Cross-validation of Quantitative Analytical Software Using 18F-florbetapir PET Imaging

Abstract

18F-florbetapir is an amyloid β (Aβ) -targeted 18F-labeled positron emission tomography (PET) tracer for the clinical diagnosis of Alzheimer's disease. The standardized uptake value ratio (SUVR) serves as a tool with which to differentially diagnose. The present study aimed to cross-validate and compare SUVR derived from Amygo neuro and MIMneuro software. We injected 40 individuals with 18F-florbetapir and then acquired PET images from 50 to 60 minutes later. All images were separately normalized to the standard 18F-florbetapir PET template using Amygo neuro and MIMneuro. Volumes of interest (VOIs) were automatically placed on six target regions each in Amygo neuro and MIMneuro. The composite SUVR (cSUVR) and regional SUVR (rSUVR) were calculated from mean values measured in VOI. A cSUVR of>1.10 was defined as representing Aβ positivity. Correlation coefficients were calculated in the two types of software. A cSUVR>1.10 was determined by Amygo neuro and MIMneuro in 15 of the 40 individuals. The rSUVR in the posterior cingulate, parietal lobe, precuneus, and temporal lobe significantly differed between Amygo neuro and MIMneuro, whereas the cSUVR did not. The SUVR calculated by the two types of software closely correlated to each other (R=0.89-0.96, P<0.05). The cSUVR was not different between Amygo neuro and MIMneuro. We suggest that Amygo neuro is comparable to MIMneuro in quantitative analysis using SUVR for 18F-florbetapir imaging, thus facilitating the use of standardized quantitative approaches to amyloid PET imaging.