| Summary: | 碩士 === 國立陽明大學 === 生物醫學影像暨放射科學系 === 101 === Electroencephalography (EEG) is the recording of electrical activity produced from brain cell. Using these signals, we can analyze the brain signal and discriminate the differences between normal people and patients with brain defect. However, the amplitude of brain signals is weak, and can be easily influenced by noise. Besides, EEG has the properties of irregularity, non-linearity and non-stationary. It is subjective to interpret EEG signals by reading the wave directly. In recent years, many scientists used differently mathematical methods to alleviate these problems and quantify EEG signals.
Because of the defect in brain, EEG signals of patients are different from that of normal control group. In this study, we collected the EEG signals from 151 subjects, including 50 normal control subjects, 38 patients with Alzheimer's disease (AD), 18 patients with vascular dementia (VD), 15 patients with Parkinson's disease (PD), 11 patients with multiple system atrophy (MSA), and 19 patients with spinocerebellar ataxia (SCA3). EEG signals of all subjects were recorded at the resting state. In addition to the raw EEG signals excluding eye movement, we separated EEG signals into five different frequency bands: γ(30~70Hz)、β(13~29Hz)、α(8~12Hz) 、θ(4~7Hz)、δ(0.5~3Hz) for subsequent analysis. We used three methods, namely, power spectral density, entropy, and fractal dimension (FD), to quantify EEG signals, and the estimated parameters for EEG complexity were further be used in T-test to detect the significant difference between a disease group and normal control group.
In comparison with the normal control group, results demonstrated that significantly larger FD values were detected in AD group, significantly smaller power spectral densities were revealed in PD group, significant differences in values of entropy were found in VD and SCA3 groups in some channels, and significantly smaller values of power spectral density, entropy and FD were found in MSA group.
|
|---|
