A simplified and effective off-axis Winston-Lutz for single-isocenter multi-target SRS.

Abstract

To safely perform single-iso multi-target (SIMT) stereotactic radiosurgery (SRS), clinics must demonstrate SRS delivery accuracy for off-axis targets. The traditional Winston-Lutz (W-L) was widely adopted because it provides a simple and accurate solution for testing radiation-isocenter coincidence that uses a static target, enables testing arbitrary treatment angles, and does not require expensive commercial phantoms. The current noncommercial tests are cumbersome and insufficiently accurate. For an off-axis Winston-Lutz (OAWL) test, one must design MLC fields centered on off-axis targets. Unfortunately, because MLC leaf-interfaces are often misaligned with the target center, accomplishing this presents a nontrivial geometry problem that has not been previously solved in the literature. We present a solution for evaluating SIMT SRS accuracy that provides a straightforward method for creating OAWL test fields and offers all the benefits of the standard W-L test. We have developed a method to use any gantry, table, and initial collimator angles to create OAWL fields. This method calculates a series of nested coordinate transformations that produce a small collimator angle adjustment to align the MLC and create a symmetric field around an off-axis target. For an 8cm off-axis target, the described method yields OAWL results within 0.07mm of standard isocentric W-L results. Our six most recent isocentric W-L tests show max and mean errors of 0.59 and 0.37mm, respectively. For six runs of our proposed OAWL test, the average max and mean errors are 0.66 and 0.40mm, respectively. This method accurately evaluates SRS delivery accuracy for off-axis distances that span the majority of a typical human brain for a centered SIMT arc. We have made this method publicly available, so that physicists can employ it within their clinics, foregoing the need for expensive phantoms and improving access to the state-of-the-art SIMT SRS technique.