Projections onto the Pareto surface in multicriteria radiation therapy optimization

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Optimization

Background:

• Pareto surface navigation in radiation therapy planning aims to optimize multiple competing objectives.
• Approximating the Pareto surface with limited plans can lead to suboptimal treatment outcomes and errors in Pareto optimality.
• Existing methods may not fully resolve discrepancies between navigated plans and the true Pareto optimal set.

Purpose of the Study:

• To reduce or eliminate errors in Pareto optimality during Pareto surface navigation.
• To improve the quality of radiation therapy plans by addressing limitations in Pareto surface approximation.
• To refine navigated plans to achieve better clinical goals.

Main Methods:

• A postprocessing projection step was introduced to map navigated plans onto the Pareto surface.
• The projection method seeks a Pareto optimal plan superior to the initial plan across all objectives.
• An augmented projection incorporated dose-volume histogram constraints to ensure clinical goal adherence.
• Evaluations were performed for intensity-modulated radiation therapy (IMRT) and intensity-modulated proton therapy (IMPT) techniques on prostate and head and neck cancer cases.

Main Results:

• Projections consistently improved dose conformity and organ at risk (OAR) sparing for all evaluated techniques and patient cases.
• Unconstrained projections reduced mean OAR dose by an average of 3.1 Gy.
• Constrained projections (using dose-volume histograms) reduced mean OAR dose by an average of 2.0 Gy.
• No significant improvements in target dose homogeneity were consistently observed.

Conclusions:

• Pareto navigation can leave room for improvement in OAR sparing and dose conformity, especially with coarse Pareto surface approximations or permissive constraints.
• Projection onto the Pareto surface effectively identifies inaccuracies in surface representation.
• This projection technique can enhance the quality of navigated radiation therapy plans, particularly in complex cases.