Robust optimization for HDR prostate brachytherapy applied to organ reconstruction uncertainty

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Imaging

Background:

• High-dose-rate (HDR) prostate brachytherapy planning relies on accurate 3D organ reconstruction from 2D contours.
• Reconstruction algorithms involve settings that can introduce uncertainties into dose-volume index calculations.
• Existing optimization methods may not account for these reconstruction-related uncertainties.

Purpose of the Study:

• To develop and evaluate a robust optimization approach for HDR prostate brachytherapy planning.
• To create treatment plans that are insensitive to variations in 3D organ reconstruction settings.
• To ensure clinical plan quality despite uncertainties in dose-volume indices.

Main Methods:

• A bi-objective optimization approach was augmented to incorporate robustness against 8 possible 3D organ reconstructions per patient.
• Minimax optimization was used to define plans robust to variations in interpolation, contour overlap, and organ shape.
• The robust model was tested on 26 patients and compared against the original model using dose-volume index cost and benefit metrics.

Main Results:

• Robust optimization increased planning time from 3 to 6 minutes.
• Median cost in dose-volume indices was -0.25% (range: -0.01% to -1.03%) when evaluated in the original model.
• Median benefit was 0.93% (range: 0.19% to 4.16%) when evaluated in the robust model.
• Plans optimized with the original method violated clinical protocols for 4 patients due to reconstruction variations, unlike robustly optimized plans.

Conclusions:

• The robust optimization approach yields high-quality HDR prostate brachytherapy plans consistently across different 3D organ reconstruction settings.
• This method effectively addresses dosimetry uncertainties arising from reconstruction choices.
• The approach is viable for clinical practice due to its minimal impact on total runtime.