Room: AAPM ePoster Library
Eye plaque brachytherapy commonly uses tumor models for treatment planning that are estimated to be ellipsoidal in shape, which approximate the actual tumor shape and distances to organs at risk (OAR). The purpose of this study is to dosimetrically compare such planning techniques to techniques based on contours drawn on fundus imaging.
Ocular melanoma patients are treated at Duke using Eye Physics plaques and planned to a de-escalated dose of 63Gy using the Plaque Simulator treatment planning system (Eye Physics, LLC). The plans use ellipsoidal tumor models based on surgeon hand-drawn fundus diagrams. For a subset of patients, an additional plan was retrospectively generated contouring the tumor on fundus images acquired with either the Clarus (Zeiss) or Optos (Optos) imaging systems. Tumor overall dimensions (radial and circumferential), distances from fovea and optic nerve, and several dosimetric parameters for tumor and normal tissues were extracted from both plans. Wilcoxon signed-rank tests were used to assess statistically significant differences (p<0.05).
Fundus images from 50 patients were used in this study, 20 imaged with Clarus, and 30 with Optos. Statistically significant differences were observed for tumor radial and circumferential dimensions (p=1.3e-04, p=4.4e-09), and fovea and optic nerve distances to tumor center (p=0.03, p=0.005). Dosimetrically, significance was observed for dose to optic nerve (p=0.02), opposite retina (p=0.008) and lens center (p=2.8e-04), and for the volume of the eye covered by 80% of the total dose (p=0.002). Meanwhile, the dose to tumor apex, sclera, and fovea, was shown to have no statistical significance (p=0.3, p=0.5, p=0.3).
Using fundus images to contour tumors can result in more conformal plans and the reduction of dose to some normal tissue, especially for practices that prescribe the COMS-standard of 85Gy. However, this type of planning is more time intensive and might require coordinate-based plaque insertion.