Three-dimensional, patient-specific simulation of cerebral blood flow velocities as a new approach for individualized stroke prevention and treatment.

BACKGROUND

Three-dimensional simulations of cerebral blood flow are a highly promising approach for individualized diagnostics and therapy planning in cerebrovascular disease. The aim of this study is to validate simulated blood flow velocities generated by a patient-specific model using the innovative Stroke Quantification software (StroQ, ELPIS Simulation GmbH, Germany).

METHODS

In this study, consecutive patients with transient ischemic attack (TIA) or minor ischemic stroke (defined as NIHSS ≤ 4) treated at a tertiary stroke center have been investigated. The patients underwent both computed tomography angiography (CTA) and neurovascular ultrasound (nvUS) of the extra- and intracranial arteries. Patients with relatively large artery atherosclerosis (LAA) such as significant plaques or stenosis are excluded in this study. 3D-models of the brain-supplying arteries were created using standard CTA with 0.75 mm slice thickness. Blood flow of the intracranial arteries was simulated using finite volume and fluid-structure interaction methods and employing nvUS-derived flow velocities and systemic blood pressure of the extracranial internal carotid arteries (ICA) and vertebral arteries (VA). The simulated flow velocities of all intracranial arteries were then compared to the transcranial color-coded duplex sonography TCCD-derived measurements using intraclass correlation coefficients, Pearson correlations, Bland-Altman, and scatter plots.

RESULTS

Patient-specific artery models were created for 53 consecutive ischemic stroke patients without relevant LAA. The results demonstrated a high accuracy between TCCD- and simulation-derived velocities for peak systolic velocities (PSV). Intraclass correlation coefficients were moderate for three arteries and good to excellent for all remaining intracranial arteries (ICC ≥ 0.750, p < 0.001). Overall, only small differences between simulation-derived velocities and TCCD were observed (PSV: mean difference -2.6 cm/s, -4.5 %, end-diastolic velocities (EDV): mean difference 0.5 cm/s, 1.4 % and mean flow velocity (MFV): mean difference -0.3 cm/s, -3.3 %). Pearson correlations between TCCD- and simulation-derived PSV, EDV and MFV velocities were significant for all intracranial arteries (p < 0.001).

CONCLUSIONS

The study revealed accurate flow velocity values derived from the patient-specific artery simulation of the intracranial arteries as compared to the clinical gold standard represented by TCCD. Simulation of cerebral perfusion may support personalized medicine by helping to tailor blood pressure targets and guide decisions on invasive reperfusion strategies (e.g., carotid endarterectomy or stenting), based on individual vascular anatomy and physiological parameters.