Designing a local seizure detection system using brain mapping, image processing techniques, and artificial neural networks

The general method for paraclinic diagnosis of epilepsy is electroencephalography that is performed by visual analysis by experienced neurologist. However, due to false detection and impossibility of evaluating electrodes and brain areas coherence, it is not uniquely used for seizure detection. In recent years, Quantitative Electroencephalogram (QEEG) has become a strong instrument for detection of brain disorders. Hence, studies in the field of EEG performance improvement and brain mapping images analysis corresponding to new methods that contain 2-D and 3-D output images and automatic epilepsy diagnosis are necessary. In this cross-sectional study, through extracting epilepsy feature by computing the energy of each EEG channel, brain map pattern of each patient was plotted using cubic interpolation and generalized and partial patterns and potential center of epilepsy were diagnosed by LVQ artificial neural network using image processing combination methods. In the proposed algorithm, 11 epilepsy brain mapping patterns, including 1 generalized and 10 partial seizure patterns, were automatically diagnosed. Since seizure detection in the EEG signals is a complex procedure and the number of expert neurologists is small, this schema can be used for epilepsy diagnosis as an intelligent diagnosis method so that generalization of this method can help detect various brain disorders.