Independent component analysis for the inverse problem in the multidipole model of magnetoencephalogram’s sources

This paper is devoted to a challenging task of brain functional mapping which is posed due to the current techniques of noninvasive human brain investigation. One of such techniques is magnetoencephalography (MEG) which is very potent in the scientific and practical contexts. Large data retrieved from the MEG procedure comprise information about brain processes. Magnetoencephalography data processing sets a highly ill-posed problem consisting in spatial reconstruction of MEG-signal sources with a given accuracy. At the present moment, there is no universal tool for accurate solution of such inverse problem. The same distribution of potentials on the surface of a human head may be caused by activity of different areas within cerebral cortex. Nevertheless, under certain assumptions, this task can be solved unambiguously. The assumptions are the following: signal sources are discrete, belong to distinct functional areas of the brain, and have superficial location. The MEG-signal obtained is assumed to be a superposition of multidipole signals. In this case, the solution of the inverse problem is a multidipole approximation. The algorithm proposed assumes two main steps. The first step includes application of independent component analysis to primary/basic MEG-signals and obtaining independent components, the second step consists of treating these independent components separately and employing an analytical formula to them as for monodipole model to get the isolated signal source location for each component.