ELECTRONICS’ 2005 21 – 23 September, Sozopol, BULGARIA SPECTRAL ANALYSIS OF LIFE-THREATENING CARDIAC ARRHYTHMIAS Vessela Tzvetanova Krasteva, Irena Ilieva Jekova Centre of Biomedical Engineering “Prof. Ivan Daskalov” - Bulgarian Academy of Sciences Acad.G.Bonchev str. Bl.105, 1113, Sofia, Bulgaria, e-mail: vessika@clbme.bas.bg The presented work is aimed at investigation of power spectra characteristics of ECG life-threatening cardiac arrhythmias by applying Fourier transform on 10 s ECG epochs. A total number of 240 ECG signal segments (selected from a large out-of-hospital ECG database to cover the variety of specific ECG waves, which appear for six arrhythmia categories) were involved into statistical analysis in order to assess the possibility for separation between shockable and non-shockable arrhythmia categories, using the calculated power spectra characteristics. Their combination with ECG morphology parameters could be a useful approach for improvement of the shock-advisory decision accuracy of automatic external defibrillators, in order to match the diagnosis of trained medical personnel. Keywords: ECG arrhythmia, power frequency spectrum, ECG characteristics 1. INTRODUCTION The reliable recognition and adequate electrical shock therapy of life-threatening cardiac states depends on the electrocardiogram (ECG) descriptors, which are used by the defibrillator-embedded automatic arrhythmia analysis algorithms. Our previous experience with real-time analysis of ECG morphology parameters and band-pass filter output parameters [1] proved that the value range of each parameter overlaps for the two general classification groups of shockable and non-shockable rhythms. More detailed analysis [2] demonstrated that this overlapping results from shared distributions of the parameter values for specific arrhythmias, which belong either to the shockable rhythms or to the non-shockable rhythms. Since we had recently studied a wide set of morphology parameters and several parameters extracted in a narrow frequency band, it was of interest to investigate the spectral characteristics of the different arrhythmia types in the whole range of spectral frequencies. The power spectrum of the ECG signal could provide useful information about the QRS complex and the other ECG waves (P, T, fibrillation). Thakor et al. [3] designed a band-pass filter with a central frequency at 17 Hz, which effectively detected the QRS complexes in the ECG signal. Minami et al. [4] studied the QRS frequency spectra in 4 Hz bands for classification of three arrhythmia types. Other authors [5,6] used the ECG frequency characteristics to apply wavelet transformations for detection of life-threatening cardiac arrhythmias. However, the achieved accuracy for making an adequate automatic shock advisory decision should be improved in order to match the diagnostic accuracy of trained medical personnel. The presented study was aimed at assessment of the spectral characteristics of a wide variety of arrhythmia types in the whole range of spectral frequencies in order to investigate the possibility for better separation between shockable and non- shockable rhythms.