Applications of multilayer feedforward neural networks in electrogastrography

Artificial neural networks have been widely used in biomedical signal processing. This paper introduces applications of multilayer feedforward neural networks (MFNN) in surface electrogastrogram (EGG) which is a noninvasive measurement of the electrical activity of the stomach. These applications focus on identification and classification of EGG signals, including identification of motion artifacts in the EGG recordings, identification of gastric contractions from surface EGG, and classification of normal and abnormal EGG. Related theories on MFNN and feature extraction are given before the presentation of the applications in EGG. Further studies may lead to clinical applications of the surface EGG.