Comparison of algorithms for multileaf collimator field segmentation

In the "stop and shoot" method of intensity modulated radiation therapy, it is desirable to use an efficient multileaf collimator (MLC) field segmentation algorithm in the sense that it translates beam intensity maps into the least number of MLC field segments. In this work, we compare the performance of eight different algorithms, including the ones by Bortfeld et al., Galvin et al, Xia and Verhey, the Siemens IMFAST algorithm, and four other algorithms which have not been studied before. We find that the algorithm of Xia and Verhey is most frequently the algorithm that needs the least MLC field segments. However, no single algorithm is the most efficient for all clinical cases or intensity maps. This suggests that it is desirable to have multiple algorithms available in a clinical treatment planning system which will search through all algorithms automatically and find the most efficient delivery sequence for a given treatment. Each intensity map in a treatment could be delivered by a different algorithm, whichever is the most efficient for that map. It is pointed out that when the background intensity level is not zero, it is not always efficient to deliver a segment to bring the background level down to zero. (C) 1999 American Association of Physicists in Medicine. [S0094-2405(99)01911-2].