Seizure prediction using polynomial SVM classification

Zisheng Zhang, Keshab K. Parhi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

14 Scopus citations

Abstract

This paper presents a novel patient-specific algorithm for prediction of seizures in epileptic patients with low hardware complexity and low power consumption. In the proposed approach, we first compute the spectrogram of the input fragmented EEG signals from a few electrodes. Each fragmented data clip is ten minutes in duration. Band powers, relative spectral powers and ratios of spectral powers are extracted as features. The features are then subjected to electrode selection and feature selection using classification and regression tree. The baseline experiment uses all features from selected electrodes and these features are then subjected to a radial basis function kernel support vector machine (RBF-SVM) classifier. The proposed method further selects a small number features from the selected electrodes and train a polynomial support vector machine (SVM) classifier with degree of 2 on these features. Prediction performances are compared between the baseline experiment and the proposed method. The algorithm is tested using intra-cranial EEG (iEEG) from the American Epilepsy Society Seizure Prediction Challenge database. The baseline experiment using a large number of features and RBF-SVM achieves a 100% sensitivity and an average AUC of 0.9985, while the proposed algorithm using only a small number of features and polynomial SVM with degree of 2 can achieve a sensitivity of 100.0%, an average area under curve (AUC) of 0.9795. For both experiments, only 10% of the available training data are used for training.

Original languageEnglish (US)
Title of host publication2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2015
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages5748-5751
Number of pages4
ISBN (Electronic)9781424492718
DOIs
StatePublished - Nov 4 2015
Event37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2015 - Milan, Italy
Duration: Aug 25 2015Aug 29 2015

Publication series

NameProceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS
Volume2015-November
ISSN (Print)1557-170X

Other

Other37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2015
CountryItaly
CityMilan
Period8/25/158/29/15

Fingerprint Dive into the research topics of 'Seizure prediction using polynomial SVM classification'. Together they form a unique fingerprint.

Cite this