A Monte Carlo simulation framework for performance evaluation of a proton imaging system without front trackers

Abstract

Today, radiotherapy is one of the main methods for cancer treatment and it is used to irradiate a tumour with a prescribed dose according to a dose plan designed to irradiate the tumour cells while sparing surrounding healthy tissue and organs at risk as much as possible. As radiotherapy has improved and increased the survival rates for several types of cancer over the years, a reduction of long and short term side-effects has become an important focus in modern radiotherapy. One of the most severe side-effect is secondary cancer that can occur decades after treatment.

Particle therapy has a potential to reduce the risk of long and short term side-effects in radiotherapy by reducing the irradiated volume and enable more sparing of healthy tissue surrounding the tumour. This can be achieved due to the physics of charged particles stopping in matter and depositing an increased dose at the end of their range. To ensure accurate treatment with particles it is imperative to have the particles stop inside the tumour volume. Today, particle therapy dose plans are based on X-ray computed tomography (CT) images of the patient that are converted to Relative Stopping Power (RSP) to calculate how the particle will deposit dose inside the patient. This conversion, due to the calibration and difference between photon and charged particle interactions in matter, is associated with uncertainties, up to 3.5% in some cases, this can result in misplacement of the distal dose of several mm and necessitate the inclusion of treatment margins around the tumour volume.

Proton CT is an imaging method circumventing this conversion step by applying protons as the imaging particle and directly calculate the RSP for dose planning purposes, proton CT can potentially make treatment with particle therapy even more accurate. Proton CT uses a high energy proton beam with sufficient initial energy to pass through the patient and enter a detector that measures the proton residual energy. The energy-loss of each proton is thus used to reconstruct a volumetric stopping power map over the patient to be used for dose planning. Due to the physics of charged particle interactions, protons will scatter in matter and this necessitates path estimations, i.e. most likely path, of the individual protons as they traverse the patient to achieve more accurate distribution of the energy-loss locations. This typically requires two sets of position sensitive detector systems (tracker planes), one upstream (front) and one downstream (rear) of the patient to measure the proton entrance and exit position for Most Likely Path (MLP) estimations.

Since proton imaging does not exist in the clinics today, an idea to adapt a proton CT detector assembly to use in proton therapy treatment rooms and bringing proton imaging a step closer to a clinical implementation is to remove the front trackers and instead rely on pencil beam scanning and rear trackers (single-sided imaging setup) for path estimations. A GEANT4/GATE based Monte Carlo (MC) simulation environment was designed to create the necessary MC framework for investigating proton imaging setups both with and without front trackers. The MC calculated proton positions on position sensitive tracker planes is used to reconstruct proton radiographs and proton CT images. The MC simulation environment is based on pencil beam scanning irradiating standardized Catphan® phantoms for spatial resolution and RSP accuracy investigations, including a clinically relevant paediatric head phantom. The pencil beam spot-size and spot-spacing parameters were varied in the single-sided setup to identify and study their effect on MLP and image quality, while a conventional proton imaging setup consisting of both front and rear trackers (double sided) was used as a gold standard in comparisons. The reconstructed proton radiography and proton CT image quality in terms of spatial resolution and RSP accuracy was quantified to evaluate the proton imaging setups being investigated. The impact on most likely path estimations and image quality in radiographs were also investigated when modifying the pencil beam spot size, e.g. 7 mm and 3 mm full width at half maximum (FWHM), and spot-spacing (spot spacing of 0.5, 1, and 2 times the FWHM) when performing pencil beam scanning.

The practical use of the MC simulation framework was exemplified by modelling the proton CT Digital Tracking Calorimeter (DTC) prototype that is designed and under construction by the Bergen proton CT collaboration. The DTC is a single-sided imaging setup consisting of multiple layers of ALPIDE pixels sensors and aluminium energy absorbers and was modelled in the MC simulation framework with accurate material budgets. The DTC was investigated in terms of the resulting MLP accuracy and image quality using the expected tracker position resolution and RSP resolution of the DTC.

Simulations of the radiation environment using the FLUKA MC code was performed to investigate the radiation environment the detector assembly is expected to be exposed to during irradiation of the patient. Potential radiation damage and effects such as single event upsets in the radiation sensitive FPGA readout electronics of the DTC were estimated based on FLUKA calculated particle fluence and dose deposited in the FPGAs. When the sensitive FPGAs were placed at a distance of 100 cm or more perpendicular from the DTC it was found to be radiation hard enough to be operational for over 30 years without considerable radiation effects during operation. The ALPIDE in terms of its documented design limitations were also found to be radiation hard enough to survive in the radiation environment for over 30 years.

Image quality analysis in the form of spatial resolution and RSP accuracy revealed that the single-sided proton imaging setup, such as the Bergen DTC, has the potential to be used for dose planning purposes. The spatial resolution results larger than 3 line pairs per cm from the Catphan® CTP528 phantom module, and the less than 0.5% RSP deviation from reference RSP values of materials involved in the Catphan® CTP404 phantom module showed this. However, investigation into the proton CT reconstructed image of a paediatric head phantom revealed that more studies focused on dose plans based on proton CT images should be performed in the future to better evaluate the impact of using a single-sided proton imaging setup.

The MC simulation framework for proton imaging and image analysis is expected to be usable in future proton imaging studies by modifying the proton imaging setups and evaluating resulting proton radiographs and proton CT image qualities.