| With the aging of the population, the incidence of Alzheimer’s disease (AD) is keeping increasing, researches on biomarkers of AD have drawn more and more attention. In recent years, researches on gene, pathology, and imaging of AD have made some progress. EEG, as a kind of neurophysiological curve, accompanies with whole person’s life. It records brain electrical activities of perception, cognition, thinking and execution instantaneously, and contains rich information. For research of cognitive function, EEG is one of the best tools. However, the analysis of EEG in clinic still relies on the naked eye recognition and empirical judgment of the doctor. EEG, as a tool for the diagnosis of AD. has not yet played its due role. Using new mathematical methods for the quantitative analysis of the EEG signal, extracting valuable objective quantitative information and parameters has important value for the diagnosis and assessment of AD. At the same time it may provide new ideas for the study of the electrophysiological mechanisms in AD.Objective:In this study, continuous wavelet transform (CWT) is introduced for multi-scale analysis of spontaneous EEG in mild, moderate and severe AD patients, comparing with normal controls. Wavelet power spectrum is used to characterize the EEG time-frequency multi-scale features in AD patients. Wavelet entropy, as a quantitative parameter, is introduced to measure the complexity of EEG. Conditional sampling and phase-average technique obtain the phase-average wavelength, synchronous phase difference and amplitude difference from leads at different brain area. Time-frequency characteristics, such as wavelet power spectrum, wavelet entropy, average wavelength of scale9, the lead phase difference and amplitude differences, were compared among light, moderate, severe AD patients and normal controls. The aim of this study is to find useful quantitative electrophysiological parameters of spontaneous EEG for the diagnosis and assessment of AD.Methods:1. Detailed history, nervous system and systemic examination was underwent for light, moderate and severe AD patients. Assessment of MMSE, CDT, HIS, CDR, ADL were made for all AD patients.temporal lobe atrophy(MTA) of MRI were classified by Visual assessment of MTA. Detailed history, nervous system and systemic examination were underwent, Assessment of MMSE, MoCA and ADL were evaluated for normal controls.2. Raw digital EEG signals were cllected at quiet, sober eye-close state for light, moderate, severe AD patients and normal controls. Sampling frequency was200Hz, Artifact-free digital EEG data was selected and storaged as20seconds segment, for quantitative analysis.3. Quantitative analysis of the EEG is divided into five parts(1) Multi-scale quantitative analysis was made for EEG data recored at different brain eara by CWT. Time-frequency characteristics of30scales were analyzed, the wavelet coefficient contour map of EEG.were drawn for AD patients and normal controls.(2) Sub-scale wavelet power was calculated according to wavelet coefficients for the EEG data of all the leads, maps of sub-scale power spectrum distribution with frequency were drawn, to observe the characteristics of wavelet power distribution of EEG in AD patients and normal controls.(3) Wavelet entropy, as a quantitative parameters to measure the complexity of the EEG was extracted according to the percentage of the sub-scale power distribution with frequency of EEG in AD patients and normal controls.(4) EEG signal of scale9(corresponding to the frequency center of10Hz) was tested by wavelet coefficient, and detected similar incidents using by conditional sampling was superimposed and averaged in the phase alignment. EEG phase average waveform of scale9was obtained.(5) The phase-average waveform of the different lead simultaneously acquisied by synchronous cross-correlation method, the differents of phase and amplitude of phase averaged waveforms between the leads at different brain area were obtained.4. SPSS13.0statistical package was used for statistical analysis. Measurement data were expressed as (X±S). Count data were analyzed by χ2test.measurement data≥3groups were compared using single factor analysis of variance, and homogeneity of variance, pairwise comparisons of LSD. Single factor correlation analysis between measurement data using Pearson correlation analysis. Significant test for a=0.05.Results:(1) Rich scale, stable rhythmic activities near the three frequency bands of10Hz,1Hz,0.1Hz, and closed association between neighber scales was the time-frequency characteristics of EEG of normal controls. However time-frequency feather of AD patients was characterized by single scale, instability rhythmic activities, obvious rhythmic activity near1Hz, and loss of the normal EEG of multi-scale interconnected time-frequency structural. The time-frequency characteristics of EEG in AD patients evolved with the worsen of the disease.(2) There were three low and wide power peaks near the frequecy of0.1Hz,1Hz,10Hz on wavelet power spectrum distribution of spontaneous EEG in normal controls. While the wavelet power spectrum distribution in patients with AD was characteristic for a narrow power peak near1Hz, with the aggravation of the disease evolved, the power peak near1Hz increased, and power peak near0.1Hz,10Hz decreased.(3) Wavelet entropy of spontaneous EEG recorded from all leads in Mild, moderate and severe AD patients were lower than normal controls (P<0.01). Wavelet entropy and the MMSE score was positively correlated (P<0.01). This shows that the more severe cognitive impairment, the lower wavelet entropy, the lower complexity of the EEG.(4) Wavelength of phase average waveform of scale9recorded from all leads is greater than the normal controls (P<0.01), Wavelength of phase average waveform of scale9and MMSE score was negtivly correlated (P<0.01). This shows that the more severe cognitive impairment, the longer the wavelength of phase average waveform of scale9, the slower EEG frequency within the range of this scale.(5) There exist significant phase and amplitude difference of phase average waveform at scale9for spontaneous EEG between occipital leads and forehead leads in normal controls with phase difference between occipital leads and forehead leads is about π/2phase. Amplitude of phase average waveform of occipital leads at scale9is higher than that of forehead lead. Phase difference and amplitude difference of phase average waveform at scale9between occipital leads and forehead leads in AD patients decrease when compared with normal controls while phase difference and amplitude difference in sever AD patients are almost negligible. The result is consistent with alpha rhythm diffused and forward transfer in AD patients on visual EEG.Conclusion:The new EEG multi-scale analysis methods including wavelet transform, wavelet power spectrum, wavelet entropy, conditional sampling and phase average, synchronous cross-correlation analysis technology is suitable for EEG analysis. It can privide new quantitative parameters for diagnosis and assesment of AD.Spontaneous EEG signals of AD patints and normal controls were quantitative analyzed by new methods. The results showed time-frequency characteristics, wavelet power spectrum distribution, wavelet entropy, wavelength of phase average waveform of scale9, Phase difference and amplitude difference of phase average waveform at scale9between occipital leads and forehead leads, as quantitative characteristics and parameters can be used for clinical diagnosis and assessment of AD. |
PDF Full Text Request