The use of calibrated phantoms to assess the accuracy of PET-CT image quantification for ⁶⁴Cu.

Objectives: For applications in which the activity value is important, such as dosimetry for treatment planning, or when the comparability of data across multiple clinical sites is critical, such as in a clinical drug trial, absolute image quantification is desired. This requires both the scanner and activity calibrator to be calibrated on an absolute basis against National standards of activity. Copper-64 is an example of a radionuclide with potential applications for both diagnosis and therapy. Image-based dosimetry is essential for optimized treatment planning with these applications, and would therefore benefit from absolute quantification. The effect that the decay properties of a particular radionuclide have on the quantitative accuracy of PET, particularly for nuclides with complex decay schemes such as Cu-64, has not been systematically studied. Additionally, a metrological approach to assessing the uncertainties associated with extracting activity values from PET imaging data for one radionuclide when the system is calibrated for another is needed but has not yet been reported. In the first part of such a study, we have used liquid phantoms with activity concentrations traceable to National standards to calibrate the NIST PET-CT scanner for F-18 and evaluate the image quantification accuracy for a uniform phantom containing Cu-64.

Methods: The scanner was calibrated according to the manufacturer’s SUV calibration protocol using a gravimetrically-prepared and calibrated uniform phantom containing F-18. The calibration was monitored over the course of about 8 months using both calibrated Ge-68 solid phantoms and calibrated liquid uniform phantoms prepared with F-18. Calibrated Cu-64 phantoms of different configurations and having combined standard uncertainties on the contained activity of less than 1.5 % were gravimetrically prepared and scanned at different time points to assess how accurately the Cu-64 activity values could be extracted from the images. Reconstructions were performed using standard OSEM-based routines provided by the manufacturer.

Results: Biases in the recovery of the calibrated Cu-64 activity concentration of about -9 % were revealed. Longitudinal data from the calibrated F-18 and Ge-68 phantoms indicated similar biases, even after calibration, and suggested that this information could be used to renormalize the Cu-64 data. Applying this approach reduced the bias on the activity, albeit with a larger uncertainty.

Conclusion: With this method, which is only possible because of the application of the calibrated F-18 and Ge-68 phantoms, systematic biases in the extracted activity values for uniform phantoms of Cu-64 could be removed, giving absolute activities consistent with 100 % recovery. Research Support: Internally-supported.