sajjadrezvani/ecg_ai

ECG Signal Processing using Neural Networks

Electrocardiogram(ECG) Signal processing using DeepLearning and AI in Python

My Bachelor Thesis Documents

Abstract

Heart diseases are a major cause of death worldwide that take many lives each year. Doctors use ECGs to analyze the heart's electrical signals and diagnose these diseases, either to prevent them or treat them after they occur. In comparison with humans, computer systems, including artificial intelligence, offer more accurate and faster diagnosis of heart diseases. Many models have been developed to improve the accuracy and speed of diagnosing heart diseases or suggest new approaches; but there are some challenges like complex signals or lack of data that cause model overfitting. In our research, we tested LSTM and CNN models on the PTB-XL dataset and found a one-dimensional CNN-based model with an accuracy of 79.24% in predicting disease across five categories. This accuracy performs better previous models and is comparable to the best accuracy in similar dataset.

Keywords

Electrocardiogram, Cardiac signal classification, Neural networks, Deep learning