Abstract
MRI is increasingly being used in radiotherapy of the liver. The purpose of this study was to develop and validate a strategy to acquire MR images for treatment planning and image guidance in the presence of respiratory motion. By interleaving two navigator triggered MRI sequences, a fast but low-resolution image in mid-ventilation (midV) and a high-resolution image in exhale were acquired efficiently. Deformable registration was applied to map the exhale image to the midV anatomy. Cine-MRI scans were acquired for motion quantification. The method was validated with a motion phantom, 10 volunteers and 1 patient with a liver tumor. The time-weighted mean position of a local structure in a cine-scan was defined as the midV-position ground truth and used to determine the accuracy of the midV-triggering method. Deformable registration accuracy was validated using the SIFT algorithm. Acquisition time of the midV/exhale-scan was 3-5min. The accuracy of the midV-position was ⩽0.5±0.5mm for phantom motion and ⩽0.9±1.2mm for the volunteers. Mean residuals after deformable registration were ⩽0.2±1.8mm. The accuracy and reproducibility of the method are within inter- and intra-fraction liver position variability (Case et al., 2009) and could in the future be incorporated in a conventional liver radiotherapy or MR-linac workflow.