Abstract: Feature selection is one of the most important aspects of classification problem. Due to the nonlinear characteristic of the ECG signal, the time-frequency transformation is proposed for better feature extraction to classify different ECG arrhythmia beats. In our study Choi-Williams time-freqency distribution and short time Fourier transform are used to extract features from the ECG signals. Total sixteen features are extracted which are used as the input to the feed-forward backpropagation neural network using the conjugate gradient optimization algorithm. ECG signals of MIT-BIH arrhythmia dataset are used in this work. Six different classes of ECG beats, Normal (N) beat, Left Bundle Branch Block (L) beat, Right Bundle Branch Block(R) beat, premature ventricular contraction (V) beat, paced (PA) beat and fusion of paced and normal (f) beat are classified and the performance is compared.
Keywords: Time-Frequency Distribution, Backpropagation neural network, Pattern Classification, ECG Arrhythmia