Digital Signal Processing 18 (2008) 892–899 Contents lists available at ScienceDirect Digital Signal Processing www.elsevier.com/locate/dsp Efficient solution for frequency band decomposition problem using wavelet packet in HRV Suleyman Bilgin a,d , Omer H. Çolak b,d,∗ , Etem Koklukaya c , Niyazi Arı d a Golhisar Vocational School of Higher Education, Mehmet Akif Ersoy University, Burdur, Turkey b Faculty of Engineering, Department of Electrical and Electronics Engineering, Akdeniz University, Antalya, Turkey c Faculty of Engineering, Department of Electrical and Electronics Engineering, Sakarya University, Sakarya, Turkey d Department of Applied Research and Development, University of Applied Sciences, Zurich, Switzerland article info abstract Article history: Available online 25 April 2008 Keywords: Heart rate variability Discrete wavelet transform Wavelet packet Autonomic nervous system Sympathovagal balance Heart rate variability (HRV) is a very significant noninvasive tool for assessment of sympathovagal balance (SB) that reflects variation of parasympathetic and sympathetic activities in autonomic nervous system (ANS). Low frequency/high frequency (LF/HF) power ratio provides information about these activities. Because of nonstationary characteristic of HRV, analyses based on wavelet transform were typically preferred in previous studies. There is an important problem that required frequency ranges for LF and HF cannot be obtained using discrete wavelet transform (DWT). Different sampling frequencies do not remove this problem. In this study, a solution based on wavelet packet (WP) is presented for removing this problem. In addition, effect of WP on SB values is investigated. Method was applied to spontaneous ventricular tachyarrhythmia database and variation of energy values and LF/HF energy ratios were compared for DWT and WP. WP provides absolutely excellent approximation to required frequency bands and exposes different and impressive SB results.  2008 Elsevier Inc. All rights reserved. 1. Introduction Electrocardiogram (ECG) is a powerful tool for assessment of cardiac disturbances. It provides significant information about electrical conduction in the myocardium. Sinoatrial (SA) node that is the pacemaker tissue of heart is associated with autonomic nervous system (ANS) [1–3]. ANS consists of two opposite branches which are called as sympathetic nervous system (SNS) causing the cardio-acceleration and parasympathetic nervous system (PNS) providing a cardio-deceleration effect [1]. This association between SNS and PNS is very important for balance of body. HRV obtained from ECG is a tool for assessment of this relation [1–3]. And also, it provides a measurement of the relationship between cardiac events and ANS or other cardiac arrhythmias [4,5]. In addition, instantaneous changes on HRV related to different drugs were noted in the literature [6]. There are several methods for analyzing HRV [1–3]. By using various power spectral density methods, four main spectral components are occurred and these components are distinguished in a spectrum calculated from short term recordings: ultra low frequency (ULF) ( 0.003 Hz), very low frequency (VLF) (0.003–0.04 Hz), low frequency (LF) (0.04–0.15 Hz), and high frequency (HF) (0.15–0.4 Hz) components [1–3]. The HF component of HRV can be accepted as a marker of parasympathetic modulation and conversely, it is noted that LF component of HRV can be a marker of sympathetic modulation [7]. So, LF/HF power ratio can be considered as the index of SB [1–3,7]. * Corresponding author. E-mail address: omercol@akdeniz.edu.tr (O.H. Çolak). 1051-2004/$ – see front matter  2008 Elsevier Inc. All rights reserved. doi:10.1016/j.dsp.2008.04.007