Wide complex tachycardia discrimination using dynamic time warping of ECG beats. (October 2018)

• Record Type: Journal Article
• Authors: Mazandarani, F. Niknejad; Mohebbi, M.
• Is Part Of: Computer methods and programs in biomedicine. Volume 164(2018)
• Page Start: 238
• Page End: 249
• Publication Date: 2018-10
• Subjects: Wide complex tachycardia -- Dynamic time warping -- Ventricular tachycardia -- Supraventricular tachycardia; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2018.04.009
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: The first study for automatic wide complex tachycardia arrhythmia discrimination. Using dynamic time warping of ECG beats. This method includes pre-processing, best template search and 1NNclassifier modules. Good performance in discrimination between VT, SVT-A and normal subjects. Abstract: Background and objective : Automatic processing and accurate diagnosis of wide complex tachycardia (WCT) arrhythmia groups using electrocardiogram signals (ECG) remains a challenge. WCT arrhythmia consists of two main groups: ventricular tachycardia (VT) and supraventricular tachycardia with aberrancy (SVT-A). These two groups have similar morphologies in the realm of ECG signals. VT and SVT-A arrhythmias originate from the ventricle and atrium, respectively. Hence, inaccurate diagnosis of SVT-A instead of VT can be fatal. Methods : In this paper, we present a novel algorithm using dynamic time warping (DTW) to discriminate between VT and SVT-A arrhythmias. This method includes pre-processing, best template search (BTS), and classifier modules. The first module, pre-processing, is responsible for filtering, R-wave detection, and beat detection of ECG signals. The second module, BTS, automatically extracts the minimum possible number of signals as a template from the entire training dataset using an intelligent algorithm. These template signals have the greatest morphological difference, which leads to accurate WCT discrimination. Finally, a 1NN classifier categorizes the test data using DTW distance. Results : Our proposed method was evaluated on an ECG signal database consisting of 171 subjects. The results showed that the proposed algorithm can accurately discriminate between VT, SVT-A, and normal subjects, and appears to be suitable for future use in clinical application. The obtained accuracy, sensitivity, specificity, and positive predictive values were 93.22%, 88.68%, 96.98%, and 90.27%, respectively. Conclusion : The presented diagnostic method for discriminating VT and SVT-A, using only one ECG lead, is suitable for future clinical use. It can reduce needless therapeutic interventions and minimize risk for patients.