Dosimetric Impact of CBCT-Based Setup Corrections In G-Frame Gamma Knife Radiosurgery for Post-Craniotomy Patients
Abstract
Purpose
Stereotactic G-frame immobilization is the reference standard for Gamma Knife radiosurgery (GKRS); however, post-craniotomy anatomical alterations may compromise setup accuracy and reproducibility. This study is on CBCT–based setup corrections during treatment delivery in post-craniotomy GKRS patients immobilized with G-frames. We evaluated the dosimetric impact of residual setup uncertainty on target coverage and plan quality.
Methods
Thirty seven post-craniotomy patients treated with frame-based GKRS using CBCT image guidance were retrospectively reviewed, comprising 58 treatment targets. Pre-treatment CBCT images were rigidly registered to the plan reference images. The geometric translation and rotation corrections and the calculated maximum displacement were obtained from the Leksell Gamma Plan (LGP) treatment planning system and the predicted deviation in dosimetric coverage, as reported by the planning system if these corrections were not applied, were studied.
Results
The maximum setup correction was 1.52 ± 0.74 mm, with systematic and random errors of Σ = 0.63 mm and σ = 0.41 mm, respectively. Twenty-seven percent of treatments required translational corrections ≥1.0 mm. Rotational corrections demonstrated a maximum magnitude of 0.71 ± 0.38°, with 11% of treatments exceeding 1.0°. Increasing translational setup error was associated with progressive degradation in target coverage and conformity. Targets requiring ≥0.5 mm correction demonstrated a mean reduction in V100% of 3.1% and a CI degradation of 0.12, compared to reductions of 0.9% and 0.03 for corrections <050 mm (p < 0.01). GI remained relatively stable for translational corrections <1.5 mm but increased by 6.4% for larger deviations.
Conclusion
Further, we find that overall maximum displacement corrections greater than 0.5 mm resulted in appreciable dosimetric coverage degradation and can be used as criteria for rejecting G-frame based registration and replanned with CBCT based registration. Earlier work (1) by previous authors had suggested a 1 mm correction tolerance.