Adaptive feature extraction for QRS classification and ectopic beat detection

Abstract

Automatic procedures to classify the QRS complex are very useful in diagnosis of cardiac disfunctwns. In this work we present an adaptive system to extract, in real time, the features that characterize the QRS with the Hermite functions model.The adaptive system is based on the multiple-input adaptive linear combiner, where the primary input signal is the succession of the QRS complexes, and the reference inputs are the Hermite functions. The weight vector becomes an estimation of the coefficients that represent the QRS complex in the Hermite function base. To adapt these weights we w e the LMS algorithm. We have incorporated a procedure to aLiaptively estimate a width parameter (b) that best fits each QRS complex. We present applicatww of this system to classify QRS in case of ECG signals affected by the phenomenon of bigeminy and to detect ectopic beats using the b parameter. In both cases we have obtained correct pattern classification.

Truncated orthogonal expansions of recurrent signals: equivalence to a linear time-variant periodic filter

“…When the input signal is corrupted with uncorrelated noise, adaptive techniques [14]- [17] are often used to estimate the orthogonal expansion coefficients. The reference inputs to the adaptive linear combiner [15] …”

“…ECG data compression systems based on orthogonal expansions, like [17], [19], and [25], get better rate-distortion tradeoff than methods based on interpolation techniques [4]. With the shown description of orthogonal expansion of recurrent signals as a linear time-variant periodic filter, we can quantitatively predict which frequency components are well represented at every occurrence time using a variable number of basis functions.…”

“…Two classical and well-known techniques are often used to estimate the expansion coefficients: inner product (IP) [12], [13] when the value of the signal to noise ratio (SNR) is high, and adaptive estimation with the LMS algorithm [14]- [17], when the uncorrelated noise with the signal must be attenuated. Both techniques have been previously analyzed by many authors, but to the best knowledge of the authors, the analysis with a variable number of basis functions has yet to be addressed.…”

Denoising of ECG Signal by using Adaptive Filter and Non Adaptive Filter

“…In the LMS algorithm works on an instant basis such that weight vector is renewed for each sample. The calculation complexity can be decrease by providing the sign based algorithms, such as, the signed regressor algorithm, the sign algorithm and the sign-sign algorithm [14,15]. In order to work with both the difficulty and convergence issues we used various adaptive filter structure based on standardized sign regressor LMS (NSRLMS) algorithm, normalized sign LMS (NLMS) algorithm and normalized sign-sign least mean square error (NSSLMS) algorithm.…”

Comparison of four methods for premature ventricular contraction and normal beat clustering

“…Some arrhythmias appear infrequently, and in order to capture them the clinicians use Holter devices. The use of specific algorithms for automatic analysis of ECG recordings may facilitate the analysis of the very long Holter ECG recordings.Several algorithms for the discrimination between normal beats (N) and premature ventricular contractions (PVC) have been proposed in literature, some of them using heart beat morphology parameters [1][2][3][4][5][6] or frequency-based parameters [7,8].In addition numerous classification methods have been studied, and they include: adaptive signal processing for on-line estimation of non-stationary signals that present a recurrent behaviour [9][10][11][12][13], linear discriminants [4,5], neural networks [14,15,3,8], fuzzy adaptive resonance theory mapping [16], operation on vectors in the multidimensional space [6] and self-organized maps [17].A particular aspect of the learning strategy is studied, paying attention to the organization of the classifiers' training set, and considering two main strategies: local learning set and global learning set [18,4,6]. In the first case the learning set is customized to the tested patient, while in the latter it is built from a large ECG database.…”