A two-shift optimisation of the 'no action level' setup correction protocol

Full text: As electronic portal imaging equipment becomes more common, many radiotherapy centres now have the ability to collect patient treatment position deviation values. One commonly used off-line set-up correction protocol for calculating patient setup corrections is the 'no action level' (NAL) protocol. This paper proposes a two-shift approach and calculates the number of images required for minimum systematic error. Patient data is used in a simulation to confirm this approach. Patient treatment position deviations were available for all treatment sessions for a large group of patients undergoing radiation therapy for prostate. Thirty of these patients were selected. The patient position at treatment and all isocentre shifts made were recorded in the treatment notes. These were used to simulate the effect of the NAL protocol using a range of image numbers as the basis of the set-up correction. As Bortfeld et al noted, there is an error minimum that can be observed beyond which the mean radial systematic set-up error increases slowly with an increase in the number of images used. An enhancement to the NAL was proposed in which the patient's position is corrected on two occasions; once early in the treatment schedule, and again after more images have been collected. The expectation value of the set-up error for this two-shift NAL was found and minimised. The optimum staging for the two-shift NAL for the prostate patients was to image for a total of 9 sessions and to shift the patient after 3 sessions and 9 sessions. The thirty patients showed an uncorrected mean radial setup error of 0.65cm. In this simulation this was corrected to 0.26cm by application of the NAL using 5 images and to 0.17 cm using the two shift NAL with shifts after three and nine images. In situations where staff can manage the workload of collecting and analysing portal images for nine sessions for each patient, the two-shift NAL will result in a high level of set-up accuracy. Copyright (2004) Australasian College of Physical Scientists and Engineers in Medicine