AbstractID: 9494 Title: A graph searching method for MLC leaf sequencing under
constraints
We have developed a new leaf sequencing method for step-and-shoot IMRT delivery
based on a graph searching technique. Variable intensity levels and two collimator
angles with a 90º rotation are used for minimizing the number of segments. In an
iterative process, one segment is selected at a step as guided by a quantitative complexity
measure for the remaining intensity pattern. Under leaf position constraints disallowing
leaf interdigitations, we first assume that deliverable field segments form simply
connected areas. For each intensity level, the field area with equal or larger intensity is
segmented by graph searching into the least number of simply connected fields with each
of the two MLC directions. The segment that results in the most reduced complexity of
the remaining intensity distribution is selected. For further simplification of the
remaining intensity pattern, the shapes of selected segments are modified by leaving out
certain beamlets for later delivery with out changing the deliverability of current
segment. The deliverable segments are then combined under appropriate constraints for a
specific type of MLC. Finally, combined segments are sequenced to minimize the leaf
travel. Our algorithm has been tested using randomly generated intensity distributions
and clinical cases for the Varian, Siemens, and Elekta MLC systems. The segment
number increases only moderately while the MU numbers almost unchanged with
increasing the number of intensity levels. Using two collimator angles can considerably
decrease the number of segments and thus advantageous if the beam delivery can be
automated.