A Subject-Specific Cerebrovascular CFD Modeling Approach Based on a Multimodal Data-Driven Boundary Calibration Framework: A Proof-of-Concept Study

Cerebrovascular computational fluid dynamics (CFD) models often rely on generic boundary conditions, which may limit their ability to represent subject-specific hemodynamics and cerebral autoregulation (CA). We propose a multimodal data-driven boundary calibration (MDBC) framework integrating transcranial color-coded Doppler and continuous blood pressure monitoring to optimize individualized outlet resistances. As a proof-of-concept, we evaluated the MDBC framework in a single healthy volunteer at resting baseline and enhanced external counterpulsation (EECP)—a hemodynamic perturbation potentially triggering CA. Compared with conventional open boundary (OB) and static Murray allocation boundary (SMAB) strategies, MDBC achieved closer agreement with in vivo middle cerebral artery (MCA) velocity waveforms under both states. At rest, MDBC’s left MCA relative root mean square error (rRMSE) was 7.19%, versus 22.85% (OB) and 30.89% (SMAB). During EECP, conventional models yielded rRMSEs > 32%, whereas MDBC maintained 11.24%. Meanwhile, MDBC reproduced inter-hemispheric perfusion imbalance, an EECP-induced flow surge in the right MCA, and pronounced wall shear stress increases that were masked by generic boundary strategies. Moreover, MDBC estimated a 25.8% increase in global cerebrovascular resistance during EECP, suggesting the capability of the framework to characterize subject-specific impedance adaptations potentially associated with CA during intervention. These single-subject findings support the technical feasibility of integrating multimodal physiological measurements into cerebrovascular CFD boundary calibration and warrant further validation in larger cohorts and patient populations.