We investigated the quantitative accuracy of the model-based dual-isotope single-photon emission computed tomography (DI-SPECT) reconstructions that use Klein-Nishina expressions to estimate the scattered photon contributions to the projection data. Our objective was to examine the ability of the method to recover the absolute activities pertaining to both radiotracers: Tc-99m and I-123. We validated our method through a series of phantom experiments performed using a clinical hybrid SPECT/CT camera (Infinia Hawkeye, GE Healthcare). Different activity ratios and different attenuating media were used in these experiments to create cross-talk effects of varying severity, which can occur in clinical studies. Accurate model-based corrections for scatter and cross-talk with CT attenuation maps allowed for the recovery of the absolute activities from DI-SPECT/CT scans with errors that ranged 0-10% for both radiotracers. The unfavorable activity ratios increased the computational burden but practically did not affect the resulting accuracy. The visual analysis of parathyroid patient data demonstrated that our model-based processing improved adenoma/background contrast and enhanced localization of small or faint adenomas.

We investigate the quantitative accuracy of the reconstruction of absolute (99m)Tc and (111)In activities from (99m)Tc/(111)In dual-isotope SPECT studies. The separate reconstruction of two images is achieved by applying Monte Carlo simulation-based corrections for self-scatter and cross-talk between energy windows. For method evaluation, a series of (99m)Tc/(111)In physical phantom experiments was performed using a clinical SPECT/CT camera. The containers were filled with different ratios of (99m)Tc and (111)In activities to create cross-talk with varying severity levels. In addition, we illustrate the performance of our method by reconstructing images from four simultaneous (99m)Tc/(111)In SPECT/CT studies of neuroendocrine patients. Similarly to the phantom experiments, clinical cases provide examples with different severity of cross-talk. Phantom experiments showed that Monte Carlo simulation-based corrections improved both quantitative accuracy and visual properties of (99m)Tc and (111)In images. While the errors of absolute activities for both tracers in six containers ranged from 16% to 75% if no corrections for self-scatter and cross-talk were applied, these errors decreased to below 10% when images were reconstructed with the aforementioned corrections. These activities were measured using regions of interest larger than the true sizes of the containers in order to account for the spill-out effect. Analysis of patient studies confirmed that accurate simulation-based compensations improved resolution and contrast for both (99m)Tc and (111)In images.