Author: Yang Sheng, Qingrong Jackie Wu, Qiuwen Wu, Xin Wu, Dongrong Yang 👨🔬
Affiliation: Duke University Medical Center 🌍
Purpose:
Treatment planning for intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT) relies on inverse planning, an iterative and non-intuitive process of adjusting dose-volume objectives to achieve the desired dose distribution. We aim to introduce a novel foresight planning strategy to directly guide optimization toward the desired dose distribution by modeling the inverse optimization process.
Methods:
The proposed strategy involves two stages. First, a Deep-Dose-Predictive (DDP) model was trained to predict the dose response based on historical plan states and dose-volume objective (DVO) adjustments. The training of the DDP model mirrored that of a clinical planner, during which it gained intelligence in understanding dose-response principles by witnessing extensive plan state transitions. The training dataset was generated using Monte Carlo sampling without any intervention. Then the trained DDP model was employed for automatic DVO adjustments through model predictive control. A cost function evaluated the plan quality based on the predicted dose responses for all potential adjustments, and the adjustment that maximizes the cost function was selected. The planning strategy can be tailored to specific clinical priorities by modifying the cost function without retraining the model. The efficacy was evaluated in head-and-neck cancer IMRT, using 40 cases for training and 30 for testing.
Results:
The proposed strategy achieved equivalent or improved dose sparing for all OARs compared to clinical plans, with superior homogeneity index (0.0571 vs. 0.0599) and conformity index (1.167 vs. 1.475). By encoding clinical priorities into the cost function, the model flexibly adjusted its strategy, enabling unilateral or bilateral parotid sparing while consistently maintaining clinically acceptable plan quality.
Conclusion:
The proposed foresight planning strategy offers an effective and flexible solution for automating radiation therapy treatment planning, providing a fresh perspective on the process.