Design and Comparison of Deep Learning Model for ECG Classification using PTB-XL Dataset

Abstract

The classification of distinct types of electrocardiogram (ECG) signals, such as Normal, ST/T Change, Hypertrophy, Conduction Disturbance, and Myocardial Infarction using deep learning (DL) techniques is crucial to the area of cardiology. The ECG is a common diagnostic tool because of the data that it offers about the electrical function of the heart. However, due to the sheer volume of data and the small differences in waveform sequences, precise and prompt interpretation of ECG signals might be problematic. Incredibly, DL approaches have shown excellence in extracting beneficial features and correlations from massive datasets. ECG signal categorization might be performed automatically with the use of DL techniques, resulting in a more quick and accurate diagnosis. Here, we work on automatic ECG signal classification using DL models like AlexNet and LeNet. At first, we gather data from the PTB-XL ECG database, which contains recordings from patients with a wide range of cardiac diseases. To ensure high-quality data for further analysis, the raw ECG signals are pre-processed to reduce noise and baseline drift. The DL model is then instructed to classify the ECG signals based on their pre-processed data. Metrics including accuracy, precision, recall, and F1-score are used to assess the DL models' effectiveness. In the simulation, the AlexNet method was shown to be successful in accurately categorizing the various cardiac conditions.