Purpose: This study was designed to examine the utility of visual inspection of medial temporal lobe (MTL) metabolism in the diagnosis of mild cognitive impairment (MCI) and Alzheimer's disease (AD) using FDG-PET scans.
Methods: Seventy-five subjects [27 normal controls (NL), 26 MCI, and 22 AD] with FDG-PET and MRI scans were included in this study. We developed a four-point visual rating scale to evaluate the presence and severity of MTL hypometabolism on FDG-PET scans. The visual MTL ratings were compared with quantitative glucose metabolic rate (MR(glc)) data extracted using regions of interest (ROIs) from the MRI-coregistered PET scans of all subjects. A standard rating evaluation of neocortical hypometabolism was also completed. Logistic regressions were used to determine and compare the diagnostic accuracy of the MTL and cortical ratings.
Results: For both MTL and cortical ratings, high intra- and inter-rater reliabilities were found (p values <0.001). The MTL rating was highly correlated with and yielded a diagnostic accuracy equivalent to the ROI MR(glc) measures (p values <0.001). The combination of MTL and cortical ratings significantly improved the diagnostic accuracy over the cortical rating alone, with 100% of AD, 77% of MCI, and 85% of NL cases being correctly identified.
Conclusion: This study shows that the visual rating of MTL hypometabolism on PET is reliable, yields a diagnostic accuracy equal to the quantitative ROI measures, and is clinically useful and more sensitive than cortical ratings for patients with MCI. We suggest this method be further evaluated for its potential in the early diagnosis of AD.