Dosimetric Impact of Using a Virtual Couch Shift for Online Correction of Setup Errors for Brain Patients on an Integrated High-Field Magnetic Resonance Imaging Linear Accelerator

Abstract

Purpose: To quantify the dosimetric impact of using virtual couch shift (VCS) for correcting setup errors in glioblastoma multiforme (GBM) patients treated on a magnetic resonance imaging (MRI)-linac.

Methods and materials: Six GBM patients treated with 60 Gy (30 fractions) were selected for this simulation study. For each case, 2 reference plans were generated in the MRL treatment planning system: With (WIB) and with no (NOB) MRI B field present. Subsequently, 2-mm, 4-mm, and 6-mm translational errors were simulated and corrected for using a VCS method based on shift-only, warm start segment weight (SWO), and segment weight and shape (SSO) optimization. The resulting distributions were compared with the reference plan using planning target volume (PTV) homogeneity index (HI), conformity index (CI), organs at risk (OAR) maximum dose (D_0.01cc), and OAR median dose (D50). A simulated 30-fraction treatment was constructed to evaluate the cumulative effect of daily corrections. Feasibility and workflow for correcting rotations were also assessed.

Results: All reference plans were deemed clinically acceptable with respect to PTV and OAR objectives. The difference in HI (ΔHI) between corrected and reference was not statistically significant between WIB and NOB (P=.89). The average ΔHI was +0.8%, -0.1%, and -1.0% for shift-only, SWO, and SSO, respectively, with a statistically significant difference (P<.001) for shift-only versus SWO and SSO. The CI remained unchanged (mean ΔCI = -0.01) between the corrected and reference plans, with no statistically significant dependence on magnetic field presence, correction method, or shift magnitude or orientation. The brainstem D50 on average decreased with SWO and SSO; however, D_0.01cc increased by a median value of 1.2%, 1.9%, and 2.0% for shift-only, SWO, and SSO, respectively. For other OARs, D_0.01cc decreased using SWO or SSO. For the simulated treatment and rotational corrections, similar trends were measured.

Conclusion: For translational errors in brain MRI-linac radiation therapy, the VCS method is an acceptable correction strategy, but caution must be used in particular for serial organs where maximum doses are most relevant. The effect of the magnetic field on relative changes between corrected versus reference plans is not clinically relevant.