Source localization for gastric electrical activity using simulated magnetogastrographic data.

Abstract

In this study, the use of magnetic dipole (MDP) approximation to localize the underlying source of magnetogastrographic (MGG) data was investigated. An anatomically realistic torso and a stomach model were used to simulate slow wave (SW) activities and magnetic fields (MFs). SW activity in the stomach was simulated using a grid-based finite element method. The SW activity at each time sample was represented by the dipoles generated for each element and MFs were computed from these dipoles including secondary sources in the torso. Gaussian noise was added to the MFs to represent experimental signal noise. MDP fitting was executed on the time samples of selected 2-second time frames, and goodness of fit (GOF) and the distance from the fitted MDP to the center of gravity (COG) of active dipoles were computed. Then, for each time frame, the spatial changes of COG and MDP positions in x-, y-, and z-directions and correlation scores were computed. Our results showed that MDP fitting was capable of identifying propagation patterns with mean correlation scores of 0.63 ± 0.30, 0.71 ± 0.19, and 0.81 ± 0.24 in x-, y-, and z-directions, respectively. The mean distance from COGs to the identified MDPs was 49±4 mm. The results were similar under the noise conditions as well. Our results suggest that source localization using MDP approximation can be useful to identify the propagation characteristics of SWs using MGG data.