The segregation between maternal and fetal heart signals by a convolution deep learning based on image processing

Nowadays, the fetal heart monitoring is an increasingly important issue in medicine, because statistics show that ۱ out of ۱۲۵ neonates are born with congestive heart failure. This paper proposes a deep learning approach based on convolution neural network to separate maternal electrocardiogram signals from fetal without isolating maternal ECG signal. The method put forward in the present research showed how to monitor fetal heart rate signals and separate them from that of mother via the deep neural system. With regard to the marked role of separating fetal heart rate from that of mother in order to determine embryo's abnormalities and finding a treatment for them, it is necessary to take an accurate measurement of fetal heart rate and analyze its morphology. Despite some limitation, like mother's fast heart rate or other available noises, a deep neural system-centric method was tested successfully. By combining signals received from mother and fetal heart rate and without the necessity of extracting features or doing pre-process steps, it could successfully monitor fetal heart rate than other methods. It not only did not feel the need to taking steps of feature extraction, but also it showed high precision. Finally, by comparing this works with earlier ones, its efficiency was conformed.

Keywords:

Convolution deep learning , Deep learning , Foetal heart signal , Image processing