Patient-Specific Cerebral Flow Model Using Regional Flows and Multi-Objective Optimization

Background: Some cerebral flow models have good accuracy in predicting patient outcome, but are too complicated to be readily duplicated by others. Others are simpler, but lack accuracy in utilizing patient-specific boundary conditions. Methods: A new patient-specific cerebral flow model aimed at both simplicity and accuracy was designed and applied to predict stump pressure (SP) during balloon test occlusion (BTO). The new model simulates both a baseline and an internal carotid artery (ICA) occlusion flow model. The former involves building a novel patient-specific cerebral flow model with regional flows, where the resistances of all inlet and internal vessels were obtained using a multi-objective optimization algorithm; regional blood flows were calculated using vessel flows measured from quantitative magnetic resonance angiography (QMRA). The ICA occlusion flow model computes the new blood flows and pressures of efferent, inlet and internal vessels with the simulated occlusion of the ICA, while keeping the resistances of the peripheral, inlet and internal vessels constant. Results: The model was applied to predict SPs of four patients undergoing BTO. When aortic pressures are used, the simulated SPs demonstrate -11% to 7% error when compared to actual clinical measurements. When cuff pressures are used to approximate aortic pressures, the errors of the corresponding SPs becomes -19% to 1%. Conclusions: The proposed model flow was validated with both clinically measured blood flows and SPs. Even when cuff pressures were used to approximate aortic pressures, the reliable predicted SPs were achieved. The model may be promising for clinical use. J Neurol Res. 2021;000(000):000-000 doi: https://doi.org/10.14740/jnr671