Author ORCID Identifier

0009-0005-2522-972X

Date of Graduation

12-2025

Document Type

Thesis (MS)

Program Affiliation

Medical Physics

Degree Name

Masters of Science (MS)

Advisor/Committee Chair

S. Cheenu Kappadath, Ph.D.

Committee Member

Armeen Mahvash, M.D.

Committee Member

Richard Wendt III, Ph.D.

Committee Member

James Long, Ph.D.

Committee Member

Rajat Kudchadker, Ph.D.

Abstract

This work interrogates absorbed dose metrics for spherical tumors under various configurations with emerging theragnostic radionuclides to understand the true dose being delivered to key points in and around the tumor and how real imaging systems effect the accuracy of these doses. Dose volume kernels (DVK) were generated in TOPAS Monte Carlo for 32P, 64Cu, 67Cu, 90Y, 131I, 153Sm, 161Tb, 166Ho, and 177Lu. These were generated with a fine resolution of (0.1 mm)3, meaning they can be used to generate clinically useful DVKs using the proposed upsampling algorithm. Spherical tumor geometries were generated in TOPAS MC for diameters from 1 mm to 100 mm with varying tumor-to-background ratio (TBR) and then discretized into cubic voxels using MATLAB to generate activity maps. Spatial resolution blurs and different voxel sizes were computed from the discretized MATLAB activity maps. Absorbed dose maps were generated using both DVK and local deposition (LDM) dose algorithms with the different radionuclides for all tumor configurations (diameters and TBR) that additionally included varying spatial resolutions ( 3 mm, 5 mm, 10 mm, 15 mm FWHM Gaussian blurs) and voxel sizes ([3 mm]3 and [5 mm]3). The mean dose, point-dose to rim, and point-dose to the 1 mm and 2 mm outside margins were extracted from these dose maps and compared to the corresponding ground truth. Additionally, an additional phantom scan on a clinical PET/CT system for validation was conducted to compare acquired image data with synthetic data outputs. vi Analytical models were determined for the mean doses as a function of tumor diameter and TBR for ideal, spherical geometries. Voxel size, spatial resolution, radionuclide, TBR, and dose algorithm all influence the accuracy of the inferred dose metrics. Mean tumor doses for all radionuclide affected by the diameter and spatial resolution with increasing accuracy as the tumor diameter increase and spatial resolution improved but was largely unaffected by the voxel size. The point-dose metrics were most susceptible to error for radionuclides with average beta energy exceeding 0.5 MeV (32P, 90Y, 166Ho) with the DVK dose algorithm outperforming the LDM for realistic imaging voxel sizes and resolutions. The LDM dose algorithm slightly outperforms the DVK for mean dose calculation for the energetic beta emitters, but the DVK and LDM-derived doses converge as spatial resolution increases. Across the board, the metrics scale linearly with respect to 1/TBR. Hypoenhancing geometries for key combinations of radionuclide, diameter, TBR, voxel size, and spatial blur were also generated from the solid sphere geometries using a combination of core diameter and core-to-tumor uptake fraction to interrogate the dose consequences of underdosing these regions.

Keywords

dosimetry theragnostic theranostic therapy radionuclide nuclear medicine imaging

Available for download on Thursday, December 10, 2026

Share

COinS