Study On Pacing ECG Real-time Detection And Analysis For Cardiac Pacemaker

At present, more and more patients with heart disease have been implanted artificialcardiac pacemaker. After working for a period of time of the device implanted in body, wefind that its pacing and sensing functions must be detected and evaluated. The existingdetecting methods require patients to go to hospital regularly in order that conventionalECG and24-hour Holter system could be respectively used to achieve static detection anddynamic detection, however, scientific evaluation of the pacemaker function could not beachieved because of the limited monitoring time. This paper studies the real-time detectionand analysis of pacing ECG so that the technical support for real-time monitoring ofpacemaker could be provided anytime and anywhere.First of all, preprocessing and analysis for pacing ECG is proposed, using differentialfiltering and sliding weighted filtering to eliminate common noise signal, such aspower-frequency interference, baseline drifting and high frequency noise and so on inorder to facilitate latter process. In the procedure of waveform detection, the position ofpulse is acquired through edge enhancement algorithm and threshold analysis ofpreprocessed signal and the examination of missed and false detection. The positiondetection of autonomous QRS wave complex is similar to the position detection of pulse.After the process of zero-value (amplitude of pulse is replaced with zero), Re-thresholdanalysis, missed and false detection are used to acquire the position of autonomous QRSwave complex. For the detection of pacing QRS wave complex, the process of zero-valuefor autonomous QRS wave complex is applied again, and then64-point smoothing averagefilter is adopted to eliminate high frequency components in order to make signal smoothand easy to handle, finally threshold analysis and missed and false detection is applied sothat the position of pacing QRS wave complex is obtained.For feature extraction, based on the above detection of waveform, calculate amplitudeand delay of each unjudged waveform, four intervals (RS, SS, RR, SR) of each waveformis obtained, we regard them as feature dataset in order to facilitate real-time classification.In real-time classification, based on waveform detection and the feature dataset,according to characteristics of pacing function and perceptual function of patient’spacemaker, ID3algorithm is used to construct pacing ECG real-time analysis decision treethrough established pacing decision table and a kind of corresponding software is developed to realize a system.The experimental results on an open benchmark datasets from pacing data in MIT-BIHdatabase and hospital show that the accuracy rate of waveform detection is more than98%，according to the ID3algorithm, the accuracy rate of classification is more than96%.Furthermore, the proposed detection and analysis approach of pacing ECG has a highanalysis accuracy and lower computational complexity and it is suitable in the real pacingECG automatic analysis of telemonitoring system.