ABSTRACT
Electroencephalogram signals or EEG signals are highly complex, nonlinear signals which clearly reveal the information regarding the brain working and crucial clues of various neuropathologies. Linear features are not sufficient for classifying disease case EEGs from that of normal healthy subjects. This chapter summarizes the emphasis of various chaotic and nonlinear features for EEG for the diagnosis of various neurological diseases.
EEG signal is decomposed using discrete wavelet transform (DWT) to generate EEG sub-bands. Chaotic analysis of signal calculates correlation dimension (CD) and Lyapunov exponent, which gives a measure of the complexity and chaoticity of the signal after reconstructing the phase space. The fractal dimension (FD) also gives a measure of the signal complexity and shows the significant difference between the disease group and the normal group. Entropy gives a measure of randomness or irregularity in the signal.
68All these nonlinear features have proven to be good biomarkers for the diagnosis of epilepsy, Alzheimer’s disease, encephalopathy, depression, etc. Thus, chaotic and nonlinear features have proven to be good biomarkers of EEG in the diagnosis of various neurological diseases.
