基于可穿戴设备的血氧饱和度检测及算法的研究

发布时间：2018-04-11 17:56

本文选题：智能穿戴 + PPG信号　；参考：《上海师范大学》2017年硕士论文

【摘要】：光电容积脉搏波信号(PPG)反映了人体的重要信息,包含了丰富的微循环生理病理信息,是我们研究人体循环系统重要的信息来源。其应用领域由人体循环系统发展到呼吸系统,在人体血压、血流、血氧、脑氧、肌氧、血糖、脉率、呼吸率和呼吸容量等的无创检测中都有很好的应用前景。从PPG信号中提取血氧饱和度的特征已成为目前最主要的无创测量血氧饱和度的方式;而在智能穿戴方面,PPG信号的检测主要用于提取人体的心率信号。在智能穿戴的设备上采集PPG信号存在很多噪声,最主要的噪声来源于运动伪差.目前尚未有有效的方法完全去除PPG信号中的运动伪差,这使得具有血氧饱和度的测量智能穿戴的设备迟迟不能走入人们的生活。本文针对从穿戴设备上采集的光电容积脉搏波信号进行移除运动伪差,特征提取等来检查血氧饱和度等问题展开讨论,主要内容有:介绍基于光电容积脉搏波描记法(PPG技术)的无创血红蛋白浓度检测技术的光学与生理学的相关基础知识。并根据Lambert-Beer定律从透射式血氧饱和度测量仪的数学模型出发,推导出适用于反射式的血氧饱和度的数学模型。基于能够用于穿戴设备上的AFE4403芯片搭建的光电容积脉搏波信号采集设备作为光电容积脉搏波信号的采集设备。主要包括,两路LED灯的驱动方式和电流大小控制,AFE4403芯片放大参数的设置,PPG信号采集的频率、空占比等问题的研究。利用该采集设备完成了135组数据的采集,为后续PPG信号处理和血氧饱和度的预测提供了可靠的数据。研究PPG信号的信号处理算法,包括:基线漂移、滤除高频噪声、消除运动伪差以及PPG信号的特征信息的提取等。对于PPG信号的基线漂移问题,本文提出了零相移的IIR无限脉冲响应滤波器,通过对具有基线漂移的PPG信号处理,表明该滤波器能够有效消除PPG信号的基线漂移。针对PPG信号中的运动伪差噪声的消除,提出了基于FastICA的独立成分分析算法从测量信号中分离出运动伪差信号和PPG信号,将PPG原始信号作为PPG信号与运动伪差等噪声信号的混合信号,分离出PPG信号与噪声信号。通过对PPG信号的处理实验,证明这种盲信号处理算法能够有效地移除PPG信号中的运动伪差。采集实验数据,根据反射式的PPG信号的数学模型从PPG信号中提出特征信息,以标准血氧仪检测量的血氧饱和度的结果作为参考值,对实验结果进行回归分析,得到适用于反射式的血氧饱和度测量的经验函数。
[Abstract]:PPG, which reflects the important information of human body and contains abundant physiological and pathological information of microcirculation, is an important information source for us to study the human circulatory system.Its application field is developed from human circulatory system to respiratory system. It has a good prospect in noninvasive detection of blood pressure, blood flow, blood oxygen, cerebral oxygen, muscle oxygen, blood glucose, pulse rate, respiration rate and respiratory volume.Extracting the characteristic of oxygen saturation from PPG signal has become the most important non-invasive method to measure the saturation of blood oxygen, while the detection of PPG signal is mainly used to extract the human heart rate signal.There is a lot of noise in the acquisition of PPG signal on the smart wearable device, and the main noise comes from the motion pseudo-error.At present, there is no effective method to completely remove the motion artifacts in PPG signal, which makes the intelligent wearable device with oxygen saturation can not be used in people's life.In this paper, the problems of removing motion artifacts and feature extraction from photoelectric volumetric pulse wave signals collected from wearable devices to check blood oxygen saturation are discussed in this paper.The main contents are as follows: the basic knowledge of optics and physiology of noninvasive hemoglobin concentration detection technology based on photoelectric volumetric pulse wave tracing (PPG) is introduced.According to Lambert-Beer 's law, the mathematical model of reflex oxygen saturation is derived from the mathematical model of transmission oxygen saturation meter.Based on the AFE4403 chip which can be used in wearable equipment, the photoelectric volumetric pulse wave signal acquisition equipment is used as the photoelectric volume pulse wave signal acquisition equipment.This paper mainly includes the study of the driving mode of two LED lamps and the setting of the amplifier parameters of AFE4403 chip, the frequency and duty cycle of the signal acquisition.The data collection of 135 groups has been completed by using this acquisition equipment, which provides reliable data for the subsequent PPG signal processing and the prediction of blood oxygen saturation.The signal processing algorithms of PPG signal are studied, including baseline drift, filtering high frequency noise, eliminating motion artifacts and extracting characteristic information of PPG signal.For the baseline drift of PPG signal, a zero-phase-shift IIR infinite impulse response filter is proposed. By processing the PPG signal with baseline drift, it is shown that the filter can effectively eliminate the baseline drift of PPG signal.An independent component analysis (ICA) algorithm based on FastICA is proposed to separate the motion pseudo signal and the PPG signal from the measured signal in order to eliminate the motion pseudo noise in the PPG signal.The original PPG signal is used as the mixed signal of the PPG signal and the noise signal such as the motion pseudo difference, and the PPG signal and the noise signal are separated.Experiments on PPG signal processing show that the blind signal processing algorithm can effectively remove the motion artifacts in PPG signals.The experimental data are collected and the characteristic information is presented from the PPG signal according to the mathematical model of the reflective PPG signal. The results of the oxygen saturation measured by the standard oximeter are taken as the reference values, and the experimental results are analyzed by regression analysis.An empirical function suitable for reflex oxygen saturation measurement is obtained.
【学位授予单位】：上海师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318.6;TN911.7

【相似文献】