Characterization of Delayed Imaging Quantitative Accuracy on a Long-Axial Field-of-View Total-Body PET Scanner
Abstract
Purpose
To quantify radiotracer- and reconstruction-dependent feasibility of delayed PET imaging on a long axial field-of-view (LAFOV) system, and to identify pediatric-relevant delayed imaging windows (or equivalent low-activity protocols) while maintaining small-lesion quantitative performance.
Methods
A NEMA NU-2 image quality phantom (six spheres: 10–37 mm) was imaged using five PET radiotracers (F-18, C-11, N-13, Ga-68, Cu-64) with 30-min acquisitions at serial delays through 10 half-lives (analysis at HL1/2/4/6/8/10). Reconstructions included TOF+PSF OSEM (3 iterations, 10 subsets, no post-filter), vendor-provided Hyper-Iterative (HI), and Deep Progressive Reconstruction (DPR) when available (F-18, C-11, Ga-68). Metrics included target-to-background ratio (TBR), contrast recovery coefficient (CRC), background variability (BV%), lesion variability (LV%), and pooled contrast-to-noise ratio (CNR_p). Uncertainty-based 95% confidence intervals (CI) for TBR and CRC were computed by error propagation.
Results
Small spheres (10–13 mm) governed feasibility. Under a balanced pediatric decision rule (CRC 95% CI ≥ 0 .70, BV% ≤ 25, and size-dependent CNR_p thresholds), the maximum feasible delays for 10–13 mm targets were: C-11 up to HL4 (HI), N-13 up to HL6 (HI), Ga-68 up to HL6 (DPR), F-18 up to HL8 (HI), and Cu-64 up to HL4 (OSEM). These correspond to remaining activity fractions of 1/16, 1/64, 1/64, 1/256, and 1/16, respectively.
Conclusion
LAFOV PET enables extended delayed (low-count) imaging windows that are tracer- and reconstruction-dependent. Uncertainty-based constraints support conservative pediatric protocol optimization and provide objective bounds for activity reduction while preserving small-lesion quantitative performance.