ComBat-Centiloid: A Calibration-Free Method for Quantifying Centiloid Values in Amyloid PET Imaging

Objective: Amyloid-β (Aβ) positron emission tomography (PET) imaging is essential for diagnosing and monitoring Alzheimer’s disease (AD). The Centiloid (CL) scale standardizes Aβ quantification across centers and tracers, but its limitations include calibration requirements and the inability to capture regional A| heterogeneity. Methods: The ComBat-Centiloid method harmonizes A| PET data without calibration by combining a 11C-Pittsburgh Compound B standard reference database with the ComBat algorithm and CL framework to generate harmonized CL (HCL) and harmonized regional CL (HRCL) values. Pearson correlation analysis was used to evaluate the relationship between HCL and CL values within the same tracer/protocol combinations. Paired t -tests were used to assess differences in HCL and CL values between two same-subject scans using different tracers taken within 1 year. Multicenter analyses were performed with combined datasets with different tracer/protocols to compare the consistency of HRCL, regional CL (RCL), and SUVR for differentiating patients with AD from cognitively normal (CN) individuals. Results: HCL values strongly correlated with CL across all tracer/protocol combinations and effectively eliminated inter-tracer biases, showing no significant differences in paired tests. In multicenter analyses, HCL values outperformed SUVR and RCL, demonstrating superior consistency for distinguishing patients with AD from CN individuals. Conclusion: The ComBat-Centiloid method eliminates calibration requirements and supports robust harmonized assessments in multicenter multitracer studies.