A patch source model for treatment planning of ruthenium ophthalmic applicators

Beta-ray emitting Ru-106/Rh-106 ophthalmic applicators have been used for close to 4 decades in the treatment of choroidal melanoma. The form factor of these applicators is a spherically concave silver bowl with an inner radius of curvature between 12 and 14 mm, and a total shell thickness of 1 mm. The radioactive nuclide is deposited in a layer 0.1 mm below the concave surface of the applicator. Calculation of dose distributions for clinical treatment planning purposes is complicated by the concave nature of the distributed source, the asymmetric shape of the active region of some applicators, imperfections in the manufacturing process which can result in an inhomogeneous distribution of activity across the active surface, and absorption and scatter in the 0.1 mm layer of silver which seals and protects the radioactive layer. A semi-empirical method of calculating dose distributions for these applicators is described which is fundamentally compatible with treatment planning systems that use the AAPM TG43 brachytherapy formalism. Dose to water is estimated by summing a "patch source" dose function over a discrete number of overlapping patches uniformly distributed over the active surface of the applicator. The patch source dose function differs conceptually from a point source dose function in that it is intended to represent the macroscopic behavior of a small, disk-like region of the applicator. The patch source dose function includes an anisotropy term to account for angular variation in absorption and scatter as particles traverse the 0.1 mm silver window. It geometrically models the nearfield of a patch with properties akin to both a small disk and infinite plane, and models the farfield as if the patch were a point. This allows a manageable number of discrete patches (300 to 1000) to provide accuracy appropriate for clinical treatment planning. This approach has the advantages of using familiar concepts and data structures, it is computationally quick, and it readily adapts to asymmetric applicator shapes and inhomogeneities in the radionuclide distribution. A method for optimizing the patch source dose function parameters is presented, and the dosimetric calculations are compared with published Monte Carlo calculations and measurements.