Isovolumic Contraction Time (ICT) is the interval from mitral closing to aorta opening. Fetal ICT can be noninvasively measured from Doppler Ultrasound (DUS) signal. Automated identification of opening and closing of mitral and aortic valves from DUS signal was proposed in recent studies. Fetal electrocardiogram (fECG) has a crucial role as a reference in automated methods by identifying the onset of each cardiac cycle. However simultaneous recording of abdominal ECG and DUS and separation of fECG from the noisy mixture of ECG complicate this technique. In this study the automated identification of valve motion events without using fECG was investigated. The DUS signal was decomposed by Empirical Mode Decomposition (EMD) to high and low frequency components linked to valve and wall motion, respectively. The peaks of the latter were used for segmentation of the high frequency component as a substitute for fECG. The mitral and aortic valve motion was then automatically identified by hybrid Support Vector Machine (SVM)-Hidden Markov Model (HMM). Results show a significant positive linear correlation between average ICT obtained with and without using fECG (r=0.90, p<0.0001) with the mean absolute difference of 1.4 msec.
|ジャーナル||Computing in Cardiology|
|出版ステータス||Published - 2014|
|イベント||41st Computing in Cardiology Conference, CinC 2014 - Cambridge, United States|
継続期間: 2014 9月 7 → 2014 9月 10
ASJC Scopus subject areas
- コンピュータ サイエンス（全般）