Queen Mary University of London researchers develop AI technique to rapidly reconstruct blood flow patterns in coronary arteries

15 March 2024


New method eliminates time-consuming computational fluid dynamics calculations.

Researchers at Queen Mary University of London have developed a simple and rapid technique to reconstruct 3D pressure and shear stress fields in coronary arteries. This new method, published in Frontiers in Cardiovascular Medicine, could revolutionise the diagnosis and treatment of atherosclerosis, by providing cardiologists with critical information about blood flow patterns within minutes.

Atherosclerosis is a condition that causes plaque build-up in the arteries, which can lead to heart attacks and strokes. Abnormal blood flow patterns are strong predictors of atherosclerotic plaque rupture, which is a serious complication of the disease.

Traditionally, blood flow patterns have been analysed using 3D-imaging-based computational fluid dynamics (CFD), which is a computationally expensive process. However, Professor Sergey Karabasov and Professor Rob Krams from Queen Mary’s School of Engineering and Materials Science have developed a new AI method that combines data-reduction techniques and non-linear regression tools from the machine learning paradigm to expedite the reconstruction of 3D pressure and shear stress fields.

“Our new method is able to reconstruct 3D blood flow patterns in less than 10 seconds, a significant improvement over traditional CFD methods,” said Prof Krams. “This could have a profound impact on the diagnosis and treatment of cardiovascular diseases, as it will allow cardiologists to obtain critical information about blood flow patterns much more quickly.”

The new method involves generating a large dataset of blood flow simulations in an atherosclerotic pig coronary artery with random perturbations introduced for the mesh geometries. This dataset is then processed using the Proper Orthogonal Decomposition (POD) method to obtain Eigen functions and principal coefficients, which are then used to train a Random Forest Regressor model. The model can then be used to predict the pressure and shear stress fields in unseen arterial geometries with an accuracy of over 75%.

“This is a major breakthrough in the field of coronary artery hemodynamics,” said Prof Karabasov. “Our new method is not only faster, but it is also more accurate than traditional CFD methods. This could be a game-changer for the diagnosis and treatment of cardiovascular diseases.”

The ability to rapidly calculate biomechanical parameters in blood vessels could lead to the development of new diagnostic tools and treatments for atherosclerosis and other cardiovascular diseases, giving patients a better chance of a longer and healthier life. The method could also be used to optimise the design of stents and other medical devices for coronary artery interventions.

Contact:Ayden Wilkes
Research Centre: