The increasing incidence of epilepsy has led to the need for automatic systems that can provide accurate diagnoses in order to improve the life quality of people suffering from this neurological disorder. This paper proposes a method to automatically classify epilepsy types using EEG recordings from two databases. This approach uses the spectral power density of intrinsic mode functions (IMFs) that are obtained through the empirical mode decomposition (EMD) of EEG signals. The spectral power density of IMFs has been applied as features for the classification of focal and non-focal, as well as of focal and generalized EEG signals. The data are then classified using K-nearest Neighbor (KNN) and Naïve Bayes (NB) classifiers. The focal and non-focal data were classified with high accuracy, with KNN and NB classifiers achieving a maximum classification rate of 99.90% and 99.80%, respectively. Focal and generalized epilepsy data were classified with high rates of accuracy during wakefulness and sleep stages, with KNN achieving a maximum rate of 99.49% and NB achieving 99.20%. This method shows significant improvements in the classification of EEG signals in epilepsy compared to previous studies. It could potentially aid clinical decisions for epilepsy patients.

No abstract

No abstract

The paper proposes an approach based on higher order statistics and phase synchronization for detection and classification of relevant features in electroencephalographic (EEG) signals recorded during the subjects are performing motor tasks. The method was tested on two different datasets and the performance was evaluated using k nearest neighbor classifier. The results (classification rates higher than 90%) have shown that the method can be used for discriminating right and left motor imagery tasks as an offline analysis for EEG in a brain computer interface system.

Epilepsy is a neurological disorder characterized by recurrent seizures and has a high incidence rate. The aim of this research is to classify EEG signals as either focal and non-focal in order to identify the epileptogenic area of the brain, which can be surgically treated to manage epilepsy. In this paper, was proposed a classification method based on higher order spectra (HOS) parameters and four different classifiers: linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), k-Nearest neighbors (KNN), and Mahalanobis distance (MD). The method was evaluated using a public dataset that consists in EEG recordings from epileptic patients. The classifiers performances were evaluated and it was shown that KNN classifier achieves a maximum classification rate of 99.55%, sensitivity of 100%, and specificity of 99.09%. The data classification was performed with maximum values of 0.96 for F1-score, and 0.91 for both Kappa and Matthews Coefficient. The results demonstrate the efficiency of the proposed method to identify the type of EEG signals.