Rapid pressure-to-flow dynamics of cerebral autoregulation induced by instantaneous changes of arterial CO2

Continuous assessment of CA is desirable in a number of clinical conditions, where cerebral hemodynamics may change within relatively short periods. In this work, we propose a novel method that can improve temporal resolution when assessing the pressure-to-flow dynamics in the presence of rapid changes in arterial CO2. A time-varying multivariate model is proposed to adaptively suppress the instantaneous effect of CO2 on CBFV by the recursive least square (RLS) method. Autoregulation is then quantified from the phase difference (PD) between arterial blood pressure (ABP) and CBFV by calculating the instantaneous PD between the signals using the Hilbert transform (HT). A Gaussian filter is used prior to HT in order to optimize the temporal and frequency resolution and show the rapid dynamics of cerebral autoregulation. In 13 healthy adult volunteers, rapid changes of arterial CO2 were induced by rebreathing expired air, while simultaneously and continuously recording ABP, CBFV and end-tidal CO2 (ETCO2). Both simulation and physiological studies show that the proposed method can reduce the transient distortion of the instantaneous phase dynamics caused by the effect of CO2 and is faster than our previous method in tracking time-varying autoregulation. The normalized mean square error (NMSE) of the predicted CBFV can be reduced significantly by 38.7% and 37.7% (p < 0.001) without and with the Gaussian filter applied, respectively, when compared with the previous univariate model. These findings suggest that the proposed method is suitable to track rapid dynamics of cerebral autoregulation despite the influence of confounding covariates