论文部分内容阅读
J波是心电图ECG(Electrocardiogram)上心室复极的一种新指标,它的出现可能预示着心脏发生室速甚至猝死的风险大大增加。近年来关于区分J波高危与良性形态的研究在医学上备受关注,但仅限于在临床上观察心电图上J波的时域波形,凭医生的经验作出判断,缺乏系统的评判标准。为了进一步研究J波信号的时频域特性,从ECG中准确地提取到J波信号是研究重点。分析了正常ECG信号及病变J波信号的循环平稳特性,采用以高阶循环平稳度HODCS(High-order Degree of Cyclostationarity)作为分离准则的盲源分离BSS(Blind Source Separation)算法,在利用小波包变换(Wavelet Packet Transform)定位S、T点和白化预处理之后使用吉文斯矩阵分离源信号。仿真结果表明,算法较经典Fast-ICA算法在提取J波信号的实际应用中有更好的分离性能。
J-wave is a new indicator of ventricular repolarization on the ECG (Electrocardiogram), its appearance may indicate a significant increase in the risk of ventricular tachycardia or sudden death in the heart. In recent years, the research on distinguishing the high risk and benign morphology of J wave has drawn much attention in medicine. However, it is limited to observe the time-domain waveform of J wave on the ECG, and it is judged by the doctor’s experience. There is no systematic evaluation criteria. In order to further study the time-frequency domain characteristics of J-wave signals, it is important to extract J-wave signals accurately from ECG. The cyclostationary characteristics of normal ECG signals and lesion J-wave signals are analyzed. Blind Source Separation (BSS) algorithm based on High Order Degree of Cyclostarity (HODCS) is used as the separation criterion. Transform (Wavelet Packet Transform) Selects the source signal using the Reversi’s matrix after locating the S, T points and pre-whitening. The simulation results show that the proposed algorithm has better separation performance than the classical Fast-ICA algorithm in the practical application of J-wave signal extraction.