Diagnostic utility and characteristics of CT-based attenuation correction in brain perfusion SPECT/CT in predicting the exacerbation of Alzheimer changes from mild cognitive impairment utilizing voxel-based statistical analysis in comparison with Chang's method.

OBJECTIVE: We examined the diagnostic value of brain perfusion single-photon emission computed tomography (SPECT) using voxel-based statistical analysis with CT-based attenuation correction (CT-AC) by comparing it to that with Chang's AC in mild cognitive impairment (MCI) patients and attempted to locate brain areas that are good indicators predicting the progression of MCI. METHODS: Twenty-six individuals matched for age, educational background and initial Mini-Mental State Examination (MMSE) score of more than 24 underwent SPECT with N-isopropyl-4-[123I]iodoamphetamine and were assigned to 2 groups: the stable MCI (S-MCI) group comprising 11 subjects who maintained their MMSE score (mean 27.0) during at least a 1-year follow-up period (mean 37.2 months) and the progressive MCI (P-MCI) group comprising 15 subjects whose MMSE scores decreased by 3 or more points (from 26.4 to 21.4, mean). The diagnostic values of the two AC methods for discriminating P-MCI from S-MCI were compared using voxel-based statistical analysis in the lobe (Level 2) and lobule/gyrus levels (Level 3). RESULTS: Receiver operating characteristic analysis revealed that the area under the curve (AUC) was higher with CT-AC than with Chang's AC in the left temporal and limbic lobes in Level 2. In Level 3, the AUC in the left middle temporal gyrus was higher with CT-AC (0.852) than with Chang's AC (0.827). There were differences between the gyri/lobules that showed higher AUCs with CT-AC and those that showed higher AUCs with Chang's AC. When the gyri with the 4 highest AUCs were combined, AUC (0.897) and accuracy (84.6%) were better with CT-AC than with Chang's AC (0.806 and 80.8%). Surprisingly, the AUCs in the posterior cingulate gyrus and precuneus, excluding the AUC in the right precuneus with Chang's AC (0.715), were no more than 0.70 and less useful. CONCLUSIONS: CT-AC may allow brain perfusion SPECT to reflect more exact neuropathic changes in MCI that would cause progression of early AD. CT-AC in conjunction with voxel-based statistical analysis could possess higher diagnostic accuracy for exacerbation of disease implying early Alzheimer changes in MCI patients, with decreases in cerebral perfusion in the left temporal and limbic lobes representing good indicators.