Title: Analysis on effect of resampling techniques on cardiac arrhythmia classification using convolutional neural networks

Authors: Rekha Rajagopal; V. Shyam Kumar

Addresses: Department of Information Technology, PSG College of Technology, Coimbatore, India ' Department of Information Technology, PSG College of Technology, Coimbatore, India

Abstract: Cardiac arrhythmia is a condition in which the heart beats at a faster rate or a slower rate instead of a regular rhythm. Medical professionals identify the category of cardiac arrhythmia by manually viewing the electrocardiogram (ECG) signal which is more time-consuming. There are possibilities of incorrectly identifying the arrhythmia categories due to practical difficulties in manually assessing the slight variations in the amplitude of ECG signals. This research work focuses on automatically categorising the heartbeats as normal beat, supraventricular beat, ventricular beat, fusion beat, and unknown beat using convolutional neural network. The class imbalance problem that arises because of a few numbers of heartbeats in certain arrhythmia categories is resolved using techniques such as synthetic minority oversampling technique (SMOTE), borderline SMOTE, SVM-SMOTE, and adaptive synthetic sampling (ADASYN). The proposed model demonstrates an average accuracy of 97.76% in classifying arrhythmia classes using ADASYN technique. This model can help medical professionals in accurately diagnosing the arrhythmia classes.

Keywords: cardiovascular disease; electrocardiogram; ECG; arrhythmia; convolutional neural network; CNN; class imbalance.

DOI: 10.1504/IJMEI.2023.134540

International Journal of Medical Engineering and Informatics, 2023 Vol.15 No.6, pp.516 - 524

Received: 10 May 2021
Accepted: 03 Aug 2021
Published online: 27 Oct 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article