Time-Dependent Monte Carlo Simulation for Patient-Specific Dosimetry in Radioactive Iodine Therapy
J Shin1*, Y Seo2, Y Li2, R R Flavell2, C Kitahara1, C Lee1, (1) National Cancer Institute, Rockville, MD, (2) University of California San Francisco, San Francisco, CA
Presentations
TH-B-202-6 (Thursday, 7/14/2022) 8:30 AM - 9:30 AM [Eastern Time (GMT-4)]
Room 202
Purpose: Accurate estimation of dose to tumor and normal tissues in radiopharmaceutical therapy requires the quantification of the time-dependent biodistribution of radiopharmaceuticals. We developed the methods to derive time-dependent source distribution from patient specific time-activity curves and to calculate time-dependent dose distribution using Monte Carlo (MC) techniques.
Methods: We developed two time-dependent functions for source particle generation and dose scoring using Geant4. Pre- or post-therapy images and multi time point planar images were combined to produce a pair of 3D location and the temporal activity of each location in the particle generator. To reduce memory usage, the source location with activity greater than the minimum counter rate were only considered so that the reconstruction process stopped when the total activity exceeded 0.1% compared to that of previous time step. In the dose calculation step, only non-zero dose pixels were stored to reduce the memory usage. We evaluated the performance of the methods for an advanced thyroid cancer patient with a metastasis in the lungs. A SPECT and three whole-body gamma camera images were combined using the conjugate-view method to generate the spatiotemporal particle generator. Time-dependent MC simulations were performed to calculate dose distribution.
Results: For 5-hour time step, a source particle generator was reconstructed for the 21 time points (equivalent to 110 hours) but only 7% of locations was used to build the particle generator. For 1e6 histories, only 10% of the four-dimensional array (patient’s image matrix size: 512x512x107 with 21 time points) were used for time-dependent dose calculations. We confirmed that our methods handled memory efficiently enough for dealing with four-dimensional dosimetry.
Conclusion: We developed time-dependent MC-based dosimetry methods for radiopharmaceutical therapy allowing for patient-specific dose calculation. The methods should be useful to accurately estimate tumor and normal tissue doses for clinical and epidemiological research of radiopharmaceutical therapy patients.
Keywords
Taxonomy
Not Applicable / None Entered.
Contact Email