Clinical implementation and evaluation of a patient-specific surface-guided clearance mapping system for collision
Siqiu Wang1, Eric Chambers1, Ruiqi Li1
1Department of Radiation Oncology, University of Texas Southwestern Medical Center, Dallas, Texas, USA.
Purpose/Objectives:
Collision avoidance is critical in external beam radiotherapy to ensure patient safety and plan deliverability. Limited understanding of the collision-free treatment space risks both patient safety and unnecessary exclusion of useful beams-particularly in noncoplanar setups-resulting in suboptimal plans. Conventional methods (manual clearance checks or CT-based assessments, etc.) are either labor-intensive or fail to account for collision-prone anatomy outside the scan. We investigated and clinically implemented a virtual patient-specific clearance mapping system and evaluated its utility as a noncoplanar beam selection tool to improve plan quality.
Materials/Methods:
The system integrates full-body, patient-specific surfaces-acquired during simulation using near-infrared imaging-with interactive 3D linac/couch models. Clearance mapping accuracy was validated through phantom measurements and a comparative analysis with manual clearance checks of 60 patients across treatment sites. Workflow efficiency data were reported over three years of clinical implementation. A workflow for patient-specific non coplanar beam selection was proposed and evaluated in 20 lung stereotactic body radiation therapy (SBRT) and 18 breast stereotactic partial breast irradiation (sPBI) cases.
Results:
The clearance mapping accuracy was within ± 1° (gantry/couch rotation) of phantom measurements. For 60 patients, the virtual predictions accurately identified all potential clearance issues, while manual verification missed 5 collision events. Virtual checks saved approximately 15 min of linac and therapist time per plan and eliminated an average 6.2-clinical hour planning delay. With the proposed beam selection workflow, noncoplanar replans for lung SBRT improved target conformality (Paddick Conformity Index from 0.89 to 0.91, p < 0.01) and reduced low dose spillage. For breast sPBI, heart mean dose was lowered (103 cGy to 68 cGy, p < 0.01). Delivery time increased by approx. 30s per plan.
Conclusions:
The virtual clearance mapping system outperformed manual verification, streamlined clinical workflow, and could significantly improve plan quality through efficient noncoplanar beam selection. It has replaced manual verification at our institution.
