PT - JOURNAL ARTICLE AU - Hongki Lim AU - Yuni Dewaraja TI - Impact of Tc-99m SPECT reconstruction methods on lung shunt and lesion/normal liver activity quantification in radioembolization DP - 2017 May 01 TA - Journal of Nuclear Medicine PG - 1302--1302 VI - 58 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/58/supplement_1/1302.short 4100 - http://jnm.snmjournals.org/content/58/supplement_1/1302.full SO - J Nucl Med2017 May 01; 58 AB - 1302Objectives: 99mTc-MAA SPECT/CT can potentially be used for more accurate determination of lung shunt and for lesion/normal liver dosimetry based treatment planning in Y-90 radioembolization. Our goal is to determine optimal reconstruction methods/parameters for quantitative Tc-99m SPECT/CT.Methods: A digital equivalent liver/lung torso phantom with multiple spherical and ovoid shaped hepatic ‘lesions’ (14 to 30 mL) was first used in a Monte Carlo simulation (SIMIND code) study followed by measurement of the physical phantom on a Symbia SPECT/CT with typical clinic acquisition parameters. A clinically realistic activity distribution for 99mTc-MAA (200 MBq, a 5% lung shunt and a 5:1 lesion-to-normal-liver uptake ratio) was used. In addition, to assess visibility of extra-hepatic uptake, objects with very low activity concentration (0.01 to 0.1 MBq/mL) were placed in the 'cold' torso region of the phantom. Simulated and measured phantom images were reconstructed with our in-house 3D-OSEM without and with CT-based attenuation correction (AC), window based (5% lower window adjacent to 15% main window) scatter correction (SC) and depth-dependent collimator-detector response (RR). For scatter correction, the optimal scaling factor k (typically assumed to be equal to one) of the scatter estimate S = k [asterisk](w_main/w_low) [asterisk]0.5, was investigated with the gold standard considered to be the ‘primary’ only reconstruction in the simulation study. In the physical phantom study, reconstructions were evaluated by comparing error in lung shunt and lesion/normal liver activity quantification (with a liver relative calibration approach) as well as noise in the background (normal liver). Visibility of extra hepatic uptake was assessed using an index for detectability based on the contrast-to-noise ratio and object size. Comparisons were made at 40 EM equivalent iterations (5 iterations, 8 subsets), but the impact of increasing the number of iterations was also investigated.Results: Optimal k value for accurate count recovery was 1.5 for lesions, 1.4 for normal liver and 0.9 for lung, but considering all regions in the optimization process, a value of k = 0.9 was selected as there is more gain in count recovery for lung than loss for lesion/normal liver when we set k = 0.9. All reconstructions (see Table) gave good quantification accuracy for the normal liver because of the (liver) relative calibration. Among correction methods, AC has biggest impact on the lung shunt accuracy and RR and SC on the quantification accuracy for lesions. Reconstruction with AC + SC +RR and a scatter scale factor of 0.9 gave quantification accuracy within 6% in lung and normal liver and 13% in lesions. Visibility of extra-hepatic uptake was high for all methods but the index increased considerably with SC and RR. The number of iterations to approach convergence was less than 30 without RR and around 50 when RR was included in the reconstruction.Conclusion: With a liver relative calibration approach, quantification accuracy within 11% was achieved for the normal liver with all methods evaluated here. However, to achieve high quantification accuracy in all regions that are of potential interest in 99mTc-MAA based treatment planning (lung, lesion, normal liver) all corrections (AC+SC+RR) are needed. Research Support: NIH R01EB022075 View this table:Evaluation of SPECT reconstructions using various correction methods