A Changepoint Detection-Based General Methodology for Robust Signal Processing: An Application to Understand Preeclampsia's Mechanisms.

Abstract

Motivated by illuminating the underlying mechanisms of preeclampsia, we develop a changepoint detection-based general and versatile methodology that can be applied to any experimental model, effectively addressing the challenges of high uncertainty produced by experimental interventions, intrinsic high variability, and rapidly and abruptly varying time dynamics in perfusion signals. This methodology provides a systematic and reliable approach for robust perfusion signal analysis. The main innovation of our methodology is a highly efficient automatic data processing system consisting of modular programming components. These components include a signal processing tool for optimal segmentation of perfusion signals by isolating their "genuine" vascular response to experimental interventions, and a novel and suitable normalization to evaluate this response concerning an experimental reference state, typically basal or pre-intervention. In this way, we can identify anomalies in an experimental group compared to a control group by disaggregating noise during the transitions just after experimental interventions. We have successfully applied our general methodology to perfusion signals measured from a preeclampsia-like syndrome model developed by our research group. Our findings revealed impaired brain perfusion in offspring from preeclampsia, particularly dysfunctional brain perfusion signals with inadequate perfusion signal vasoreactivity to thermal physical stimuli. This general methodology represents a significant step towards a systematic, accurate, and reliable approach to robust perfusion signals analysis across various experimental settings with diverse intervention protocols.