The aim of this study is to provide a practical method to quantify the dosimetric effects on organs-at-risk (OARs) due to systematic uncertainties in linear accelerator treatment delivery in order to aid external beam treatment planning and raise warnings about additional risks to critical OARs.Methods
A dose approximation method, based on geometrical transformations, was developed to automatically estimate perturbations to dose volumes arising from five important potential uncertainties at the time of treatment delivery, including (a) systematic isocenter misalignment between image guidance and beam delivery systems, and systematic errors in, (b) collimator, (c) gantry, (d) couch table, and (e) multi-leaf collimator (MLC) leaf bank positions. The agreement between the estimated dose volume using the dose approximation method and the re-calculated dose volume obtained from the treatment planning system (TPS) was verified using a dose difference test (2% threshold and 0 mm distance-to-agreement). For each type of uncertainty, the worst-case maximal dose values to the most critical OARs (brainstem, chiasm, optic nerves, and spinal cord) were quantitatively evaluated, and compared with the maximal dose values to the corresponding OARs from clinical plans.Results
Six brain and six T-spine IMRT plans were used for evaluation. The average passing rates of 2% dose difference test were calculated to be 98.9% ± 1.3% for the uncertainties considered in this paper. The average time per patient to automatically analyze the dosimetric effects of all systematic uncertainties is 5.8 s. The worst-case scenarios for each plan, i.e., the largest changes in maximal doses to the OARs, were identified and confirmed to be in agreement with those calculated using the TPS.Conclusion
For a given external beam plan, the proposed dose approximation method allows efficient evaluation of the dosimetric effects of potential patient positioning uncertainties and systematic machine delivery errors on maximal dose to critical OARs. While the same uncertainties can be manually analyzed using the TPS, the proposed method is automatic and computationally inexpensive, and therefore significantly more practical. The proposed method could be useful to provide insights about otherwise unquantified risks and plan robustness during the stage of treatment planning.