A protocol for the reduction of systematic patient setup errors with minimal portal imaging workload.

Abstract

To evaluate a new off-line patient setup correction protocol that minimizes the required number of portal images and perform a comparison with currently applied protocols. We compared two types of off-line protocols: (a) the widely applied shrinking action level (SAL) protocol, in which the setup error, averaged over the measured treatment fractions, is compared with a threshold that decreases with the number of measurements, to decide if a correction is necessary; and (b) a new "no-action-level" (NAL) protocol, which simply calculates the mean setup error over a fixed number of fractions, and always corrects for it. The performance of the protocols was evaluated by applying them to (a) a database of measured setup errors from 600 prostate patients (with, on average, 10 imaged fractions/patient) and (b) Monte Carlo-generated setup error distributions for various values of the population systematic and random errors. The NAL protocol achieved a significantly higher accuracy than the SAL protocol for a similar workload in terms of image acquisition and analysis, as well as in setup corrections. The SAL protocol required approximately three times more images than the NAL protocol to obtain the same reduction of systematic errors. Application of the NAL protocol to measured setup errors confirmed its efficacy in systematic error reduction in a real patient population. The NAL protocol performed much more efficiently than the SAL protocol for both actually measured and simulated setup data. The resulting decrease in required portal images not only reduces workload, but also dose to healthy tissue, if dedicated large fields are required for portal imaging (double exposure).