BACKGROUND: Proton beam therapy offers comparable tumor control to conventional photon therapy while reducing radiation-related toxicity due to its finite range and minimal exit dose beyond the Bragg peak. However, the accuracy of proton dose delivery is highly dependent on precise knowledge of tissue stopping power along the beam path. Small deviations in tissue composition or density can result in clinically meaningful proton range errors, potentially compromising target coverage and normal tissue sparing. A primary contributor to this uncertainty is the conversion of computed tomography (CT) Hounsfield Units (HU) to relative stopping power using a bilinear calibration curve, which provides only an approximation of patient tissue properties. Although robustness optimization strategies in modern spot-scanning proton therapy help mitigate the dosimetric impact of range uncertainty, uncertainties in HU-to-stopping-power calibration remain a fundamental limitation of CT-based proton treatment planning. Consequently, there is an ongoing need for practical, efficient methods to verify and validate bilinear calibration curves under clinical conditions, ensuring accurate modeling of proton range and distal dose falloff while maintaining compatibility with routine quality assurance workflows. PURPOSE: To present an end-to-end test for verifying proton range calculations derived from a recently established CT bilinear calibration curve, using animal tissues and standard QA tools. This study provides a robust framework for preclinical verification of proton dose calculations in complex, heterogeneous tissues and supports the safe delivery of proton therapy. METHODS: PMMA container was filled with combinations of animal muscle, adipose, and bone tissues to simulate heterogeneous anatomy. Proton treatment plans were optimized with RayStation's Monte Carlo (MC) algorithm to deliver a uniform, single-fraction dose of 200 cGy(RBE). Range accuracy was assessed by measuring 2D planar dose distributions at multiple depths with an ionization chamber array. Proton range parameters including R90, R80, R50, and R20 were determined from measured depth-dose curves and compared with the range parameters calculated by both MC and Pencil Beam Convolution Superposition (PBCS) algorithms. RESULTS: Measured and calculated range parameters showed excellent agreement across all tissue configurations. Regarding R80, for the first three scenarios, percentage differences were noticeably minimal: 0.24% (MC) and 0.16% (PBCS) for a water-equivalent thickness (WET) of 18.3 cm (muscle/adipose); 0.47% (MC) and 0.47% (PBCS) for 10.7 cm (muscle/adipose) plus 3.5 cm (bone); and 0.65% (MC) and 0.56% (PBCS) for 9.4 cm (muscle/adipose) plus 4.7 cm (bone). Even in the most challenging configuration, with 3 cm of lateral heterogeneity (bone intersecting muscle+adipose) plus 14.3 cm of muscle/adipose, differences remained low at 1.21% (MC) and 1.08% (PBCS). CONCLUSIONS: This end-to-end method provides an effective approach for verifying proton range calculations in heterogeneous tissues, supporting accurate and robust treatment planning. This approach is especially valuable when implementing proton treatment planning systems, providing increased confidence in the reliability of proton therapy delivery.
Boria et al. (Fri,) studied this question.