Evaluation of Two Asynchronous Envelope Detection Techniques for Photoplethysmography

Photoplethysmography (PPG) is widely used in both clinical and consumer settings. It is often used to measure heart rate using methods that rely on heart beat detection. However, additional features of the waveform, including the relative levels of the signal envelopes and the resulting amplitude (PPGA), encode valuable physiological information.This work describes and compares two asynchronous techniques for envelope detection in PPG that do not rely on peak detection: the Asymmetric Root-Mean-Square (RMS) detector and the Peak Injection Low-Pass detector. These methods are developed to enhance real-time patient monitoring and facilitate decision-making systems. Both methods are simple to implement and incur minimal time delay. Further, the detectors offer values between beats, simplifying downstream processing by removing an interpolation step.The envelope detectors are evaluated on the CapnoBase TBME benchmark dataset. Across all subjects, the absolute errors for the upper envelope at beat times are 0.60 and 0.02 (arbitrary units) for the half-wave RMS and peak injection methods, respectively. A simple method using envelopes to facilitate peak detection is also presented. This work demonstrates the feasibility of extracting additional information from a widely used physiological measurement.