Method for the Measurement of the Dicrotic Notch in the Photoplethysmography Signal

A critical feature in photoplethysmography (PPG) signals is the diastolic (reflective) pressure wave following the systolic (forward) pressure wave during each compression cycle. The shape of these can often result in a feature known as the dicrotic notch. The PPG signal’s attributes provide valuable insights into cardiovascular health related to arterial vessel health, yet its measurement and characterization remain challenging due to variability and noise. We introduce an improved Gaussian decomposition algorithm to identify and quantify the forward and reflected pressure waves in individual cycles of the PPG signal. The algorithm iteratively decomposes the signal for each heartbeat into two Gaussian components, isolating the rising edge portion of the signal to model the systolic peak, followed by a model of the remaining diastolic peak. This enables robust characterization of the systolic and diastolic peaks across varying signal classes. Validation was performed on synthetic data, with parameter sweeps to assess robustness under different noise levels, and a wide range of signal morphologies then revalidated on patient data from the MIMIC-III database. Results demonstrate that the algorithm accurately identifies the PPG waveform measurements including dicrotic notch in most PPG signal classes, outperforming traditional peak detection and derivative-based methods, particularly under moderate noise conditions.