Document Type

Journal Article


Science Domain International


Faculty of Health, Engineering and Science


School of Medical Sciences / Centre of Excellence for Alzheimer's Disease Research and Care




Villemagne, V. L., Doré, V., Yates, P., Brown, B., Mulligan, R., Bourgeat, P., ... & Williams, R. (2014). En Attendant Centiloid, 2(12). pp. 723-729. Available here


Aims: Test the robustness of a linear regression transformation of semiquantitative values from different Aβ tracers into a single continuous scale. Study Design: Retrospective analysis. Place and Duration of Study: PET imaging data acquired in Melbourne and Perth, Australia, between August 2006 and May 2014. Methodology: Aβ imaging in 633 participants was performed with four different radiotracers: flutemetamol (n=267), florbetapir (n=195), florbetaben (n=126) and NAV4694 (n=45). SUVR were generated with the methods recommended for each tracer, and classified as high (Aβ+) or low (Aβ-) based on their respective thresholds. Linear regression transformation based on reported head-to-head comparisons of each tracer with PiB was applied to each tracer result. Each tracer native classification was compared with the classification derived from the transformed data into PiB-like SUVR units (or BeCKeT: Before the Centiloid Kernel Transformation) using 1.50 as a cut-off. Results: Misclassification after transformation to PiB-like SUVR compared to native classification was extremely low with only 3/267 (1.1%) of flutemetamol, 1/195 (0.5%) of florbetapir, 1/45 (2.2%) of NAV4694, and 1/126 (0.8%) of florbetaben cases assigned into the wrong category. When misclassification occurred (Conclusion: While a definitive transformation into centesimal units is being established, application of linear regression transformations provide an interim, albeit robust, way of converting results from different Aβ imaging tracers into more familiar PiB-like SUVR units.



Creative Commons License

Creative Commons Attribution 3.0 License
This work is licensed under a Creative Commons Attribution 3.0 License.