Simultaneous crosstalk compensation for a ROI activity estimation in dual-isotope SPECT imaging

Simultaneous model-based compensation for dual-isotope SPECT (DI-SPECT) imaging considers data acquired in both energy windows as a single set, models crosstalk effects from both isotopes and includes them simultaneously into the reconstruction. Being very appealing from the theoretical point of view, it requires substantial computer resources. However, if the investigated object can be divided into regions of interest (ROI) and if only the total activity in each ROI needs to be reconstructed, the dimension of the imaging problem tremendously decreases. In this work, we propose and validate a new strategy for reconstructing total activities pertaining to both radiotracers inside ROI from DI-SPECT/CT studies. This approach (ROI-DI-SPECT) can be applied to any clinical DI-SPECT study where the boundaries of investigated organ(s) are determined from structural modalities. In our technique, the elements of imaging matrix modeling contribution of each photon to each energy window are calculated by uniformly filling each ROI with unity activity and projecting it. For the projection step, the Analytical Photon Distribution (APD) method incorporating effects of attenuation, scattering, and resolution loss was employed. The computed noiseless APD-based projections were re-scaled using measured calibration factors. For validation, we performed two phantom experiments on clinical SPECT/CT system (Infinia-Hawkeye, GE Healthcare), where four 30 ml plastic containers (cylindrical bottles modeling ROI) were filled with different concentrations of Tc-99m and In-111 activities. From these phantom experiments, our algorithm recovered the absolute activities (MBq) in each ROI with errors 1.5-9.0% (In-111 activities) and 0.8-21.3% (Tc-99m activities). The decreased number of unknowns in the ROI-based DI-SPECT reconstruction allowed us to use data from only 6 projections without affecting the quantitative accuracy of the results. The improvement of the (i) down-scatter estimation and (i- - i) CT-based ROI determination are key elements of our plan to optimize the presented method.