Author: Ananta Raj Chalise, Matthias K Gobbert, Zhuoran Jiang, Sina Mossahebi, Stephen W. Peterson, Jerimy C. Polf, Lei Ren, Ehsan Shakeri, Vijay Raj Sharma, Jie Zhang 👨🔬
Affiliation: University of Maryland School of Medicine, University of Maryland Baltimore County, University of Maryland, Baltimore County, Stanford University, University of Maryland, School of Medine, Department of Physics, University of Cape Town, M3D, Inc 🌍
Purpose: Prompt gamma (PG) imaging is a promising modality for proton dose verification. Currently, there is a lack of effective tool to investigate the PG emission during proton therapy and optimize the imaging process in patient anatomy. To address this gap, we developed a Monte-Carlo package that simulates the entire PG emission and imaging workflow in patient anatomy to evaluate and optimize this novel technique.
Methods: We utilized Geant4 classes and G4-ancillary tools, employing the DCMTK external tool with G4PhantomParameterisation to convert CT data into voxelized geometries. Proton beams were modeled based on medical physics commissioning data. A two-stage POLARIS-J Compton-Camera (CC) positioned for recording total-double-triple-scattered PG signals. A parallel world approach was implemented for percentage depth dose (PDD) calculations. Simulations were executed on a high-performance computing facility consists of 51-node cluster server. The detected PG signals data were used to reconstruct PG images using Kernel-Weighted-Back-Projection (KWBP) algorithm.
Results: Gamma energy distribution analysis across various proton energies reveals an exponentially decaying continuum with discrete line emissions from nuclear reactions involving elements like oxygen, carbon, nitrogen, and calcium. PG emissions from proton-induced (pI) contribute more significantly to the total PG signal than neutron induced (nI). For instance, simulating a 198.7 MeV energy layer for a single spot delivering 1.36 monitor units (MU) in the Geant4 engine resulted in 1,945,111 emission of total PGs with 1057812 of pI and 148666 of nI. Positioning a two-stage-CC beneath the patient couch detected only 8,115 total PG signals. Further, using the KWBP algorithm and plugin detected PG signals generates gamma hotspots at range as close as 1.6 cm from the proton dose maximum.
Conclusion: The developed simulation package effectively elucidates the origin of PG signals within patient anatomy with CC positioned under patient couch. KWBP algorithm reconstructs gamma hot-spots with accuracy of < 2 cm.