Uncertainty of Skull Tracking for Cervical Spine Radiosurgery
Abstract
Purpose
ZAP-X is a self-shielded radiosurgery system featuring a compact gyroscopic gantry design with kV imaging for patient setup and tracking. Its well-engineered treatment table and treatment-planning system allow lesions located inferiorly to remain treatable. However, at present, skull tracking is the only available tracking method. This study investigates positioning errors arising from the rigid-body assumption between the skull-base and C2 during skull-tracked treatment.
Methods
Five head-and-neck (HN) patients and three cranial patients treated with conventional fractionation were included in this study. Treatment setup was performed using planar kV image alignment, consistent with the workflow used for ZAP-X treatments. Following each treatment setup, a mobile CT scan was acquired with the aid of a robotic couch while maintaining identical patient positioning. Head-and-neck patients represent setups in which chin extension and shoulder positioning were carefully reproduced, whereas cranial patients represent setups in which these factors were not explicitly controlled. For each setup, alignment was achieved by registering kV-pair images to digitally reconstructed radiographs (DRRs). Decomposition of region-based registration matrices from the C2 relative to the skull-aligned images were analyzed to quantify translational and rotational deviations of the cervical spine, assuming skull-based tracking during treatment.
Results
Gaussian-distributed two-sigma translational uncertainties for HN patients were ±4.0 mm in the left–right (LR), ±4.2 mm in the anterior–posterior (AP), and ±4.8 mm in the superior–inferior (SI) directions. In contrast, cranial patients exhibited substantially larger uncertainties in the AP (±10.6 mm) and SI (±8.0 mm) directions, while LR uncertainty remained comparatively small (±3.2 mm). Angular deviations were less than 2° for HN patients and up to 3° for cranial patients.
Conclusion
Skull tracking for C2 cervical spine targeting is challenged by differential motion between the skull and upper cervical vertebrae, the resulting uncertainty remains controllable. Appropriate treatment margins are therefore recommended during planning.