Classification of Doppler Ultrasound signal quality for the application of fetal valve motion identification

Faezeh Marzbanrad, Yoshitaka Kimura, Miyuki Endo, Marimuthu Palaniswami, Ahsan H. Khandoker

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Citations (Scopus)

Abstract

One dimensional Doppler Ultrasound (DUS) is a commonly applied technique for fetal heart rate monitoring, but it can also be used to identify the timings of fetal cardiac valve motion. These timings are required to estimate the fetal cardiac intervals, which are fundamental and clinically significant markers of fetal development and well-being. Several methods have been proposed in previous studies to automatically identify the valve movement timings using 1-D DUS and fetal Electrocardiography (fECG) as a reference. However DUS is highly susceptible to noise and variable on a beat-to-beat basis. Therefore it is crucial to assess the signal quality to ensure its validity for a reliable estimation of the valve movement timings. An automated quality assessment can provide the operator with an online feedback on the quality of DUS during data collection. This paper investigates automated classification of the DUS signal quality using Naive Bayes (NB) classifier. The quality of 345 beats of DUS signals collected from 57 fetuses was assessed by four independent annotators and used for training and validation of the classifier. Using Fleiss kappa test, a fair agreement was found between the raters with overall κ = 0.3. The performance of the classification was tested by 10-fold cross validation. Results showed an average classification accuracy of 86% on training and 84% on test data.

Original languageEnglish
Title of host publicationComputing in Cardiology Conference 2015, CinC 2015
EditorsAlan Murray
PublisherIEEE Computer Society
Pages365-368
Number of pages4
ISBN (Electronic)9781509006854
DOIs
Publication statusPublished - 2015 Feb 16
Event42nd Computing in Cardiology Conference, CinC 2015 - Nice, France
Duration: 2015 Sep 62015 Sep 9

Publication series

NameComputing in Cardiology
Volume42
ISSN (Print)2325-8861
ISSN (Electronic)2325-887X

Other

Other42nd Computing in Cardiology Conference, CinC 2015
CountryFrance
CityNice
Period15/9/615/9/9

ASJC Scopus subject areas

  • Computer Science(all)
  • Cardiology and Cardiovascular Medicine

Fingerprint Dive into the research topics of 'Classification of Doppler Ultrasound signal quality for the application of fetal valve motion identification'. Together they form a unique fingerprint.

Cite this