Application of Wavelet Transform for Analysis of
QRS Complex in Intraventricular
Conduction Abnormalities

Hisa Shimojima^a, Takeshi Tsutsumi^a, Fumiko Yanagisawa^a, Masae Komukai^a, Naoko Zenda^a,
Yukei Higashi^a, Youichi Takeyama^a, and Yoshiwo Okamoto^b

^aDivision of Cardiology, Showa University Fujigaoaka Hospital, Yokohama 227-8501 Japan
^bDepartment of Electric Engineering, Chiba Institute of Technology, Narashino 275-0016 Japan

1.    Introduction

It has been pointed out by several authors (Flowers, 1969; Maehara, 1999; Zenda 2000) that the scattered myocardial damages such as focal necrosis may be related to the production of high frequency component, but the clinical significance of these findings has not been established.　Compared with widely used FFT algorithm, the wavelet transform (WT) is based on special functions with time-widths adapted to each frequency and is suitable for transient signal detection within the QRS (Morlet, 1993; Figliola, 1997), which can be applied for exploring the deformation of cardiac excitation wave front. It has been known that the progress of normal activation of the ventricular wall is nearly radial and that of intraventricular conduction abnormalities (IVCA) is more tangential. Different findings are expected in the power spectrum of these conditions, as well as in cases with conduction barriers such as myocardial infarction (MI). The present study was performed to evaluate the alteration of the frequency power during QRS in IVCA with or without MI by the new method using WT.

2.    Material and Methods

Standard 12 leads electrocardiogram (ECG) were recorded at sampling rate of 10 kHz from normal volunteers (n=12) and the patients with IVCA (n=6), and with IVCA plus MI. These signals were inputted into Window based PC as binary files. To assess the frequency power limited to QRS, WT was utilized for the assessment of signals within the QRS interval. The PC software was newly developed for this purpose (BIOMAS ver 1.0, Elmec Co., Inc., Tokyo, Japan). The Morlet wavelet was used as a mother wavelet. The time course of the signal power was obtained from 40 scales corresponding to frequency bands from 10-400Hz. In addition, the time integral of the power spectrum corresponding to each frequency was calculated over 100msec period from the onset of the QRS complex. These calculation steps were repeated on 10 consecutive beats and the results were averaged and compared in cases with and without MI.

3.    Results

Two representative examples of frequency analysis of QRS in V₃ are illustrated in figure 1. A normal case is shown in the left panel, where greater power, indicated by red zone, was observed at frequency range from 20 to 80Hz. The red zone was surrounded by middle frequency power (green zone). A case with abnormal QRS under right ventricular pacing was illustrated in the right panel. In this case, the red zone is narrower, and the green zone was distributed over wide range QRS interval. Moreover there are irregular upward projections of the light blue zone, denoting smaller power. The time integral of the power spectrum against the frequency were shown in figure 2. As seen in the left panel, the time integral was larger at low frequency (100Hz or less) with exponential decay at increasing frequencies in normal case (N). In IVCA, the integral values were smaller in low frequency range compared with those in normal cases. In IVCA with MI, the integral values were greater in high frequency range (150-250Hz), as indicated by an arrow in the right panel of figure 2, values from IVCA (ventricular paced beat) was smaller in low frequency range than it in normal. In contrast, the integral values from IVCA (CLBBB) plus MI was greater in high frequency range (200-250Hz), indicated by the arrow in right panel of figure 2. In addition, the number of sigma peak was larger in IVCA with MI was significantly larger than that in IVCA without MI (15.4±3.1 vs. 8.6±2.3).

4.    Discussion

The results of this study indicate that the frequency analysis with WT is a beneficial tool for elucidating the frequency profiles of QRS complex. The reduction of the low frequency power in IVCA may be due to more tangential activation pattern of the ventricular wall. The increased power and number of peaks in high frequency range (150-250Hz) were observed in IVCA with MI, which may reflect the transformation of the excitation front passing throughout MI lesions. It is suggested that the methods provides a new approach to reveal the myocardial damage concealed under IVCA.

References

Flowers NC, Horan LG, Thomas JR, Tolleson WJ: The anatomical basis for high-frequency components in the electrocraiogram. Circulation XXXIX:531-539, 1969.

Figliola A. and Serrano E: Analysis of physiological time series using wavelet transforms. IEEE Engineering in Medicine and Biology. May/Jun:74-79,1997.

Langner PH, Geselowitz DB, Mansure FT: High-frequency components in the electrocardiogram of normal subjects and of patients with coronary heart disease. Am Heart J. 62:746-755, 1961.

Morlet D, Peyrin F, Desseigne P, Rubel P: Wavelet analysis of high-resolution signal-averaged ECGs in postinfarction patients. J Electrocardiol 26:311-320, 1993.

Maehara K, Kokubun T, Awano N, Taira K, Ono M, Furukawa T, Shimizu Y, Maruyama Y: Detection of abnormal high-frequency components in the QRS complex by the wavelet transform in patients with idiopathic dilated cardiomyopathy. Jpn Circ J 63:25-32, 1999.

Zenda N, Tsutsumi T, Sato M, Takeyama Y, Harumi K, Wei D: Computer simulation of notches on initial part of QRS complex in patients with anterior myocardial infarction, in 2000 Electrocardiology. Ambroggi LD, Editor. Casa Editrice Scientifica Internazionale,Roma,2001,117-120.