Measuring Heart Rate Variability Using Facial Video

Gerardo H Martinez-Delgado, Alfredo J Correa-Balan, José A May-Chan, Carlos E Parra-Elizondo, Luis A Guzman-Rangel, Antonio Martinez-Torteya

Research output: Contribution to journalArticlepeer-review

6 Citations (Scopus)

Abstract

Heart Rate Variability (HRV) has become an important risk assessment tool when diagnosing illnesses related to heart health. HRV is typically measured with an electrocardiogram; however, there are multiple studies that use Photoplethysmography (PPG) instead. Measuring HRV with video is beneficial as a non-invasive, hands-free alternative and represents a more accessible approach. We developed a methodology to extract HRV from video based on face detection algorithms and color augmentation. We applied this methodology to 45 samples. Signals obtained from PPG and video recorded an average mean error of less than 1 bpm when measuring the heart rate of all subjects. Furthermore, utilizing PPG and video, we computed 61 variables related to HRV. We compared each of them with three correlation metrics (i.e., Kendall, Pearson, and Spearman), adjusting them for multiple comparisons with the Benjamini-Hochberg method to control the false discovery rate and to retrieve the q-value when considering statistical significance lower than 0.5. Using these methods, we found significant correlations for 38 variables (e.g., Heart Rate, 0.991; Mean NN Interval, 0.990; and NN Interval Count, 0.955) using time-domain, frequency-domain, and non-linear methods.

Original languageEnglish
Article number4690
JournalSensors
Volume22
Issue number13
DOIs
Publication statusPublished - 1 Jul 2022

Bibliographical note

Funding Information:
Funding: This research was funded by the Proyectos de Investigación e Innovación and the Fondo de Publicaciones grants from Universidad de Monterrey

Publisher Copyright:
© 2022 by the authors. Licensee MDPI, Basel, Switzerland.

Fingerprint

Dive into the research topics of 'Measuring Heart Rate Variability Using Facial Video'. Together they form a unique fingerprint.

Cite this