Sci‐Fri PM: Radiation Therapy, Planning, Imaging, and Special Techniques ‐ 01: On the use of proton radiography to reduce beam range uncertainties and improve patient positioning accuracy in proton therapy

Abstract

To present two related developments of proton radiography (pRad) to minimize range uncertainty in proton therapy. The first combines a pRad with an X‐ray CT to produce a patient‐specific relative stopping power (RSP) map. The second aims to improve the pRad spatial resolution for accurate registration prior to the first. The enhanced‐pRad can also be used in a novel proton‐CT reconstruction algorithm.Monte Carlo pRad were computed from three phantoms; the Gammex, the Catphan and an anthropomorphic head. An optimized cubic‐spline estimator derives the most likely path. The length crossed by the protons voxel‐by‐voxel was calculated by combining their estimated paths with the CT. The difference between the theoretical (length×RSP) and measured energy loss was minimized through a least squares optimization (LSO) algorithm yielding the RSP map. To increase pRad spatial resolution for registration with the CT, the phantom was discretized into voxels columns. The average column RSP was optimized to maximize the proton energy loss likelihood (MLE).Simulations showed precise RSP (<0.75%) for Gammex materials except low‐density lung (<1.2%). For the head, accurate RSP were obtained (µ=−0.10%1.5σ=1.12%) and the range precision was improved (ΔR80 of −0.20±0.35%). Spatial resolution was increased in pRad (2.75 to 6.71 lp/cm) and pCT from MLE‐enhanced pRad (2.83 to 5.86 lp/cm).The LSO decreases the range uncertainty (R80σ<1.0%) while the MLE‐enhanced pRad spatial resolution (+244%) and is a great candidate for pCT reconstruction.