|
A Deep Learning Approach for Detection and Classification of QRS Contours using Single-lead ECG
|
|
| Author:
|
A. ANUHYA, VENKATA RATNAM KOLLURU, RAJESH KUMAR PATJOSHI
|
| Abstract:
|
Early detection of cardiovascular diseases can prevent millions of deaths every year worldwide. The aim of this research is to present a novel approach for the automatic diagnosis of Electrocardiogram (ECG) abnormalities based on detection of QRS complexes. This paper proposes an efficient Global QRS-Deep Neural Network (GQRS-DNN) model for ECG beat classification. Pre-processing and feature extraction is done with the use of traditional GQRS algorithm to extract the QRS complexes. The performance of algorithm is tested on MIT-BIH Arrhythmia database and have obtained an average Sensitivity of 99.52%, Positive Predictivity of 99.80%, F1-Score of 0.977 and Accuracy of 99.26%. The extracted features are given as inputs to the classifier to boost the performance of the system. Further, a comparison is made between the proposed DNN with SVM (Support Vector Machine) and KNN (K-Nearest Neighbors) and observed that proposed DNN method achieved the highest accuracy of 99.7%. It is noticed that the proposed network results in improved classification performance with less number of nodes and is applicable in real time for the development of smart health.
|
| Keyword:
|
Deep Neural Network, ECG, K-Nearest Neighbours, QRS Complex, Support Vector Machine
|
| EOI:
|
-
|
| DOI:
|
https://doi.org/10.31838/ijpr/2020.12.02.0001
|
| Download:
|
Request For Article
|
|
|