Designing Robust and Automated Treatment Plans Utilizing Dynamic Couch Motion With Automated Treatment Planning: Delivering Dose to a New Dimension ― Part B

Abstract

<h3>Purpose/Objective(s)</h3> To develop a novel treatment technique, with automated treatment planning, that will yield superior dose distribution with dynamic couch motion at static gantry positions. <h3>Materials/Methods</h3> Modern treatment planning systems (TPS) cannot design, calculate, or optimize treatment plans utilizing continuous couch-based delivery. Based on the current limitation of modern TPS, the need to plan dynamic couch-based dose delivery arises for our investigation. We implement an automatic method utilizing the TPS scripting application programming interface (ESAPI). Additionally, we introduce a novel gantry static couch motion technique (GsCM) with static gantry positions and dynamic couch motion for brain treatment. The script interface allows selection of PTV, adding MLCs and PTV margin, choosing the number of fields, and initiation of the dose calculation within the TPS environment. We can automatically design custom treatment plans with simple 3-dimentional (3D) planning and conformal arc. This script based automated planning was retrospectively applied to brain patients (n = 5) using 200 cGy for 23 fractions. Initial development required implementing a conversion from the TPS mechanical axes to that of the machine. The readable Extensible markup language (XML) files are generated in the TPS environment using ESAPI by combining the fields at each couch position into a single deliverable field for each gantry angle utilizing continuous couch motion. The entire workflow is feasible within our script interface. The last part of this study focused on verifying the dose delivery. We used an ion chamber array positioned in the vertical direction to avoid side beam entrance. Trajectory log files were also analyzed using root mean square error (RMSE) to verify the MLC, gantry, and couch position accuracy. <h3>Results</h3> All treatment plans satisfy the criteria of V100% ≥ 95% and respect the dose to organs-at-risk (OARs). Automated script-based plans took less than a minute (excluding dose calculation) to design as compared to manual planning (2-8 hours per plan). The delivery was found to be less than 2 minutes for two 90 degrees partial arcs as compared to field-by-field delivery (15-30 minutes based on the plan type). All script-based plans (n = 5) passed patient-specific QA (> 95%) using the criteria of 2 mm distance to agreement and 2% dose difference. The couch rotational RMSE error for step and shoot and continuous delivery was 0.1 degrees and 0.12 degrees, respectively. MLC RMSE was 0.005 mm, and 0.006 mm for step and shoot and continuous delivery. <h3>Conclusion</h3> The results of our study indicate that the automated custom planning with couch motion is feasible and comparable with clinical planning in terms of dosimetric quality. The developed methodology provides a tool that is integratable with the current TPS without any additional resources. It designs script-based plans for dynamic couch motion-based dose delivery.