A New Marker of Alzheimer’s Disease for the Elderly Based on Sleep EEG
Date Issued
2015
Date
2015
Author(s)
Huang, Po-Hao
Abstract
The objective of this study is to develop a new marker of Alzheimer’s disease (AD) for the elderly using sleep EEG signals. An experiment was conducted on 18 subjects over 75 years old. Four of them were AD patients, and the remaining subjects were normal. The all-night sleep EEG signals were recorded from electrodes C3-A2, C4-A1, O1-A2 and O2-A1. The signals were firstly transformed into the δ, θ, α, σ and β band average magnitude time series using the moving window Fourier transform. The band average magnitude time series in sleeping status was extracted based on the report of sleeping stages. Then, the time series were normalized and the similarity between each pair of band average magnitude time series was calculated. The measure of similarity may be defined as the Euclidean distance between two time series. However, two perfectly similar time series may nonzero Euclidean distances if one of the time series undergoes a vertical shift. Hence, the time series with low average magnitude was shifted so that its average magnitude equaled that of the other time series. To make the similarity comparable, the Euclidean distances is further normalized by the lengths of the time series. The normalized Euclidean distance, denoted Sd, is studied in this research for its possible application in the diagnosis of Alzheimer’s disease. We discover that the Sd values of the series combinations δ and α, δ and β, θ and α, α and σ in C3 channel, θ and σ, α and σ, α and β in O1 channel, α and σ in C4 channel, and α and σ in O2 channel are conspicuously larger among AD patients than normals (p < 0.01). Moreover, the series combinations δ and δ between O1 and O2 channels is conspicuously smaller among AD patients than normals (p < 0.01). Most of the Sd values in the previous mentioned are highly correlated with MMSE scores. This indicates that Sd seems to reflect neuropsychology and cognitive performance, thus Sd may be used as a the marker for diagnosis of Alzheimer’s disease. Compared to previous research, the similarity index Sd is more direct and easily calculated. Moreover, only one channel measurement is sufficient for the purpose of diagnosis. With such advantages, the proposed method seems to provide great flexibility and potential in future applications.
Subjects
Alzheimer’s disease
Sleep EEG
Similarity
Euclidean Distance
Cognitive Performance
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-104-R02543055-1.pdf
Size
23.54 KB
Format
Adobe PDF
Checksum
(MD5):c7f547828525a4eb85b821d259400a63