A maximum-entropy method for the planning of conformal radiotherapy

The maximum entropy method (MEM) is a powerful inverse analysis technique t hat is used in many fields of science and engineering to perform tasks such as image reconstruction and processing of nuclear magnetic resonance signa ls. Unlike other methods, MEM naturally incorporates a priori knowledge of the problem into the optimized cost function. This feature is especially im portant in radiotherapy planning, because some knowledge is usually availab le about the stage of tumor development and about the prescription doses, i ncluding some dose constraints to the surrounding normal organs. Inverse pl anning is inherently consistent with the ability of MEM to estimate paramet ers inversely. In this investigation, an entropy function determines the ho mogeneity of dose distribution in the planning target volume; a least-squar es function is added to the maximum entropy function as a constraint to mea sure the quality of reconstructed doses in organs at risk; and an iterative Newton-Ralphson algorithm searches for the optimization solution. Here we provide two examples that validate this application of MEM and the results were compared with manual plans. Although the examples involve conformal ra diotherapy, we think MEM can be adopted to optimize intensity-modulated rad iation therapy. (C) 2001 American Association of Physicists in Medicine.