Research Information System
The 28th International Conference Mechanika 2024, Kaunas, Lithuania, 31 May 2024, pp.132-136, (Full Text)
The aim of this study is to predict post-op and pre-op hemodynamic characteristics of the blood flow in the carotid artery with the machine learning (ML) and computational fluid dynamic analysis (CFD). The use of artificial intelligence algorithms has a significant impact on the accurate prediction and assessment of carotid artery blood flow for hemodynamic parameters, both pre- and post-operations. Machine learning predictions provide clinicians with valuable information about the impact of carotid artery disease on blood flow, which will be important in clinical decision-making. The augmentation of patient data sets enables more precise predictive models to be developed for machine learning. The integration of ML and computational fluid dynamics (CFD) analyses significantly enhances clinical decision-making and patient outcomes. This is achieved by enabling the analysis and prediction of hemodynamic parameters, such as blood flow rate, pressure and wall shear stresses, both before and after surgical operations. Furthermore, ML provides a solution as an alternative to CFD, reducing the burden on CFD resources. In this study, CFD results of average velocity compared with the machine learning results of the patient-specific carotid artery for the pre-operation case as contour view. The velocity values for the contour graphics are obtained by point cloud method. Also, for the post-operation case the predicted average velocity values compared with CFD analysis for eight different cross-sections. The DNN model predicts the average velocity values with a good accuracy. The model was written in Python and used the Keras library. The open-source software OpenFOAM was used for analyzing computational fluid dynamics.