University of Twente Student Theses

Login

Weighted convolutional neural networks rare electrocardiogram detection for real-time heart monitoring

Jonker, J.H.C. (2021) Weighted convolutional neural networks rare electrocardiogram detection for real-time heart monitoring.

[img] PDF
424kB
Abstract:The electrocardiogram(ECG) plays a vital role to reduce the high mortality rate from cardiovascular diseases in computer-aided arrhythmia detection. Many arrhythmia classification research is devoted to developing classifiers that attain high prediction accuracies from the MIT-BIH Arrhythmia dataset. However, the complex variations and high imbalance in this dataset makes this a demanding issue. Current state-of-the-art research achieves high overall accuracy in classifying regular ECG beats but receives less satisfactory results in the classification of rare classes. This research proposes to apply weights to the minority classes in the loss function of convolutional neural networks. The study will show that the proposed method can achieve high performance on rare ECG beat detection on embedded devices. The research will be compared to the state-of-the-art using the Supraventricular Ectopic beats(SVEB) and Ventricular Ectopic Beats(VEB) evaluation metrics, the research will be compared to the state- of-the-art. The best performing ECG classifier presented in this paper achieved an SVEB- and VEB-accuracy of 99.7% and 99.6%, respectively. The proposed classifier required around 2ms for classification per sample, which is suitable for real-time application.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business & IT BSc (56066)
Link to this item:http://purl.utwente.nl/essays/86913
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page