Proton CT on biological phantoms for x-ray CT calibration in proton treatment planning

Abstract

Abstract Objective. To present and characterize a novel method for x-ray computed tomography (xCT) calibration in proton treatment planning, based on proton CT (pCT) measurements on biological phantoms. Approach. A pCT apparatus was used to perform direct measurements of 3D stopping power relative to water (SPR) maps on stabilized, biological phantoms. Two single-energy xCT calibration curves—i.e. tissue substitutes and stoichiometric—were compared to pCT data. Moreover, a new calibration method based on these data was proposed, and verified against intra- and inter-species variability, dependence on stabilization, beam-hardening conditions, and analysis procedures. Main results. Biological phantoms were verified to be stable in time, with a dependence on temperature conditions, especially in the fat region: (−2.5 土 0.5) HU °C−1. The pCT measurements were compared with standard xCT calibrations, revealing an average SPR discrepancy within ±1.60% for both fat and muscle regions. In the bone region the xCT calibrations overestimated the pCT-measured SPR of the phantom, with a maximum discrepancy of about +3%. As a result, a new cross-calibration curve was directly extracted from the pCT data. Overall, the SPR uncertainty margin associated with this curve was below 3%; fluctuations in the uncertainty values were observed across the HU range. Cross-calibration curves obtained with phantoms made of different animal species and anatomical parts were reproducible with SPR discrepancies within 3%. Moreover, the stabilization procedure did not affect the resulting curve within a 2.2% SPR deviation. Finally, the cross-calibration curve was affected by the beam-hardening conditions on xCTs, especially in the bone region, while dependencies below 2% resulted from the image registration procedure. Significance. Our results showed that pCT measurements on biological phantoms may provide an accurate method for the verification of current xCT calibrations and may represent a tool for the implementation of a new calibration method for proton treatment planning.