Analytical probabilistic modeling for radiation therapy treatment planning

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Quantifying uncertainties in radiation therapy is crucial for treatment plan quality.
• Current methods may not fully capture complex input parameter correlations and uncertainty types.

Purpose of the Study:

• Introduce Analytical Probabilistic Modeling (APM) for quantifying uncertainties in radiation therapy quality indicators.
• Develop analytical methods for dose calculation and inverse planning under uncertainty.

Main Methods:

• Assumed Gaussian probability densities for input parameters.
• Derived closed-form expressions for expectation values and covariances of dose distributions (photon/proton) and objective functions.
• Incorporated complex correlation models and distinct random/systematic uncertainties.

Main Results:

• APM enables analytical calculation of moments for quality indicators.
• Variance calculations on phantom data highlight the impact of correlation and uncertainty types.
• APM shows potential for robust inverse planning with comparable computational cost to conventional methods.

Conclusions:

• APM provides a framework for uncertainty quantification in radiation therapy planning.
• The method effectively incorporates complex uncertainty models.
• APM offers a feasible approach for robust and efficient treatment planning.