Characterizing Complexity of Electroencephalograms in Alzheimer's Disease at Multiple Temporal Scales

Entropy is a widely-used feature and a basic framework for electroencephalogram (EEG) analyses. Their corresponding multiscale methods were recently developed, adopting multiple time scales for entropy estimation. In this paper, towards Alzheimer's disease (AD) EEG recordings, we employed weighted-permutation entropy combining multiscale analysis, so that multiscale complexity information could be exposed across different scale factors. Our experiment and off-line analyses results confirmed that multiscale weighted-permutation entropy (MWPE) method could distinguish AD pattern and control pattern in terms of the evolution trend of MWPE profiles. The consistency of 16 channels were broken in AD brain. The entropy topographic distribution in AD patients distinctly differed from healthy subjects at large-scale.