Respiratory impacts on dosimetric measurements for Tc-99m-MAA SPECT/CT

1444 Objectives: Respiration compromises Tc-99m-MAA SPECT image quality and quantitation accuracy especially near the diaphragmatic region not only by respiratory motion (RM) blurring, but also inaccurate attenuation correction (AC) and volume-of-interests (VOI), i.e., lungs, liver and tumor segmentation from mismatched CT. This study aims to evaluate these effects on Tc-99m-MAA SPECT/CT dosimetric measurements for Y-90 microsphere radioembolization treatment planning. Methods: The XCAT phantom was used to simulate a phantom population of 60 phantoms, modelling 9 different anatomical variations, 8 tumor-to-normal uptake ratios (TNR, 2 to 13.2), 8 tumor sizes (2 to 6.7 cm diameter), 7 tumor locations, 3 axial motion amplitudes of 1, 1.5 and 2 (cm) and lung shunt fractions (LSF) of 5%, 10%, 15% and 20%. Activity maps at end-expiration (End-EX) phase were used to simulate a respiratory gated SPECT study to exclude the RM blurring and evaluate the effects of AC and VOI segmentation, while averaged activity maps were used to evaluate the RM and combined (RM+AC+VOI) effects. An analytical projector for low energy high resolution parallel-hole collimator was used to simulate 60 noisy projections over 360o, modeling attenuation and geometric collimator-detector-response (GCDR). The OS-EM algorithm with 50 updates was employed for image reconstruction with AC and GCDR modelling. For AC effect, two attenuation maps at end-inspiration (HCT-IN) and mid-respiration (HCT-MID) were used for AC respectively and VOIs were mapped out based on the organ maps of HCT-EX. For VOI effect, HCT-EX was used for AC and VOIs were mapped out using organ maps of HCT-IN and HCT-MID respectively. For RM effect, the averaged attenuation map was used for AC and VOIs were mapped out using averaged organ maps. For combined effects, HCT-IN, HCT-MID and HCT-EX were used for AC and mapping out VOIs correspondingly. Noisy projections of activity maps at End-EX were generated and then reconstructed with AC using HCT-EX, which was then applied to map out VOIs from the reconstructed images as a reference of no RM, AC and VOI errors. The LSF, TNR, absorbed dose of tumor, lungs and normal liver (NL) were calculated based on the measured counts from different VOIs and the partition model. Results: Generally, along with the increase of breathing motion amplitude, the absolute errors of all indices increase. AC effect has the least errors among all indices where VOI segmentation with maximum mismatched CT, i.e., HCT-IN in this case, has the largest errors. For a common motion amplitude of 1.5 cm, the LSF errors from AC, VOI, RM and combined effects can reach 5.2±2.7%, 156.3±70.2%, 75.0±36.7% and 53.6±24.8% respectively. The tumor absorbed dose errors from those effects can reach -6.4±6.5%, -45.3±18.8%, -23.7±12.9% and -21.1±9.6% respectively while NL absolute absorbed dose errors were