EEG Source Localization Using a Genetic Algorithm-Based Artificial Neural Network
Rukiye Karakis, Irem Capraz, Erhan Bilir and Inan Guler
Pages 188-194 (7)
Electroencephalogram source localization (ESL) attempts to detect the sources of the brain activities by electroencephalography
(EEG). ESL has been an important research area for both clinical and basic brain research such as
neurology, psychiatry, and psychopharmacology since the 1950s. Current dipoles are used to model the electrical activities
in the ESL. These dipoles can be determined by the artificial neural network (ANN) model instead of numerical
methods. A genetic algorithm (GA) can be used to reinforce and optimize the architecture of the ANN model and increase
the localization accuracy. The purpose of this study is to detect the sources of brain activities using a GA-based ANN
model by reviewing recent patents and literature. EEG data was simulated and three concentric spheres (scalp, skull,
brain) were used for modeling the head as a volume conductor. The preliminary results indicated that the localization accuracy
was 0.01 mm. According to preliminary results, the proposed GA-based ANN model can be used to obtain the
sources of equivalent current dipoles.
Artificial neural network, dipole source, EEG source imaging, genetic algorithm.
Department of Computer Engineering, Faculty of Technology, Gazi University, 06500 Teknikokullar, Ankara, Turkey.