Author: Ananta Raj Chalise, Matthias K Gobbert, Hajar Moradmand, 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, University of Maryland, Baltimore County, University of Maryland, School of Medine, Department of Physics, University of Cape Town, M3D, Inc 🌍
Purpose: We are developing a novel Compton Camera (CC) system mounted orthogonally on the treatment nozzle for prompt gamma (PG) imaging-based proton range verification during treatment. This study investigates the efficacy of the imaging system and methods to improve the image quality.
Methods: Monoenergetic point sources (ranging: 511 keV to 5 MeV) were positioned within a water phantom and patient models using the Geant4-toolkit to assess depth-dependent gamma-counts. The CC was mounted on the treatment nozzle in a fixed configuration. Data-acquisition focused on recording double-and-triple event interactions, where gamma photons underwent one or two Compton scatters followed by absorption, respectively. These specific interactions were selected to reconstruct gamma emission hotspots accurately while avoiding noise from higher-order interactions. A kernel-weighted back-projection (KWBP) method with fixed parameters was employed to process the recorded data. Proton simulations were conducted at three distinct clinical energies within the same experimental setup, generating a comprehensive dataset of PG emissions from inelastic interactions. An iterative convolution post-processing method was introduced to correct distorted gamma images, significantly enhancing the clarity, ranges, and shape of the experimental reconstructed PG data. This method involved iterative convolution operations using point source data to refine the proton-simulated gamma images.
Results: Lower-energy gamma sources showed higher counts near the phantom surface, with an exponential attenuation behavior observed with depth. Reconstructed images demonstrated the robustness of KWBP algorithm for gamma imaging with known point sources. However, when applied to prompt gamma event data from proton simulations, the KWBP algorithm struggled to accurately reconstruct the images, resulting in shallower images. The introduction of the convolution-based post-processing point source correction method significantly improved range accuracy, shifting gamma spots from 115 mm to 33 mm from proton max-range.
Conclusion: The simulation results strongly support the new experimental-design for the Snout-Mounted-CC System, provided that image corrections are applied.