MO-D-BRB-06: JUNIOR INVESTIGATOR WINNER - Fast and Accurate Patient Specific Collision Detection for Radiation...

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Geometry

Background:

• Radiation therapy requires precise patient positioning.
• Hardware collisions can disrupt treatment delivery and compromise safety.
• Current methods for collision detection are often time-consuming or limited in scope.

Purpose of the Study:

• To develop a rapid and broadly applicable method for identifying all potential hardware collisions specific to a patient's setup prior to treatment planning.
• To create a virtual mapping of collision-free spaces in radiation therapy.

Main Methods:

• Simulated an anthropomorphic phantom in a typical breast setup using CT data.
• Created 3D polygon meshes for phantom, table, and gantry geometry.
• Developed a software tool using a hierarchical bounding box (HBB) algorithm for virtual collision detection.

Main Results:

• Mapped 47K polygons across 6480 states in 35.6 seconds on a standard computer.
• Achieved 96.35% accuracy and 91.2% positive predictive value in computed collision space.
• Demonstrated a fast and efficient collision detection process.

Conclusions:

• The framework provides a fast, accurate map of collision-free space for patient-specific setups.
• Discrepancies between physical and simulated collision spaces are due to geometrical model inaccuracies.
• Future work will focus on algorithm efficiency, model enhancement, and expanded search dimensions.