DWT Feature Extraction and Classification Using ANN for Detection of Cardiomyopathy

Abstract

The Proposed System Detects a Heart Disease called Cardiomyopathy from ECG signal by using Image and Signal processing techniques Discrete Wavelet Transform and Artificial Neural Network. Cardiomyopathy is a disease of Heart muscle in which the muscle get enlarged or thickened and the heart lacks it’s working. Electrocardiogram or ECG is a diagnostic tool used for finding any heart related problem the ECG is Visually Analyzed by Cardiologist as a Human he is prone to make wrong interpretation that may leads to wrong diagnosis hence automation is required to help medical practitioners therefore the system is developed to Analyze the ECG signal, extract the valuable information and detects the disease automatically. ECG signals for Cardiomyopathy and Healthy are obtained from PTB diagnostic ECG database. DWT is used for feature Extraction and Artificial Neural Network is used for classification. Classification accuracy obtained with Levenberg-Marquardt algorithm is 93.65% which is better when compared with gradient descent giving 88.71%. Keywords: Cardiomyopathy, Discrete Wavelet Transform, Artificial Neural Network, Classification accuracy