PET scanner calibration using a small Ga68/Ge68 source

2052 Objectives A small Ga68/Ge68 source for easy and accurate routine cross-calibration of clinical PET scanners and dose calibrators is presented. Partial volume effect (PVE) is minimized by joint estimation of calibration factor and image resolution. Methods A Ga68/Ge68 source (10cc, 11g, 0.58-1.02mCi during study period) indirectly traceable to NIST at the center of a modified ACR phantom top was imaged on 4 PET/CTs (Biograph6, Biograph64, mCT, Discovery STE). Source activity was measured in the respective clinical dose calibrators. The phantom was aligned on the scanner axis. Images were reconstructed using filtered backprojection; average of 5 consecutive central slices through the source was used for cross-calibration-factor (CCF) calculation. CCF was computed as C_true (source activity measured in dose calibrator / source volume) divided by I_true (source image intensity in the absence of PVE). Reconstructed images were modeled as the convolution of a scaled binary image [I_true inside a 10.8mm radius circle, 0 outside] and a point spread function (PSF), assumed to be Gaussian. I_true and the full-width-at-half-maximum (FWHM) of the PSF were estimated by minimizing the mean square difference between the convolution and the reconstructed image. Standard F18 cross calibrations were done for comparison, using a standard cylindrical phantom (D=20cm) and large regions of interest. Results Difference of CCFs calculated from our method and F18 phantom is less than 4% without systematic bias, showing that our method is accurate (typical repeated F18 CCF variation is within 4%). Estimated image FWHM (mm): Biograph6 7.5, Biograph64 8.2, mCT 8.7, Discovery STE 7.5. Conclusions Data from 4 clinical scanners in 3 institutions confirm that easy and accurate cross calibration of PET scanners and dose calibrators can be done successfully using a small Ga68/Ge68 source and joint estimation of calibration factor and image resolution.