网络化无伤血氧监测系统的研究与设计

发布时间：2018-05-18 04:10

本文选题：光电容积脉搏波 + 运动伪差　；参考：《南京邮电大学》2017年硕士论文

【摘要】：随着人口老龄化现象加剧,同时伴随着各种疾病在人群中的传播,人们对自身的健康日益关注,越来越希望能够实时的了解自身身体状况,建立健全的网络化医疗监测体系显得尤为重要。血氧饱和度是反应人体生理状态的重要参数之一,本文主要研究了针对血氧提取的网络化监测系统。首先,对峰谷值提取算法进行了介绍,提出了改善的峰谷提取算法——变窗长滑窗法,改进后的滑窗法采用了跳跃式检测,而不是对每个点都检测,因此运算量减少了,同时跳跃步长和检测窗长都是根据实时心率周期进行更新使得该算法不仅运行效率高而且对变心率有很好的适应性,这对之后的运动伪差消除算法和血氧提取奠定了基础。然后,针对PPG信号中的运动干扰,本文研究了两种伪差消除算法:第一种为基于ICA与自适应滤波的PPG信号运动伪差消除算法,应用在需要还原并显示波形情况下,算法主要内容为先从干扰PPG信号中合成可适应滤波器需要的参考信号,即结合周期移动平均和独立成分分析来分离PPG成分和运动伪差成分,并通过独立成分选择算法来提取出PPG成分,接着通过可适应滤波器来恢复PPG信号的幅值信息,该算法可有效降低运动伪差的干扰并恢复PPG信号波形形态;第二种为基于统计分析的PPG信号运动伪差检测与剔除算法,应用于干扰剧烈且只需要提取出血氧值而不需显示波形的情况下,该算法采用设定阈值的方法对分段后信号的统计量进行检测,剔除含有干扰的波段,将剩余的优质波段提取出来,用提取出的优质波段代替整段信号计算血氧饱和度可以屏蔽掉运动伪差的干扰,从而提高了血氧检测的精确性。其次,介绍了基于线性回归的特征值提取算法。验证了红光与红外光PPG信号的相关性;研究出了参与运算样本数在采样频率附近时会使R值更加稳定的结论;证明了线性回归法提取R值比峰谷值法的稳定性更高;最后进行了标定试验,表明了R值与血氧值的高度相关性。最后,详细地阐述了整个网络化血氧检测系统的实现方法,介绍了血氧检测系统的网络化以及硬件、软件构成:首先简要介绍了网络化实现的背景意义以及实现方式,然后介绍了硬件系统,包括血氧传感器的选择、WiFi模块和采集处理模块的功能;最后在软件方面,介绍了基于Android系统编写的信号分析软件,主要功能包括数据通信,实时波形显示和生理参数计算。
[Abstract]:With the increasing aging of the population and the spread of various diseases in the population, people are increasingly concerned about their own health, and more and more hope to be able to understand their physical conditions in real time. The establishment of a sound network medical monitoring system is particularly important. The saturation of blood oxygen is one of the important parameters of the body's physiological state. In this paper, the network monitoring system for blood oxygen extraction is mainly studied. First, the peak valley extraction algorithm is introduced, and the improved peak valley extraction algorithm, the variable window long sliding window method, is proposed. The improved sliding window method uses jumping detection, not the detection of every point, so the computation is reduced, jumping step length and inspection are also taken. The length of the window is updated according to the real time heart rate period. The algorithm not only has high efficiency but also has good adaptability to the rate of change of heart rate. It lays the foundation for the motion artifact elimination algorithm and blood oxygen extraction. Then, two kinds of pseudo difference elimination algorithms are studied in this paper for the motion interference in the PPG signal: the first is based on the ICA The algorithm of PPG signal motion pseudo difference elimination with adaptive filtering is applied in the case of the need to restore and display the waveform. The main content of the algorithm is to synthesize the reference signals that can be adapted to the needs of the filter first from the interference PPG signal, that is to separate the PPG components and the motion artifact combining the periodic moving average and the independent component analysis. The PPG component is extracted by the selection algorithm, and then the amplitude information of the PPG signal is recovered by the adaptive filter. The algorithm can effectively reduce the interference of the motion artifact and restore the form of the PPG signal waveform. The second kinds of PPG signal motion pseudo difference detection and elimination method based on statistical analysis are applied to the intense interference and only need to extract the bleeding. In the case of oxygen value without showing the waveform, the algorithm uses a set threshold method to detect the statistics of the segmented signal, remove the bands containing the interference, extract the remaining high quality band, and use the extracted high quality band instead of the whole signal to calculate the saturation degree of blood oxygen, so that the interference of the motion artifact can be shielded. Thus, the interference of the motion artifact can be shielded. Thus, the interference of the motion artifact can be shielded. Thus, the interference of the motion artifact can be shielded. Thus, the interference of the motion artifact can be shielded and thus the interference of the motion artifact can be shielded. Secondly, the characteristic value extraction algorithm based on linear regression is introduced. The correlation between red light and infrared PPG signal is verified. The conclusion that the R value will be more stable when the number of the operation samples is near the sampling frequency is studied. It is proved that the linear regression method is more stable than the peak valley value method; finally, the R value is more stable; finally, the linear regression method is more stable than the peak valley value method; finally, the result shows that the linear regression method is more stable than the peak valley value method; finally, the results are more stable than the peak valley value method; finally, the results are more stable than the peak valley value method. Finally, the results are proved to be more stable than the peak valley value method. Finally, the results show that the linear regression method is more stable than the peak valley value method; finally, the results are more stable than the peak valley value method; finally, the results are more stable than the peak valley value method. Finally, the results are proved to be more stable than the peak valley value method. The calibration test is carried out, and the high correlation between the R value and the blood oxygen value is shown. Finally, the realization method of the whole network blood oxygen detection system is elaborated in detail. The network of the blood oxygen detection system and the hardware and software are introduced. First, the background meaning and the realization method of the network realization are briefly introduced. Then the hardware system is introduced. It includes the selection of blood oxygen sensor, the function of the WiFi module and the acquisition and processing module. Finally, in the software aspect, it introduces the signal analysis software based on the Android system. The main functions include the data communication, the real-time waveform display and the calculation of the physiological parameters.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318;TP274

【参考文献】