ARTICLE IN PRESS Pattern Recognition ( ) – www.elsevier.com/locate/patcog Quantitative evaluation of a wavelet-based method in ventricular late potential detection Ali Shahidi Zandi ∗ , Mohammad Hassan Moradi Faculty of Biomedical Engineering, Amirkabir University of Technology, PO Box: 15875-4413, 424 Hafez Ave., Tehran, Iran Received 1 July 2005; received in revised form 19 December 2005; accepted 17 January 2006 Abstract Ventricular late potentials (VLPs) are low-amplitude, high-frequency waveforms appearing in the terminal part of the QRS complex in electrocardiogram (ECG) of patients who are susceptible to ventricular tachycardia and sudden cardiac death, after surviving myocardial infarction.Accordingly, VLP detection presents a prominent non-invasive marker for some cardiac diseases clinically. This paper proposes a VLP detection method based on the wavelet transform and investigates its performance. In this method, a modified vector magnitude waveform is formed using discrete wavelet transform for each high-resolution ECG (HRECG) record; then, by applying the continuous wavelet transform to the QRS complex end part in this waveform, a feature vector is extracted from the resultant time-scale plot. This wavelet-based feature vector is processed by principle component analysis to reduce its dimensionality. Finally, a supervised feedforward artificial neural network, trained by a proper set of these feature vectors, is employed as a classifier. To evaluate the proposed method performance, a HRECG database consisting of the real VLP-negative and simulated VLP-positive patterns is used. In a comparative approach, different VLP detection techniques including the conventional time-domain method, developed by Simson, and some methods utilizing distinct diagnostic features are also applied to this database to investigate the capability of the proposed method in VLP analysis more completely. The results show the proposed method, employing the wavelet transform in both pre-processing and feature extraction stages, reveals high evaluation criteria (accuracy, sensitivity, and specificity) and is qualified to detectVLPs.  2006 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved. Keywords: Ventricular late potentials (VLPs); High-resolution ECG (HRECG); Wavelet transform (WT); Artificial neural network (ANN); Principal component analysis (PCA) 1. Introduction Cardiovascular diseases cause a considerable part of mor- tality all over the world annually, and myocardial infarc- tion (MI) is responsible for many of these deaths. Although some patients survive MI, many of them will be prone to some serious types of cardiac arrhythmias, such as ventric- ular tachycardia (VT) and ventricular fibrillation (VF), after the heart attack. Whereas ventricles beat too rapidly in VT ∗ Corresponding author. Tel.: +98 21 64542399, +98 21 88890419; fax: +98 21 66495655. E-mail addresses: ashahidi@bme.aut.ac.ir, ali_shahidi_zandi@yahoo.com (A. Shahidi Zandi), mhmoradi@aut.ac.ir (M.H. Moradi). 0031-3203/$30.00  2006 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.patcog.2006.01.012 in comparison with the normal heart activity, VF causes the ventricles to contract in a disorganized fashion resulting in loss of blood-pumping action. If VF is not interrupted by an electrical shock, it kills the patient immediately. VT is, also, a potentially lethal arrhythmia because not only does VT cause ventricles not to fill adequately and loss of blood pres- sure, but it also may progress into VF. Accordingly, most sudden cardiac deaths (SCDs) are thought to be initiated by VT or VF [1]. As the appearance of ventricular late poten- tials (VLPs) is associated with VT in many cases, there is a clinical interest in detection of VLPs in order to anticipate the occurrence of this arrhythmia. In other words, the VLP detection is a non-invasive diagnostic marker of VT for post-MI patients. VLPs are low-amplitude, high-frequency signals appearing at the end part of the QRS complexes in the electrocardiogram (ECG) signal. These potentials arise