TY - JOUR T1 - On the Accuracy of Voxel-Based Kidney Dosimetry JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 263 LP - 263 VL - 60 IS - supplement 1 AU - Johannes Tran-Gia AU - Michael Lassmann Y1 - 2019/05/01 UR - http://jnm.snmjournals.org/content/60/supplement_1/263.abstract N2 - 263Aim: Due to continuous improvements in quantitative SPECT/CT imaging, voxel-based dosimetry for radionuclide therapies has aroused growing interest as it promises the visualization of differences in absorbed doses on a sub-organ level. In this work, a non-uniform kidney distribution was mimicked by a 3D-printed 2-compartment kidney phantom filled with Lu-177 solutions of different activity concentration. Subsequently, SPECT/CT-based voxel-based dosimetry was performed to assess the potential of voxel-based absorbed dose assessments. Additionally, the potential of the PETPVC partial-volume correction tool was investigated [1]. MATERIALS AND METHODS: Phantom: 3D-printed kidney filled with Lu-177 solution (cortex: 100 mL @ 1.89 MBq/mL, medulla: 50 mL @ 0.36 MBq/mL), placed inside a water-filled NEMA body phantom [2]. Activity concentrations were measured with a calibrated HPGe detector. Acquisition: Siemens Intevo Bold SPECT/CT, medium-energy collimator, 2×60 views of 30 s, continuous non-circular Orbit. Reconstruction: xSPECT Quant (voxel-size: 2 mm, output: Bq/mL). Different combinations of iterations (12/24/36/48) and Gaussian post-filters (0/1/2/4/8/16 mm FWHM). CT-based attenuation correction, triple-energy-window scatter correction. Partial-Volume Correction (PVC): Interpolation of VOIs to CT resolution (1 mm isotropic), application of PETPVC with iterative Young and predetermined SPECT resolutions (11-22 mm depending on the reconstruction [3]). Absorbed Dose Distributions: Convolution of SPECT/CT images [MBq/mL] with a Monte-Carlo based, pre-tabulated Lu-177 dose kernel [mGy/MBq/s] [4]. Multiplication with voxel volume [mL] and time-integrated activity coefficient (assuming 51h tissue-specific kidney half-life [5]) yields absorbed dose [Gy] in each voxel. Total Absorbed Doses: Mean over dose distribution in CT-based cortex and medulla VOIs (post-processing in [6]). Nominal Absorbed Doses: Calculation based on 2 uniform activity distributions placed in a voxel representation of the 3D printing designs. RESULTS AND DISCUSSION: Absorbed Dose Distributions: SPECT/CT image reconstruction blurs the 2 discrete absorbed dose values into a continuous distribution, shifting the absorbed voxel doses towards smaller values. While this effect is slightly improved by applying more iterations, it is further enhanced by post-filtering. After PETPVC, the distribution of the absorbed dose voxels is separated into 2 continuous parts, leading to a better agreement between SPECT/CT-based and nominal values. Total Absorbed Doses: The nominal absorbed doses were calculated as 7.7/1.6 Gy (cortex/medulla). SPECT/CT imaging resulted in total absorbed doses ranging from 4.0-5.3 Gy (cortex) and 4.4-3.2 Gy (medulla, 12-48 iterations). PETPVC improved the accuracy to 7.3-9.4 Gy (cortex) and 3.9-1.2 Gy (medulla). The best agreement to the nominal absorbed doses was reached after PVC for 12 iterations (cortex, −5.2%) and for 36 iterations (medulla, −4.5%). View this table:Table 1: Nominal and SPECT/CT-Based Sub-Organ Absorbed Doses (Gy) CONCLUSION: Our study shows that Lu-177 quantitative SPECT/CT imaging of the kidneys leads to voxel-based dose distributions largely differing from the real organ distribution. Despite large errors in the absorbed voxel doses, an accuracy of 5% was achieved for total sub-organ absorbed doses by means of well-calibrated SPECT/CT imaging in combination with well-defined VOIs. PETPVC not only improved the visual match between SPECT/CT-based and nominal dose distributions, it also largely decreased the error in the total sub-organ absorbed doses. In conclusion, the concept of voxel-based dosimetry should be treated with caution. Specifically, it should be kept in mind that the absorbed dose distribution is only a convolved version of the underlying SPECT reconstruction. LITERATURE: [1] Thomas; PhysMedBiol; 2016; 61(22). [2] Tran-Gia; JNuclMed; 2016; 59(4). [3] Tran-Gia; JNuclMed; 2019; 60(1). [4] Kletting; ZMedPhys; 2015; 25(3). [5] Hänscheid; JNuclMed; 2018; 59(1). [6] Fedorov; MagnResonImaging; 2012; 30(9). ER -