3159 - Simulation-Driven Motion Emulation for Adaptive MR-Guided Lung Radiotherapy

Purpose/Objective(s): Gated delivery with 1.5T MR-linac continues to be promising for several sites. However, for challenging cases such as ultra-central lung, users still schedule a ‘day0’ fraction to set-up and evaluate motion tracking surrogates. This study aims to compare MR-simulation motion emulation with delivered respiratory gating performance for ultra-central lung adaptive treatments on a 1.5T MR-linac. We hypothesize motion emulation performed at MR-simulation can prospectively identify robust tracking surrogates that remain effective throughout adaptive MR-guided radiotherapy and can replace the 'day0' simulation.

Materials/Methods: Five ultra-central lung patients treated with gated adaptive radiotherapy on a 1.5T MR-linac were analyzed. Respiratory motion was characterized at MR-simulation using cine imaging, and motion emulation was performed using tracking surrogates selected to isolate tumor motion. All motion emulation data were acquired during the MR-simulation session, allowing gating feasibility and tracking structure selection without requiring an additional MR-linac visit. During treatment, real-time MR tracking recorded delivered duty cycle, tracking success, and motion amplitudes. Emulator predictions were compared with patient-averaged treatment metrics. Adaptive plans were evaluated to assess PTV changes during treatment.

Results: Emulated duty cycle agreed with delivery within 2 percentage points in two patients and underestimated delivery by 10 to 32 percentage points in three patients, with treatment duty cycles exceeding 90% in most cases despite motion amplitudes exceeding 4 to 6 mm. Motion magnitude alone did not predict gating efficiency, and patients with large motion excursions were successfully tracked with optimized surrogate selection. Tracking success remained high across treatments. Adaptive replanning demonstrated substantial target volume reduction. In one representative case, the PTV decreased from 141 cc to 62 cc during treatment. Despite this substantial anatomical change, tracking remained feasible.

Conclusion: MR-simulation motion emulation provides a practical method to define robust tracking surrogates without requiring additional MR-linac sessions. Appropriate surrogate selection at simulation supports effective tracking even in the presence of substantial anatomical and target volume changes during adaptive treatment.