EyeDose: An Open Source Tool For Calculating The Radiation Dose Delivered To The Tumor And Critical Ocular Structures For 125I COMS Eye Plaques

Abstract

Radiation side effects and visual outcome for uveal melanoma patients managed with plaque radiotherapy are dependent on the radiation dose administered to the tumor and nearby healthy tissues. The AAPM Task Group Report 129 recommends that the dose to critical normal structures be recorded, preferably using heterogeneity corrected dose distributions. We have developed an open source software tool, EyeDose, to simplify and standardize tumor and critical structure dose reporting for COMS eye plaques. This work builds upon the accomplishments of previous researchers and aims to combine the best features of prior tools, including the accuracy of Monte Carlo, the elimination of manual contouring and the treatment planning system, the standardization of eye geometry, and precise plaque localization using high resolution fundus diagrams. The software may be used for prospective clinical decision making, for example to help with plaque size selection or a comparison of plaque radiotherapy versus proton therapy, or to rapidly process large retrospective data sets, without repetitive manual entry, for reporting on the relationships between dosimetry and outcomes. EyeDose is a Matlab program that calculates point and volume (DVH) dose metrics for the tumor and critical ocular structures. The three-dimensional dose distributions for 125I Collaborative Ocular Melanoma Study (COMS) eye plaques, calculated with Monte Carlo methods, are oriented with respect to the eye using the tumor’s position on a fundus diagram. A standard model for the ocular structures was created using published measurements and patient CT scans. EyeDose reports radiation statistics for the fovea, optic disc, lens, lacrimal gland, retina, and tumor. The dosimetric margin for implant placement uncertainty is also calculated. EyeDose calculations were validated against published Monte Carlo results for nasal, temporal, inferior, and superior tumor positions, including the dose to the fovea, optic disc, lacrimal gland, lens, and along the central axis. The EyeDose program demonstrated close agreement with published Monte Carlo results. The mean, median, and maximum point dose differences between EyeDose and previous Monte Carlo studies were 0.0, 0.0, and 0.5 Gy. The mean difference, reported as %volume, among the DVH calculations was -0.2% ± 0.2% (Std. Dev.) with a range from -1.08% to 0.05%. The execution time averaged <40 sec. per patient on an Intel i-5-6500 CPU @ 3.20 GHz with 32 GB RAM. EyeDose is published as an open source tool for easy adaptation at different institutions. EyeDose calculates radiation statistics for COMS eye plaque patients with Monte Carlo accuracy and without a treatment planning system. EyeDose streamlines data collection for large retrospective studies and can also be used prospectively to assess plaque applicability.