On the Root Causes of Linac Off-Isocenter Geometric Errors
Abstract
Purpose
To perform a root cause analysis (RCA) of LINAC geometric error (LGE) dependency on target distance from isocenter for multitarget radiosurgery (SIMT) treatments and develop efficient and accurate clinical QA workflows
Methods
An RCA identified factors contributing to distance-dependent LGE, separating intrinsic LINAC geometric errors from phantom setup errors. Phantom pose uncertainty was eliminated using machine vision to position a newly developed multitarget phantom, determining whether geometric errors arise from phantom misalignment or intrinsic LINAC inaccuracy. Experiments utilized a LINAC coordinate system based on physical isocenter determined via optical tracking and a novel pose-based dynamic baselining approach. CBCT registration and 6DOF correction accuracy were evaluated by comparing phantom pose from clinical CBCT workflows against optical tracking measurements. Intrinsic errors were reduced by identifying relationships between error distributions and LINAC sub-module failure modes, then using these mappings to isolate and eliminate the distance-error relationship. All experiments were performed on two Varian and two Elekta LINACs. Clinical QA methods emphasizing root-cause analysis included an efficient daily test, an optical tracking test isolating intrinsic LINAC geometric errors, and an end-to-end test including CBCT registration and correction.
Results
Machine vision eliminated phantom setup errors, revealing geometric errors showing weak distance dependence for Varian (0.002 slope) and stronger dependence for Elekta (0.005 slope). Intrinsic errors were systematically reduced on Elekta LINACs to achieve Varian-comparable performance (0.003 slope). Similar CBCT registration and 6DOF correction errors were observed for both LINACs. The daily workflow produced intrinsic error determination comparable to machine vision. End-to-end testing achieved consistent passing results (< 1mm) for both platforms
Conclusion
RCA methodology and novel QA strategies successfully quantified and reduced LGE contributing factors in SIMT QA, enabling efficient, precise, and accurate clinical workflows