Analysis of Spectral Features of EEG signal in Brain Tumor Condition

Abstract

The scalp electroencephalography (EEG) signal is an important clinical tool for the diagnosis of several brain disorders. The objective of the presented work is to analyze the feasibility of the spectral features extracted from the scalp EEG signals in detecting brain tumors. A set of 16 candidate features from frequency domain is considered. The significance on the mean values of these features between 100 brain tumor patients and 102 normal subjects is statistically evaluated. Nine of the candidate features significantly discriminate the brain tumor case from the normal one. The results encourage the use of (quantitative) scalp EEG for the diagnosis of brain tumors

Keywords

• Brain tumor
• electroencephalography
• power spectrum
• bispectrum
• bicoherence
• spectral entropy