Sabrina Ammar, Omar Trigui, and Mohamed Senouci
Epilepsy, seizures detection, empirical mode decomposition, genetic algorithm, linear discriminant analysis
This paper presents a novel fully generic automated patient-specific seizures detection system. The aim is the detection of the seizure epochs with a high precision. For this end, the empirical mode decomposition is used to overcome the system limitations caused by the non-linear and non-stationary characteristics of the electroencephalography (EEG) signal. The genetic algorithm allows selecting the best parameters combination of each patient without the need of any prior information. For instance, it can estimate the relevant features for each subject from the list of 14 features extracted from the intrinsic mode functions. Thus, the proposed system is able to automatically self-adapt to increase its accuracy rate. The experimental results found using the benchmark CHB-MIT scalp long-term EEG database prove the effectiveness and the reliability of the proposed system with an average sensitivity of about 93.4% and an average specificity of about 99.9%.
Important Links:
Go Back
