The optimization and inherent limitations of 3D conformal radiotherapy treatment plans of the prostate.


Journal Article

This paper describes the applications of an inverse planning optimization algorithm to the real clinical problem of prostate cancer. The algorithm has been designed to compute optimized beam-weights taking full account of three-dimensional spatial information of dose inside the patient. The algorithm is based on fast simulated annealing, utilizing a cost-function containing both linear and quadratic terms. The linear part of the cost function allows for the implementation of "short-cuts" in the cost-function computation, which reduces the calculation time by a factor of about 30. It has been applied to compute optimized beam-weights for a three-field and a seven-field prostate treatment plan. It is shown for the three-field plan that the optimization algorithm can reproduce, and even slightly improve on, the results of an experienced human planner. For the seven-field plan, the human planner experienced difficulty finding beam-weights that gave an acceptable dose distribution. It is shown that the optimization algorithm can achieve good results in this case. The outcome of the optimization of the seven-field plan prompted an investigation into the best results that could be achieved by an "ideal" conformal radiotherapy technique. The results of this investigation are presented and it is shown that the limiting factor for conformal therapy of the prostate is the size of the overlap volume between the planning target volume (PTV) and the rectum. Finally, the efficiency and accuracy of fast simulated annealing is compared with that of classical simulated annealing. The former was found to be at least 10 times faster for the problem studied.

Full Text

Duke Authors

Cited Authors

  • Oldham, M; Webb, S

Published Date

  • August 1995

Published In

Volume / Issue

  • 68 / 812

Start / End Page

  • 882 - 893

PubMed ID

  • 7551787

Pubmed Central ID

  • 7551787

International Standard Serial Number (ISSN)

  • 0007-1285

Digital Object Identifier (DOI)

  • 10.1259/0007-1285-68-812-882


  • eng

Conference Location

  • England