RESUMO
OBJECTIVES: We introduce a novel analysis method of spatio-temporal pattern analysis of EEG. Using it, we presents a results discriminating severe Alzheimer's disease patients from normal subjects. METHOD: To the groups of 10 Alzheimer's disease patients and 10 normal aged subjects, we applied the Karhunen-Loeve decomposition method and observed their principal patterns and time-varying dynamics. RESULTS: First, there was a first primary pattern of the eigenvector in Alzheimer's disease patients that showed opposite polarities at the left and the right hemispheric regions. Second, the Alzheimer's disease patients had significantly different average values of the eigenvector at the left parieto-temporal area than the normal controls did. Third, some Alzheimer's disease patients had one, coherent frequency component in the Fourier amplitude spectrum of the first pattern expansion coefficient. CONCLUSION: Our results suggest that the pathophysiologic site of Alzheimer's dementia may be left temporo-parietal area in brain and further studies need to evaluate whether the spatio-temoral pattern analysis could be used to diagnose Alzheimer's dementia patients.