Validation System for Models of Intracranial EEG Signals

This paper presents a methodology for validation of signals obtained from a physiological model of intracranial EEG signals (iEEG) with epileptic activity. The goal is to obtain a system that can be implemented in a DSP platform, a methodology that can be easily adjusted to models of different biomedical signals. The validation system uses real iEEG signals of patients suffering of Mesial Temporal Lobe Epilepsy (MTLE) as reference. Once a model of iEEG was selected and implemented, simulated and real signals were classified into ictal or interictal by an Evolutionary Artificial Neural Network (EANN). The simulated and real signals were classified and compared using the proposed methodology obtaining a sensitivity of 0.8163 and a specificity of 0.9592 for the testing signals, and a sensitivity of 0.9385 and a specificity of 0.9 for the simulated signals.