Analysis of simulated electrogastrograms using parameters associated with correlation for detection of phase shift between signals

This paper presents a parameter obtained from the difference in the correlation of simultaneously electrogastrogram signals (EGG) recorded along abdominal wall, as a possible guideline for the determination of abnormal phase shift associated with stomach pathologies. A database of 105 simulated EGG signals using a bipolar recording, with pacemaker frequency selected randomly in three possible cases (namely bradygastria, typical gastric frequency and tachygastria), seven different phase shifts and five levels of white noise were used in testing the suggested parameter. From the database were taken two signals, the auto-correlation of the first signal and the cross-correlation of this and the second signal were computed and plotted together; vectors from origin to peaks of each plot, average of differences in magnitude between peaks and standard deviation of all of them were obtained. These values were compared with previous associated ranges obtained from one half of the database, and a detection of 100 % cases of 8 randomly selected test signals from the remaining available data was achieved. This new parameter named AvD, along with the pacemaker frequency are considered as possible inputs in a neural network for automatic identification of normal phase shift in future research.