| The detection and analysis software developed by the present graduate project can record real-time long-lasting dynamic electrocardiogram and analyses it. As soon as any emergent cardiac illness happens, the software will give an alarm to prompt patients/doctors to take necessary measures. When the record and analysis missions have been completed, an ultimate medical report can also be displayed and printed simultaneously. The real-time analysis task mainly includes waveform detection and arrhythmia classification. The waveform detection part fulfills QRS complex wave, P wave, T wave detection, ST segment measurement and HRV analysis. Based on the aforementioned wave detection, the arrhythmia classification part use clinical knowledge to judge SVA (Super-ventricular arrhythmia) and VA, and perform real-time monitoring function to check alarm events. In waveform detection, the newly arising technology of wavelet transforms has been employed to detect QRS complex waves. Detection accuracy is greater than 98% verified by MIT/BIH ECG database. Wavelet transforms method has multitude advantages including simple operation, reliable performance and precise measurement for characteristic parameters.. And accurate QRS detection has played a solid foundation for the other waveform identification. P and I waves detection can be realized by convenient magnitude-depended method. ST segment can be measurement by classical J+X method, and HRV analysis can be implemented by the combination of linear and nonlinear methods. All these detection approaches have met the requirement of real-time analysis and clinical diagnosis. In arrhythmia classification, SVA includes tachycardia, bradycardia, asystole, SVPB (super-ventricular premature beat) and rhythm disorders. VA includes RonT, ventricular tachycardia, VPB, coupled VPB, bigeminy and trigeminy. Other abnormal QRS named as unclassified waves are selected and will be confirmed by doctors to avoid missed ventricular arrhythmia. The real-time monitoring function will give out alarm information when serious tachycardia, bradycardia, asystole and frequent VPB are found. ?II ? Abstract The real-time implementation software above is programmed with Visual C++ 6.0. It is well-known for fast operation, high accuracy, perfect function, friendly interface and convenient usage. The software has presented good performance tested by MIT/BIH ECG database. The clinical experiment in the People抯 Hospital of Qinhuangdao shows that the total detection accuracy for SVA is greater than 95%, for VA is greater than 92%. |
