Characterizing a deformable registration algorithm for surface-guided breast radiotherapy.

Area of Science:

• Medical Physics
• Radiation Oncology
• Image-guided therapy

Background:

• Surface-guided radiation therapy (SGRT) uses external surface imaging to guide radiation treatments, assuming surface anatomy correlates with internal anatomy.
• Patient posture variations and anatomical deformations can disrupt this correlation, potentially affecting treatment accuracy.
• Current methods lack the ability to quantify the impact of these deformations on SGRT system performance.

Purpose of the Study:

• To develop a framework for assessing how a commercial optical surface imaging system (C-RAD) with a nonrigid registration algorithm handles patient surface rotations and deformations.
• To investigate the impact of these transformations on the accuracy of isocenter positioning in radiation therapy.

Main Methods:

• A workflow using a female torso phantom and software-introduced surface transformations was created.
• Rigid transformations (translations and rotations) were applied to benchmark the system.
• Breast radiotherapy-relevant deformations (e.g., breast size changes, abdominal distension, hunching/arching) were simulated to evaluate performance under realistic conditions.

Main Results:

• The C-RAD system achieved sub-millimeter and sub-degree accuracy for rigid body transformations, irrespective of isocenter location.
• Yaw and pitch setup errors showed differences in shift corrections (2.3 mm/1° and 2.1 mm/1°, respectively) depending on isocenter position.
• Simulated surface deformations led to significant discrepancies (up to 16 mm and 7°) in calculated shifts between different isocenter locations, potentially causing treatment field mismatches.

Conclusions:

• The developed methodology is suitable for quality assurance of SGRT systems.
• While the C-RAD algorithm accurately handles rigid transformations, surface deformations introduce significant isocenter position dependency.
• To optimize SGRT accuracy, particularly with multiple fields, treatment planning should favor on-axis isocenters to leverage the deformable registration algorithm's capabilities.