The results demonstrate that our mathematical framework and modest training cohorts successfully predict achievable OAR DVHs based on individual patient anatomy. The models correctly identified suboptimal plans that demonstrated further OAR sparing after replanning. This modeling technique requires no manual intervention except for appropriate selection of a training set with identical evaluation criteria. Clinical implementation is in progress to evaluate impact on real-time IMRT QC.

No abstract

No abstract

Purpose: The purpose of this work was to investigate the impact of a predictive DVH (pDVH) model developed at one institution on IMRT plan quality control (QC) at an unrelated radiotherapy facility. Methods: DICOM‐RT datasets for twenty randomly selected intact prostate cancer patients treated at a small clinic were analyzed with knowledge based pDVH models developed at a larger institution. Previously validated rectum and bladder pDVH models used in this study were derived based on the correlation of expected dose to the distance from a voxel to the PTV surface. A sum of residuals (SR) metric quantifying the integrated difference between clinical DVHs and the pDVHs was used to rank the 20 evaluated plans. The 5 plans with greatest deviation from the prediction were replanned by the small clinic to evaluate the ability to achieve rectum and bladder DVH predictions while maintaining PTV quality metrics per institutional standard. Metrics used to evaluate plan quality (V65 and V40) quantified the clinical gains in the rectum and bladder DVHs. The pDVHs were compared to the replan DVHs using dV65=V65(replan)‐V65(pred), dV40=V40(replan) ‐ V40(pred), and mean SR to measure DVH prediction accuracy. Results: The significance of IMRT QC using pDVHs was demonstrated by an average reduction in V65 and V40 in the 5 replanned patients of 4.8% ± 2.3% and 17.9% ± 10.3% for the rectum and 3.4% ± 2.1% and 6.0% ± 2.8% for the bladder, respectively. The pDVH models demonstrated excellent prediction accuracy with an average dV65 and dV40 of 0.9% ± 1.1% and 0.7% ± 1.4% for the rectum and 0.4% ± 0.5% and 0.6% ± 0.9% for the bladder, respectively. Conclusion: DICOM‐based pDVH modeling methods based on patient geometry accurately predict achievable rectum and bladder DVH parameters that are clinically relevant and may facilitate improved IMRT plan quality across multiple institutions. L.Appenzoller, S. Mutic, K. Moore: Patent Application # 13, 486,809 : Developing Predictive Dose‐Volume Relationships for a Radiotherapy Treatment.

Purpose: The goals are to develop a computer system that performs an automatic and comprehensive plan quality check (QC) according to institutional guidelines, QUANTEC, and RTOG standards, and to standardize and streamline the evaluation, approval, and review of radiotherapy treatment plans. Methods: A computer software program (and backend database) was developed to automatically 1) read the plan data directly from the TPS, 2) process data, 3) match ROI names, 4) compute DVHs, 5) compute plan quality metrics from the DVHs, 6) evaluate the metrics against the QC rules, and 7) create a report. A rule defines a plan metric that should be satisfied, e.g. V95%>95% for PTV. Rules are organized in groups, and multiple groups are defined for each disease site corresponding to different prescriptions or treatment modalities. On‐the‐fly management of groups and rules is implemented to allow creating, editing, copying, and deleting. The tool can create new isodose structures and composite ROIs on‐the‐fly to support comprehensive yet flexible rules. The tool supports a number of plan quality metrics based on DVH or volumetric data and checks complex metrics including hot spots and conformity ratio. OARs that are not covered by the rules are automatically checked against standard OAR tolerances defined in QUANTEC or RTOG protocols. The report is presented to users with multiple tabulated forms containing color coded results and references. Results: This plan QC tool has been successfully tested. The computed DVH and plan quality metrics were verified to be within 0.5% of the TPS computation. It takes less than 1 minute to load data from a TPS, check all the rules, and generate a report, demonstrating the efficiency of this tool. Conclusion: This software program efficiently checks plan quality and is useful for assisting dosimetrists/physicists in treatment planning and physicians in plan evaluation, approval, and chart review.