UC Berkeley

Proton beam usage to treat cancer has recently experienced rapid growth, as it offers the ability to target dose delivery in a patient more precisely than traditional x-ray treatment methods. Protons stop within the patient, delivering the maximum dose at the end of their track—a phenomenon described as the Bragg peak. However, because a large dose is delivered to a small volume, proton therapy is very sensitive to errors in patient setup and treatment planning calculations. Additionally, because all primary beam particles stop in the patient, there is no direct information available to verify dose delivery. These factors contribute to the range uncertainty in proton therapy, which ultimately hinders its clinical usefulness. A reliable method of proton range verification would allow the clinician to fully utilize the precise dose delivery of the Bragg peak.

Several methods to verify proton range detect secondary emissions, especially prompt gamma ray (PG) emissions. However, detection of PGs is challenging due to their high energy (2–10 MeV) and low attenuation coefficients, which limit PG interactions in materials. Therefore, detection and collimation methods must be specifically designed for prompt gamma ray imaging (PGI) applications. In addition, production of PGs relies on delivering a dose of radiation to the patient. Ideally, verification of the Bragg peak location exposes patients to a minimal dose, thus limiting the PG counts available to the imaging system.

An additional challenge for PGI is the lack of accurate simulation models, which limit the study of PG production characteristics and the relationship between PG distribution and dose delivery. Specific limitations include incorrect modeling of the reaction cross sections, gamma emission yields, and angular distribution of emission for specific photon energies. While simulations can still be valuable assets in designing a system to detect and image PGs, until new models are developed and incorporated into Monte Carlo simulation packages, simulations cannot be used to study the production and location of PG emissions during proton therapy.

This work presents a novel system to image PGs emitted during proton therapy to verify proton beam range. The imaging system consists of a multi-slit collimator paired with a position-sensitive LSO scintillation detector. This innovative design is the first collimated imaging system to implement two-dimensional (2-D) imaging for PG proton beam range verification, while also providing a larger field of view than compared to single-slit collimator systems. Other, uncollimated imaging systems have been explored for PGI applications, such as Compton cameras. However, Compton camera designs are severely limited by counting rate capabilities. A recent Compton camera study reported count rate capability of about 5 kHz. However, at a typical clinical beam current of 1.0 nA, the estimated PG emission rate would be 6 x 108 per second. After accounting for distance to the detector and interaction efficiencies, the detection system will still be overwhelmed with counts in the MHz range, causing false coincidences and hindering the operation of the imaging system.

Initial measurements using 50 MeV protons demonstrated the ability of our system to reconstruct 2-D PG distributions at clinical beam currents. A Bragg peak localization precision of 1 mm (2σ) was achieved with delivery of (1.7 ± 0.8) x 108 protons into a PMMA target, suggesting the ability of the system to detect relative shifts in proton range while delivering fewer protons than used in a typical treatment fraction. This is key, as the ideal system allows the clinician to verify proton range when delivering only a small portion of the prescribed dose, preventing the mistreatment of the patient. Additionally, the absolute position of the Bragg peak was identified to within 1.6 mm (2σ) with 5.6 x 1010 protons delivered.

These promising results warrant further investigation and system optimization for clinical implementation. While further measurements at clinical beam energy levels will be required to verify system performance, these preliminary results provide evidence that 2-D image reconstruction, with 1–2 mm accuracy, is possible with this design. Implementing such a system in the clinical setting would greatly improve proton therapy cancer treatment outcomes.