Localization accuracy of robotic radiosurgery in 1-view tracking

Area of Science:

• Medical Physics
• Radiation Oncology
• Image-Guided Therapy

Background:

• Fiducial-less tracking for lung lesions with CyberKnife is often limited to a single X-ray view (1-view), posing challenges for accurate tumor targeting.
• This limitation affects a significant portion of lung treatments (34%) and necessitates careful consideration of localization uncertainties.

Purpose of the Study:

• To determine the necessary internal margin for lung treatments utilizing 1-view CyberKnife tracking.
• To quantify the out-of-plane (blind view) localization uncertainty.
• To establish patient-specific margin recommendations for improved treatment accuracy.

Main Methods:

• Retrospective analysis of 36 patients (127 fractions) treated with 2-view tracking.
• Tumor positions were determined from logfiles and projected onto 2D image planes.
• A custom Matlab software calculated the internal margin required to cover 95% of target position variability in the out-of-plane direction.
• Validation was performed using an XLT Phantom in both 1-view and 2-view scenarios.

Main Results:

• The validation test confirmed the method's reliability for margin estimation.
• Estimated margins were 5 mm for upper lobe targets and 6 mm for lower lobe targets.
• Biphasic breath-hold CT scans underestimated target movement in the out-of-plane direction.
• Inter-fractional variability in spine-target distance was a key uncertainty factor for 1-view treatments.

Conclusions:

• A graphic comparison method preserving metric can be integrated into clinical workflows for 1-view treatments.
• This approach provides patient-related data for customized margin definition.
• Accurate margin determination is crucial for optimizing radiation delivery and patient outcomes in fiducial-less lung treatments.