One instrument to record the activity of brainwave in a specific time is called Electroencephalography (EEG). EEG signal can be used to analyze the epilepsy disease. Brainwave of seizure patient has a low frequency with a tighter pattern than brainwave of normal people. We use data from Temple University Hospital Seizure Corpus (TUSZ) that represents an accurate clinical condition characterization. Based on neurologist report, several types of seizure can be found in the dataset. In this research, we classify three types of seizure, Generalized Non-Specific Seizure (GNSZ), Focal Non-Specific Seizure (FNSZ) and Tonic-Clonic Seizure (TCSZ). We added a normal EEG signal, so we have four classes to be classified using Support Vector Machine (SVM). The training dataset consists from 120 data (20 GNSZ, 50 FNSZ, 25 TCSZ and 25 Normal), while the evaluation dataset is 90 datasets (20 GNSZ, 50 FNSZ, 5 TCSZ and 15 Normal). We observe the combination of three feature extraction method, Mel Frequency Cepstral Coefficients (MFCC), Hjorth Descriptor and Independent Component Analysis (ICA). The best result obtained by combining MFCC and Hjorth descriptor that can detect seizure type with 90.25%, 97.83%, and 91.4% of average sensitivity, average specificity, and accuracy respectively.

Electroencephalogram (EEG) is a bioelectric signal produced by brain activity. The abnormalities that occur in the brain, such as epilepsy, can be seen through a particular pattern on the EEG signal. A recurrent unprovoked seizure occurs in epilepsy patients as a result of excessive brain cell activity. EEG is a non-linear and non-stationary signal, so a visual interpretation is difficult to conduct. One method to measure EEG characteristics is the entropy that quantifies the signal complexity. Several studies have been conducted to classify epileptic EEG signal using entropy as the feature set. Previous studies has shown a promising result for epileptic EEG signal classification. However, to achieve effectiveness for the classification process, we propose a new method to reduce the number of features with a competitive accuracy. In this research, we propose a wavelet-based entropy method named multilevel wavelet packet entropy (MWPE) for automatic EEG signal analysis. MWPE is calculated from the wavelet packet entropy (WPE) which performed at some decomposition level. WPE was calculated from wavelet packet decomposition (WPD) which give more informations in every signal subbands compared to discrete wavelet transform (DWT). Using MWPE, we got informations about the distribution of subband energy in every level of signal decomposition. MWPE and support vector machine (SVM) are used as the feature extraction and classifier respectively. The result showed that the method is able to classify three classes of the EEG data set (normal, interictal, seizure). The best accuracy is 94.3% which achieved by using a 1-5 decomposition level with biorthogonal 2.8 wavelet, and cubic or quadratic SVM. MWPE provides high accuracy with relatively few features.

The electroencephalogram (EEG) examination provides information on the brain’s electricity, especially in cases of epilepsy. Since the characteristics of EEG signals are nonlinear and nonstationary, visual inspection becomes very difficult. To overcome this problem, digital EEG signal processing was developed. Automatic epileptic EEG recognition is an area of interest on which much research focuses. The complexity approach to EEG signal analysis is interesting to be used as feature extraction, referring to the nonlinear characteristics of the signal. This study proposed an automatic epileptic EEG classification method based on the multiscale Hjorth descriptor measurement. EEG signals consisting of normal, interictal, and seizure (ictal) were simulated. The signal is scaled into new signals using the coarse-grained procedure on a scale of 1–20. Then, the Hjorth parameter which consists of activity, mobility, and complexity is calculated on the new signal. This process produces a feature vector that is used in the classification stage. Support vector machine (SVM) is used to evaluate the proposed feature extraction method. Simulation results showed that the Hjorth parameter on a scale of 1–15 yields 99.5% accuracy. The proposed method is expected to be applied to digital EEG for seizure detection and prediction.

When an epileptic seizure occurs, the neuron's activity of the brain is dynamically changed, which affects the connectivity between brain regions. The connectivity of each brain region can be quantified by electroencephalography (EEG) coherence, which measures the statistical correlation between electrodes spatially separated on the scalp. Previous studies conducted a coherence analysis of all EEG electrodes covering all parts of the brain. However, in an epileptic condition, seizures occur in a specific region of the brain then spreading to other areas. Therefore, this study applies an energy-based channel selection process to determine the coherence analysis in the most active brain regions during the seizure. This paper presents a quantitative analysis of inter- and intrahemispheric coherence in epileptic EEG signals and the correlation with the channel activity to glean insights about brain area connectivity changes during epileptic seizures. The EEG signals are obtained from ten patients’ data from the CHB-MIT dataset. Pair-wise electrode spectral coherence is calculated in the full band and five sub-bands of EEG signals. The channel activity level is determined by calculating the energy of each channel in all patients. The EEG coherence observation in the preictal (Cohpre ) and ictal (Cohictal ) conditions showed a significant decrease of Cohictal in the most active channel, especially in the lower EEG sub-bands. This finding indicates that there is a strong correlation between the decrease of mean spectral coherence and channel activity. The decrease of coherence in epileptic conditions (Cohictal <Cohpre ) indicates low neuronal connectivity. There are some exceptions in some channel pairs, but a constant pattern is found in the high activity channel. This shows a strong correlation between the decrease of coherence and the channel activity. The finding in this study demonstrates that the neuronal connectivity of epileptic EEG signals is suitable to be analyzed in the more active brain regions.