ESTRO 2021 Abstract Book

S220

ESTRO 2021

Conclusion We propose a simple recipe to replace the PRV and correct planning dose constraints due to changes in RT uncertainties. First estimates show that published constraints can be relaxed with on average 1.3 Gy for current treatments. OC-0311 Integrating data envelopment analysis into radiotherapy treatment planning for head and neck cancer J. Simpson 1,2 , A. Raith 3 , F. Fauzi 4 , K. Lin 3 , A. Macann 5 , P. Rouse 6 , M. Ehrgott 7 1 Calvary Mater Hospital, Radiation Oncology, Newcastle, Australia; 2 University of Newcastle, School of Physics and Mathematics, Newcastle, Australia; 3 University of Auckland, Department of Engineering Science, Auckland, New Zealand; 4 The National University of Malaysia, Faculty of Information Science and Technology, Bangi Selangor, Malaysia; 5 Auckland City Hospital, Department of Radiation Oncology, Auckland, New Zealand; 6 University of Auckland, Department of Accounting and Finance, Auckland, New Zealand; 7 Lancaster University Management School, Department of Management Science, Lancaster, United Kingdom Purpose or Objective We describe how a productivity method, Data Envelopment Analysis (DEA), is used as a real-time decision support tool to assess quality of treatment plans for head and neck (H&N) cancer patients. Materials and Methods DEA is a nonparametric method used widely in operations research for the estimation of best-practice frontiers. A benchmark library was established from 81 previously treated H&N plans using a DEA model for OAR and PTV doses. Patient geometry was considered using the overlap volume histogram for the two principal volumes that relate to dysphagia and mucositis (swallowing structures and oral cavity). Other OAR were considered as a single input derived from the empirical cumulative distribution function of dose violations from the library of plans. DEA determines a multi-dimensional best practice frontier against which any plan can be compared. A DEA score is generated for each plan as a measure of its distance from the best practice frontier. In order to assess the utility of this approach, five randomly selected plans from the benchmark library were re-planned in several iterations, each iteration accessing the DEA tool integrated within a Raystation planning system. Plan quality was independently assessed by an expert radiation oncologist. Various plots as shown in Figure 1, allow comparison of a current plan (red cross) and its target (green dot) to other plans in the benchmark library (histogram). DEA target values are shown to the planner (bottom screenshot).