Electrogastrographic Signals Analyzed by Independent Component Analysis

Electrical activity of the stomach pacemaker structure, the network of interstitial cells of Cajal, can be noninvasively recorded as an electrogastrogram (EGG). Power of the EGG signal is, however, much weaker in comparison with the other biological signals (e.g. ECG, respiratory signal). EGG signal is therefore usually contaminated by other biological signals and artifacts, which are difficult to separate from the EGG signal by conventional filtering methods. In the present study, independent component analysis (ICA), as a method for blind source separation, is used for artifact removal and biological signal separation. EGG signals were recorded using 4channel custom made recording device controlled by ADμC812 8-bit microcontroller and connected to a notebook computer. All signals were measured relatively to a single reference electrode. The present study shows that ICA application on a set of EGG signals approved potential of this method for artifact removal and signal separation. The algorithm is useful for separation of different types of signals (motion artifact, oscillating transients) from the original EGG recordings. We can conclude that ICA improves further analysis of separated signals and their biological interpretation. It could potentially improve also analysis of relations between signals originating from different biological structures.