Abstract
Background: In positron emission tomography (PET) scan reporting, both quantitative and qualitative interpretations are used. Qualitative and quantitative interpretations depend upon PET/CT image quality that, along with many biological factors, strongly depends upon image reconstruction parameters. This study intends to evaluate one of the key reconstruction parameters, i.e., the number of reconstruction iterations on standardized uptake value (SUV) and the image quality of PET/CT scan.
Methods: Images of National Electrical Manufacturers Association International Electrotechnical Commission Image Quality Phantom (with a tumor to background ratio of 4:1 and 2-Fluoro-2-Deoxy-Gulocose as a radiotracer) were acquired on Discovery STE PET/CT scanner for acquisition time of 5 minutes. Acquired data were reconstructed using 2-7, 10, 15, and 20 ordered subset expectation maximization (OSEM) iterations of iterative reconstruction.
Results: Quantitative measure SUV and qualitative measure hot and cold lesion contrast, image's spatial resolution and background variability showed an increasing trend with OSEM iterations. A maximum increase in 20.25%, 16.33%, 9.79%, and 6.88% was observed in SUV for 10, 13, 17, and 22 mm lesions for OSEM iterations varying from 2 to 3. Percentage of hot lesion contrast showed a rapid increase as the number of iterations changed from 2 to 7 and increased slowly afterward. The percentage background variability showed an increasing trend for each lesion with OSEM iterations.
Conclusion: Optimized number of OSEM iterations is 3 with an image matrix size of 128 x 128 with filters full width at half maximum (FWHM) of 6 mm. High-resolution studies require three iterations with an image matrix size of 256 x 256 with filters FWHM of 6 mm. The extended reconstruction time limits the use of a higher number of iterations.

Key words: SUV, PET/CT, hot lesion contrast, cold lesion contrast, OSEM, iterations, FDG, background variability