A Study On The Method Of Discrimination Of Malignant Ventricular Arrhythmia For AED And The Design Of System Control Software

With the improvement of living standards of people, malignant ventricular arrhythmias such as ventricular tachycardia or ventricular fibrillation have become a major cause of human deaths. In China, the incidence of sudden cardiac death (SCD) caused by malignant ventricular arrhythmias is about 41.8 per 100,000. Countries of the world are taking effective measures to reduce the incidence of SCD. The emergence of Automatic External Defibrillator (AED) greatly improved the survival rate of sudden cardiac arrest. A critical component in AED is the rapid and accurate discrimination of different malignant ventricular rhythms from sinus rhythms, which is the main content of this paper.There are two parts in this paper:the first part is about the discrimination between ventricular tachycardia and ventricular fibrillation in chapter two and chapter three, and the second part in chapter four is about the design of control software in AED system based on the S3C2410 ARM processor and embedded real-time operation systemμCOS-Ⅱ.In the first part, two methods were proposed for VF/VT discrimination:Entropy of Symbolic Series (SSEn) and Probability of Highly Volatile Vectors (PHVV) based on the absolute value of second derivative of ECG signals. To test the performance of a method for ECG analysis, it is essential to test with a large amount of annotated data under equal conditions. In this study, different discrimination methods were evaluated, including Fourier transform method, Hilbert transform method, complexity measure method, the two methods we proposed, etc. We used the CU Databank, the VF Databank and the AHA databank, which were widely accepted in the field of ECG analysis. The integrated receiver operator characteristic curve (IROC) and the calculation time were selected as the main performance evaluation parameters. Furthermore, we investigated sensitivity, specificity and accuracy. Experimental result shows that the proposed PHVV method obtained the best IROC value while kept short calculation time. Meanwhile, in order to further improve the performance, algorithms for pattern recognition such as support vector machine and Fisher linear discriminator were also applied.In the second part, we developed a prototype of control software in AED system, which was based on the S3C2410 ARM processor and embedded real-time operation systemμCOS-Ⅱ. The software achieved the basic functions that an AED system needed.