Measurement of T wave variability in body surface ECG Mathias Baumert, PhD ⁎ School of Electrical and Electronic Engineering, The University of Adelaide, Australia Abstract Lability in the ventricular repolarization process has been associated with an increased risk of experiencing ventricular tachycardia or fibrillation. A number of risk predictors have been devised that quantify beat-to-beat variability in the T wave morphology of body surface ECG. Initial studies have suggested that measurement of T wave variability may yield important prognostics markers of cardiac mortality, but approaches and experimental designs vary. The aim of this contribution is to provide an overview of existing techniques as well as discuss some of the methodical considerations. © 2016 Elsevier Inc. All rights reserved. Keywords: T wave variability; Repolarization; ECG Introduction Spatiotemporal lability in the ventricular repolarization process has been associated with an increased risk of experiencing ventricular tachycardia or fibrillation (VT/VF). There has been a long-standing interest in utilizing body surface ECG to quantify repolarization lability due to its non-invasiveness, wide availability and simplicity of use. The most prominent repolarization dynamics features of interest are microvolt T wave alternans and beat-to-beat QT interval variability [1]. While the former quantifies one particular phenomenon of T wave variability that typically occurs at higher heart rates and is therefore not easily measurable in ambulatory ECG, the latter does not consider the change in morphology of the T wave, but aims at quantifying beat-to-beat dynamics in the global repolariza- tion duration instead [2]. In an attempt to overcome restrictions associated with each of the two concepts, efforts have been made to develop more general descriptors of the overall beat-to-beat variability in the T wave morphology (T wave variability, TWV). A number of initial studies have suggested that TWV may yield important prognostics markers of cardiac mortality [3,4], but available evidence is limited for various reasons. Among these reasons, methods for TWV quantification vary and a comprehensive evalua- tion of existing techniques is lacking. While a systematic study of TWV methodologies that have reported in the literature is beyond the scope of this contribution, it aims to provide an overview of available techniques as well as discuss some of the technical considerations. Data acquisition and pre-processing Different ECG sources/lead systems have been used for TWV analysis, partly owing to the availability of ECG recordings and/or limitations of data acquisition systems. Given that the heart vector moves in 3-dimensional space as a function of time, multilead assessment would be generally desirable, capturing the tempo-spatial repolarization characteristics. Single lead far-field electrograms of implantable cardio- verter defibrillators have been utilized to study TWV changes before the onset of VT/VF [4]. The vector magnitude ECG of orthogonal leads has been used for TWV analysis [3]. Repolarization variability has also been studied in the vector cardiogram (VCG), obtained by transforming standard 12-lead ECG using well-known matrices or by applying singular value decomposition. In the former case, orientation with respect to a Cartesian reference system is preserved, but correct electrode place- ment is crucial, while the latter provides strictly orthogonal decomposition of the cardiac vector, but without reference to traditional frontal, transverse and sagittal planes. Beat-to-beat changes in T loop morphology or orientation have been exploited to obtain measures of repolarization variability [5]. Pre-processing of ECG, including removal of baseline wander and high frequency noise, is typically carried out using band pass filters. Although filter settings can have significant effects on the T wave morphology, by either Available online at www.sciencedirect.com ScienceDirect Journal of Electrocardiology 49 (2016) 883 – 886 www.jecgonline.com ⁎ School of Electrical and Electronic Engineering, The University of Adelaide, Adelaide, SA 5005, Australia. E-mail address: mathias.baumert@adelaide.edu.au http://dx.doi.org/10.1016/j.jelectrocard.2016.07.014 0022-0736/© 2016 Elsevier Inc. All rights reserved.