An epoxy containing 68GeCl4 with an activity concentration of about 70 kBq·g-1 was prepared according to the procedure given in  and gravimetrically dispensed into 10 of the cylinders (two complete sets of 1, 2, 4, and 6 mL plus one each of the 1 mL and 2 mL), which were then solvent-sealed. An additional three sets of cylinders (total of 12) were gravimetrically filled with a calibrated  solution of 68GeCl4 having the same approximate activity concentration as the epoxy. These solution-filled cylinders were used to determine efficiency calibration factors for three high-purity germanium (HPGe) gamma-ray spectrometry systems that were later used to calibrate the individual epoxy sources.
To check for uniformity, the sources were scanned on a Philips Gemini-TF PET scanner in a standard Jaszczak phantom in air, in water alone ("cold" background), and with a "hot" 18F background. The images were analyzed using a MATLAB  based program developed in-house specifically for this purpose. The reconstructed image DICOM files were read into the program and the sum of counts in a 5 pixel x 5 pixel (2 cm x 2 cm) region centered on the middle of each cylinder was calculated for all 45 axial slices. A similar procedure was used to calculate the sum of counts in a background region away from the 68Ge sources.
For the sources with volumes greater than 1 mL, the combined standard uncertainty (k = 1) on the efficiency calibration factor was less than 0.6 %, while the increased uncertainty on the peak fitting routine used to analyze the HPGe spectra (partially due to lower counting statistics) drives the uncertainty on the smallest source to 0.74 %. Future work will focus on lowering the magnitude of this component by increasing the counting times and implementation of improved peak fitting techniques. The maximum combined standard uncertainty on the activity measurement for any epoxy source using these calibration factors was 0.89 %. The average massic activity was 68.7 kBq of 68Ge per gram of epoxy, with a standard deviation on the 10 sources of 1.5 %, which is in agreement with the theoretical value of 70 kBq·g-1 .
A plot of the PET image intensity in the 6 mL cylinder as a function of axial distance is given in figure 3. Within the cylinder, the standard deviation on the sum in each slice was 1.7 %, while the standard deviation on the background counting rates in an identically-sized region was 1.1 % and therefore comparable to that of the cylinder sources. One notable feature of this plot is the slightly enhanced intensity observed at the walls of the cylinders, which is due to scattering effects at the interface of the wall and the water. This effect is also observed at other interfaces within the phantom (e.g., where the cold rods are placed at the bottom of the phantom).
As a result of the experiments with the prototype sources, we have designed a new set of cylinders having more uniform walls on all sides that will hopefully reduce the edge scattering effects that are observed in Figure Z. We have also included a larger sized (23 mL nominal) cylinder into the set to provide a source that should be free of partial volume effects.
1. Zimmerman, B.E. and Cessna, J.T. "Development of a Traceable Calibration Methodology for Solid 68Ge/68Ga Sources Used as a Surrogate for 18F in Radionuclide Activity Calibrators," J. Nucl. Med., 51, 448-453 (2010).
2. Zimmerman, B.E., Cessna, J.T. and Fitzgerald, R. "Standardization of 68Ge/68Ga using three liquid scintillation counting based methods," J. Res. Nat. Inst. Stand. Technol., 113, 265-280(2008).
3. MathWorks, Inc., MATLAB 2009b (2009).