An automatic signal detection algorithm for the digital volume pulse

Abstract

Many previous studies suggested that the digital volume pulse (DVP) can be used to estimate arterial stiffness. The DVP waveform can be rapidly acquired by measuring reflection or absorption of red light (or infrared light) across the photoplehtysmography (PPG). Because of the nature of its measuring principle, the DVP is often contaminated by noise from light sources, such as sunlight, and disturbances due to movement of the recording electrodes. This phenomenon makes us hard to extract pulse features and thus provides incorrect clinical information. In order to screen out this problem, we have developed the automatic signal detection algorithm for the DVP and have evaluated its performance. In the performance evaluation, we have obtained the false negative rate (FNR) of 0.33% and the predictive accuracy rate (PAR) of 97.19%. From the results, we conclude that our proposed signal detection algorithm can be used as a preprocessor before the extraction of pulse features with reasonable performance.