Optimization of image reconstruction conditions with phantoms for brain FDG and amyloid PET imaging.

Abstract

The purpose of this study was to optimize image reconstruction conditions for brain (18)F-FDG, (11)C-PiB, (18)F-florbetapir and (18)F-flutemetamol PET imaging with Discovery-690 PET/CT for diagnosis and research on Alzheimer's disease (AD) based on the standard imaging protocols and phantom test procedures and criteria published by the Japanese society of nuclear medicine (JSNM). A Hoffman 3D brain phantom and a cylindrical pool phantom were scanned according to the JSNM procedure, and the reconstruction conditions (iteration, subset, post-filter) were optimized so that the images satisfy the JSNM criteria regarding spatial resolution (FWHM ≤ 8 mm) and gray/white matter contrast (%contrast ≥ 55%) on the Hoffman phantom and uniformity (SD of small ROIs ≤ 0.0249) and image noise (coefficient of variation ≤ 15 %) on the pool phantom. Human images were acquired with (18)F-FDG (15-min scan starting at 30 min post-injection [p.i.] of 185 MBq), (11)C-PiB (20-min scan starting at 50 min p.i. of 555 MBq), (18)F-florbetapir (10-min scan starting at 50 min p.i. of 370 MBq) and (18)F-flutemetamol (30-min scan starting at 90 min p.i. of 185 MBq) on 1 or 2 subjects for each tracer and reconstructed with thus determined conditions to evaluate the image quality visually. The effect of reconstruction parameters on the standardized uptake value ratio (SUVR) was also evaluated on 5 amyloid-positive and 5 amyloid-negative PiB images. A sufficient image quality was obtained at an iterative update (product of iteration and subset) of 64 for (18)F-FDG. The same reconstruction parameters with an additional Gaussian filter of 5 mm FWHM was optimal for (11)C-PiB, (18)F-florbetapir and (18)F-flutemetamol to achieve the phantom criteria. Those optimal reconstruction conditions were confirmed with human images. The SUVR value was stable over a wide range of iterative updates around the optimal parameters both for positive and negative amyloid images. Optimal image reconstruction conditions were determined for brain (18)F-FDG and amyloid PET imaging with Discovery-690 PET/CT for diagnosis and research on AD based on the JSNM phantom criteria. This supports feasibility of the phantom criteria for standardization and harmonization of brain (18)F-FDG and amyloid PET for multicenter studies.