HRVCam: robust camera-based measurement of heart rate variability

Area of Science:

• Biomedical Engineering
• Signal Processing
• Computer Vision

Background:

• Non-contact, camera-based estimation of heart rate variability (HRV) is valuable across medical, automotive, and entertainment fields.
• Current camera-based HRV methods face challenges with accuracy and reliability, particularly due to lower signal-to-noise ratio (SNR) in darker skin tones and motion artifacts.

Purpose of the Study:

• To develop a robust algorithm, HRVCam, for accurate camera-based HRV estimation that overcomes limitations of low SNR and motion artifacts.
• To improve the practical applicability of imaging photoplethysmography (iPPG) for HRV measurement.

Main Methods:

• HRVCam computes HRV from the instantaneous frequency of the iPPG signal.
• The algorithm employs automatic adaptive bandwidth filtering and discrete energy separation for instantaneous frequency estimation.
• Algorithm parameters are optimized based on the characteristics of HRV and iPPG signals.

Main Results:

• A new dataset with 16 participants of diverse skin tones was collected.
• HRVCam demonstrated a significant reduction (over 50%) in error for camera-based HRV metrics in videos featuring dark skin and facial motion.
• The algorithm shows improved robustness against common artifacts in iPPG signals.

Conclusions:

• HRVCam provides a practical solution for robust camera-based HRV measurements.
• The algorithm can be integrated with existing iPPG estimation techniques to enhance HRV accuracy.
• This advancement broadens the potential applications of non-contact physiological monitoring.