Comparative evaluation of quantitative analysis using two phantom-based software programs for 18F-FDG-PET/CT.

Abstract

Variability in image quality and quantitative accuracy of Fluorine-18-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG-PET/CT) has been reported across institutions and devices. The Japanese Society of Nuclear Medicine (JSNM) introduced two guidelines to facilitate image quality assurance. Recently, a fully automated software, Arimaru (PDRadiopharma Inc.), has been developed to streamline this process. However, its performance relative to conventional software has not yet been fully validated. This study aimed to compare the physical parameters calculated using Arimaru and conventional software (PETquact IE, Nihon Medi-Physics Co., Ltd.) from PET images acquired using two PET/CT systems: Discovery MI and Discovery IQ (GE Healthcare). Images of the NEMA IEC Body phantom were acquired in list mode for 1800 s and reconstructed at multiple time points (30-1800 s) to simulate different noise levels. Five physical parameters (Q_H,10, N₁₀, Q_H,10/N₁₀, CV_BG, and SUV_max) were calculated using both the methods. The results showed that the automated method accurately positioned the region of interests (ROIs) and had a strong correlation with the conventional method across all parameters (r > 0.85, p < 0.05). However, some physical parameter values from the automated method were significantly different from those obtained using conventional software program. In conclusion, the automated software showed strong concordance with the conventional method and met JSNM guidelines. Nevertheless, systematic differences in the calculated values highlight the need to understand software-specific characteristics. The adoption of such tools may promote a broader and more consistent implementation of standardized PET imaging practices.