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

Abstract

The learning capacity and the classification ability for normal beats and premature ventricular contractions clustering by four classification methods were compared: neural networks (NN), K-th nearest neighbour rule (Knn), discriminant analysis (DA) and fuzzy logic (FL). IntroductionDetection and classification of different types of heartbeats in the electrocardiogram (ECG) is of major importance in the diagnosis of cardiac dysfunctions. 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. The local learning set requires a cardiologist to annotate a set of normal (N) and premature ventricular contraction (PVC) heart beats of the patient under consideration. On the other hand, the capacity of the global learning set to classify new records without additional training is balanced by a lower accuracy, since the morphology of N and PVC differ not only from patient to patient, but also from lead to lead of a same individual.In the present work, the local and global classifiers were investigated, considering 26 morphology heart beat parameters for the classification of normal beats and premature ventricular contractions in the electrocardiogram. For this purpose, the learning capacity and the classification ability of four classification methods were compared: neural networks (NN), K-th nearest neighbour rule (Knn), discriminant analysis (DA) and fuzzy logic (FL). 2.Methods and material ECG databaseAll 48 ECG recordings from the Massachusetts Institute of Technology -Beth Israel Hospital (MIT-BIH) arrhythmia database were used. Each recording has a duration of 30 min and includes two leads -the modified limb lead II and one of the modified leads V1, V2, V4 or V5 [19]. The sampling frequency is 360 Hz and the resolution is 200 samples per mV. Two cardiologists have annotated all beats in the database. Approximately 70% of the beats have been a...

Classification of Premature Ventricular Contraction in ECG

“…However, the size of the learning set was very small (260 beats for the global set, 76 beats for the local set). The best accuracy achieved was 88.5% from the global set with a DA classifier and 98.7% from the local set with a k-NN classifier [8]. Similarly, Christov et al used a k-NN algorithm to classify PVC beats in all files in the Database and achieved sensitivity and specificity rates of 96.9% and 96.7%, respectively [9].…”

“…Researchers attempting to classify PVC arrhythmias have mostly used time-frequency analysis techniques, statistical measurements, and hybrid methods. The most recently published works are those presented in [6][7][8][9][10][11][12][13][14][15]. In [6], the authors applied a dynamic Bayesian network for PVC classification.…”

“…Simple decision rules were used in the classifier process, which was suitable for embedded applications. Another study [8] compared the learning capability and classification skill for normal heartbeats with PVC clustering using four classification techniques: neural networks (NN), the k-nearest neighbour method (k-NN), discriminant analysis (DA) and fuzzy logic (FL). In [9], the authors used the k-NN method to classify PVC beats and normal beats, while the authors in [10] tried to detect PVC using a neural networkbased classification scheme and extracted 10 ECG (electrocardiogram) structural features and one timing interval feature.…”

“…In [15], Independent Component Analysis (ICA) was used for feature extraction and k-means and Fuzzy C-Means(FCM) classifiers were employed to recognize the PVC beat. All of these studies [6][7][8][9][10][11][12][13][14][15] used ECG records from the MIT-BIH Arrhythmia Database.…”

Pro-Detection of Atrial Fibrillation Using Mixture of Experts

“…The Global dataset is built from a large database that most automatic ECG analysis research works refer to [19], [20]. Simply, there are training and testing datasets with different percentages through which the classifier is trained using the training dataset and later predicts the unseen group of data through the testing dataset.…”