Determination of HU override for gold fiducial markers in proton therapy.

backgroundIn radiotherapy, it is not uncommon to treat patients with metallic implants. Those implants often cause significant amount of artifacts on the simulation CT images, and the intrinsic Hounsfield unit (HU) values cannot be directly used as those of the implant without appropriate evaluations. The alternative solution is to delineate the implant and assign certain HU values to the delineated volume to represent the implant. However, the artifacts make it almost impossible to precisely delineate the sizes and shapes of the metallic implants on the CT images, especially for implants with irregular shapes. It is almost always the case that the delineated "implant" contains both implant itself and adjacent tissues. If the inaccurately delineated implant is considered as the pure implant material, dose calculation around the implant is not accurate, leading to distorted dose distribution (from the real) and misguided treatment plan, especially in proton therapy. PURPOSE: This study presents a generalizable method for properly calculating the HU override of small metal implants for proton therapy and evaluated the dosimetric effects of this method as applied to gold fiducial markers in a prostate proton treatment plan.

methodsThe HU value for gold fiducial markers was calculated using a mass-weighting approach to determine the relative stopping power of the delineated fiducial volumes, taking into account the amounts of gold and tissue within the volumes. This stopping power was used to select a corresponding HU number from the CT calibration curve for dose calculation. To evaluate the effects of this HU override on dosimetry calculation for proton therapy, dose distributions were calculated for a prostate plan with gold fiducial markers in four different ways: with the fiducial contours overridden using the method described in this work, with no override using the intrinsic HU values across the CT images, with delineated fiducial structure overridden based on the assumption that the entire contoured volume was gold, and with modeled fiducial structure which matched as the vendor's design inserted onto the CT images. The fourth method was considered as the closest to reality and as the reference. Further investigations were also conducted to determine the validity of the method and its robustness to inter-user contouring variation.

resultsCompared to the reference, the non-overridden and overridden delineated fiducial volumes with pure gold material resulted in inaccurate distortions of the isodose lines in and around the target and considerable unreal reductions in target coverage, while the tissue weighted approach presented in this study yielded the closest resemblance to the reference. This method was also shown to be robust to delineation variations caused by different users.

conclusionA method is proposed and validated for tackling the dose calculation challenges imposed with the presence of metallic implants inside patients during radiotherapy planning process. This developed method has the potential to be generalized for other types of metallic materials.