Chaos in the Brain: Novel Methodologies for Epilepsy Diagnosis and Seizure Detection

Abstract

This article is a summary of the authors’ research on the application of the chaos theory along with wavelets and neural networks for automated EEG-based epilepsy diagnosis and seizure detection. The authors developed a novel wavelet-chaos-neural network methodology for classification of EEGs into healthy, ictal, and interictal EEGs based on the premise that EEG subbands may yield more accurate information compared with EEGs about constituent neuronal activities underlying the EEG. Certain changes in the EEGs that are not evident in the original full-spectrum EEG may be amplified when each sub-band is analyzed separately. Wavelet analysis is used to decompose the EEG into delta, theta, alpha, beta, and gamma sub-bands. Three parameters are employed for EEG representation: standard deviation (quantifying the signal variance), correlation dimension, and largest Lyapunov exponent (quantifying the non-linear chaotic dynamics of the signal). It was discovered that a particular mixed-band feature space consisting of nine parameters and a Levenberg-Marquardt Backpropagation Neural Network (LMBPNN) resulted in a high classification accuracy of 96.7%. In addition, a novel principal component analysis-enhanced (PCA) cosine radial basis function neural network classifier has been developed that yields comparable classification accuracy but is more robust with a lower standard deviation to changes in training data compared to LMBPNN. These statistics are especially remarkable because even the most highly trained neurologists do not appear to be able to detect interictal EEGs more than 80% of the times.