期刊名称：International Journal of Computer Science Issues
印刷版ISSN：1694-0784
电子版ISSN：1694-0814
出版年度：2012
卷号：9
期号：2
出版社：IJCSI Press
摘要：In this paper we go about the segmentation and analysis of an electrocardiographic (ECG) signal. Firstly, it consists in working out the locations of different characteristic waves of this signal: the QRS complex and the waves P and T, successively and in the order of magnitude and their amplitude. Secondly, we go about the analysis of this signal by the linear principal component analysis (LPCA) based on the results found in the first part. The importance of this algorithm comes in the context of the ECG supervision and then the cardiovascular system. This algorithm integrates the multi-scale wavelet analysis and principal component analysis. This analysis allows us, first, to apply the continuous wavelet transform (CWT) on the totality of the signal. After that, based on the property of the CWT regularity, the waves will be detected after a thresholding operation. To evaluate the segmentation algorithm two parameters are introduced, the sensitivity Se and the predictive value P+. The results have given an average of Se=99.93% and P+=99.96%, which indicates that our segmentation algorithm is sufficiently reliable in comparison to the real database. In the supervision of the ECG, we use the detected parameters to construct the data-matrix of the LPCA. The defects are detected and located by this tool in order to determine the existing failure in the used signal. Comparing our results with those of the expert, we find that the LPCA gives a concrete state of the cardiovascular system.
