Integral proton radiography scatter reduction through pencil beam pixel weighting and thresholding
MO-0214
Abstract
Integral proton radiography scatter reduction through pencil beam pixel weighting and thresholding
Authors: Daniel Robertson1, Chinmay Darne2, Charles-Antoine Fekete3, Sam Beddar2
1Mayo Clinic, Radiation Oncology, Phoenix, USA; 2The University of Texas MD Anderson Cancer Center, Radiation Oncology, Houston, USA; 3University College London, Medical Physics and Biomedical Engineering, London, United Kingdom
Show Affiliations
Hide Affiliations
Purpose or Objective
The purpose of this study is to develop a new integral proton imaging approach using beam-by-beam processing with pixel weighting and thresholding (PWT) for proton scatter reduction.
Proton radiography can improve image guidance, decrease stopping power uncertainty, and streamline adaptive radiotherapy workflows. Most proton radiography research focuses on single proton tracking detectors, but technical challenges including high count rates and the expense and complexity of detector assemblies have slowed clinical adoption of these systems. Integrating detectors employing scintillators and cameras circumvent these challenges, but proton scattering increases image noise and decreases contrast and water-equivalent thickness (WET) accuracy. The PWT imaging approach can improve image quality while maintaining the benefits of integrating proton imaging.
Material and Methods
The detector
comprises a cubic block of plastic scintillator with 20 cm side length and a
camera facing the beam nozzle (Fig. 1a). A collection of calibration curves is
formed by imaging a single pencil beam as a function of penetration depth in
the scintillator and distance from the beam center (Fig. 1b-c). An object is imaged
by interposing it between the nozzle and the scintillator and scanning a proton
pencil beam across the object. The camera acquires one image per proton pencil beam.
The intensity of each camera pixel is converted to a proton WET via the
calibration curves (Fig. 1d). Several proton beams may contribute to a single
pixel. The WET for each pixel is determined by weighting all pixel
contributions by the fraction of the peak intensity of the reference pencil
beam. Proton scattering is decreased through this weighting process and by
rejecting pixel contributions whose baseline intensity is below 50% of the peak
intensity of the reference pencil beam.
A Monte
Carlo simulation of the detector was implemented in Geant4, including an
aluminum “Las Vegas” contrast-detail phantom with a WET of 47.7 mm. Proton
radiographs were reconstructed via the PWT method, by integral proton
radiography without spot-by-spot PWT processing, and also by a simulated single
particle tracking detector comprising two tracking planes and an energy
calorimeter.
Results
Image contrast increased
with the PWT method relative to integral imaging
without PWT. Using a contrast to noise
ratio of 3.0 for the detection threshold, 18, 22, and 30 phantom holes were
detectable with the integral, PWT, and single particle tracking image formation
methods, respectively (Fig. 2a-c). PWT imaging decreased scattering artifacts
at the phantom edge by 77% compared to integral imaging (Fig. 2d-e). The WET measurement
error was 14% with integral imaging and <1% with the PWT method.
Conclusion
Single pencil beam
imaging with pixel weighting and intensity thresholding provides improved image
quality relative to standard integrating proton radiography. Some scattering
artifacts persist with this approach, and further work on scatter reduction is needed.