Multiobjective approach to morphological based radiation treatment planning.

摘要

Radiation therapy is a widely used treatment modality for cancer management. The state-of-art intensity-modulated-radiation-therapy (IMRT) technology is capable of controlling radiation intensity on a voxel-by-voxel basis. However, planning an IMRT treatment is a challenging multi-objective optimization process. The main criteria used in the optimization are the dose-volume-histogram (DVH) constraints of the target, adjacent organ-at-risk (OAR) and normal tissues. The DVH is a graphic representation of the amount of radiation that a given structure receives. The planning system generates a mathematically optimized plan by minimizing the deviation of dose between the planned dose and the desired dose. One of the problems in this process is the objective function may not adequately encapsulate the clinical requirements. Consequently, the optimal plan generated may not be clinically deliverable. For example, DVH constraints lack spatial information. If the plan has a hot spot (although within the target) close to the OAR, patient movements during treatment may result the hot spot being shifted to the OAR. Incorporating morphological constraints into the optimization can yield robust plans against patient movement. The core, a medial line structure of an object, is used to capture the morphological information of target and OAR. Individual voxel desired dose level is calculated and assigned using the space scale from the core and the input prescription, and then these voxels are incorporated into the dose-volume objective functions to steer the local dose distribution. Another problem is the current treatment planning systems rely on the gradient search method which does not guarantee to find the optimal solution. Goal programming is another optimization method based on linear programming. Therefore, the optimization guarantees to find the optimal solution if the solution exists. Our experiments demonstrate that integrating morphological information to the objective function coupled with a robust optimization method can significantly improve the quality of a treatment plan.