On the Value of Testing the Null Hypothesis of Uncoupling in Cerebral Autoregulation Analysis

Dynamic cerebral autoregulation (dCA) can be characterized by assessing transfer function gain (TFG) from mean cerebral blood velocity (MCBv) to mean arterial pressure (MAP). TFG theoretically requires the significance of squared coherence ($\mathrm{K}^{2}$). However, it is unclear whether this test is necessary in practical computation of dCA. In this study a running approach for testing the significance of the MCBv-MAP association was proposed and TFG was computed over all the segments and after exclusion of those segments with $K^{2}$ below the threshold of significance. The study was performed in 24 subjects (age: $61 \pm 14 \mathrm{yrs}$) at supine resting and during active standing before and after undergoing cardiac surgery. $K^{2}$ was computed in the low frequency (LF) band ($\mathrm{K}^{2}{ }_{\mathrm{LF}}$) and the threshold of $\mathrm{K}^{\mathbf{2}}{ }_{\mathrm{LF}}$ significance was calculated as the $95^{\text {th }}$ percentile of $K^{2}$ LF computed over 100 uncoupled surrogate pairs. The analysis shows that a biased estimation of TFG was obtained when segments with low $\mathrm{K}^{2} \mathrm{lf}$ were included, thus stressing the relevance of the surrogate test in dCA analysis.