Comparison of two semi‐quantitative amyloid PET analysis methods with and without MRI in Alzheimer’s Disease Spectrum Disorder

Abstract

AbstractBackgroundStandardized uptake value ratio (SUVR) is a standard method for amyloid PET analysis. 3D T1 MRI imaging is required for accurate spatial normalization in SUVR measurement to overcome regional cortical atrophy in neurodegenerative disorders like Alzheimer’s disease spectrum disorders. The SUVR method without 3D MRI has recently been developed for clinical application. We compared two SUVR methods with and without 3D MRI for the semi‐quantitative analysis of global and regional amyloid loading in Alzheimer’s disease spectrum disorders.Method75 subjects (normal cognition 8, subjective cognitive decline 20, mild cognitive impairment 23, Alzheimer’s disease dementia 24) were selected in the single dementia clinic for the analysis. For the SUVR measurement with 3D T1 MRI, in‐house software with SPM8 and MATLAB was used. For the SUVR measurement without 3D T1 MRI, AI‐based automated PET quantitative analysis software (called ‘BTXBrain‐Amyloid‐Research’) was developed by Brightonix Imaging Inc. With the visual PET reading as a gold standard, we compared regional and global SUVR between groups. Intra‐class correlation coefficient(ICC) was used to test the validity of the two methods.ResultGlobal SUVR was higher in amyloid PET positive SCD group (SUVR with MRI (1.205 V.S. 1.635), SUVR without MRI (1.210 V.S. 1.599)). The mean global SUVR with MRI and without MRI of 75 subjects were 1.400 (S.D. 0.264) and 1.390(S.D. 0.237). ICC of global SUVR between the two methods was 0.950, p = 0.0001. Regional SUVR ICC were 0.941 in the prefrontal cortex, 0.953 in the orbitofrontal cortex, 0.947 in the anterior cingulate, 0.844 in the posterior cingulate, 0.955 in the lateral parietal cortex, 0.962 in the precuneus, 0.959 in the lateral temporal cortex, and 0.94 in the occipital cortex. The area under the curve of ROC curve analysis of both two methods to diagnose amyloid PET positivity was 0.98 and 0.97.ConclusionAI‐based automated PET quantitative analysis software of Brightonix imaging without MRI was valid as an MRI‐based SUVR method in Alzheimer’s disease spectrum disorders with variable brain atrophy. Short processing time (30 ‐40 seconds) and reliable SUVR measurement without 3D MRI benefit the Alzheimer’s disease diagnosis in clinical practice.Grant from the Ministry of Health and Welfare, HI18C0530, Seoul R&BD Program(BT200151)