Standardisation of Quantitative PET Measures in Alzheimer’s Disease Using the Centiloid Scale

Abstract

Amyloid-beta (Aβ) positron emission tomography (PET) is an established neuroimaging technique for the diagnosis of Alzheimer’s disease (AD). Quantitative PET measures can be heavily influenced by particular methodological choices. The Centiloid (CL) scale has been proposed as a metric to standardise these measurements. Nonetheless, the calibration to this scale requires the acquisition of PET scans using the tracer [11C]-Pittsburgh Compound-B (PiB), which is not widely available. The aim of this work is to process and quantify a set of 96 Aβ PET images and test whether a conversion to the CL scale can be done without using PiB acquisitions. Using preprocessed [18F]-Florbetapir (FBP) PET images from the ADNI dataset, we performed quantification using a software tool from Siemens. Results were converted to CL using an externally calibrated formula. These values were compared with standard CL values, also available in the dataset. A high degree of linear correlation was found between the estimated CL values and the standard ones. Nonetheless, the linear regression showed a considerable amount of associated uncertainty. As a conclusion, both methods are comparable in measuring brain pathology. However, further work should explore the origin of uncertainty in the conversion, which should be corrected