Spatially oriented segmentation allows researchers to break down the continuous stream of the ongoing EEG into microstates with stable topography of the brain electrical landscapes. The resulting microstates were shown to be related to conscious mental experience as well as to psychiatric disorders typically associated with thought disorders. In the present study, the microstates of the resting EEG of patients presenting with mild or moderate probable dementia of the Alzheimer type (DAT) were investigated. A significant anteriorisation of the centers of gravity of the microstate fields, an increase of the microstates' optimal window size and a reduced duration of sustained microstates were found. These differences were statistically more robust than the typical changes in the frequency domain (diffuse slowing) and were significantly correlated with the cognitive decline. The adaptive spatial segmentation into microstates is discussed as a method to extract meaningful EEG parameters for the early diagnosis and staging of Alzheimer's disease.