OSCAR presents a method for automated signal quality assessment of non-invasive foetal electrocardiogram recordings
Scientists from OSCAR’s Digital Health group have recently developed an algorithm to assess foetal electrocardiogram (fECG) recordings for prenatal diagnosis and monitoring. The approach is expected to improve the reliability of foetal monitoring and contribute to healthier births.
Foetal arrhythmia has a 1%-to-2% incidence among pregnancies, of which about 10% are severely fast or slow foetal arrhythmias, often accompanied by foetal heart failure. Foetal arrhythmia can further lead to secondary damage to important organs, and even premature birth or death of the fetus.
Most fetuses with arrhythmia have a good prognosis, with a 0.3% incidence of cardiac malformations. Continuous foetal bradycardia is associated with severe congenital heart structural defects. When an arrhythmia is observed in foetal heart rate, there is a 1% likelihood that it is linked to abnormalities in the foetal heart structure. Eight to twelve in a thousand infants in China are born with foetal congenital heart diseases, and more than 20% of them fail to receive proper treatment or even die in the early period of infancy, making congenital heart disease one of the main causes of death in new-borns.
Foetal ECG is a useful tool for monitoring foetal heart activities, providing a reliable basis for the clinical diagnosis of congenital heart conditions. Nevertheless, it is challenging to accurately identify and extract foetal ECG signals as they are weak and blended with maternal ECGs, maternal respiration, and muscle motions.
Abdominal foetal ECG monitoring
The algorithm developed by OSCAR researchers helps derive high-quality fECG signals desirable for the monitoring of foetal development in uterus, the detection of foetal hypoxia and other pathological conditions during pregnancy or delivery, as well as the screening of foetal congenital heart diseases.
High-quality fECG signals facilitated by the algorithm
The technology is relevant for hospital scenarios where figure recordings are taken as part of a routine prenatal monitoring screening for foetal defects, in particular heart defects or abnormalities. The assessment results will inform the technician whether the recording is of sufficient quality or if a rerecording is required. The invention is equally useful in home-monitoring fECG devices, assessing the quality of routinely collected data.
The research is led by OSCAR PI Prof. David Clifton and led to a Chinese patent application being filed in April.
Professor of Clinical Machine Learning
University of Oxford