External Beam Radiotherapy (EBRT) of Intraocular Tumors

Introduction

Ocular anatomy and radiation-associated toxicities provide unique challenges for the treatment of intra-ocular tumors with radiation therapy (RT). Tumor control is the primary goal, but side effects such as dry eye, eyelash loss, cataract, neovascular glaucoma, radiation retinopathy, and optic neuropathy are potential local complications of ocular RT. Over the last 2 decades, several new RT delivery techniques like stereotactic fractioned RT, intensity modulated RT, and proton beam RT, have been developed that provide a more conformal radiation dose distribution to the tumor while minimizing radiation exposure to the surrounding healthy tissue in order to decrease side effects. Many ocular neoplastic diseases respond favorably to RT, which plays an important role as both primary treatment and adjuvant therapy. External beam RT (EBRT) has a curative potential in children treated for retinoblastoma. Uveal melanoma is the most common primary eye cancer in adults and is associated with significant morbidity and mortality. Consequently, uveal melanoma requires sharply delimited irradiation treatment to achieve the necessary high tumor dose while sparing healthy tissue. To fulfill these requirements, heavy particle radiation techniques (proton and helium ion beam) have been applied with a control rate of 96 % at 5 years. However, regardless of which form of EBRT is applied, treatment planning is a prerequisite for all of them. High-accuracy treatment planning is used to optimally spare the structures at risk while respecting the safety margins for the target coverage. Structures at risk are optic nerve, macula, ciliary body, and lens. Recently, there has been a trend toward the treatment of smaller melanomas, leading to an increased demand for reliable high-accuracy treatment planning.

State of the Art Planning and Treatment

For treatment planning, precise modeling of organs at risk and tumor volume is crucial. The development of a precise eye model as well as the adaptation of this model to the patient’s anatomy remains problematic because of organ shape variability. Today, to support the planning procedure and to enable referencing of the tumor during treatment, several tantalum clips are sutured onto the patient's sclera in a preceeding clip surgery. State of the art treatment planning is then based on a parametric eye model (spherical or elliptical model), which is adapted to the patient's anatomy based on distance values extracted from CT, MRI or Ultrasound images. During treatment, the established treatment plan is registered with the patient's moving eye using X-ray projections of the diseased eye in order to detect the positions of the sutured clips.

Objectivs

The objectives of this project are twofold, first to improve state of the treatment planning using robust segmentation and registration methods. Second, to develop a non-invasive tumor referencing technique, which allows for a clip-free planning and treatment of intra-ocular tumors.

Obviously, a clip-free planning and treatment would bring huge benefits to the patients as well as to the radiation oncologists.

Clip-Free Treatment Planning

This project introduces the first steps toward a multimodal planning system for the clip-free treatment of intraocular tumors. We present the development and application of a 3D statistical shape model (SSM) as a novel method for precise eye modeling for external beam radiotherapy of intraocular tumors.

http://www.youtube.com/watch?v=neTS8XJMOJA

Active Shape Model (ASM)

  • Automatic fitting of SSM based on intensity profiles extracted perpendicular to the model surface.
  • Building of intensity model profiles by the computation of the mean profile and covariance matrix for each profile point over all training shapes.
Fig. 3. Iterative model fitting scheme: Search profiles are extracted perpendicular to the model surface in 3D and matching points are computed by minimizing the mahalanobis distance to the model profiles.
Fig. 4. A 3D model of the retina is established by mapping a patient specific fundus image onto the model surface.

Validation

  • Measurement of the performance of the SSM and ASM based on a leave-one-out cross validation.
  • Result: average fitting error of 0.3 mm, which is in the order of the CT in-plane resoultion of 0.39 mm.
Fig. 5. Distribution of mean model fitting error over model surface. The mean fitting error was determined for 4 quadrants (Q1-Q4) within 4 layers (cornea, lens, sclera front and sclera back).

Publications

M. B. Rüegsegger et al., "Statistical Modeling of the Eye for Multimodal Treatment Planning for External Beam Radiation Therapy of Intraocular Tumors." Int J Radiation Oncol Biol Phys, Vol. 84, No. 4, 2012
M. B. Rüegsegger et al., "Statistical modeling of the eye for multimodal treatment planning for external beam radiotherapy of intraocular tumors." Strahlentherapie Und Onkologie, Volume 188,  2012
M. B. Rüegsegger et al., "Precise Modeling of the Eye for Proton Beam Radiotherapy of Intraocular Tumors." Invest. Ophthalmol. Vis. Sci. 53, 2012