ESTRO 2021 Abstract Book

S1427

ESTRO 2021

within the radiotherapy department. According to the systemic analysis, a set of measurements was derived as key the performance assessments as they support the objectives and mission of the department. Radiotherapy capacity, timely access to radiation therapy, loco-regional tumor control, and patient’s quality of life were the main measures that were derived if they met the goals. Results The process map supported us with faster decision-making and helped us to improve the efficiency and effectiveness of the processes by finding specific parts that need changes. When we understand how different parts of the department are connected as a whole entity from a systemic point of view, we would not only blame or judge staff with recommending refresher training or punishments in case of any misperformance or error incident in the radiotherapy workflow. Regarding organizational decision-making, the system behavior at a specific period may be influenced by the RT capacity, treatment demand, human and equipment resources, and other limitations. Analyzing causal relationships and feedbacks in CLD, made the quality management team capable to explain how changes come about both temporarily and spatially in common problems and explore why decisions were made in special situations. Conclusion Systems thinking is essential for improving treatment quality and safety in radiotherapy. Causal tracing also enables fast and accurate analysis of system dynamics and supports the managerial team to understand what are the reasons for poor performance and what are the possible actions to solve the current problems. For future studies, a system dynamic simulation with the support of a multi-criteria analysis method can be effective for adding value to the management process in radiotherapy. PO-1700 Automated plan quality assessment tool for head and neck cancer VMAT plans I. van Bruggen 1 , E. de Ronde 1 , R. Steenbakkers 1 , A. van den Hoek 1 , T. van Zon-Meijer 1 , E. Oldehinkel 1 , J. Langendijk 1 , R. Kierkels 2 , S. Both 1 , E. Korevaar 1 1 UMCG, Radiotherapy, Groningen, The Netherlands; 2 Radiotherapy Group, Radiotherapy, Deventer, The Netherlands Purpose or Objective In the near future, adaptive and automated planning will be increasingly used, which puts a larger time burden on the radiation oncologist which has to approve all individual plans. In this study, we developed and validated, a fully automated quality assessment (QA) tool for scoring target coverage, target conformity and hotspots of Volumetric Modulated Arc Therapy (VMAT) plans of head and neck cancer (HNC) patients. The assessment of treatment plan quality is subject to inter-observer variability. Although recommendations on plan quality provide clear thresholds on target coverage (e.g. ICRU83), the spatial characteristics of those dose distributions are not taken into account explicitly but play a definitive role in the approval of treatment plans for clinical use. We hypothesize that the proposed QA tool has the ability to flag clinically unacceptable treatment plans, similar with the radiation oncologists (RO). Materials and Methods The QA tool was developed with a training set of 121 clinically accepted oropharyngeal cancer VMAT plans with RO expert’s opinion about target coverage, conformity and hotspots. Multiple measures were developed to quantify these characteristics, and corresponding thresholds were selected based on the highest area under the receiver operating characteristic curve while maintaining maximum sensitivity. For example, target coverage is assessed by measures using planning target volume (PTV) 7000 and PTV 5425 contracted by 1 mm. The QA tool automatically identified the regions of failure as regions of interest derived from isodose lines that exceed the given thresholds. The QA tool was then back-tested in 12 treatments plans (test set) which were independently scored by three other ROs using a fail/pass grading scale (fail: plan is not acceptable & minor or major revision is necessary, pass: plan is acceptable). The RO indicated the region of failure. The plans were also scored by the QA tool, which was implemented into the treatment planning system via scripting. The endpoints were sensitivity and specificity measures for all three categories of the QA tool.

Sensitivity=(True flags fail by QA tool)/(Flags fail by ROs)

Specificity=(True flags pass by QA tool)/(Flags pass by ROs)

Results The QA report, generated by the QA tool, indicated the criteria (table 1) and location of potential failure (i.e. the location at which a plan can be improved). The QA tool correctly identified all the regions of failure in 7/12 plans, as marked by the ROs (table 2). Target conformity in PTV 5425 and maximum dose were erroneously flagged as pass by the QA tool in patient 3/9 and 2/7/11 respectively. In 5/12 patients (3,6-9) all pass flags were correct.