International Journal of Artificial Intelligence and Neural Networks
Author(s) : SANG-HONG LEE
This study proposes feature extraction using wavelet transform (WT), sequential increment method, and phase space reconstruction (PSR) to classify normal sinus rhythm (NSR) and ventricular fibrillation (VF) from ECG episodes. We implemented four pre-processing steps to extract features from ECG episodes. In the first step, we use the WT for multi-scale representation and analysis, and then we extract wavelet coefficients from ECG episodes. In the second step, we use sequential increment method to extract peaks from the wavelet coefficients. In the third step, we make a threedimensional phase space reconstruction (PSR) using the successive peaks. In the final step, we calculate the Euclidean distance between the peaks that are plotted in a threedimensional phase space diagram and origin (0, 0), and then extract 20 features from the Euclidean distances by using statistical methods, including frequency distributions and their variabilities. We apply the 20 features as inputs to a neural network with weighted fuzzy membership functions (NEWFM), and recorded sensitivity, specificity, and accuracy performances of 100%, 100%, and 100%, respectively.