Untangling discrepancies between cerebrovascular autoregulation correlation coefficients: An exploration of filters, coherence and power.

Abstract

Cerebrovascular autoregulation (CAR) maintains stable cerebral perfusion by adjusting arteriole diameters in response to slow pressure fluctuations. Various CAR correlation coefficients-PRx (based on intracranial pressure-ICP), Mx (based on transcranial doppler-TCD), and COx/THx (based on near-infrared spectroscopy-NIRS)-are used interchangeably despite fundamental differences. 566 hours of ICP, NIRS, and TCD data from 38 traumatic brain injury (TBI) patients were assessed. The intercorrelation between CAR correlation coefficients was compared in relation to: (1) The impact of different filtering methods (to minimize noise); (2) The impact of slow wave power (i.e., magnitude of incoming trigger); (3) The impact of coherence (i.e., to what extent can the power of slow waves explain the change in power within the cerebral biosignal). Only coherence stratification consistently increased the metric intercorrelation to PRx (high vs. low) when evaluating Mx (0.43 vs. 0.08, p < 0.01) and THx (0.36 vs. 0.05, p < 0.01). Additionally, high coherence and ABP power were associated with fewer correlation coefficients around 0. Coherence increases the intercorrelation between the different CAR metrics. These sections might be regarded as more reliable, since they are derived from different biosignals that are all affected by CAR through different mechanisms.