From ECG signals to images: a transformation based approach for deep learning

• Published in: PeerJ. Computer science 2021-02, Vol.7, p.e386-e386, Article e386
• Main Authors: Naz, Mahwish, Shah, Jamal Hussain, Khan, Muhammad Attique, Sharif, Muhammad, Raza, Mudassar, Damaševičius, Robertas
• Format: Article
• Language: English
• Subjects: Accuracy; Artificial Intelligence; Automation; Cable television broadcasting industry; Cardiac arrhythmia; Cardiology; Cardiovascular system; Classifiers; Computer Vision; Convolutional neural networks; Data Mining and Machine Learning; Deep features; Deep learning; Disease; ECG; Electric properties; Electric pulses; Electrocardiogram; Electrocardiography; Feature extraction; Feature fusion; Feature selection; Heart; Heart beat; Heart diseases; Image processing; Machine learning; Patients; Pattern recognition; Signal processing; Support vector machines; Tachycardia
• ISSN: 2376-5992; 2376-5992
• DOI: 10.7717/peerj-cs.386

Provocative heart disease is related to ventricular arrhythmias (VA). Ventricular tachyarrhythmia is an irregular and fast heart rhythm that emerges from inappropriate electrical impulses in the ventricles of the heart. Different types of arrhythmias are associated with different patterns, which can be identified. An electrocardiogram (ECG) is the major analytical tool used to interpret and record ECG signals. ECG signals are nonlinear and difficult to interpret and analyze. We propose a new deep learning approach for the detection of VA. Initially, the ECG signals are transformed into images that have not been done before. Later, these images are normalized and utilized to train the AlexNet, VGG-16 and Inception-v3 deep learning models. Transfer learning is performed to train a model and extract the deep features from different output layers. After that, the features are fused by a concatenation approach, and the best features are selected using a heuristic entropy calculation approach. Finally, supervised learning classifiers are utilized for final feature classification. The results are evaluated on the MIT-BIH dataset and achieved an accuracy of 97.6% (using Cubic Support Vector Machine as a final stage classifier).