Evaluation of cosine radial basis function neural networks in detection of artifacts in neonatal EEG

This paper presents the results of a study aimed at the development of an automated system for the detection of artifacts in neonatal EEG. The study relied on conventional feed-forward neural networks and cosine radial basis function (RBF) neural networks, which were trained to detect artifacts and distinguish them from seizure segments. This study focused on the extraction of features from EEG segments, the formation of training and testing sets, and the neural network models trained to function as classifiers. The feasibility of the proposed approach is validated by testing the trained networks on the testing set, which contain selected artifact and seizure segments, and on long EEG recordings.