Moving from gamma passing rates to patient DVH-based QA metrics in pretreatment dose QA.

PURPOSE

The purpose of this work is to explore the usefulness of the gamma passing rate metric for per-patient, pretreatment dose QA and to validate a novel patient-dose∕DVH-based method and its accuracy and correlation. Specifically, correlations between: (1) gamma passing rates for three 3D dosimeter detector geometries vs clinically relevant patient DVH-based metrics; (2) Gamma passing rates of whole patient dose grids vs DVH-based metrics, (3) gamma passing rates filtered by region of interest (ROI) vs DVH-based metrics, and (4) the capability of a novel software algorithm that estimates corrected patient Dose-DVH based on conventional phantom QA data are analyzed.

METHODS

Ninety six unique "imperfect" step-and-shoot IMRT plans were generated by applying four different types of errors on 24 clinical Head∕Neck patients. The 3D patient doses as well as the dose to a cylindrical QA phantom were then recalculated using an error-free beam model to serve as a simulated measurement for comparison. Resulting deviations to the planned vs simulated measured DVH-based metrics were generated, as were gamma passing rates for a variety of difference∕distance criteria covering: dose-in-phantom comparisons and dose-in-patient comparisons, with the in-patient results calculated both over the whole grid and per-ROI volume. Finally, patient dose and DVH were predicted using the conventional per-beam planar data as input into a commercial "planned dose perturbation" (PDP) algorithm, and the results of these predicted DVH-based metrics were compared to the known values.

RESULTS

A range of weak to moderate correlations were found between clinically relevant patient DVH metrics (CTV-D95, parotid D(mean), spinal cord D1cc, and larynx D(mean)) and both 3D detector and 3D patient gamma passing rate (3%∕3 mm, 2%∕2 mm) for dose-in-phantom along with dose-in-patient for both whole patient volume and filtered per-ROI. There was considerable scatter in the gamma passing rate vs DVH-based metric curves. However, for the same input data, the PDP estimates were in agreement with actual patient DVH results.

CONCLUSIONS

Gamma passing rate, even if calculated based on patient dose grids, has generally weak correlation to critical patient DVH errors. However, the PDP algorithm was shown to accurately predict the DVH impact using conventional planar QA results. Using patient-DVH-based metrics IMRT QA allows per-patient dose QA to be based on metrics that are both sensitive and specific. Further studies are now required to analyze new processes and action levels associated with DVH-based metrics to ensure effectiveness and practicality in the clinical setting.