Phantom test procedures for a new neuro-oncological amino acid PET tracer: [¹⁸F]fluciclovine

Amino acid positron emission tomography (PET) examinations using anti-1-amino-3-[¹⁸F]-fluorocyclobutane-1-carboxylic acid ([¹⁸F]FACBC) were allowed for routine clinical use in July 2024. However, phantom test procedures for [¹⁸F]FACBC reconstruction parameters have not yet been established. The present study aimed to establish new phantom test procedures for [¹⁸F]FACBC brain PET imaging to determine optimal reconstruction parameters.

Background (BG) activity as well as hot sphere and target-to-background ratios (TBRs) of [¹⁸F]FACBC were estimated based on brain activity and tumor-to-normal tissue ratios (TNR) in a Japanese clinical trial of [¹⁸F]FACBC. Phantom experiments proceeded under [¹⁸F]FACBC or L-[methyl-¹¹C]-methionine ([¹¹C]MET) conditions. The number of iterations and the Gaussian filter parameters were determined from the reconstruction parameters %contrast_mean and coefficients of variation (CVs) in ordered subset expectation maximization (OSEM) and time-of-flight (TOF) with or without point-spread-function (PSF) correction.

The amounts of activity in the hot spheres and BG were 1.1 and 5.5 kBq/mL, respectively, and the TBR was 5.0 at the start of acquisition. The %contrast_mean of all hot spheres was higher with [¹⁸F]FACBC than [¹¹C]MET, and %contrast_mean converged between 4 and 6 iterations in hot spheres with diameters < 10 mm. We used four iterations for OSEM + TOF and five for OSEM + TOF + PSF correction for [¹⁸F]FACBC and [¹¹C]MET images. The CV was higher for [¹⁸F]FACBC than [¹¹C]MET. The optimal sizes of Gaussian filters for OSEM + TOF and OSEM + TOF + PSF correction of image reconstruction were 5 mm for [¹⁸F]FACBC, and 4 and 3 mm, respectively, for [¹¹C]MET images.

We estimated phantom activity and TBR based on brain activity in a Japanese clinical trial and established new phantom test procedures for [¹⁸F]FACBC. We recommend that the optimal reconstruction parameters for [¹⁸F]FACBC should be set to the same number of iterations as [¹¹C]MET and that the FWHM of Gaussian filter should have a few mm higher than [¹¹C]MET to reduce image noise from brain normal tissue.