Fluid-Structure Interaction Analysis of Carotid Artery Blood Flow with Machine Learning Algorithm and OpenFOAM

Abstract

In this study, a patient-specific carotid artery model was analyzed with an open source program foam-extend. The research includes the effect of arterial wall deformation by fluid-structure analysis. Pulsatile velocity cycle is trained for 144 patients with different hemodynamic parameters, by machine learning algorithm using blood flow velocity measured from 337 points of the carotid artery. Data used for training is obtained from an open source in the literature. Here, the machine learning algorithm was created by the help of an open source code Phyton. Then, using trained values of machine learning, and the known systole and diastole blood pressures for a specific chosen patient, the patient-specific pulsatile velocity cycle was estimated. The estimated pulsatile velocity cycle was then fitted to Fourier series. This pulsatile velocity cycle is used as the input boundary condition for the model analyzed in foam-extend. The outlet boundary condition, pulsatile pressure cycle is found by 4-Element Windkessel algorithm. Wall shear stresses and time averaged wall shear stresses were obtained for both the rigid and fluid structure interaction models, and variation of displacement throughout the pulsatile cycle was found for the FSI model. Wall shear stresses, velocity, and displacements were obtained high at peak systole, consistent with pulsatile cycles. Like the wall shear stresses, the time averaged wall shear stresses for the FSI model were also found lower than the rigid model. The wall shear stresses showed an increase towards the exit of internal and external carotid artery.