We present a calibration phantom for multi-center and longitudinal PET/CT studies of assessing response to therapy. There is recognized potential for PET/CT imaging to assess response to therapy. However, there is significant variability in estimating tracer uptake due to differences in scanner design, reconstruction algorithms, partial volume effects, and method of reporting (e.g. max vs. mean SUV). At recent workshops there have been proposals for a standardized phantom relevant to the imaging task. In June 2006 the Society of Nuclear Medicine (SNM) formed a standards validation task force that has designed and constructed such a phantom.
The phantom is based on the NEMA NU-2 IQ phantom, with six hot spheres (internal diameters of 10, 13, 17, 22, 28, and 37 mm) as target lesions. The phantom uses Ge-68 to remove filling variability and allow for estimating standard errors and reproducibility. Absolute quantitation is obtained by using an NIST traceable source for the Ge-68 calibration phantom. The effect of lesion size is measured by the six spheres. The overall activity level is based on a 10-15 mCi injection in a 70 Kg patient, with sphere:background ratio set to 4:1. Phantom scans are reconstructed with a full range of reconstruction methods and parameters for each scanner, and the resulting images analyzed in terms of how mean and max absolute activity and SUV vary with sphere diameter.
The lesion phantom allows for the determination of measured versus true activity (and SUV) as a function of sphere diameter, reconstruction parameters, and offers insights into cross platform variability.
For PET to realize its potential as a biomarker, it is critical that issues in quantitatively assessing response to therapy from different sites be addressed. The calibration phantom will allow for direct comparison of quantitative results from sites using different scanners and/or different processing methods. Improving the accuracy of comparative results from multiple sites will improve patient treatment and safety, and lead to faster, more accurate, and less expensive therapy trials using imaging biomarkers.