A Recurrence Plot Based Method for the Detection of End of T-wave in Abnormal Non-invasive Fetal Electrocardiogram Signals

Namareq Widatalla1, Ahsan Khandoker2, Yoshiyuki Kasahara1, Yoshitaka Kimura1
1Tohoku University, 2Khalifa University


Aims: This research aimed at developing an automatic algorithm to iden-tify ends of T-waves within abnormal non-invasive fetal electrocardiogram (nfECG) signals. Methods: Simultaneous Doppler Ultrasound (DUS) and nfECG record-ings were collected in Tohoku University Hospital from 9 pregnant women at gestational age of 24-41 weeks. Data were collected from healthy and unhealthy fetuses who were diagnosed with at least one cardiac abnormality that included heart failure, heart anomaly, tachycardia and umbilical artery dysfunction. nfECG signals were filtered and segmented into intervals based on the location of the R peaks and then recurrence plots of these intervals were obtained to determine the ends of T-waves. In order to validate the results of the detection, QT and QTc of the nfECG signals were calculated based on the detected points and compared with the Q-Aortic closing (Q-Ac) and Q-Acc of the doppler signals. In addition, two correlation tests between QT and Q-Ac and between QTc and Q-Acc were performed in Matlab for the abnormal cases for further validation of the results. Results: T-end points were detected within 321 beats in abnormal nfECG signals and 126 beats in normal nfECG signals and the results of QT and QTc were as follows: doppler normal QT(s): 0.238±0.027 QTc(s): 0.37± 0.15, doppler abnormal: QT(s) 0.25±0.043 QTc(s): 0.40±0.060. nfECG normal QT(s): 0.237±0.030 QTc(s): 0.37±0.16, nfECG abnormal: QT(s) 0.23± 0.069 QTc(s): 0.37±0.11. The correlation between QT and Q-Ac was 0.66 and was 0.63 between QTc and Q-Acc; also, the value of P was 0 in both cases. Conclusion: Based on the results, recurrence plots proved to be effective in identifying location of end of T-waves within abnormal nfECG signals. However, more challenging nfECG signals need to be analyzed for further validation of the method.