Optimizing computerized treatment planning for the gamma knife by source culling

Purpose: A good plan is crucial to the success of gamma knife treatment, wh ich depends not only on parameters such as the number of shots, shot positi on, collimator sizes, and shot weight, but also on the number of blocked co balt sources. However, during treatment, a plug is generally used to block those cobalt sources, so the beam cannot reach critical tissues. We present here an automated method to optimize all of those parameters, and to choos e a source set, although the beams of some blocked sources do not hit any c ritical tissue. This strategy is used to achieve a high dose that better co nforms to the tumor shape, and at the same time, avoids healthy tissue. Methods and Materials: Using a workstation that integrates the gamma knife treatment planning system, we developed a two-step optimization algorithm. First, we used a modified Powell's method to optimize the location of the s hot, collimator size, and shot weight; we used simulated annealing to deter mine if the number of shots was adequate using this parameter. Then, simula ted annealing was used to determine which cobalt sources we needed to block . Results: Application of this optimization method in two cases showed that t he treatment plan can be much improved when the set of blocked cobalt sourc es has been taken into consideration. Conclusion: Determining the set of blocked sources is necessary in certain cases. This technique better conforms the desired isodose curves to the out line of the target volume and minimizes damage to the surrounding normal ti ssues. (C) 1999 Elsevier Science Inc.