Related Experiment Videos
A diagnostic approach in Alzheimer's disease using three-dimensional stereotactic surface projections of
S Minoshima1, K A Frey, R A Koeppe
1Department of Internal Medicine, University of Michigan, Ann Arbor, USA.
Unlabelled:
To improve the diagnostic performance of PET as an aid in evaluating patients suspected of having Alzheimer's disease, we developed a fully automated method which generates comprehensive image presentations and objective diagnostic indices.
Methods:
Fluorine-18-fluorodeoxyglucose PET image sets were collected from 37 patients with probable Alzheimer's disease (including questionable and mild dementia), 22 normal subjects and 5 patients with cerebrovascular disease. Following stereotactic anatomic standardization, metabolic activity on an individual's PET image set was extracted to a set of predefined surface pixels (three-dimensional stereotactic surface projection, 3D-SSP), which was used in the subsequent analysis. A normal database was created by averaging extracted datasets of the normal subjects. Patients' datasets were compared individually with the normal database by calculating a Z-score on a pixel-by-pixel basis and were displayed in 3D-SSP views for visual inspections. Diagnostic indices were then generated based on averaged Z-scores for the association cortices.
Results:
Patterns and severities of metabolic reduction in patients with probable Alzheimer's disease were seen in the standard 3D-SSP views of extracted raw data and statistical Z-scores. When discriminating patients with probable Alzheimer's disease from normal subjects, diagnostic indices of the parietal association cortex and unilaterally averaged parietal-temporal-frontal cortex showed sensitivities of 95% and 97%, respectively, with a specificity of 100%. Neither index yielded false-positive results for cerebrovascular disease.
Conclusion:
3D-SSP enables quantitative data extraction and reliable localization of metabolic abnormalities by means of stereotactic coordinates. The proposed method is a promising approach for interpreting functional brain PET scans.