Abstract:
As a noninvasive neuroimaging method, the dipole source localization of brain electrical activity has a much higher temporal resolution when compared with the functional magnetic resonance (fMRI) or Positron Emmision Tomography (PET) Imaging. It gives a direct image of the electrical events occuring in the brain. In this study, a user friendly computational system is developed for routine analysis of EEG activity, to perform electrical Dipole Source localization. The forward problem which is an essential part of source localization is solved by both the analytical and numerical methods. For the inverse problem, the Multiple Signal Classification algorithm (MUSIC) is used. The three concentric spherical shell and realistic head models which lead to analytical and numerical forward solutions, respectively are performed for different dipole parameters for evaluation and comparison. The center of gravity (COG) approximation is used for the forward slution of the Boundary Element Method. The head model is obtained by the T1 weighted average head image issued by the Montreal Neurogical Institute. The graphic user interface is extremely used on epileptic data obtained from mesial temporal sclerotic patients. The results obtained are in agreement with the clinical diagnoses reached by MRI and other neurological tests.|Keywords:EEG, Dipole Source Localization, Inverse Problem, Forward problem, Boundary Element Method, MUSIC algorithm.
