BIGDOSE: Software for 3D Personalized Targeted Radionuclide Therapy Dosimetry

Objectives: Absorbed dose for targeted radionuclide therapy (TRT) is usually calculated by the product of cumulated activity obtained based on sequential planar acquisitions and organ-based S-values from standard MIRD phantoms in conventional dosimetric software. Advance 3D quantitative radionuclide imaging techniques combined with voxel-based S-values boost the TRT dosimetry accuracy to voxel level, which is realized in several latest software packages. The goal of this work is to develop a comprehensive 3D software program, BIGDOSE, with new features of image registration and virtual CT for patient-specific image-based voxel level internal dosimetry.

Methods: BIGDOSE includes a portable GUI written in Python, integrating (i) input of sequential ECT/CT images; (ii) whole-body or organ-based, ECT or CT registration; (iii) curve fitting of time activity curves and voxel-based integration to obtain cumulative activity; (iv) dose conversion via convolution with voxel-based S values kernels; and (v) 3D dose analysis. The subroutines of image segmentation and non-rigid registration are executed via open source programs of ITK-SNAP and Elastix respectively. When only 1 CT is available, virtual CT can be generated at other time points to aid organs segmentation on ECT images and image registration. The effectiveness of the software was demonstrated using a simulation study with 9 XCAT phantoms and a patient study. We simulated SPECT/CT acquisitions at 1, 12, 24, 72 and 144 hrs post In-111 Zevalin injection using an analytical MEGP/CT projector, modeling attenuation, scatter and collimator-detector response. The patient was injected with 222 MBq In-111 DTPAOC and underwent SPECT/CT scans at 24, 48, and 72 hrs post-injection. SPECT data were then reconstructed using OS-EM method with full compensations. CT organ-based registration was performed before the dose calculation. Organ doses for the corresponding Y-90 therapeutic agents were calculated on target organs and compared with those obtained from OLINDA/EXM v1.1 with weight adjustment. For the simulation, organ dose estimated by a Monte Carlo code (GATE v6.1) was used as the gold standard.

Results: In the simulation, the organ dose errors of BIGDOSE were -9.59±9.06%, -8.36±5.82%, -23.41±6.67%, 14.13±5.32% for liver, spleen, kidneys and lungs, while they were -25.72±12.52%, -14.93±10.91%, -28.63±12.97% and -51.02±5.21% for OLINDA/EXM. Cumulative dose volume histograms (CDVH), dose maps and iso-dose contours provided 3D dose distribution information of the target organs. The total computational time for 3 organs-of-interest was about 3 hrs in the patient study.

Conclusion: BIGDOSE provides a one-stop platform for voxel-based dose estimation which can substantially alleviate the tissue inhomogeneity problem in dose calculation, with enhanced function to reduce the mis-registration and segmentation errors. It is a promising tool to streamline the current clinical TRT dosimetric practice with high accuracy and 3D personalized information for improved treatment outcome. Research Support: This work was supported by research grants from Macau Science and Technology Development Fund (079/2011/A3), University of Macau (MYRG2016-00091-FST) and National Natural Science Foundation of China (81601525).