Study On Cross Analysis Of ECG-PW Signals Based On Multivariate Multiscale Entropy

In modern society, all kinds of cardiovascular system diseases have become a major killer of human beings. To find out the incipient symptoms of cardiovascular system diseases and to take steps to prevent these diseases according to the symptoms have become an extraordinary important issue. Facing this social issue, researchers have been developing munificent equipment to test the condition of cardiovascular system. Among all the problems related to such equipment, to acquire and analyze the signals generated from cardiovascular system non-invasively and non-destructively is one of the most important problems. Since ECG signal and pulse wave signal contain the most important and basic biomedical information of the cardiovascular system, the cross complexity of these signals is studied in this paper.As a project supported by State863Project (grant number:2009AA02Z408) and the national natural science foundation of china (grant number:61201049), the cross complexity of heart rate variability and heart diastolic time variability (noted as HRV-DIV cross complexity) are studied based on Multivariate Multiscale Sample Entropy in this paper in order to provide more comprehensive understanding of human cardiovascular system. Based on the study, a biomedical signal research workbench is constructed using LabVIEW2011. The main contributions of this paper are as follows:(1) Based on the fact that biomedical signal are nonlinear and non-stationary data, we apply empirical mode decomposition which is a fully data driven method to extract the characteristic points of ECG signal. In this paper, we first decompose the original ECG signal using EMD method. Then only the intrinsic mode functions that contain the QRS wave information are processed. And at the same time the noises contained in the original signals are filtered out. After some nonlinear transform and integration calculation, the R points are detected. A fast EMD method is also introduced in this paper to improve the calculation time of the method.(2) In this paper multivariate multiscale entropy is used to measure the cross complexity of multiple biomedical signals. Based on the definition of MMSE, we studied the performance of this method, such as how the data length and spike trains influence the results of MMSE. Our experiment shows that spike trains can low the results of the MMSE at each scale and more spike trains shows a stronger influence. Hence, the spike trains must be removed before the analysis.(3) Based on the preprocessed ECG data and pulse wave cited from the Fantasia database, MMSE is applied to measure the cross complexity of these two cardiovascular signals. Actual ECG and pulse wave signals acquired from our lab are used to further illustrate the studied problem. Experiment results show that the two signals have some cross correlativity and diseases and age can low down this correlativity. This conclusion can help to evaluate the condition of cardiovascular system.(4) A cardiovascular system signals complexity analysis workbench is developed using LabVIEW2011. We can easily extract characteristic points of cardiovascular system signals and then analyze the complexity of single or multiple signals using this workbench.