基于IPPG的非接触式血氧饱和度检测技术研究  

IPPG-based research on non-contact blood oxygen saturation detection technology

作　　者：柴荣轩 赵津发 吴航 张文昌 张广 陈炜[1,2] CHAI Rong-xuan;ZHAO Jin-fa;WU Hang;ZHANG Wen-chang;ZHANG Guang;CHEN Wei(Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control,School of Mechanical of Tianjin University of Technology,Tianjin 300384,China;National Demonstration Center for Experimental Mechanical and Electrical Engineering Education,Tianjin University of Technology,Tianjin 300384,China;Institute of Medical Support Technology,Academy of Systems Engineering of Academy of Military Scienceof Chinese PLA,Tianjin 300161,China)

机构地区：[1]天津理工大学机械工程学院先进机电系统设计与智能控制重点实验室,天津300384 [2]天津理工大学机械工程学院机电工程国家级实验教学示范中心,天津300384 [3]军事科学院系统工程研究院卫勤保障技术研究所,天津300161

出　　处：《医疗卫生装备》2021年第12期23-28,共6页Chinese Medical Equipment Journal

基　　金：天津市科技支撑重点项目(19YFZCSN01150)。

摘　　要：目的:提出一种非接触式血氧饱和度检测的新方法,以提高血氧饱和度的检测精度。方法:该方法基于成像式光电容积描记(imaging photoplethysmography,IPPG)原理提出,采用多任务卷积神经网络的算法进行人脸检测,采用欧拉视频放大中的运动放大和颜色放大相结合的算法实现视频图像放大的功能,使用集合经验模态分解和小波变换联合滤波的方法完成降噪滤波,最后使用经验公式计算血氧饱和度。开展“金标准”(指夹式血氧仪检测)对比实验,以验证该方法的检测准确性。结果:该方法检测血氧饱和度精度高,多次实验的平均误差均不高于2%,与指夹式血氧仪的检测结果有较好的一致性。结论:提出的方法可以完成血氧饱和度的非接触检测,且误差较小,能够达到预期要求。Objective To propose a new method of non-contact blood oxygen saturation detection to improve the detection accuracy of blood oxygen saturation.Methods A non-contact blood oxygen saturation detection was put forward based on the principle of imaging photoplethysmography(IPPG),which used a multi-task convolutional neural network algorithm for face detection,a combination of motion amplification and colour magnification in Eulerian video magnification to achieve the function of video image magnification,a joint method using ensemble empirical modal decomposition and wavelet transform to complete noise reduction filtering.The blood oxygen saturation was calculated using an empirical formula.The accuracy of the method was verified through comparing its results with those by a finger clip oximeter(gold standard).Results The method had high accuracy in the detection of blood oxygen saturation,with an average error not higher than 2%in multiple experiments,and good agreement with the results of the finger clip oximeter.Conclusion The method proposed can be used for non-contact blood oxygen saturation with low error.[Chinese Medical Equipment Journal,2021,42(12):23-28]

关 键 词：血氧饱和度 非接触检测 成像式光电容积描记 欧拉视频放大 集合经验模态分解 小波变换 

分 类 号：R318[医药卫生—生物医学工程]