314 ANALYSIS OF ELECTROCARDIOSIGNALS WAVELET-SPECTRUM FACTOR DISTRIBUTION AIMED AT CLASSIFYING CARDIOVASCULAR DISEASES 1 D.G. Khafizov2, N.V. Parsaev2 2Mari State Technical University, Lenin Sq. 3, 424000 Yoshkar-Ola, Russian Federation. tel.: (8362) 455412 e-mail: [email protected] Histograms of factor distribution of electrocardiosignals wavelet-spectrum are determined and analyzed in normal state and under various pathologies. Probabilities of electrocardiosignals correct recognition by histograms of wavelet-spectrum factor distribution are obtained. Introduction Cardiovascular diseases (CVD) are the most widespread reasons of lethality. The analysis of the electrocardiogram allows the skilled expert to judge a damage rate of a myocardium. Thus computer diagnostics of cardiovascular system condition is meant to help the cardiologist with the analysis of electrocardiograms. Wavelet transform is the ideal instrument for the analysis of non-stationary signals such as electrocardiosignal [1]. Wavelet transform Wavelet transform allows to localize the characteristic features of electrocardiosignals by time and frequency, resolving the problem of signal particularization. Time-scale localization of characteristic features of electrocardiosignals by the results of wavelet transform depends on the principle of construction the orthogonal basis of decomposition. Unlike the Fourier transform traditionally applied for signal analysis the wavelet transform provides two-dimensional presentation of onedimensional signal under study, for this purpose frequency and coordinate are considered independent variables. Eventually there is an opportunity to analyze the properties of a signal simultaneously in both physical and frequency spaces. By the results of research, given in [2], a wavelet «Mexican hat» was chosen as a parent wavelet, 1 1 mhat t exp t 2 t 2 exp t 2 , 2 2 which is more adaptive to the reference electrocardiosignal by criterion of resolution in the field of low frequencies where the greatest self-descriptiveness of electrocardiograms is incorporated. The power spectrum of "Mexican hat" wavelet without stretch is located within the frequencies ranging from 0 up to 4 Hz, therefore to study the whole frequency range of electrocardiosignal it is necessary to apply stretching and compression of parent wavelet. 1 t mhat . a, t a a Since the useful spectrum of electrocardiosignal is located within the range between 0,05 and 100 Hz, if we conduct a wavelet analysis of electrocardiosignal with the change of wavelet stretch factor а "Mexican hat" within the range between 1/64 and 16 with gradual increase as much again the whole useful spectrum of the signal will be studied. Factors of electrocardiosignals wavelet– notation are defined as follows: N 1 1 k T j T Сi, j f k mhat . j i a 0 2 a0 2 k 0 _______________________________________________________________________ 1 Work is executed under the financial support of RFFI, project 07-01-00058а 315 Thus, wavelet notations of electrocardiosignals (ECS) were obtained (Fig. 1-5). Fig. 4. Wavelet plane of the electrocardiogram at hypertrophy of the left ventricle Fig. 1. Wavelet plane of the normal electrocardiogram Fig. 5 Wavelet plane of the electrocardiogram at an acute cardiac infarction Fig. 2. Wavelet plane of the electrocardiogram at WPW syndrome Fig. 6. Histogram of factor distribution of normal ECS wavelet-notation Fig. 3. Wavelet plane of the electrocardiogram at hypertrophy of the right auricle The analysis of wavelet-spectra On the basis of the obtained wavelet-spectra, the histograms of factor distribution of ECS wavelet notation were calculated and built (Fig. 6-10). Fig. 7. Histogram of factor distribution of waveletnotation at WPW syndrome 316 Fig. 8. Histogram of factor distribution of waveletnotation at hypertrophy of the right auricle (RA) Fig. 9. Histogram of factor distribution of waveletnotation at hypertrophy of the left ventricle (LV) Fig. 10 Histogram of factor distribution of waveletnotation at an acute cardiac infarction (CI) On the basis of the criterion 2 widely used in biometry the degree of distinction of these histograms (table) was evaluated. Values 2 for distributions of wavelet-notation fac- Cardiovascular diseases Normal ECS WPW syndrome Hypertrophy of RA Hypertrophy of LV tors of ECS at CVD WPW Hyper- HyperNormal Acute syn- trophy trophy ECS CI drome of RA of LV 0 287,7 287,7 1,3·103 1,8·103 1,7·103 0 1,3·103 1,4·103 1,4·103 1,3·103 1,9·103 0 1,8·103 1,3·103 1,1·103 1,1·103 2,9·103 0 1,7·103 1,9·103 2,9·103 2,5·103 2,5·103 0 From the table it can be seen that for all pairs of compared histograms actual values of 2 exceed the critical value 2кр 37,65 on the confidence level 0,05 . Taking this into account it is possible to draw a conclusion about the opportunity of electrocardiosignals recognition of the considered cardiovascular diseases, on the basis of distribution of wavelet-notation factors. Conclusion For classification of CVD diseases we chose diseases with infringement of cardiac conduction and the possibility of their diagnostics with the help of II standard lead: WPW syndrome (syndrome of preexcitation of ventricles), hypertrophy of the right auricle, hypertrophy of the left ventricle and acute cardiac infarction. To find a wavelet-transform we chose a parent wavelet MHAT ("Mexican hat") and defined its stretch boarders taking into account useful spectrum of electrocardiosignals. On the basis of electrocardiosignal models of these cardiovascular diseases their waveletnotations were obtained and histograms of their factor distribution were calculated. On the basis of 2 method the algorithm of cardiovascular diseases classification by factor distributions of electrocardiosignals waveletnotations was developed. References 1. Dobeshi I. Ten lectures on wavelets. – Izhevsk: Scientific center «regular and chaotic dinamics», 2001, 464 p. 2. Badin О.N., Burukina I.P. Peculiarities of electrocardiographic information analysis with the use of velevet-transform // Medical equipment. – 2004. – № 4. – p.26 – 29. 3. Astafeva N.М. wavelet-analysis: basics of theory and examples of application (actual problems reviews). – М.: – 1996. 26 p. 4. Bala Y.М. et. al. Atlas of practical electrocardiography/ Y.М. Bala, А.V. Nikitin, V.B. Fuki. – Voronezh: Publishing house of Voronezhskiy University. – 1983. – 176 p.
© Copyright 2026 Paperzz