Session Item

The number of artificial intelligence applications in radiotherapy treatment planning (TP) is growing. This results in different required competencies and less opportunities in daily practice to keep up the complex TP skills. In this project, we evaluated if in-depth reflection among peers could improve TP skills in a sustainable way.

A group of 12 RTTs with variable level of experience in TP for lung cancer (3 months–20 years) was asked to create a plan (T0) for stage III non-small cell lung cancer patient according to our standard clinical protocol (30x1.5Gy). The results were discussed in detail among the RTTs in a peer session using a plan QA tool. In this tool, dosimetric and technical parameters can be plotted for a group of participants and compared to an expert. This improves objectivity and due to visual representation facilitates detailed reflection on clinical decision making among RTTs (Fig. 1). During the peer session, discussion focused on: differences in DVH and technical parameters, priorities in the optimization box and reasons behind those choices. After the peer, RTTs created a plan for a different patient (T1), with similar anatomy and tumor stage, to see how well the skills they learnt were transferred to another case. This plan was repeated after 6 months (T2) to evaluate sustainability of competencies. DVH parameters of the RTTs’ plans were compared to expert plans. Different timepoints were compared: T0-T1 and T1-T2. For the relevant DVH parameters and pairs of timepoints, differences in agreement among the group (GA) and with the expert (EA) were tested using Flinger and Wilcoxon tests (p-value<0.05), respectively. In addition, the RTTs were asked on their perception and experience of the peer session.

Significant differences in DVH parameters were observed between T0 and T1 (Fig 2). GA increased for esophagus Dmean, heart Dmean and heart V45Gy, showing better agreement at T1. EA increased for heart Dmean, V30Gy and V45Gy at T1. GA for PTV D0.03cc decreased at T1, which can be explained by a larger PTV volume for patient 2. GA and EA also decreased for lungs V5Gy. In this case constraint was easily reached and RTTs differently prioritized further decrease of V5Gy as it was not reflected in lungs Dmean decrease.
At T2 RTTs showed good agreement in technical choices, dose to organs at risk remained equally good compared to T1 and PTV coverage improved significantly.

The peer session guided by the plan QA tool increased knowledge and competencies, it also decreased variation in decision making among RTTs. The TP outcomes for lung cancer were more uniform directly after the peer session which sustained after six months. The RTTs experienced the peer discussions as helpful, especially when their prior level of experience was limited. The peer session guided by the plan QA tool had positive impact on the TP skills of RTTs for lung cancer treatment and can be a valuable tool to improve training of TP skills in general.