Feature extraction and recognition of ictal EEG using EMD and SVM

Abstract Automatic seizure detection is significant for long-term monitoring of epilepsy, as well as for diagnostics and rehabilitation, and can decrease the duration of work required when inspecting the EEG signals. In this study we propose a novel method for feature extraction and pattern recognit...

Full description

Saved in:
Bibliographic Details
Published in:Computers in biology and medicine 2013-08, Vol.43 (7), p.807-816
Main Authors: Li, Shufang, Zhou, Weidong, Yuan, Qi, Geng, Shujuan, Cai, Dongmei
Format: Article
Language:English
Subjects:
Algorithms
Databases, Factual
Decomposition
EEG
Electroencephalography - methods
EMD
Epilepsy
Epilepsy - diagnosis
Epilepsy - physiopathology
Feature extraction and recognition
Humans
Internal Medicine
Neural networks
Other
Reproducibility of Results
Seizure detection
Seizures - physiopathology
Sensitivity and Specificity
Signal Processing, Computer-Assisted
Support Vector Machine
SVM
Wavelet transforms
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Automatic seizure detection is significant for long-term monitoring of epilepsy, as well as for diagnostics and rehabilitation, and can decrease the duration of work required when inspecting the EEG signals. In this study we propose a novel method for feature extraction and pattern recognition of ictal EEG, based upon empirical mode decomposition (EMD) and support vector machine (SVM). First the EEG signal is decomposed into Intrinsic Mode Functions (IMFs) using EMD, and then the coefficient of variation and fluctuation index of IMFs are extracted as features. SVM is then used as the classifier for recognition of ictal EEG. The experimental results show that this algorithm can achieve the sensitivity of 97.00% and specificity of 96.25% for interictal and ictal EEGs, and the sensitivity of 98.00% and specificity of 99.40% for normal and ictal EEGs on Bonn data sets. Besides, the experiment with interictal and ictal EEGs from Qilu Hospital dataset also yields a satisfactory sensitivity of 98.05% and specificity of 100%.
ISSN:0010-4825
1879-0534
DOI:10.1016/j.compbiomed.2013.04.002