s and TPDs of 42.5 mmHg, aligning well with clinical and experimental data. Additionally, the model measures wall shear stress (WSS), a factor that cannot be directly observed in patients. High WSS levels are linked to severe AS, particularly in the aortic valve and ascending aorta, potentially damaging red blood cells and activating platelets, which raises the risk of blood clots.
Finally, the model has been exploited to study heart performance after an ischemic event (heart attack), where part of the heart muscle is damaged due to a lack of oxygen. This damaged region affects the heart muscle’s ability to contract and conduct electrical signals. The study examines various scenarios by changing the size and location of the damaged area, depending on which coronary artery is blocked. It evaluates heart efficiency in terms of blood pressure differences, cardiac output, and WSS, aiming to identify key factors influencing disease progression and detection.
In summary, this advanced heart model offers valuable insights into cardiac disorders and therapies, helping to improve the understanding and treatment of conditions like electrophysiology disfunctions, aortic stenosis, and ischemic heart disease.
We have created a semi-automatic procedure for segmenting CT scans to get cardiac anatomies timely. The procedure combines automatic algorithms (running in a few seconds) and human operations for better accuracy and assessment of the results.
The resulting cardiac anatomies are the input computational domain of our multiphyics solver. The latter is a groundbreaking computational model of the whole human heart, including all four chambers, valves, and the initial sections of major arteries and veins. The model simulates the complex mechanics of heart dynamics, including blood flow, muscle motion, and electrical signals, using high-performance-computing. To ensure high accuracy, it handles up to one billion spatial degrees of freedom and half a million-time steps per heartbeat.
First, we have used the model to replicate the cardiac dynamics of a healthy heart. Then, to simulate the left bundle branch block condition by disrupting the heart’s electrical signals between the atrioventricular node and the left bundle branch. This causes a decline in cardiovascular performance, reflecting real-life clinical symptoms. Pacemaker therapy is then simulated to resynchronize the heart. By testing different positions for the pacemaker lead in the left ventricle, the study generates a small-scale clinical trial to analyze therapy outcomes.
The model is also applied to study the effects of aortic stenosis (AS), a condition where the aortic valve narrows, restricting blood flow. The simulations reveal how AS increases blood velocity and pressure differences across the valve, known as transvalvular pressure drop (TPD). Severe cases result in peak jet velocities of 4.9 m/s and TPDs of 42.5 mmHg, aligning well with clinical and experimental data. Additionally, the model measures wall shear stress (WSS), a factor that cannot be directly observed in patients. High WSS levels are linked to severe AS, particularly in the aortic valve and ascending aorta, potentially damaging red blood cells and activating platelets, which raises the risk of blood clots.
Finally, the model has been exploited to study heart performance after an ischemic event (heart attack), where part of the heart muscle is damaged due to a lack of oxygen. This damaged region affects the heart muscle’s ability to contract and conduct electrical signals. The study examines various scenarios by changing the size and location of the damaged area, depending on which coronary artery is blocked. It evaluates heart efficiency in terms of blood pressure differences, cardiac output, and WSS, aiming to identify key factors influencing disease progression and detection.
In summary, this advanced heart model offers valuable insights into cardiac disorders and therapies, helping to improve the understanding and treatment of conditions like electrophysiology disfunctions, aortic stenosis, and ischemic heart disease.
This research shows advances in simulating the mechanics of heart tissues, electrical activity in the heart, and blood flow. These parts have been combined to create a tool to predict heart functioning in healthy and diseased conditions. The heart model includes three main parts: (i) the mechanics of heart tissues, influenced by blood flow and electrical signals; (ii) the electrical network that drives heart contractions; and (iii) blood flow, which has complex and turbulent behaviour that requires advanced simulations. The resulting model can accurately reproduce heart dynamics by combining these parts with sophisticated numerical techniques.
Hence, how these parts are linked and solved together in the model is an important research achievement as it opens the way to reach the objectives of the CARDIOTRIALS project. Indeed, the computational model has been and is currently exploited to unravel the Physics of cardiac dynamics as well as to investigate pathologies, devices and surgical procedures. Consequently, this precious tool opens for investigations that were beyond imagination just a few years ago. The small clinical trials we have carried out are just a proof-of-concept on how the model can push further the state-of-the-art.
Hence, how these parts are linked and solved together in the model is an important research achievement as it opens the way to reach the objectives of the CARDIOTRIALS project. Indeed, the computational model has been and is currently exploited to unravel the Physics of cardiac dynamics as well as to investigate pathologies, devices and surgical procedures. Consequently, this precious tool opens for investigations that were beyond imagination just a few years ago. The small clinical trials we have carried out are just a proof-of-concept on how the model can push further the state-of-the-art.