Room: AAPM ePoster Library
Purpose: To quantify the relative impact of patient setup, immobilization and machine performance on targeting accuracy of conventionally-fractionated (54Gy-60Gy in 30 fractions) brain patients treated on a 1.5-T MRI-Linac (MRL).
Methods: Four factors contributing to geometrical accuracy were identified: (1)residual setup error; (2)patient immobilization; (3)machine MRI-to-MV alignment; (4)machine MLC positional accuracy. For residual setup error, daily MR images of fifteen brain patients (269 fractions) clinically treated on MRL were retrospectively analyzed. The clinical workflow involved translation-only co-registration of daily MRI to the reference image. Anatomical landmarks were retrospectively identified by a radiologist and geometrically tracked on each co-registered MRI. For intra-fraction motion, the landmarks on post versus pre-treatment MRI were compared in a subset of 7 patients (45 fractions). Routine MRI-to-MV isocenter and MLC position tests were assessed for 5 consecutive months. The standard deviations in each factor were calculated for random and systematic components and presented in 3 directions: X(lateral), Y(superior-inferior), Z(anterior-posterior).
Results: For anatomical landmark position variation, the random component was 0.7mm, 1.0mm, 1.1mm and the systematic component was 0.7mm, 1.0mm, 0.8mm. For intra-fraction motion, random was 0.4mm, 0.6mm, 0.3mm and systematic was 0.2mm, 0.4mm, 0.2mm. The standard deviation of the MRI-to-MV isocenter test was 0.1mm, which counted as purely systematic error in all directions. The positional variation in MLC was 0.2mm and 0.3mm for random and systematic, respectively. Summing in quadrature over the 4 factors, the combined random uncertainty was 0.8mm, 1.2mm, 1.2mm and the systematic uncertainty was 0.8mm, 1.1mm, 0.9mm. Residual setup error was the dominant source of uncertainty.
Conclusion: Residual error of daily plan adaptation using a translation-only correction was the largest source of uncertainty in targeting accuracy for brain MRL treatments. Uncertainty mitigation methods are being investigated, including offline correction strategies and incorporating rotational corrections in order to reduce planning target volume margins.
Funding Support, Disclosures, and Conflict of Interest: Dr. Sahgal: Advisor/consultant with Abbvie, Merck, Roche, Varian, Elekta, BrainLAB, and VieCure (Medical Advisory Board); Board member for ISRS; Research grant with Elekta AB; Travel accommodations/expenses by Elekta, Varian, BrainLAB; Belongs to Elekta MR Linac Research Consortium Dr. Ruschin: Owns intellectual property associated with image-guidance system on Gamma Knife Icon