Abstract
BACKGROUND: When validating intracranial multi-lesion CyberKnife M6/S7 plans with SRSmapcheck, setting the array to a fixed 0° measures only one target dose distribution, leaving the other lesions unmeasured. Moreover, the CyberKnife treatment planning system does not support roll verification tools, and testing confirms that X-sight fiducial marker guidance is incompatible with free array roll. A novel method and workflow are required to validate multi-lesion plans with random positions. METHODS: A geometric model was established based on the relationship between SRSmapcheck and the tumor location. For two tumors spaced 77 mm apart (each 20 mm in diameter, or one 40 mm apart and the other infinitesimally small), the corresponding array roll angle interval was approximately 15.05°. The SRSmapcheck and StereoPHAN computed tomography (CT) images were acquired at 15° intervals, starting at 0°, and preprocessed into phantom plans for verification. A total of 101 intracranial multi-lesion plans were verified using the fixed 0° and pre-set roll angle methods to optimize the dose distribution, particularly in high-dose and rapidly varying areas. A two-sample test compared the results of the 0° versus pre-set roll angle verification and assessed the performance under different criteria to determine suitable criteria for pre-set roll angle verification. RESULTS: The equivalent diameter of the 296 tumors ranged from 5 to 45 mm (average: 21.86 mm). Each plan had an average of 2.97 lesions. Using the pre-set roll angle method, 2.34 targets were assessed on average (89.83% of lesions had diameters ranging from 10 to 40 mm), compared to 1.32 targets on average in 0° plans. Statistically significant differences occurred at 2 mm/1% and 2 mm/2% in the γ analysis, showing that plan pass rates were stable between the fixed 0° and pre-set roll angle methods. Relaxing either the distance to agreement or dose deviation significantly increased the pass rates during pre-set roll angle verification, whereas cross-transforming criteria had minimal impact. For pre-set roll angle methods, it is recommended to use 1 mm/1% (action limit: 86.0% ± 13.3%) and 1 mm/2% (action limits: 91.6% ± 7.9%) for γ analysis. CONCLUSION: SRSmapcheck with the pre-set roll angle method can verify intracranial multi-lesion CyberKnife plans by measuring multiple targets in a single validation and comparing the 1 mm/1% and 1 mm/2% γ analysis criteria.