Complexity-Measure-Based Sequential Hypothesis Testing for Real-Time Detection of Lethal Cardiac Arrhythmias
- PDF / 517,056 Bytes
- 8 Pages / 600.03 x 792 pts Page_size
- 114 Downloads / 189 Views
Research Article Complexity-Measure-Based Sequential Hypothesis Testing for Real-Time Detection of Lethal Cardiac Arrhythmias Szi-Wen Chen Department of Electronic Engineering, Chang Gung University, 259 Wen-Hwa 1st Road, Kwei-Shan, Tao-Yuan 333, Taiwan Received 15 June 2006; Revised 17 September 2006; Accepted 17 September 2006 Recommended by Tan Lee A novel approach that employs a complexity-based sequential hypothesis testing (SHT) technique for real-time detection of ventricular fibrillation (VF) and ventricular tachycardia (VT) is presented. A dataset consisting of a number of VF and VT electrocardiogram (ECG) recordings drawn from the MIT-BIH database was adopted for such an analysis. It was split into two smaller datasets for algorithm training and testing, respectively. Each ECG recording was measured in a 10-second interval. For each recording, a number of overlapping windowed ECG data segments were obtained by shifting a 5-second window by a step of 1 second. During the windowing process, the complexity measure (CM) value was calculated for each windowed segment and the task of pattern recognition was then sequentially performed by the SHT procedure. A preliminary test conducted using the database produced optimal overall predictive accuracy of 96.67%. The algorithm was also implemented on a commercial embedded DSP controller, permitting a hardware realization of real-time ventricular arrhythmia detection. Copyright © 2007 Szi-Wen Chen. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1.
INTRODUCTION
Ventricular fibrillation (VF) and ventricular tachycardia (VT) are life-threatening cardiac arrhythmias [1]. Reduction of mortality from such cardiac causes depends on rapid detection and accurate classification of these arrhythmias. Thus, the development of accurate noninvasive techniques for identifying patients at risk of lethal arrhythmias is essential to reducing mortality from cardiac deaths. For this reason, a number of quantitative analysis techniques for electrocardiogram (ECG) arrhythmia recognition have been proposed previously [1–9]. While all these algorithms show advantages in versatile aspects of performance evaluation, some of them are still too difficult to implement and compute for defibrillators. On the other hand, for computational convenience, some algorithms utilized in either surface ECG monitoring-based automated external defibrillators (AEDs) or in implantable cardiovertor/defibrillators relied only on simple heart rate for arrhythmia detection. In fact, this might be problematic since it has been indicated that simply using heart rate as the sole feature might always unavoidably lead to a certain error rate in the detection since while both VF and VT have significantly higher rates than normal sinus rhythm, the rate range of VF overlaps with that of VT. Therefore,
using the heart rate as a single feature might achieve
Data Loading...
