Methodological problems and clinical relevance of structural neuroimaging in dementia research.

Structural neuroimaging and dementia are conceptually different being only loosely correlated. Computed tomography or magnetic resonance imaging can never "prove" a clinical syndrome such as dementia, but yield clues as to its cause and the site and extent of pathological changes. Conversely, the type and degree of intellectual deterioration can hardly predict neuroradiological findings. The value of structural neuroimaging lies in detecting or excluding possible causes of dementia and quantifying linear or volumetric parameters of tissue and fluid volume. If based on a presumed or established etiology, however, specific neuropsychological and dementia syndromes may correspond to focal pathological changes seen in well-defined cerebral areas as recent investigations have shown with hippocampal atrophy in Alzheimer's dementia.