Heartbeat Classification of Arrhythmia using Hybrid Features Extraction Techniques

An electrocardiogram (ECG) is the noninvasive method for arrhythmia recognition used to monitor the heart electrical activity. The hidden information present in ECG data and the irregularity of the heartbeat is difficult to determine. Hence, this hidden information can be used to detect abnormalities. This paper proposes hybrid feature extraction techniques for raw ECG heart beat classification. Many feature extraction techniques and machine learning algorithms are used f that combined features of time domain, frequency domain and time frequency domain gives the best accuracy. These features were combined to create a hybrid features with huge amount of data. In addition, some of the extracted features have been reduced using feature reduction techniques. The suggested method was accomplished to be able to acquire better features of the ECG data. This approach has the ability to recognize the discriminant features between heartbeats classes of the input signals using two types of classifiers, namely, the neural network methods(NN), and support vector machine (SVM-RBF) with tenfold cross-validation technique. The Association for Advancement of Medical Instrumentation analyzed five types of beat classes of arrhythmia such as: non-ectopic beats (N), supra-ventricular ectopic beats (S), ventricular ectopic beats (V), fusion beats (F) and unclassifiable and paced beats (U). Our proposed method is able to classify the N, S, V, F and U arrhythmia classes with high accuracy (98.95%) but with (39) feature dimension.