Negatif Olmayan Matris Ayrıştırma Yöntemlerinde Fetal Elektrokardiyogram İşaretin Ayrıştırılması

Abstract

Electrocardiogram (ECG) sign is the signals produced by the heart muscles during heart beats and taken with the help of electrodes placed on the surface of the body to learn the electrical activity of the heart. After the obtained signals are strengthened, they can be analyzed by digital signal processing methods. The signs obtained as a result of the analysis will have a decisive effect especially in the diagnosis and treatment of heart diseases or in terms of the health of the person. As a result of the measurement, the mixed form of at least two marks in the marks obtained from the human body will be obtained. In this study, non-negative matrix separation method from blind source decomposition algorithms was used in signal decomposition to obtain the ECG of the baby in the mother's womb. As a source sign, the signal decomposition process was performed in the data that was a mixture of maternal ECG, Fetal ECG (baby’s) and noise signals. Performance analysis and transaction costs were compared using the Multiplicative Update method and Hierarchical Least Squares method, which are non-negative matrix decomposition algorithms. Signal / noise ratio criterion was used to determine the appropriate method.