Video pulse rate variability analysis in stationary and motion conditions

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Signal Processing

Background:

• Recent studies explore video cameras for heart rate (HR) and heart rate variability (HRV) measurements via skin color changes.
• Existing research primarily focuses on stationary conditions, with limited studies on HRV in motion scenarios and insufficient statistical analysis.

Purpose of the Study:

• To conduct a video pulse rate variability (PRV) analysis, measuring pulse-to-pulse (PP) intervals in both stationary and motion conditions.
• To propose and evaluate a selective tracking method using Viola-Jones and KLT algorithms for robust video PRV analysis.
• To address the limitations of low frame rates in commercial cameras for PRV analysis.

Main Methods:

• Implemented a selective tracking method using Viola-Jones and KLT algorithms for robust video PRV analysis.
• Analyzed two models to assess their performance in PRV measurements, considering sampling rate importance.
• Contrasted the proposed method's data and results with existing state-of-the-art approaches.

Main Results:

• Webcam performance analysis yielded better results, with high correlation (>0.9) for PRV parameters in stationary conditions.
• In stationary conditions, PP time series achieved an RMSE of 19.45 ± 5.52 ms (1.70 ± 0.75 bpm).
• Motion analysis showed good correlation for most PRV parameters, albeit lower than stationary conditions, with PP time series RMSE of 21.56 ± 6.41 ms (1.79 ± 0.63 bpm).

Conclusions:

• The proposed method demonstrated good agreement with the reference system, improving PRV parameter accuracy in stationary conditions compared to prior works.
• Comparative analysis in motion conditions was limited by the scarcity of comparable studies with sufficient data.
• The developed method presents a viable low-cost, contactless, and reliable solution for measuring HR or PRV in non-clinical settings.