A phantom study to achieve comparable ¹²⁴I PET recovery and address scanner variability for dosimetry purposes in (re)differentiated thyroid cancer

Background Accurate quantification of iodine uptake is essential for performing pre-treatment dosimetry of ¹³¹I therapy after redifferentiation of radioiodine-refractory thyroid cancer. Standardized procedures for ¹²⁴I PET/CT-based dosimetry are currently lacking. We aim to evaluate the relation between ¹⁸F and ¹²⁴I imaging for two different PET/CT scanners, the effect of bias correction on recovery across scanners and investigate the impact of clinically-encountered background (BG)-to-lesion ratios on recovery correction. Methods Cylindrical and NEMA body phantoms were scanned following vendor-recommended ¹²⁴I acquisition using clinically representative activity concentrations (5.6–5.7 kBq/mL in cylindrical phantom, 45.1–59.2 kBq/mL in NEMA spheres) and BG-to-sphere ratios (∼1:125 to 1:infinity) using two digital PET/CT scanners (Philips Vereos and GE Healthcare Omni). Additionally, BG-to-sphere ratio ∼1:10 was acquired to compare ¹²⁴I to the EARL ¹⁸F standards 1 (EARL1). Calibration accuracy and recovery coefficients (RC_max, RC_mean) were compared between scanners with and without bias correction. Results ¹²⁴I recovery was ∼20% higher for the Omni compared to the Vereos, showing calibration accuracies of 1.12–1.15 vs. 0.94, and RC_mean reaching 0.86 vs. 0.69. After bias correction, RC_mean was comparable between scanners ('1%) but below the lower limits of EARL1. A single fit for recovery correction (R² = 0.97) was obtained for different BG-to-sphere ratios for both scanners as RC_mean was comparable (p ' 0.4). Conclusion Vendor-recommended ¹²⁴I acquisition and reconstruction leads to differences in quantification but can be compensated using a bias correction. Recovery correction is minimally affected by different BG-to-lesion ratios, suggesting that one RC curve is sufficient, simplifying ¹²⁴I calibration procedures in studies requiring ¹²⁴I quantification.