Quantitative SPECT (QSPECT) at high count rates: contrasting performance of SPECT/CT systems from two major vendors

1146 Objectives: QSPECT-based dosimetry enables personalized therapy with 177Lu and other radionuclides. Widely variable administered activities and biokinetics can result in high count rate causing dead time or exceeding the camera’s quantitative capacity. In this work, we compared key characteristics of NaI crystal SPECT/CT systems from two major vendors, with focus on high count-rate behavior.\nMethods: Using 99mTc, we compared System A (Siemens Symbia T6 SPECT/CT) and System B (GE Discovery NM/CT 670), equipped with low-energy high-resolution collimators. 140-keV photopeak (20%), lower and upper scatter (10%) and additional energy windows were defined to cover a wide spectrum (18 to 504 keV). Acquisitions were performed sequentially with only detector 1, only detector 2, and both detectors activated. Twenty Eppendorf tubes were filled with 600 MBq of 99mTc. Dynamic planar acquisitions were performed by adding a tube every other 15-s frame on a cardboard between the two detectors, and this was repeated 20 h later. Calibration factor (CF; primary photons count rate per activity) and dead-time constant (τ; based on the wide-spectrum count rate) were determined by nonlinear regression of an equation derived from the Sorenson’s paralyzable model. Several SPECT/CT acquisitions of a NEMA 2012 phantom containing heterogeneously distributed 17.2 GBq of 99mTc were performed over 38 hours (96 or 90 projections, 5-10 s per projection) and reconstructed using an independent software (4i8s OSEM with attenuation and scatter correction; MIM Software). The total number of counts in the phantom was thereafter quantified, dead-time corrected, and compared to the true activity.\n Results: System A (CF=100.0 cps/MBq; τ=0.49 µs) was substantially more sensitive and less prone to dead time than System B (CF=75.4 cps/MBq; τ=1.74 µs). For System A, the maximum observed wide-spectrum count rate above which dead-time correction and quantification become impossible was dependent on detector activation mode: 700, 475 and 375 kcps with only the detector 1, only the detector 2 and both detectors activated, respectively (Fig. 1). Conversely, System B consistently exhibited a maximum count rate of 205 kcps. Planar data suggested maximum quantifiable unattenuated 99mTc activities of 6.6, 2.8 and 5.0 GBq for System A in single-detector mode, System A in dual-detector mode and System B, respectively (Fig. 1). In these settings, SPECT/CT of the heterogenous and attenuating NEMA phantom was quantified within 5% accuracy up to 5.1, 3.3 and 2.7 GBq of total activity, respectively (Fig. 2).\nConclusion: Calibration factor and dead-time constant vary considerably between SPECT/CT systems, whereas maximum quantifiable activity and corresponding count rate can also vary for a given camera depending on the detector activation mode (single vs. dual). In such case, using single-detector mode for imaging patients with very high activity retention can significantly expand the system’s quantitative capability. Accurate QSPECT at high count rate is critical to push the limits of personalized, dosimetry-based radionuclide therapy.\n\nFigure 1. Symbia T6 (left) and Discovery 670 (right) observed count rate vs. activity (top) and vs. expected count rate (bottom; only quantifiable non-grayed data points were fitted to the paralyzable model) during planar acquisitions.\n\nFigure 2. Accuracy of the NEMA phantom SPECT quantification vs. activity, without (left) and with (right) dead-time correction.