1. PET Scanner Calibration Using a Small Ga68/Ge68 Source
Bing Bai1, Magnus Dahlbom2, James Bading3, Peter S. Conti1 and Peter D. Esser4
1 Molecular Imaging Center, Department of Radiology, University of Southern California, Los Angeles, CA, USA
2 Department of Molecular and Medical Pharmacology, University of California, Los Angeles, CA, USA
3 Department of Cancer Immunotherapy and Tumor Immunology, City of Hope, Duarte, CA, U.S.A
4 Department of Radiology, Columbia University, New York, NY, USA
Introduction
Ga68/Ge68 source with simultaneous estimation of activity and resolution from image
Same advantages as method 2 but less affected by partial volume.
Reconstructed image is modeled as the convolution of a scaled binary image and a point spread function (PSF), assumed to be Gaussian.
5 consecutive central slices used for estimation.
Simultaneously estimate source image intensity and resolution (FWHM of Gaussian) by minimizing the mean square difference (MSD) between the convolution and reconstructed image.
PET Scanner Cross Calibration Factor (CCF)
𝐶𝐶𝐹= 𝐶 𝑡𝑟𝑢𝑒 𝐼 𝑡𝑟𝑢𝑒
𝐶 𝑡𝑟𝑢𝑒 : source activity measured in dose calibrator / source volume.
𝐼 𝑡𝑟𝑢𝑒 : image intensity in the absence of partial volume effect.
We have designed a small Ga68/Ge68 source for easy and accurate routine cross-calibration of clinical PET scanners and dose calibrators. Partial volume effect (PVE) is minimized by estimation of calibration factor and image resolution simultaneously. We tested the quantitative calibration of four clinical PET/CT scanners from two manufacturers in three institutions using this source. Results show that both quantitative calibration factor and image resolution can be estimated accurately using this method.
Method
Profile. Green: reconstructed image; Yellow: convolution
Relative MSD as a function of image resolution
Reconstructed image
Convolution of binary image with PSF
A Ga68/Ge68 Source for Dose Calibrator and PET Scanner Cross Calibration in PET Clinics
PET Scanner Cross Calibration
Large F18 cylindrical phantom
Standard cross calibration method used in many clinics.
Requires F18 activity and extra work (measure activity, fill phantom, measure region of interests).
Calibration results depends on dose calibrator accuracy.
Not affected by partial volume effect.
Ga68/Ge68 source with manually drawn ROI
Traceable standard sources for F18 & Ge68.
Cross calibration of ion chambers with PET scanners with the same source.
PET scanner calibration uses widely available phantom (with modified lid). Easy to use and transport, suitable for standardizing SUV measurement across multiple scanners and sites.
Affected by partial volume effect.
Indirectly traceable source: simulated F18 from (Ga68/Ge68)
10 cc Vial (RadQual BM06v)
Dose Calibrators
PET Scanners
Biograph 6, 64, mCT and Discovery STE
Ga68/Ge68 vial in phantom lid, showing ROI used for analysis
Modified ACR PET phantom lid with 2 accessible locations
Results
Scanner Model
CCF(Ge68 Est) / CCF(F18)
Estimated Resolution (FWHM)
Biograph 6
1.000
7.54 mm
Biograph mCT
0.996
8.71 mm
Biograph 64
1.033
6.36 mm
Discovery STE
0.980
6.59 mm
Dose Calibrator Measurements Using the Ga68/Ge68 Source
4 dose calibrators, 5 measurements on each.
Phantom Scan
Standard F18 20cm cylinder cross calibration
5-minute scan, FBP reconstruction, ROI D=10cm, L=10cm
Ga68/Ge68 10cc vial cross calibration
5-minute scan, FBP reconstruction
ROI D=5mm, L=10mm in the center of vial
Dose Calibrator
Max Measured Activity
Min Measured Activity
Average Measured Activity
Calibrated Activity
% Deviation 1
985.0 mCi
981.0 mCi
982.8 mCi
992.3 mCi
-0.96 2
983.5 mCi
983.7 mCi
983.4 mCi
-0.90 3
983.1 mCi
982.7 mCi
983.2 mCi
989.7 mCi
-0.66 4
855.0 mCi
852.0 mCi
853.2 mCi
859.8 mCi
-0.77
Conclusions
A 10cc indirectly traceable Ga68/Ge68 source was tested for cross calibration of PET scanners and dose calibrators. Dose calibrator measurements using F18 settings were in close agreement with the calibration vial value (deviation < 1%).
The calibrated source in the modified phantom lid generated similar results to standard PET scanner cross-calibration using a 20 cm cylinder. Difference in the cross calibration factor is within 4%. A procedure was developed to estimate cross calibration factor and image resolution simultaneously.
Our results show that the PET scanner quantitative-calibration factor, with partial volume correction, and image resolution can be accurately estimated using a 10cc calibrated Ga68/Ge68 source in a modified phantom.
Image Reconstruction Parameters
Scanner Model
Institute
Image Volume
Voxel Size (mm)
Post-smoothing Filter
Biograph 6
UCLA
168x168x55
2.04x2.04x5
All pass
Biograph mCT
200x200x109
1.63x1.63x2
Biograph 64
USC
168x168x109
1.02x1.02x2
Gaussian 5mm FWHM
Discovery STE
City of Hope
128x128x47
2.0x2.0x3.27
Rect 4.8mm
Acknowledgements
The authors would like to thank Data Spectrum (Hillsborough, NC) for the modified ACR phantom lid and RadQual (Weare, NH) for the traceable Ga-68/Ge-68 source.