Abstract
Localization of potential sources in the brain from the superficial potential distribution data, measured through surface electrodes, is one of the most promissing techniques to be used for clinical diagnosis of neurological diseases and even for the studies of the information processing mechanism of the brain. In order to get a reliable estimation for the dipole location and orientation, care has to be taken in many stages of the analysis.
In this paper we discuss and conclude (i) that electric pulse stimuli and click sounds are suitable for use in the dipole localization tests, because they can generate fairly confined potential distributions in the brain with rather short latencies; (ii) that the number of A/D converter resolution levels have little affect on the final accuracy of the data, when an averaging procedure is used for the data with fairly poor signal to noise ratio; (iii) that the amplitude measurement of waves with particular latencies requires suitable compensation of the base line instability, which could be realized via noise cancellation through linear prediction procedures; (iv) that the modeling accuracy of the 3-layered spherical model is practically acceptable for the dipole localization analyses, except for the data measured through electrodes attached close to the eyes, ears and the chin, especially when the orientation of the dipole source is pointing at those positions. These results are numerically obtained through the finite element method with an inhomogeneous model.
Finally, experimental data obtained by electrical stimulation of the median nerve are analyzed to get reasonable estimates for the dipole position along the sensory pathways in the brain.
