The Use of Established Methods to Quantify Proton Range Uncertainty Reduction When Using Proton Tomography

Abstract

The dosimetric advantage of treatment with proton beam therapy shows tremendous clinical promise. In proton treatment planning, kV CT images are converted using the stoichiometric method to obtain a model of the relative stopping powers (RSP) of the patient’s tissues. This process has inherent errors in accurately predicting the RSP values within the patient. To ensure target coverage in the presence of these uncertainties, additional margins are added to a proton field resulting in unwanted dose to healthy tissues. Several studies have systematically evaluated the magnitude of range calculation uncertainties when using kV CT imaging, with each yielding similar recommendations. In this study, we used the identical evaluation methods to calculate the potential range uncertainty if the patient model was obtained via direct proton transmission tomography (pCT) rather than kV X-ray imaging. pCT images of a phantom designed with materials of known RSP were obtained and reconstructed using a most likely path length method. The accuracy of the reconstructed images in predicting the RSP varied from 0.1% up to 0.8% for different materials. Using these uncertainties in the pCT images, evaluations for uncertainty margins were made using the identical method as in prior studies. An attempt was made to quantify potential reductions in range uncertainty in a case where pCT is used for the patient model. Table 1 summarizes the potential reduction in range uncertainty margins if pCT images were used to obtain the RSP distributions and compared to RSP distributions obtained using the stoichiometric method from prior studies. Results show that if the error evaluation methods used in the prior studies are re-evaluated using the reported errors of the pCT phantom images, considerable reductions in range uncertainty margins could be obtained.Abstract 3752Tissue TypeRange Error with kV CTRange Error with Proton CTConfidence LevelCommentsMoyers (2010)Soft Tissue3.5%1.4%2.0 SigmaCurrent TechnologySoft Tissue2.2%0.9%2.0 SigmaWith Dual Energy CT and MCYang (2012)Lung5.0%1.0%1.0 SigmaSoft Tissue1.6%0.5%1.0 SigmaBone2.4%0.6%1.0 SigmaPagganetti (2012)4.6%4.0%1.5 SigmaLocal Inhomogeneities2.4%1.4%1.5 SigmaLocal Inhomogeneities w/ MC Open table in a new tab The use of pCT images for modeling the RSP of proton therapy patients will considerably reduce the magnitude of the margins needed to account for proton range uncertainty. With the clinical implementation of more advanced proton delivery, especially multi field optimization methods, the reduction of range uncertainty has become even more critical. Further development in the clinical implementation of pCT in proton therapy departments would greatly benefit proton patients.