Classifying ECG Beats Using ICA Features and Probabilistic Neural Network

We propose a method that combines independent component analysis (ICA) and probabilistic neural network to classify electrocardiogram (ECG) signals. In this study, ICA is used to extract important features from ECG signals. A probabilistic neural network follows to classify the input ECG beats into one of eight beat types. The independent components are estimated from the training ECG beats and serve as the bases of the system. The ECG beat samples are then projected on the bases to construct the ICA features for different beat types. The features based on ICA and the time interval between successive ECG beats are then built into a feature vector and employed as inputs to the probabilistic neural network. In the study, 9800 QRS samples, including eight different ECG types, were sampled from the MIT-BIH arrhythmia database. Half of the samples were used in the training phase and the other half in the testing phase. The experiments showed an impressive classification accuracy of 98.71% under the condition that 33 independent components were used. The results show the proposed method is deserved to be an excellent tool in the computer-aided diagnosis system.