Comparison of methodological uncertainties in tissue parameter estimation for carbon ion dose calculation

Area of Science:

• Medical Physics
• Radiotherapy Physics
• Computational Biology

Background:

• Accurate dose calculation is critical for effective and safe carbon ion radiotherapy.
• Uncertainties in tissue parameter estimation for Monte Carlo dose calculations can limit clinical quality assurance.

Purpose of the Study:

• To systematically evaluate how uncertainties in tissue parameter estimation impact Monte Carlo-based carbon ion dose calculations.
• To compare three elemental decomposition approaches for their effect on dose distribution consistency and reliability.

Main Methods:

• Propagated uncertainties in physical density and elemental composition using single-energy CT (SECT), parameterized dual-energy CT (PA-DECT), and machine learning-based DECT (ML-DECT).
• Simulated dose distributions in the ICRP 110 phantom using FLUKA, evaluating physical and biological doses.
• Assessed voxel-wise uncertainty, gamma passing rate, and range uncertainty.

Main Results:

• ML-DECT reduced uncertainties in C, N, and O density by up to 77.4% compared to SECT.
• ML-DECT achieved lower average relative dose uncertainties (∼5% physical, ∼7%-9% biological) and a higher gamma passing rate (97.96 ± 0.28%).
• ML-DECT resulted in reduced range uncertainty (0.5%).

Conclusions:

• ML-DECT improves the accuracy of tissue parameter estimation, thereby reducing uncertainty in Monte Carlo-based carbon ion dose calculations.
• The findings support integrating ML-DECT into treatment planning for more quantitative and uncertainty-aware quality assurance in carbon ion radiotherapy.