【作者】 王金亮；

【导师】 刘忠国；

【作者基本信息】 山东大学 ， 生物医学工程， 2011， 硕士

【摘要】 心电信号反映了心脏电活动在体表的整体生理状态,为心脏疾病的诊断提供了重要依据。而心电信号的预处理和分析是正确诊断心脏疾病的前提。本文系统地综述了心电信号预处理及R波检测的研究现状,主要对心电信号的预处理和R波的检测进行算法研究工作,并且取得了良好的效果。本文首先介绍心电信号的特点、心电图的各个波形及其所受到的噪声。心电信号信噪比低,信号微弱,在采集信号时,由于外界的影响,采集到的信号都伴有噪声干扰,如基线漂移、工频干扰、肌电干扰等。然后根据心电信号和噪声的不同特性,对不同的噪声提出不一样的处理方法。针对工频干扰的去除,给出了自适应相干模板的方法,其基本原理是通过检测线性段计算干扰的幅度加以存储作为噪声模板,然后从原始信号中减去该模板,就得到处理后的信号,此方法能适应干扰频率和幅度的变化。对于基线漂移和肌电干扰的消除,采用数学形态学的方法,根据心电信号特征波形和噪声的时间宽度的不同,选择不同尺寸的结构元素对信号进行处理。采用多结构元素的方法,在开闭过程中对腐蚀和膨胀采用不同的结构元素,构造了广义形态开和闭运算,这种算法有效地抑制肌电干扰的同时且能保证心电波形的不失真。在矫正基线漂移时,首先选择一个较小宽度的结构元素进行形态滤波去除QRS波,再选择一个较大宽度的去除P、T波,得到基线漂移信号,最后从原始信号中减除基线漂移信号,就可以得到去除基线漂移的心电信号。由于不同的心率对应的QT间期的时间宽度不同,利用形态滤波提取的QRS波群,检测到R波,计算出平均心率,根据不同的心率选择结构元素,这种方法可以减小结构元素的变化范围,因人而异的矫正基线漂移信号,有效的滤除了干扰信号并较好的保留了ST段的信息。最后提出了将数学形态学和小波变换相结合的R波检测算法,先利用数学形态学运算提取心电信号的QRS波群,再采用mexihat小波对提取的QRS波群进行连续小波变换,利用小波变换模极大值的检测原理实现对R波的准确检测和精确定位。这种检测算法抗干扰能力较强,即使在较强噪声的干扰下,也能很好的实现对R的检测,并且采用形态学提取QRS波群,突出了特征波形,使检测更加容易。本文采用MATLAB软件读取MIT-BIH数据库中的数据,并对提出的算法进行滤波仿真,取得了预期的效果。更多还原

【Abstract】 Electrocardiograph (ECG) signal is synthetial refection of Cardiac Activity in body surface,and the ECG provides an important method to diagnoise cardiovascular diseases. Processing and analyzing ECG signal plays an important role in treating the cardiovascular diseases accurately.In this paper, ECG processing and recognition is introduced systematically. We mainly focus on the algorithm study of ECG preprocessing and R waveform detection, and obtain some good effects.The first has introduced the character of the ECG signal, the ECG waveform and the noises which corrupt the ECG signal. ECG is weak and it is extremely vulnerable to environmental impacts.Because of the mixed other biomedical signal and electromagnetic interference, the ECG signal is always with three major interference: frequency interference, muscle artifact and baseline drift.Then,with the different nature of ECG signal and noise, different methods were used to calibrate the noise.Adaptive coherent model was proposed to eliminate the power-line interference. The approach can get a noise template through detecting the section of the signal and computing the magnitude of interference, then subtract the noise template from the orginal signal,the processed signal without power-line interference can be got. The adaptive coherent model can adapt to change of frequency and amplitude of the interference. A morphological filtering approach was put forward to remove the noise of the ECG signals and to calibrate the base-line drift in this paper. Different sizes of structuring elements were used to process the signal for different duration of the characteristic wave and noise.Based on erosion and dilation,a class of generalized opening and closing morphological filter is constructed by using multiplr structuring elements.The results show that these filters can preserve the geometrical features well when they supress noise.For baseline correction,the signal is first filtered by a smaller size of structuring element for removing QRS complexes,Then the P,Twaveforms are removed by morphological filtering using a bigger size of structuring element.The final result is then an estimate of baseline drift.The correction of the baseline is then done by subtracting baseline drift from the original.As the QT intervals influenced by heart rate (HR) changing, the HR can be got through detecting R wave.Different sizes of structuring elements were used to process the signal for different HR. This method reduced the range of structuring elements, and keeps the ECG signal shape unchanged while calibrating the base-line drift effectively. Finally, a algorithm for detecting R-wave was proposed on the basis of mathematical morphological operation and wavelet transform.Mathematical morphology operation was used to extract QRS complexes, and the Mexihat wavelet transform was used to analyse the QRS complexes.Test results shows that the decetion method can detect accurately and localize precisely to R-wave of ECG signal.The ECG datebase in this paper is from MIT-BIH database, and all of algorithms are simulated on Matlab.更多还原