Estimation of Heart Rate and Respiratory Rate via Blind Estimation from Smartphone-Based Contact Image Photoplethysmography

Abstract

Heart rate (HR) and respiratory rate (RR) vital signs provide useful cardiorespiratory information, and their monitoring is routinely carried out in clinical settings via electrocardiography, photoplethysmography, and capnography. However, these and other specialized biomedical devices are not easily translated to everyday use outside clinical and research settings. Hence, there is still a need for HR and RR monitoring devices that could be used on a daily basis by the general population. In this study, we employed a contact approach to estimate both HR and RR directly from image plethysmography (iPPG) signals extracted from smartphone-acquired videos. Video recordings of the fingertips from eight (N = 8) volunteers were acquired with an Android smartphone while the subjects performed metronome breathing maneuvers. The iPPG waveform was extracted via the Extended Blind End-member and Abundance Extraction (EBEAE) algorithm. Simultaneous ECG recordings were used to compute reference HR and RR time series. We found that iPPG-based estimates are highly correlated by those from ECG-derived ones (HR: $$ \rho = 0.9953 $$, RR: $$ \rho = 0.9733 $$ with low normalized root-mean-square errors (HR: $$ NRMSE = $$ 0.0855, RR: $$ NRMSE = 0.3074 $$ for both HR and RR estimation. The obtained results corroborate the feasibility of using contact iPPG methods to accurately estimate not only HR, but also RR without specialized biomedical devices.