Addison Paul S, Foo David M H, Jacquel Dominique, Borg Ulf
Annu Int Conf IEEE Eng Med Biol Soc. 2016 Aug;2016:4747-4750. doi: 10.1109/EMBC.2016.7591788.
A method for extracting video photoplethysmographic information from an RGB video stream is tested on data acquired during a porcine model of acute hypoxia. Cardiac pulsatile information was extracted from the acquired signals and processed to determine a continuously reported oxygen saturation (SvidO2). A high degree of correlation was found to exist between the video and a reference from a pulse oximeter. The calculated mean bias and accuracy across all eight desaturation episodes were -0.03% (range: -0.21% to 0.24%) and accuracy 4.90% (range: 3.80% to 6.19%) respectively. The results support the hypothesis that oxygen saturation trending can be evaluated accurately from a video system during acute hypoxia.
一种从RGB视频流中提取视频光电容积脉搏波信息的方法，在猪急性缺氧模型获取的数据上进行了测试。从采集的信号中提取心脏搏动信息，并进行处理以确定持续报告的氧饱和度（SvidO2）。发现视频与脉搏血氧仪的参考值之间存在高度相关性。在所有八次去饱和事件中计算出的平均偏差和准确率分别为-0.03%（范围：-0.21%至0.24%）和准确率4.90%（范围：3.80%至6.19%）。结果支持这样的假设，即在急性缺氧期间可以通过视频系统准确评估氧饱和度趋势。
