article · Big Data Mining and Analytics
Extracting fetal electrocardiogram signals from a single-channel abdominal recording poses significant technical challenges due to interference from maternal cardiac activity. A novel method addresses this by combining convolutional neural networks with mathematical techniques, specifically singular value decomposition, independent component analysis, and nonnegative matrix factorisation. Because of the distinct differences in heart rate frequencies between a mother and a fetus, time-scale representations allow fetal electrical activity to be distinguished clearly by its energy profile. Disentangled signal components are fed into a convolutional neural network model, which optimises the recovery of the actual fetal electrocardiogram signal derived from the singular value decomposition and independent component analysis process. Results show that this integrated technique operates efficiently and offers the potential for deployment in real-time monitoring environments.
Monitoring a fetus's heart rate directly from a single non-invasive sensor on the mother's abdomen provides critical prenatal health information. Separating the faint fetal heartbeat from the mother's stronger signal usually requires complex multi-lead setups. Demonstrating that advanced mathematical processing and neural networks can isolate this signal from a single lead paves the way for simpler, accessible, real-time fetal monitoring.
This technique could be integrated into prenatal monitoring equipment and diagnostic software to track fetal heart activity using single-channel sensors. Potential users include medical device manufacturers and clinical healthcare teams. The abstract notes that the method demonstrates efficiency and could be deployed in real-time, indicating an applied research stage that requires further development and testing in hardware before reaching market readiness.
AI-generated from the published abstract. Always read the original work before citing.
This paper deals with detecting fetal electrocardiogram FECG signals from single-channel abdominal lead. It is based on the Convolutional Neural Network (CNN) combined with advanced mathematical methods, such as Independent Component Analysis (ICA), Singular Value Decomposition (SVD), and a dimension-reduction technique like Nonnegative Matrix Factorization (NMF). Due to the highly disproportionate frequency of the fetus's heart rate compared to the mother's, the time-scale representation clearly distinguishes the fetal electrical activity in terms of energy. Furthermore, we can disentangle the various components of fetal ECG, which serve as inputs to the CNN model to optimize the actual FECG signal, denoted by FECGr, which is recovered using the SVD-ICA process. The findings demonstrate the efficiency of this innovative approach, which may be deployed in real-time.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.26599/bdma.2022.9020035
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
