Cardiac Arrhythmia Detection through ECG Signals

ECG signals are widely used for detecting any abnormality related to heart. ECG signal has number of cardiac cycles and each cardiac cycle has P-QRS-T waves. The aim behind implementing this project is to detect cardiac arrhythmia using KNN and SVM classifiers. In this work, a total data of 48 subje...

Full description

Saved in:
Bibliographic Details
Published in2018 4th International Conference for Convergence in Technology (I2CT) pp. 1 - 5
Main Authors Patil, Anita, Edake, Ravina D.
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.10.2018
Subjects
Cardiovascular Diseases (CVDs)
Discrete Wavelet Transform (DWT)
Discrete wavelet transforms
Electrocardiogram (ECG)
Electrocardiography
Feature extraction
Finite impulse response filters
K Nearest Neighbor (KNN)
Mathematical model
Support Vector Machine (SVM)
Support vector machines
Online AccessGet full text

Cover

More Information
Summary:ECG signals are widely used for detecting any abnormality related to heart. ECG signal has number of cardiac cycles and each cardiac cycle has P-QRS-T waves. The aim behind implementing this project is to detect cardiac arrhythmia using KNN and SVM classifiers. In this work, a total data of 48 subjects ECG signals is used. Zero phase filter is used to eliminate the baseline noise. Daubechies wavelet 4 is used for feature extraction. KNN and SVM classifiers are used to classify the signals into normal and abnormal group. Performance evaluations (accuracy, sensitivity & specificity) are calculated for both the classifiers. Accuracy for KNN classifier is 76.92% where as accuracy for SVM classifier is 79.48%. Sensitivity of KNN is 82.35% and for SVM it is 71.42%. Specificity for KNN classifier is 72.72% and for SVM classifier it is 100%. Performance of both the classifiers is compared with the help of confusion matrix.
DOI:10.1109/I2CT42659.2018.9057836