Time–Frequency Approaches for the Detection of Interactions and Temporal Properties in Renal Autoregulation

Abstract

We compare the influence of time-frequency methods on analysis of time-varying renal autoregulation properties. Particularly, we examine if detection probabilities are similar for amplitude and frequency modulation for a modulated simulation signal among five time-frequency approaches, and if time-varying changes in system gain are detected using four approaches for estimating time-varying transfer functions. Detection of amplitude and frequency modulation varied among methods and was dependent upon background noise added to the simulated data. Three non-parametric time-frequency methods accurately detected modulation at low frequencies across noise levels but not high frequencies; while the converse was true for a fourth, and a fifth non-parametric approach was not capable of modulation detection. When applied to estimation of time-varying transfer functions, the parametric approach provided the most accurate estimations of system gain changes, detecting a 1dB step increase. Application of the appropriate methods to laser Doppler recordings of cortical blood flow and arterial pressure data in anesthetized rats reaffirm the presence of time-varying dynamics in renal autoregulation. An increase in the peak system gain and detection of amplitude modulation of the Myogenic mechanism both occurred after inhibition of nitric oxide synthase, suggesting a connection between the operation of underlying regulators and system performance.