Abstract
BACKGROUND AND PURPOSE: 3D Magnetic Resonance Imaging (MRI) is used in radiation therapy for reference planning and, lately, for adaptive treatments on MR accelerators. This study aimed to investigate the impact of different types of respiratory motion on the apparent target position and extent in such scans. MATERIALS AND METHODS: An MRI motion phantom with a 30 mm diameter target was used to simulate cranial-caudal (CC) motion and imaged at an MR-Linac using a standard clinically released 3D T2w sequence. Scans were acquired for each combination of functions (sin(t), sin(4)(t) and sin(12)(t)), peak-to-peak amplitudes (5, 10, 15 and 20 mm), and periods (4, 5 and 6 s). Furthermore, respiration CC motion patterns from two patients were used. Motion functions were shifted such that the time average target position would match a static reference scan at 0-position. The target was automatically identified in coronal and sagittal images using k-means clustering. The mean position and area of the target were calculated and compared to the reference scan. RESULTS: Artefacts increased with amplitude and depended on the motion type. Sin(t) and sin(4)(t) oscillations resulted in a blurring of the target, which led to an increased target area, while sin(12)(t) motion did not show significant changes in the target area. However, for the sin(12)(t) motion, the offset in apparent position was prominent, while that was not the case for the sin(t) and sin(4)(t) motion. The patient respiration motion profiles showed similar trends. CONCLUSIONS: In 3D MRI, target motion can change apparent tumour extent and apparent position. The changes increase with motion amplitude and depend on the motion type.