Integrating soft and hard dose-volume constraints into hierarchical constrained IMRT optimization.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Optimization

Background:

• Dose-volume constraints (DVCs) are crucial for intensity-modulated radiation therapy (IMRT) but pose computational challenges due to their non-convex nature.
• Existing automated IMRT planning systems struggle to efficiently incorporate these complex constraints.

Purpose of the Study:

• To develop computationally efficient methods for incorporating dose-volume constraints (DVCs) into automated IMRT planning.
• To address the non-convexity and difficulty of integrating DVCs into optimization frameworks.

Main Methods:

• A two-phase approach was proposed: phase 1 solves a convex approximation of DVCs, and phase 2 refines the solution by imposing maximum dose constraints on critical voxels.
• DVCs were categorized into hard-DVCs (strictly enforced) and soft-DVCs (encouraged in the objective function).
• The method was tested within an automated treatment planning system using various clinical cases.

Main Results:

• The algorithm successfully satisfied all hard-DVCs across multiple patient cases (paraspinal, lung, oligometastasis, prostate).
• Computational time increased modestly (10-20%) compared to baseline planning without DVCs.
• For a complex case, the proposed method achieved a near-optimal solution in 2 minutes, compared to 15 hours for traditional methods.

Conclusions:

• A computationally tractable algorithm for handling both hard- and soft-DVCs in IMRT planning has been developed.
• The algorithm effectively satisfies DVCs without requiring parameter adjustments.
• This approach is adaptable to various constrained optimization frameworks beyond the demonstrated system.