Deformable Image Registration Uncertainty-Encompassing Dose Accumulation for Adaptive Radiation Therapy

Area of Science:

• Medical Physics
• Radiation Oncology
• Image Processing

Background:

• Deformable image registration (DIR) is critical for adaptive radiation therapy but is inherently ill-posed.
• DIR quality is algorithm and user-dependent, impacting its clinical application.
• Integrating DIR uncertainty into dose accumulation is essential for accurate treatment planning.

Purpose of the Study:

• To propose an automated framework for incorporating deformable image registration uncertainty into dose accumulation.
• To enhance the reliability and precision of dose calculations in adaptive radiation therapy.
• To improve clinical decision-making and personalize radiation therapy treatments.

Main Methods:

• A hyperparameter perturbation approach was used to generate an ensemble of deformation vector fields.
• Principal component analysis on homologous points created voxel-specific DIR uncertainty confidence ellipsoids.
• Dose limits were estimated via interpolation within these ellipsoids during dose mapping.

Main Results:

• The framework is clinically feasible, with an average computational time of 30 minutes.
• Significant DIR uncertainty was observed, particularly for the gross tumor volume and lungs.
• Considering DIR uncertainty revealed cases exceeding institutional dose-volume histogram limits, emphasizing its importance.

Conclusions:

• The proposed framework facilitates clinical implementation of dose accumulation by accounting for DIR uncertainty.
• This approach can improve clinical decision-making in adaptive radiation therapy.
• It holds the potential to enable more personalized radiation therapy treatments.