Optimization of multiple-isocenter treatment planning for linac-based stereotactic radiosurgery

Objective: Computer-assisted treatment planning for linac-based radiosurgery is still an open research problem, especially for multiple-isocenter procedures, primarily due to its high complexity and computational requirements. This paper focuses on the optimization of multiple-isocenter treatment planning for linac systems, and addresses several important issues associated with multiple isocenters, such as dose conformality, homogeneity, and optimization of isocenter position and dose. Methods: The key idea behind our approach is that the desired dose distribution can be decomposed into a number of fundamental components. In the current paper, an analytical form, the so-called Ellipsoidal Dose Distribution Estimation (EDDE) model, represents each component. We establish ways (arc configurations) to achieve such ellipsoidal doses of arbitrary position, orientation, and size. Since the EDDE model is described by relatively few parameters, it allows very quick estimation of the dose distribution corresponding to a particular isocenter and thus makes the optimization of isocenter position very efficient. It is further used in a framework for optimal treatment planning, in which a number of ellipsoidal dose distributions, each corresponding to a different isocenter, are optimally placed to cover the target while sparing healthy tissue. Results: The general ellipsoidal dose distribution of linac-based radiosurgery is summarized as a mathematical model with the aid of supporting experiments. Comparisons between the EDDE-optimized and clinically implemented plans are made, revealing the superior performance of the former. In addition, a dramatic reduction in planning time is achieved using the EDDE model. Conclusion: The proposed EDDE model is a useful and effective dose model in multiple-isocenter treatment planning for linac-based radiosurgery. (C) 2000 Wiley-Liss, Inc.