A non-contact vision-based system for respiratory rate estimation.

Area of Science:

• Biomedical Engineering
• Medical Devices
• Physiological Monitoring

Background:

• Continuous respiratory rate monitoring is crucial for patient care.
• Existing methods often require physical contact, limiting long-term or at-home use.
• Non-contact methods offer a promising alternative for unobtrusive monitoring.

Purpose of the Study:

• To develop and validate a non-contact, vision-based system for continuous respiratory rate monitoring.
• To compare the performance of principal component analysis (PCA) and averaging methods for signal extraction.
• To assess the system's accuracy and suitability for simulated sleep scenarios.

Main Methods:

• Utilized a non-contact vision-based system to identify and track feature points in video feeds.
• Employed principal component analysis (PCA) and averaging techniques to derive respiratory signals from feature point trajectories.
• Selected the most representative respiratory signal using spectral analysis.
• Validated system accuracy against inductance plethysmography measurements.

Main Results:

• The vision-based system achieved high accuracy, with over 97% of recorded time showing an error within 1 breath/minute.
• Both PCA and averaging methods effectively condensed feature point data into representative respiratory signals.
• The system demonstrated consistent performance across 4 simulated sleep scenarios with 5 participants.

Conclusions:

• The developed non-contact vision-based system provides accurate, cost-effective, and simple respiratory rate monitoring.
• The system's performance suggests its potential for widespread adoption, including at-home installations.
• Non-contact respiratory monitoring using computer vision is a viable and effective approach.