2283 - Adaptive Team Training Based on Incident Learning

F. Forghani¹, E. Laugeman², D. Hobbis¹, R. Beckert³, G. D. Hugo⁴, and A. T. Price⁵; ¹Washington University in Saint Louis, Saint Louis, MO, ²Department of Radiation Oncology, Washington University School of Medicine in St. Louis, St. Louis, MO, ³Washington University School of Medicine, Saint Louis, MO, ⁴Washington University School of Medicine in St. Louis, Department of Radiation Oncology, St. Louis, MO, ⁵Department of Radiation Oncology, University Hospitals Seidman Cancer Center, Case Western Reserve University, Cleveland, OH

Purpose/Objective(s): This work aims at developing tools for training adaptive radiotherapy (ART) team members to improve the online adaptive quality assurance. Troubleshooting ART plan and delivery is a crucial skill that needs to be developed with practice but unfortunately usually occurs with a patient on the table. Using retrospective patient treatment data and identifying sources for failure-modes, one can create a testing space for individuals to practice troubleshooting methods. We take advantage of CBCT-based adaptive emulator system to help facilitate the testing space. Materials/

Methods: Retrospective ART treatment data from twelve patients who were treated with CT-guided adaptive were used in this study based on an IRB approved protocol. Patient selection was based on the institutional process improvement reports or adaptive team members’ experience. Among the treatments, there were nine pancreas, one lung, one pelvis, and one paraaortic lymph node. Three categories: Patient-, User-, and Process-related were identified to be the source of the incidents. Each incident was qualitatively graded based on deviation from treatment intent. Incidents with the highest deviation were chosen to be simulated on the emulator for team training (Table 1). The imported data to emulator included anonymized planning CTs, RT structures, and CBCT of the treatment day. The data then was used for optimization and generation of a new reference treatment plan.
Results: One example simulated a high deviation from treatment intend when a small region of target was drawn out of the true target and control ring. Dosimetric results showed abnormally higher amount of MUs in the adaptive plan compared to the reference plan. Possible strategy to mitigate this incident was to implement a constraint on dose outside of control ring in planning. In addition, visual check of the low and high dose regions in the plan during the online adaptive was added to the training QA tool. Several other scenarios from incident learning were generated and strategies to identify and mitigate them are provided (Table 1).
Conclusion: We designed an interactive adaptive training space for team members based on the incident learning reports. This tool can equip the team members with skills to identify errors and mitigate them during online ART. This tool can greatly improve the quality assurance of ART. Abstract 2283 – Table 1

Example Incidents	Related Category	Training goal
Target drawn incorrectly	User	Identify target structures drawn outside a Contour Ring
Suboptimal scan quality (bowel gas)	Patient	Improve knowledge on acceptable image quality
Patient ability to hold one arm up	Patient	Identify discrepancies in online setup from simulation
Cropping PTV_Opt from OARs 2 mm instead of 5 mm	User	Identify suboptimal planning
Suboptimal deformation of CT sim to CBCT	Process	Identify inappropriate electron density for plan calculation
Derive structure not activated	Process	Identify a derived structure in a contouring workspace