Introduction: We examined the clinical impact of commercially available quantitation software using 3-dimensional stereotactic surface projection (3D-SSP) on the diagnostic accuracy of 18F fluorodeoxyglucose positron emission tomography (18F FDG PET) in mild cognitive impairment (MCI) and Alzheimer disease (AD).
Methods: Enrollees underwent clinical evaluation to determine cognitive status and subsequent 18F FDG PET neuroimaging. Four blinded readers (2 novices and 2 experts) rated the images for degree of abnormality and interpretive confidence without and with 3D-SSP. Diagnostic accuracy was determined with area under the curve (area under the curve) of a receiver operating characteristic (receiver operating characteristic) curve analysis and change in confidence with model-based means (LSMeans).
Results: Twenty-three normal controls and 31 patients with cognitive impairment (18 MCI and 13 AD) were enrolled (28 female and 26 male; mean age 74 years). During follow-up (mean 3.6 years), all normal participants remained normal, 12 of 18 participants with MCI progressed to dementia, and all participants with baseline dementia progressed. The area under the curve with 3D-SSP (0.88; 95% CI: 0.76-0.95) was significantly higher than without it (0.72; 95% CI: 0.55-0.83). The specificity increased from 26% to 63% for novices and from 56% to 87% for experts with addition of 3D-SSP, whereas the sensitivity was essentially unchanged at 86% and 86% for the beginners and 81% and 79% for the experts. The interpretive confidence increased significantly from 3.3 to 4.0 (maximum value = 5, P = 0.048).
Conclusion: The use of commercially available 3D-SSP quantitation improved diagnostic accuracy for evaluation of MCI and AD with 18F FDG PET.