Pulse rate variability analysis by video using face detection and tracking algorithms.

Area of Science:

• Physiology
• Biomedical Engineering
• Computer Vision

Background:

• Heart rate variability (HRV) is crucial for assessing autonomic nervous system (ANS) function and cardiovascular health.
• Traditional HRV measurement often requires specialized sensors.
• Facial video analysis offers a non-contact method for physiological monitoring.

Purpose of the Study:

• To evaluate pulse rate variability (PRV) derived from facial videos as a surrogate for HRV.
• To implement and assess face detection and tracking algorithms for artifact reduction in video-based PRV.
• To validate the proposed method against a reference pulse sensor system.

Main Methods:

• Acquired facial videos to extract pulse rate signals.
• Applied face detection and tracking algorithms to mitigate motion artifacts and region of interest (ROI) variations.
• Compared the PRV derived from video analysis with data from a reference pulse sensor.
• Conducted statistical analysis to determine agreement between the two methods.

Main Results:

• The proposed method successfully extracted PRV from facial videos.
• Face detection and tracking significantly reduced artifacts caused by subject movement and ROI changes.
• Statistical analysis demonstrated good agreement between the video-derived PRV and the reference pulse sensor measurements.
• The findings support the validity of using facial video analysis for PRV assessment.

Conclusions:

• Facial video analysis, enhanced with face tracking, provides a reliable non-contact method for measuring PRV as a surrogate for HRV.
• This approach offers a practical alternative for monitoring autonomic and cardiovascular health, especially in settings where traditional sensors are inconvenient.
• Further research can explore broader applications in remote patient monitoring and health screening.