Research On Ecg Real-time Detection Algorithm And Application

Heart disease threatens one of human life main diseases, to heart disease's diagnosis and the prevention has been the medical arena important topic, ECG automatic diagnosis has a very practical significance for Doctor-assisted. Especially for long-term ECG, the use of computer analysis of auto-diagnosis can greatly reduce the burden on doctors, the accurate fast diagnosis provides the guarantee for doctor.In this thesis, the ECG Pretreatment, ECG waveform detection , diagnosis, and so on has done the quite detailed research, combined with the need for actual use, developed a real-time detection algorithm.In this thesis, we introduced the mechanism of ECG at first, and analyzed each ECG waveform and their corresponding physiological characteristics. In order to provide a reference standard for the detection of algorithm used in this thesis, We also introduce the standard MIT-BIH ECG database. Through the analysis of the sources of the ECG noise, we divided the noise into two types, one is baseline drift, the other is high-frequency interference. For baseline drift, we improved the median filter algorithm on the basis of the traditional ones. By using the algorithm of rapid sequencing and polynomial fitting with interval point, the processing speed was 45 times faster than the traditional median filter. It was obvious that the wavelet transform had its advantage in the filtering. It was particularly suited to deal with the high-frequency noise in the ECG signal. The method of wavelet threshold function filtering had well effects to deal with the serious high-frequent ECG signals. And we could have fast processing speed without too high wavelet decomposed scale in filtering. In the research of the detection of ECG waveforms, we mainly introduced the algorithm of the detection of the QRS waves, and analyzed several algorithms which were widely used in both at home and abroad. Combining with the research background in this thesis, we finally chose the improved differential threshold method to detect QRS wave. The rate of R-wave detection achieves the clinical practice level, and the processing time was 10 times faster than the detection algorithm of wavelet transform. We also gave the detection algorithm of P-wave and T-wave. In this thesis, we extracted characteristic parameters according to the results of the waveform detection, then did classification detection for the common diseases. Finally we designed the program flow chart which was applied in the actual situation and let it come true.