[OA110] A 3D-printed phantom study for quantitative 99MTC-MAA SPECT/CT imaging and dosimetry in 90Y radioembolization

Purpose Radioembolization (RE) with 90 Y-microspheres is a well-established treatment modality for treating liver malignancies. At a time of increasing evidence for dose–effect relationships in RE with 90 Y microspheres [1] , [2] , the general consensus is that there is an urgent need for accurate dosimetry in patients undergoing RE treatment. This work aimed at estimating absorbed doses to lesions and normal liver in a novel anthropomorphic set-up. Methods AbdoMan is a 3D-printed phantom provided with a fillable liver section and multiple inserts for lesion representation. A SPECT/CT Symbia Intevo provided with the proprietary xSPECT quantitative software was used to image the phantom according to the acquisition protocol currently used for RE patients. Specific regions of interest (ROIs) were drawn on MIM 6.1.7 system. A homemade tool was developed in MATLAB for image analysis and dose calculation based on two methods: I) convolution kernel and II) local deposition method. The accuracy of the two different dosimetric methods was evaluated by comparing dose-rate volume histograms (DrVHs). Moreover, γ -index was used to compare the dose distributions obtained by the two activity-to-dose methods. Results Differences calculated by the 3D γ -index are within 2%-2 mm for all AbdoMan inserts. The dose-kernel results in a γ γ as well as several image artefacts. In particular, an apparent over-dosage (about 25%) was observed in inserts with larger diameter, most likely due to spill-in and spill-out phenomena. The DrVHs for considered ROIs are within 2%-2 mm for both methods. Conclusions In RE treatment planning the dose-kernel method proved to be more accurate with respect to deposition method based on full 3D dose distributions.