A novel classification approach to detect the presence of foetal cardiac anomaly from foetal electrocardiogram. (1st June 2020)

• Record Type: Journal Article
• Authors: Anisha, M.; Kumar, S.S.; Benisha, M.
• Is Part Of: International journal of biomedical engineering and technology. Volume 33:Number 1(2020)
• Page Start: 28
• Page End: 53
• Publication Date: 2020-06-01
• Subjects: foetal electrocardiogram -- FECG -- foetal heart rate -- FHR -- support vector machine -- SVM -- foetal cardiac anomaly -- foetal cardiac activity; Biomedical engineering -- Periodicals; 610.28
• Languages: English
• ISSNs: 1752-6418

Abstract

Foetal cardiac anomaly interpretation from foetal electrocardiogram (FECG) is a challenging effort. Foetal cardiac activity can be assessed by scrutinising FECG because clinically crucial features are hidden in the amplitudes and waveform time duration of FECG, and foetal heart rate (FHR). These features are vital in foetal cardiac anomaly interpretation. Hence, here an attempt is made to detect the presence of foetal cardiac anomaly using support vector machine (SVM) classifier with polynomial kernel based on the patterns extricated from FHR, frequency domain of FECG signals, foetal cardiac time intervals and FECG morphology. Performance evaluation is done on real FECG signals with different combination of features set and the obtained results are compared. SVM showed good performance with 92% of classification accuracy when all the features are fed to the classifiers. Results evince that the proposed approach has immense prospective and guarantee in early foetal cardiac anomaly detection from FECG.