Characterization of vascular dynamics based on experimental recordings with extreme data loss

摘要

The presence of artifacts complicates analysis of physiological systems based on experimental time series. Aiming to increase the signal-to-noise ratio and to improve characterization of the system's state, bad segments are simply removed from the experimental recording, and the latter may change correlation and other properties of the resulting dataset. Here we illustrate that in the case of positively correlated processes being typical for vascular dynamics, the authentic characterization of the system's dynamics can be provided even under the condition of extreme data loss. Based on the cerebral blood flow (CBF) dynamics acquired with the laser speckle contrast imaging (LSCI) and the multiresolution analysis, we show insensitive changes of measures quantifying the system's state with the amount of missed data for both, macro- and microcerebral circulation. We also demonstrate that these results do not significantly depend on the selected basic wavelet and the resolution level.