Study Of ECG Online Detection Methods

The Electricardiolograph(ECG) has been playing an important role in diagnosing heart disease, saving lives by providing the key information about one' body. And the monitor which can analyze ECG information automatically, such as arrhythmia monitor, can do the processing, classification and alarm of ECG. This makes the ECG monitoring more reliable and easier.For the arrhythmia automation analysis, the essential procedure is the detection of QRS complex, because QRS complex is the main characteristic of ECG signal. The difficulties of QRS detection lie in two factors: firstly, the physiological variability of QRS complex. Secondly, the various types of noises that can present in the ECG, for instance, the power line noise, muscle noise, baseline drift.This thesis has researched the amplifier circuits and the access methods of the ECG. in the Windows98 OS, based on which the algorithms of the QRS detection has been studied. On account of the operation speed and accuracy, we researched two algorithms used in QRS detection: derivative-window method and wavelet-based method. For our application, the band filter of the derivative-window algorithm has been improved to fit our sampling frequency better, and the wavelet-based method has been developed by adding a nonlinear processing-multiplying the maximal modulus with its absolute one-which emphasizing the signal modulus. The eventual detection is done by combination of the information of the scale 4 and scale 1.The quantitative comparison of two algorithms has been done in this thesis, and we apply the algorithms to the MIT-BIH arrhythmia database to find that: that derivative-window method has less muscle noise sensitivity, otherwise, the wavelet-based algorithm has less step noise sensitivity. As for the resulting QRS width accuracy, the wavelet time-scale algorithm obtains exactly QRS width with an error of no more than 2 sampling points, while the other one has less accuracy of width, which is always narrower .And the calculation of the derivative-window method is much less than the wave-based one, to be exactly, the operation speed of derivative-window algorithm is almost 10 times faster than the other one in our experimental setting...