Towards anatomical robust plan evaluation using a population-based model for head and neck cancer proton therapy

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Head and neck cancer proton therapy requires robust treatment planning to account for anatomical variations.
• Current robust plan evaluation methods primarily focus on setup uncertainties, potentially underestimating anatomical changes.
• Principal Component Analysis (PCA) offers a data-driven approach to model complex anatomical variations.

Purpose of the Study:

• To evaluate the efficacy of pre-treatment anatomical robust plan evaluation (Ana-RE) using a population-based PCA model.
• To determine if Ana-RE provides advantages over setup-based evaluation methods in head and neck cancer proton therapy.
• To assess the accuracy of Ana-RE in estimating the impact of inter-fractional anatomical variations.

Main Methods:

• A PCA model was developed using deformation vector fields from planning CT and weekly CBCTs of 20 oropharyngeal cancer patients.
• The PCA model was applied to five test patients to simulate anatomical scenarios.
• Proton therapy plans were evaluated using Ana-RE, conventional robust evaluation (Conv-RE), and probabilistic robust evaluation (Prob-RE), comparing their accuracy and ranking of plan robustness against daily dose variations (Daily-Ref).

Main Results:

• Ana-RE demonstrated superior agreement with reference daily dose variations (Daily-Ref), showing the lowest median absolute errors for clinical target volume (ΔV95% = 0.24%) and organs at risk (ΔDmax = 0.54 Gy, ΔDmean = 0.93 Gy).
• Ana-RE exhibited better identification of plan robustness differences, with a mean absolute rank difference of 2.3 compared to Conv-RE (2.7) and Prob-RE (2.9).
• Combining anatomical and setup uncertainties (Ana-RE-PS) further improved performance, achieving a mean absolute rank difference of 1.20.

Conclusions:

• Population-based anatomical robust plan evaluation (Ana-RE) accurately estimates the impact of inter-fractional anatomical variations in head and neck cancer proton therapy.
• Ana-RE offers significant advantages over conventional and probabilistic setup-based evaluation methods.
• Integrating anatomical and setup uncertainties enhances plan evaluation, supporting individualized treatment decisions.