Objectives: : Previous studies showed that conventional planar Tc-99m MAA imaging overestimates lung shunt fraction (LSF) in Y-90 selective internal radiation therapy (SIRT), and SPECT/CT would provide more accurate result. However, the temporal mismatch between SPECT and conventional helical CT (HCT) may lead to respiratory artifacts from the CT-based attenuation correction. This simulation study aimed to investigate the potential improvement on SIRT treatment planning using SPECT/CT with different attenuation maps for attenuation correction, as compared to planar imaging.

Methods: The 4D Extended Cardiac Torso (XCAT) phantoms with Tc-99m MAA activity distribution and 3 anatomical variations were used in this study. For each phantom, respiratory motion of 2 and 4 cm and LSFs of 5%, 10%, 15% and 20% were modeled. We simulated a spherical liver tumor with tumor to normal liver (T/N) ratio of 11.43 and radius of 1.5 cm near the boundary of liver and lungs. An analytical projector was used to simulate anterior and posterior planar projections of the chest and abdomen region (5 min/view), and 128 projections with 30 s/view over 360^◦ for SPECT/CT, both modeling attenuation, scatter and geometric collimator-detector-response. Three different attenuation maps were evaluated for SPECT: (i) HCT captured at end-inspiration (HCT-IN), (ii) cine averaged CT (CACT) and (iii) HCT captured at end-expiration (HCT-EX). Projections were reconstructed by OS-EM algorithm with 20 iterations and 16 subsets with full compensations using different attenuation maps. The regions-of-interest of lungs and liver were manually drawn on the planar geometric mean images (2D), while they were drawn on the corresponding CT maps (3D) for SPECT to calculate the LSF, using an open-source software ITK-SNAP. The tumor was delineated from the CT maps for different SPECT protocols to calculate the T/N ratios. The errors of LSF and T/N ratio of different imaging protocols were measured.

Results: The LSF errors were generally larger for planar imaging and HCT-IN while relatively smaller for CACT and HCT-EX. For example, for 2 cm motion and 10% LSF, the LSF errors of HCT-IN, CACT, HCT-EX and planar were 77.71±3.89%, 45.19±3.76%, -29.09±3.10% and 45.22±12.36%, while the T/N errors were smallest for CACT followed by HCT-EX, and HCT-IN, i.e., 3.89±1.31%, -9.42±2.21% and -15.01±2.19% respectively. The differences among different protocols are generally higher for lower LSF and higher respiratory motion amplitude.

Conclusions: SPECT/CT protocol improves the Y-90 SIRT treatment planning accuracy as compared to planar. HCT-EX or CACT would be recommended for Tc-99m-MAA SPECT/CT attenuation correction to avoid the respiratory artifacts. Further studies to investigate how LSF and T/N ratio impact administrated dosage and radiation dose on critical organs and tumors are warranted.