Accuracy of day-to-day patient positioning in ocular proton therapy with a dedicated beamline

The purpose of this study was to determine the residual patient positioning error of ocular proton treatment based on daily setup X-ray images and tantalum clips. The residual day-to-day patient positioning accuracy in a cohort of 60 patients and 240 fractions was determined in two steps. The 3D tantalum clips' positions in the orthogonal X-ray images just prior to aperture placement were registered with 3D clips' positions derived from the reference data from treatment planning. The aperture, placed before irradiation, reduced the field of view in axial images blocking clips partially or completely in most cases. The clips' positions after aperture placement in the lateral images were compared with the reference clips' positions. Partially visible clips in the axial image were registered to the clips in the axial image prior to aperture placement. The day-to-day deviations to aperture placement were corrected with the registration results for translations only. The residual day-to-day patient position variations just prior to irradiation were -0.06 ± 0.27 mm (mean ± 1SD) for the left-right direction, -0.01 ± 0.23 mm for the anterior-posterior direction and -0.01 ± 0.19 mm for the superior-inferior direction. This data was further investigated by differentiating for treated eye, gazing eye and the use of eyelid retractor. This yielded no significant differences. The residual patient positioning errors are small for patients treated at the dedicated eyeline, indicating that the patient positioning workflow using tantalum clips is highly accurate. The reported errors can be used for ocular proton margin calculations and for the evaluation of a clipless ocular proton therapy workflow.