Purpose: An end-to-end test to evaluate the normal brain dose for stereotactic brain radiosurgery (SRS) treated with an MR-Linac using the adapt-to-position (ATP) and adapt-to-shape (ATS) workflows to account for daily setup variations.
Methods: The RTsafe Prime brain phantom was used with a delineation insert that includes two pre-defined structures mimicking gadolinium contrast-enhanced lesions. Daily adaptive plans were generated using 6 presets (3 mm and 3 degrees) and 6 random setup variations (> 10 mm). Two adaptive plans per daily MR image were generated using the ATP and ATS workflows. In the ATS workflow, GTV volumes on daily MRIs were recorded and PTVs were recreated with a 1mm margin. Target doses were measured with two ionization chambers inserted into the phantom. The adaptive plans were compared with the reference plan using the target coverage, conformity index, gradient index (GI), V12Gy, R50, optimization time, and total monitor units (MUs).
Results: For small daily setup variations (≤ 3 mm), GI and V12Gy of combined PTVs in the ATS workflow were 8% and 14% lower than those from the ATP workflow. The GTV volume change was < 1% after contour deformation in the ATS workflow. For large setup variations (> 10 mm), both GI and V12Gy from the ATS workflow were 17% lower than those from the ATP workflow but the GTVs increased by 3% for the large lesion (5 cm³) and 5% for the small lesion (1.4 cm³). The differences between the measured and planned target doses were within 1% for both treatment workflows.
Conclusion: Brain SRS on MRL is feasible but it is important to reduce daily setup variation. Normal brain dose could be lower with the ATS workflow at the cost of a longer plan optimization time (5-10 min) comparing to the ATP workflow (< 2min).