Evaluation of normal brain database impact on 3D-SSP images of patients with dementia by using 18F-FDG PET and MRI.

2622 Objectives The aim of this study was to investigate the effects of properties in normal database (DB) of FDG-PET and MRI brain images, including differences in composition of numbers, ages and genders, on the results of three-dimensional stereotactic surface projection (3D-SSP) images of patients with dementia. Methods In this study, 963 healthy adults (470 female: 53.4±9.9 yrs.; 493 male: 54.0±10.2 yrs.) were included. All subjects were diagnosed as cognitively normal by 3 doctors including a neurologist. Normal DBs for brain FDG-PET and MRI were generated for various combinations of different numbers (10, 20, 40, 50, 100 and 200), age groups (every age, 5 and 10 years) and gender groups (male, female and mixed). Each DB was evaluated for differences in mean, standard deviation (SD) and coefficient of variation (CV) of several brain regions and tested for errors in z-scores of several brain regions on 3D-SSP images of patients with mild cognitive impairment (MCI) and frontotemporal dementia (FTD). Results Normal DVs consisted of more than 50 subjects provided stable mean, SD and CV. In the posterior cingulate cortex (PCC) of a MCI patient, errors of z-scores of DBs composed of 20 subjects distributed up to 2.0 compared with DB composed of 200 subjects. In the anterior cingulate cortex (ACC), age-dependent decline in brain glucose metabolism and atrophy by about 10% were occurred from 40s to 70s. In a patient with FTD, a maximum error of 1.5 in z-score was noted in the ACC if there was a difference of 5 years between a patient age and DB age. Conclusions The 3D-SSP images of patients with dementia changed visually and quantitatively depending on the properties of DB. For the accurate diagnosis, normal DB should be composed of more than 50 subjects matching for age and gender with patients to be analyzed.