Early detection of Alzheimer's disease using nonlinear analysis of EEG via Tsallis entropy

A preliminary study by Sneddon et al. (2005) using visual working memory tasks coupled with quantified EEG (qEEG) analysis distinguished mild dementia subjects from normal aging ones with a high degree of accuracy. The present study hypothesizes that a simpler task such as having a subject count backwards mentally by ones can be coupled with qEEG to yield a similar degree of accuracy for classifying early dementia. The study focuses on participants with mild cognitive impairment (MCI) and includes both a delayed visual match-to-sample (working memory) task and a counting backwards task (eyes closed) for comparison. The counting backwards protocol included 15 normal aging and 11 MCI participants, and the working memory task included 9 normal aging and 7 MCI individuals. The EEG data were quantified using Tsallis entropy, and the brain regions analyzed included the prefrontal cortex, occipital lobe, and the posterior parietal cortex. The counting backwards task had a sensitivity of 82%, a specificity of 73%, and an overall accuracy of 77% whereas the working memory task had a sensitivity of 100%, a specificity of 89%, and an overall accuracy of 94%. The results suggest that simple tasks such as having a subject count backwards may distinguish MCI (p<;0.05) sufficiently to use as a rough screening tool, but psychophysical tasks such as working memory tests appear a potentially much more useful approach for diagnosing either MCI or very early Alzheimer's disease.