Deep Learning for Cardiac Diseases Classification

This study aims to propose a system for the classification of heart diseases based on ElectroCardioGram (ECG) signals, employing a machine learning model and deep learning models. The process is divided into four stages. First, we remove noise using a filter and identify the QRS complex. Moving to the second phase, the cleaned signal is segmented into segments based on the QRS peaks. Subsequently, from each segment, we extract three types of features namely Zero Crossing Rate (ZCR), Cepstral Coefficients (CC), and Entropy (E) that capture different aspects of the signal and combine them. Finally, these features are fed into various models for classification. These models include a traditional machine learning model and a deep learning model. For that we proposed a hybrid Convolutional Neural Network (CNN) with the Support Vector Machine (SVM) and the hybrid Convolutional Neural Network (CNN) with the Long Short-Term Memory (LSTM). The objective is to classify the ECG signal into five different classes: Normal beat (N), Left bundle-branch block beat (L), Right bundle-branch block beat (R), Premature Ventricular Contraction (V), and Paced beat (P). To assess the effectiveness of our model, we conducted tests using the Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database, achieving an impressive accuracy of 97.42% with the CNN-LSTM model.