Combined wavelet transformation and radial basis neural networks for classifying life-threatening cardiac arrhythmias

Abstract

Automatic detection and classification of arrhythmias based on ECG signals are important to cardiac-disease diagnostics. The ability of the ECG classifier to identify arrhythmias accurately is based on the development of robust techniques for both feature extraction and classification. A classifier is developed based on using wavelet transforms for extracting features and then using a radial basis function neural network (RBFNN) to classify the arrhythmia. Six energy descriptors are derived from the wavelet coefficients over a single-beat interval from the ECG signal. Nine different continuous and discrete wavelet transforms are considered for obtaining the feature vector. An RBFNN adapted to detect and classify life-threatening arrhythmias is then used to classify the feature vector. Classification results are based on 159 arrhythmia files obtained from three different sources. Classification results indicate the potential for wavelet based energy descriptors to distinguish the main features of the signal and thereby enhance the classification scheme. The RBFNN classifier appears to be well suited to classifying the arrhythmia, owing to the feature vectors' linear inseparability and tendency to cluster. Utilising the Daubechies wavelet transform, an overall correct classification of 97.5% is obtained, with 100% correct classification for both ventricular fibrillation and ventricular tachycardia.

Computerized Interpretation of Cardiovascular Physiological Signals

“…Take computerized ECG arrhythmia interpretation for an example. It has been confirmed that Hermite basis functions (HBFs) and wavelet energy descriptors are among those most competitive ones for feature characterization in discrimination analysis (Senhadii et al, 1995;Rasiah et al, 1997;AI-Farhoum & Howitt 1999;Lagerholm et al, 2000;Saxena et al, 2002;Linh et al, 2003;Engin 2007).…”

“…Actually, signal representation in time domain is legible but redundant, which may be evidenced by means of PCA (Geva 1998;Stamkopoulos et al, 1998). As a consequence, morphological analysis was generally combined with domain transformation, such as Hilbert transform (Bolton & Westphal 1981), HD (Rasiah et al, 1997;Lagerholm et al, 2000;Linh et al, 2003) and WT (Senhadii et al, 1995;AI-Farhoum & Howitt 1999;Saxena et al, 2002;Engin 2007), in those published paradigms of computerized physiological signal interpretation. Domain transformation, unlike direct morphological analysis, attempts to characterize physiological signals in an alternative space, where the genuine signal components are more discernible from noises and artefacts.…”

“…Thus a variety of successful applications have been reported on physiological signal processing and analysis, covering from adaptive denoising, compression, delineation to feature characterization, pattern recognition, and others. In terms of feature characterization, the investigators paid substantial attention on wavelet energy (Senhadii et al, 1995;AI-Farhoum & Howitt 1999), wavelet entropy (Rosso et al, 2001) and other energy-dense wavelet coefficients (Senhadii et al, 1995). In essence, wavelet transform can be considered as a kind of multirate filter banks.…”

Machine Learning:An Effective Technique in Bio-Medical Signal Analysis and Classification

“…Both the time and frequency representation can be possible due the scaling property. It has made the wavelet transform suitable for analysis of the non-stationary signals such as speech and ECG signals [17][18][19][20]. This work is confined to ECG signal which remains noisy and non-stationary using the wavelet transform.…”

Evaluation of heart rate variability by using wavelet transform and a recurrent neural network

“…Miiiami et al combined the feature extraction by Fourier transform and a back-propagation neural network (BPN) [4], and detected the tachyarrhythmia in real-time. Fahoum et al combined the WT and an R,BF neural network (R,BFN), in order to detect life-threatening cardiac arrhythmias [5]. The purpose of these reports was to detect the abnormal waveform on the ECG.…”