Room: Exhibit Hall | Forum 7
Purpose: Conventionally, leaf sequencing optimization (LSO) following optimal fluence map generation for IMRT use individual leaf pair optimization. The purpose of this work is to introduce a neighborhood-coordinated leaf motion schema for dynamic multileaf collimator leaf sequencing in order to reduce the tongue-and-groove effects in the leaf trajectories.
Methods: The neighborhood-coordinated LSO model follows the fluence optimization step. It includes the leaf motion constraints (i.e. maximum velocity), inter transmission effects and intra transmission effects. The neighborhood coordination was simulated by piecewise linear optimization functions with an objective of minimizing the delivery time and the travel distance between each leaf pair and their adjacent leaf pairs. This approach was evaluated for seven complex head and neck fluence maps having 2.5 mmÂ² bixels dimensions. To ensure delivery efficiency, we measured the ratio of delivery time with to without neighborhood coordination. Additionally, the root mean square error (RMSE) was calculated to ensure convergence of each sequenced fluence map with the reference fluence map.
Results: Incorporating the influence of neighboring leaves into the LSO model generated more synchronized leaf trajectories for complex fluence maps, reducing tongue and grove effects. The average RMSE for 7 fluence maps was 3.5. The delivery time for 5 fluence maps remained comparable in both approaches, however, for two cases, the delivery time increased by 10 and 50%.
Conclusion: Generating more synchronized leaf trajectories for complex fluence maps can reduce the tongue-and-groove effect. This technique can allow the use of jaw tracking for complex fluence maps reducing peripheral dose. Therefore, this approach may improve the IMRT delivery of complex treatment sites such as head and neck and pelvic by reducing QA failure rate.