A methodology based on constant-bandwidth TQWT filter banks and an LS-SVM classifier achieved a maximum classification accuracy of 90.01% for detecting focal epilepsy EEG signals.
A new methodology using constant-bandwidth TQWT filter-banks and LS-SVM classifier achieved 90.01% accuracy in identifying focal epilepsy EEG signals.
Epilepsy is a neurological disease that identified by reoccurrence of seizures. The economic and commonly used method for the diagnosis of epilepsy is possible with the regular monitoring of electroencephalogram (EEG) signals. These EEG signals are complex in nature and the manual identification of these EEG signals is very much tedious task for the doctors. In this paper, a new methodology based on constant-bandwidth tunable-Q wavelet transform (TQWT) filter banks has been designed for the identification of medically not curable focal epilepsy EEG signals. In this proposed methodology, the non-focal and focal EEG signals are considered to extract sub-band signals by involving constant-bandwidth TQWT filter-banks. The mixture correntropy based features are obtained from sub-band signals of the EEG signals. The least squares support vector machine (LS-SVM) classifier along with radial basis function (RBF) kernel is used for the classification of these extracted features. The feature ranking methods are also used to reduce the features space. The achieved maximum classification accuracy in this proposed methodology is 90.01% using Bern-Barcelona EEG database.
Gupta et al. (Sat,) conducted a other in Focal epilepsy. Constant-bandwidth tunable-Q wavelet transform (TQWT) filter banks with LS-SVM classifier was evaluated on Classification accuracy for focal vs non-focal EEG signals. A methodology based on constant-bandwidth TQWT filter banks and an LS-SVM classifier achieved a maximum classification accuracy of 90.01% for detecting focal epilepsy EEG signals.