Separating fetal doppler signals in pregnancy using independent component analysis : application to the extraction of fetal heart rate and global movement

Abstract

Fetal heart rate (FHR) monitoring is an integral part of fetal monitoring before and during labour. It contains important indications about the health conditions of the fetus, and could become a diagnostic tool of particular importance. Recording fetal movements requires an effective transducer with wide beam. This paper demonstrates the potential of relevant criteria selection and independent component analysis (ICA) approaches to classify the different sources. To separate the fetal heart rate, fetal breathing movement and global movement from other sources measured on the maternal body surface, a method employing ICA technique based on Maximum Likelihood (icaML) has been developed in this study. For this purpose we used Bayesian Information Criterion (BIC) to estimate the number of relevant sources in a mixture of sources. The ICA and BIC algorithm proved to be robust enough to be implemented on the ACTIFOETUS system and useful in discriminating between the numerous sources present in the Doppler signal mixture. Keywords-component; Ultrasound; Doppler; Fetal monitoring; fetal heart rate; Independent Component Analysis (ICA), BIC criteria. I. INTRODUCTION Fetal Heart Rate (FHR) monitoring captures the beating activity of the fetal heart valves opening and closing using an acoustic sensor placed on the mother's abdomen. It detects acute foetal distress. The use of fetal biophysical profiles in high risk pregnancies (gravidic hypertension, in utero infection,...) helps to distinguish healthy foetuses from those with chronic suffering. Parameters usually analysed are fetal movement rate, Fetal Breathing Movement (FBM) rate and amniotic fluid volume. However, the two first parameters are obtained by prolonged echographic observation which is difficult to propose for the routine survey of pregnancies. The motivation for monitoring the fetus through pregnancy is to recognize pathological conditions, typically asphyxia, with sufficient warning to enable clinician intervention before the occurrence of irreversible changes. Ultrasound Doppler is the most widely used method for FHR monitoring. This technique consists of directing a wide ultrasonic beam at the fetal heart. Many systems and methods have been proposed to monitor fetal heart rate and fetal activity. One of the earliest attempts was made by Hammacher et al. (1968). The purpose of the Hammacher's method was to determine heart beat frequency from the heart sounds. Hammacher also mentioned the detection of the heartbeat using the R-wave of the ECG (3). We presented last year (6) a new pulsed Doppler system based on 12 independent transducers placed against the maternal abdomen by means of a supple probe holder. Each transducer detects Doppler signals at 5 different depths between 2 and cm 15 , and all the output signals are then stored on the hard disc of a personal computer (PC). In this paper, we investigate Independent component analysis (ICA) to reduce the number signals to process. Different order estimation techniques will be compared. And the first original results concerning FHR and FBM will be showed and discussed.