Comparison of between PET-CT Scanners with Different Physical Characteristics

Objectives: PET scanner performance is often assessed based on resolution (FWHM). NEMA standards, international evaluation criteria, use FBP for image reconstruction for measurement while, in clinical practice, OSEM is often used. OSEM and FBP perform different calculations to reconstruct images and may result in different FWHM. Three scanners with different FWHM published by their manufacturers were tested under different conditions to measure and compare resolutions. Image reconstruction conditions usable in clinical practice were also studied. Methods: Equipment used PET-CT scanners Discovery ST Elite Performance (GE) Discovery 690 VCT (GE) Biograph mCT (SIEMENS) Experiment 1 A line source with a 1-mm inner diameter (for 18F injection) was installed in the center of a 30-cm diameter SPECT Phantom filled with water to collect data and reconstruct images by OSEM for the following studies. 1. Relationship between number of iterations and FWHM 2. Relationship between changes of pixel size and FWHM 3. Relationship between changes of Gaussian and FWHM 4. Comparison between FBP and OSEM Experiment 2 Comparison of resolution of defects, using a heart phantom. Acrylic cylinders, 8 mm high and 2 and 3 mm in diameter, were put in cardiac muscle to simulate defects and compare resolutions. Experiment 3 Optimal conditions for brain images, as well as for those for tumor diagnosis, were studied. Brains, breast cancers up to 2 mm and metastatic liver cancers up to 8 mm in diameter were chosen as targets. Results: Experiment 1 1, increased iterations did not change FWHM. 2, smaller pixel sizes improved FWHM, but when the pixel size was 2 mm or smaller, FMHM became 4 mm or smaller; the changes were small. 3, smaller Gaussian improved FWHM, but FWHM, also affected by pixel size, became smaller as both pixel size and Gaussian became smaller, and it reached approx. 3 mm when Gaussian was 0. 4, the above results were considered and three scanners with different FWHM , between 6.2 mm to 4.5 mm, given in brochures by their manufacturers, were compared. When images were reconstructed through FBP, FWHM were close to those given in the brochures. However, in this experiment that used OSEM, no significant differences were observed between the scanners, and they all showed much smaller values, approx. 3.0 mm. Experiment 2, all the scanners showed defects of 3 mm and 2 mm in diameter. Experiment 3, showed that iterations x subsets must be 100 or more to create high-resolution image