A non-contact camera-based method for respiratory rhythm extraction

The aim of this work is to present a non-contact video-based method for respiratory rhythm extraction. makes use consumer-grade RGB camera, and it based on computer vision algorithms detect track custom pattern placed the thorax subject. signal extracted by computing changes in position detected through time. has been validated comparing versus one obtained with reference adult population. was an inductive plethysmography system (Respiband from BioSignalsPlux™). 21 healthy subjects were measured four tests performed each signals its cycles extracted. To characterise error, assessed with: Fisher intra-class correlation (ICC), mean absolute error (MAE), percentage (MAPE) Bland-Altman plots obtained. results show >0.9 controlled respiration >0.85 unconstrained between proposed method, low (MAPE <4% constrained <6% respiration) high sensitivity when detecting (>94% all cases). From we can conclude that algorithm adequate acquire extraction, real-time performance compared could be applied real-life situations.