EEG detection of early Alzheimer's disease using psychophysical tasks

In this study, we hypothesized that a quantitative EEG (qEEG) method for measuring EEG variability combined with specific psychophysical tasks could improve the classification accuracy of subjects with normal aging vs. mild cognitive impairment (MCI) or mild dementia due to Alzheimer's Disease and Related Disorders (ADRD). The cross-sectional sample consisted of 48 subjects (32 normal aging and 16 ADRD: n=3 mild dementia, n=13 MCI FAST stage 3). During EEG recording, subjects performed two visual, delayed recognition memory tasks as well as a task that tested their ability to perceive structure-from-motion (SFM). These EEG data were used to compute qEEG measures of the (normalized) variance of posterior cortical activity during the first 150 milliseconds (ms) after stimulus onset and the variance of anterior cortical activity during the second 150 ms epoch. The ratio, anterior/posterior cerebral qEEG value, was then computed for each subject, and the optimal cutoff value identified to discriminate normal from impaired subjects. An optimal qEEG cutoff value for the delayed recognition memory tasks correctly discriminated 30 of the 32 normal aging subjects (94% specificity) and 14 of 16 MCI-to-mild ADRD subjects (88% sensitivity). On the other hand, the application of this qEEG measure to EEG data recorded while subjects performed a SFM task did not distinguish between ADRD and normal aging any better than chance. In conclusion, this qEEG measure is specific to the psychophysical task being performed by the subject. When it was combined with delayed recognition memory tasks, it yielded results that are comparable to the accuracies reported by PET scan studies of normal aging vs. AD with mild cognitive impairment. These results warrant further evaluation.