The accuracy of the clinical diagnosis of dementia is severely limited. This project proposes to develop computer-analyzed electroencephalography (CEEG) as a clinical test that will aid in the accurate diagnosis of dementia. It consists of a five-stop research plan. First, ninety geriatric subjects will be examined, divided equally among three groups: those with Alzheimer-type dementia (DAT), multi-infarct dementia (MID), and non-demented normals. All subjects will undergo rigorous clinical evaluations, including mental status and neuropsychological testing, neurological examination, and a battery of relevant laboratory tests. Second, electroencephalograms (EEGs) will be performed on each of the study subjects in the normal resting state and after administration of low-dose secobarbital. Digital EEG data will be collected with an IBM Personal Computer (IBM-PC) attached to the electroencephalograph during the procedures. Third, spectra and coherence functions will be calculated for multiple EEG channels from each subject using the IBM-PC. Fourth, the subjects will be followed to autopsy, wehre neuropathologic diagnoses will be established. Finally, multi-group, stepwise discriminant analysis will be performed under a """"""""training/testing"""""""" paradigm. Using the spectra and coherence variables from half of the subjects in each category, sets of parameters will be selected which correlate most strongly with both clinical and autopsy diagnoses (training portion). The discriminant functions developed from this first analysis will then be used to categorize prospectively the other half of the subjects into both their clinical and autopsy diagnostic categories (testing portion). The results of this discriminant analysis will be analyzed to determine the sensitivity and specificity of CEEG as a diagnostic tool for dementia.
Showing the most recent 10 out of 44 publications
