Ogbonnaya, CE;
Preston, SP;
Wood, ATA;
Bharath, K;
(2018)
Amplitude and Phase Classification of ECG data.
Presented at: 2018 CRoNoS Summer Course and Satellite Workshop on Functional Data Analysis, Iași, Romania.
Preview |
Text
CRoNoS Workshop on FDA Iasi.pdf - Accepted Version Download (860kB) | Preview |
Abstract
A functional data analysis approach to heart defect detection using heart signals recorded by electrocardiograms (ECGs) is proposed. ECGs can be thought of as continuous functions having an amplitude and phase component. However, raw heart signals are usually noisy with artifacts such as baseline wander and there are also issues with arbitrary location and scale when comparing two or more ECGs. To remove these artifacts, we propose amplitude registration models and give closed form solutions for the estimated parameters. For the classification of subjects, we propose to fit mixture Gaussian and cubic spline parametric models (which contain both amplitude and phase components) to the ECG functions. For heart conditions characterised by amplitude changes such as high peaks or inverted curves, classification is done using the estimated amplitude parameters. However, when conditions are characterised by changes in time domain, classification is done using the estimated phase parameters. The predictive accuracy of our proposed approach using leave-one-out cross-validation is 91% for the amplitude classification of myocardial infarction and 96% for the phase classification of cardiomyopathy. Our results compare favourably with state-of-the-art approaches for the classification of ECGs. Additionally, the proposed approach is applicable to the classification of other periodic biosignals.
Loading...
Loading...Loading...Loading...Archive Staff Only
 |
View Item |
