Breathing-Associated Facial Region Segmentation for Thermal Camera-Based Indirect Breathing Monitoring.

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Thermal Imaging Technology

Background:

• Non-contact breathing measurement using thermal cameras is limited by the need for clear nostril visibility.
• Existing methods struggle with diverse viewing angles and frontal views where the nostril is obscured.

Purpose of the Study:

• To develop and validate a novel algorithm for non-contact breathing signal extraction from thermal imaging.
• To overcome limitations of previous methods by utilizing a broader facial region for measurement.
• To assess the algorithm's performance across multiple viewing angles.

Main Methods:

• Defined a breathing-associated-facial-region (BAFR) for signal extraction.
• Conducted experiments on 15 healthy subjects using a thermal camera (FLIR A655sc) and Biopac as a reference.
• Acquired data from frontal, 45-degree, and 90-degree views, with and without clear nostril visibility.

Main Results:

• The proposed BAFR algorithm significantly improved breathing cycle accuracy from 65.6% to 90.9%.
• Average correlation between thermal measurements and reference data increased from 0.587 to 0.885.
• Stable breathing signals were extracted across diverse angles and views.

Conclusions:

• The BAFR algorithm enables robust, non-contact breathing monitoring from various angles.
• This method enhances the applicability of thermal imaging for physiological monitoring in diverse environments.
• The approach overcomes previous limitations, offering a more versatile solution for breathing analysis.