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

"This work was supported in part by the Korea Medical Device Development Fund grant funded by the Korean Government under Project HW20C2151; and in part by the Industrial Technology Innovation Program (Development of Emotional Cognitive and Sympathetic Artificial Intelligence (AI) Service Technology for Remote (Non-Face-to-Face) Learning and Industrial Sites) funded by the Ministry of Trade, Industry and Energy (MOTIE, South Korea) under Grant 20012603."

" shows the experimental settings. First, the best visible aspect of the nostril was determined at an angle of 30 degrees from the front (Case A). As the temperature change is largest in the nostril region, the structural features are clear in this area; thus, this view is typically used for breathing monitoring. Second, the invisible aspect of the nostril was determined at an angle of 90 degrees from the front (Case B). The thermal faces obtained at these angles do not contain stable nostrils. Third, the aspect was determined at an angle of 45 angles, not the front view (Case C). Finally, the aspect was determined at an angle of 90 angles, not the front view (Case D). The thermal faces acquired under the Case C and Case D conditions can include various BAFRs depending on the individual, which makes it difficult to define the region using structural information. 15 subjects participated in this study. The subjects consisted of healthy volunteers. The subjects participated in the experiment after hearing sufficient explanations in the laboratory and providing consent. In addition, an anesthesiologist conducted an experiment to prevent problems caused by abnormal breathing. The present study was performed at Severance Hospital from July 2021 to August 2021. This study was conducted in accordance with the principles of Good Clinical Practice and was approved by the Institutional Research Board of Severance Hospital (Reference No. 4-2020-1320); all patients gave written informed consent. The study was registered at ClinicalTrails.Gov with the number NCT04964245. The distance between the subject and the thermal camera was 50 cm, and normal breathing was performed in a stable state. Five 1-minute measurements were taken for each subject. The thermal camera was the A655sc model (FLIR systems), and the camera data were acquired at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $640\times480$\end{document} spatial resolution, 0.03 Kelvin temperature resolution, and 25 Hz. The breathing signal obtained by the chest belt sensor was used as a reference signal. A Biopac (MP150, Biopac Systems Inc) was used for reference measurements, and the chest belt sensor signal was measured at 1000 Hz. FIGURE 3.Experimental settings."