Performance evaluation of iterative PET reconstruction with resolution recovery incorporating Gallium-68 positron range correction.

Background: The distance traveled by the positron before annihilation with an electron, the so-called positron range, negatively effects the positron emission tomography (PET) image quality for radionuclides emitting high-energy positrons such as Gallium-68 (Ga).

Purpose: In this study, the effect of a tissue-independent positron range correction for Gallium-68 (Ga-PRC) was investigated based on phantom measurements. The effect of the Ga-PRC was also explored in four patients.

Methods: The positron range distribution profile of Ga in water was generated via Monte Carlo simulation. That profile was mapped to a spatially invariant 3D convolution kernel which was incorporated in the OSEM and Q.Clear reconstruction algorithms to perform the Ga-PRC. In addition, each reconstruction method included point spread function (PSF) modeling and time-of-flight information. For both Fluorine-18 (F) and Ga, the NEMA IQ phantom was filled with a sphere-to-background ratio of 10:1 and scanned with the GE Discovery MI 5R PET/CT system. Standard non-positron range correction (PRC) reconstructions were performed for both radionuclides, while also PRC reconstructions were performed for Ga. Reconstructions parameters (OSEM: number of updates, Q.Clear: beta value) were adapted to achieve similar noise levels between the corresponding reconstructions. The effect of Ga-PRC was assessed for both OSEM and Q.Clear reconstructions and compared to non-PRC reconstructions for Ga and F in terms of image contrast, noise, recovery coefficient (RC), and spatial resolution. For the clinical validation, Ga-labeled prostate-specific membrane antigen (Ga-PSMA) and Ga-DOTATOC PET scans were included of two patients each. For each PET scan, patients were injected with 1.5 MBq/kg of Ga-PSMA or Ga-DOTATOC and the contrast-to-noise ratio (CNR) was calculated and compared to the non-PRC reconstructions.

Results: For OSEM reconstructions, including the Ga-PRC improved the RC by 9.4% (3.7%-19.3%) and spatial resolution by 21.7% (4.6 mm vs. 3.6 mm) for similar noise levels. For Q.Clear reconstructions, Ga-PRC modeling improved the RC by 6.7% (2.8%-10.5%) and spatial resolution by 15.3% (5.9 mm vs. 5.0 mm) while obtaining similar noise levels. In the patient data, the use of Ga-PRC enhanced the CNR by 13.2%.

Conclusions: Including Ga-PRC in the PET reconstruction enhanced the image quality of Ga PET data compared to the standard non-PRC reconstructions for similar noise levels. Limited patient results also supported this improvement.