Reduced order constrained optimization (ROCO): clinical application to lung IMRT

Area of Science:

• Medical Physics
• Radiation Oncology

Background:

• Intensity-modulated radiation therapy (IMRT) planning for advanced lung cancer presents challenges.
• Previous applications of reduced-order constrained optimization (ROCO) were successful for prostate cancer.
• Integrating ROCO with clinical treatment planning systems requires adaptation for lung cancer specifics.

Purpose of the Study:

• To implement and evaluate a novel ROCO approach for rapid, automated IMRT plan generation in advanced lung cancer.
• To incorporate mean dose hard constraints alongside existing point-dose and dose-volume constraints.
• To validate the clinical acceptability and efficiency of the ROCO method within a clinical workflow.

Main Methods:

• ROCO involves sampling treatment plan space using penalty-based quadratic objectives.
• Principal component analysis (PCA) is employed for dimensionality reduction, creating an efficient basis.
• A final constrained optimization is solved over this basis to yield clinically acceptable IMRT plans.

Main Results:

• ROCO successfully generated clinically acceptable IMRT plans for 12 stage III non-small-cell lung cancer (NSCLC) cases.
• The generated plans met all clinical constraints and were competitive with existing clinical plans.
• The approach demonstrated a median time saving of 105 minutes per patient compared to conventional planning.

Conclusions:

• ROCO effectively addresses challenges in lung cancer IMRT planning, including site complexity and constraint management.
• The updated ROCO implementation, using full dose calculation, ensures clinical constraints are met.
• This work extends ROCO's utility for automated IMRT plan creation to advanced NSCLC, building on prior prostate cancer applications.