Comparison of different clinical models of cerebral autoregulation in time and frequency domain

Abstract

Cerebral Autoregulation (CA) is a control mechanism adjusting cerebral vasomotor tone in response to changes in arterial blood pressure (ABP), in order to ensure a nearly constant cerebral blood flow (CBF). CA is often impaired after severe craniocerebral injury or subarachnoidal hemorrhage. Patient treatment could be optimized, if monitoring of CA would be possible. Various methods of assessment of CA using spontaneous slow fluctuations of blood flow velocity (FV), arterial blood pressure, and cerebral perfusion pressure, have been used in clinical practice. This paper compares several approaches in time and frequency domain and analyses their mutual relationships. We analysed digital recordings from patients with severe head injury retrospectively. Two lumped parameter models, transfer function approach, crosscorrelation approach, and two kinds of autoregulation index (ARI) are compared. Furthermore, we analysed the relationship between above mentioned parameters and patient age, Glasgow Coma Scale (GCS) Grade and arterial pCO2 level. From 120 analysed parameter pairs we have found 23 pairs with strong correlation (Spearman’s rho > 0.281 p < 0.01), 10 pairs correlated 95 - 99% significant and 9 pairs 90 - 95% significant. Alternative parameters from Windkessel models are well correlated with clinically established ones.