Author information

Department of Pediatrics, School of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan.
Department of Pediatrics, Kaohsiung Medical University Hospital, Kaohsiung Medical University, Kaohsiung, Taiwan.
Department of Medical Imaging and Radiological Sciences, College of Health Sciences, Kaohsiung Medical University, Kaohsiung, Taiwan.
Department of Computer and Communication, National Pingtung University, Pingtung, Taiwan.
Department of Pediatrics, Changhua Christian Hospital, Changhua, Taiwan.
Department of Information Engineering, I-Shou University, Kaohsiung, Taiwan.


The life quality of patients with refractory epilepsy is extremely affected by abrupt and unpredictable seizures. A reliable method for predicting seizures is important in the management of refractory epilepsy. A critical factor in seizure prediction involves the classification of the preictal and interictal stages. This study aimed to develop an efficient, automatic, quantitative, and individualized approach for preictal/interictal stage identification. Five epileptic children, who had experienced at least 2 episodes of seizures during a 24-hour video EEG recording, were included. Artifact-free preictal and interictal EEG epochs were acquired, respectively, and characterized with 216 global feature descriptors. The best subset of 5 discriminative descriptors was identified. The best subsets showed differences among the patients. Statistical analysis revealed most of the 5 descriptors in each subset were significantly different between the preictal and interictal stages for each patient. The proposed approach yielded weighted averages of 97.50% correctness, 96.92% sensitivity, 97.78% specificity, and 95.45% precision on classifying test epochs. Although the case number was limited, this study successfully integrated a new EEG analytical method to classify preictal and interictal EEG segments and might be used further in predicting the occurrence of seizures.


classification; epilepsy; feature selection; scalp EEG; support vector classifier

10.1177/1550059416649076                                                                                                                                             ncbi
Go to top