Ponente
Descripción
The problem that models the propagation of electrical potentials from the cerebral cortex to the scalp, observable by electroencephalographic (EEG) signals, is known as the electroencephalographic direct problem. This model assumes that these potentials can be represented as a linear combination of a spatial function-defined by a symmetric three-dimensional Gaussian bell-and a temporal function. The spatial function provides information about the location and degree of dispersion of the cortical source through its parameters. On the other hand, the electroencephalographic inverse problem seeks to estimate the cortical sources that originate the potentials observed in the scalp. This can be approached by means of mathematical formulations or using artificial intelligence techniques. In this work, a synthetic database is generated from the direct model, in order to train a neural network whose resolution capacity is compared with that of traditional mathematical methods.
