Room: Exhibit Hall
Purpose: Motion-induced distortion of organs could cause target delineation errors in the application of motion tracking using a pre-clinical MR-linac. The purpose of this study was to preliminarily quantify object distortion during respiratory cycles using an MR-compatible motion phantom. Results were compared to 4D CT-simulation, the current gold standard.
Methods: A Quasar motion phantom was used in this study with an acrylic insert reciprocating sinusoidally mimicking the respiratory motion of thoracic patients. The acrylic insert was filled with deionized water and MR contrast, and an air-filled sphere with a 3.3cm outer diameter was placed at the center to mimic tumor motion. MR-linac and large bore CT systems were used for motion tracking. Four dimensional images of the motion phantom were retrospectively reconstructed from 10 phases of MR and CT images. The amplitude of target motion and respiratory frequency was 15mm and 15bpm, respectively, which was common for patient simulation motion. An image registration-based tracking algorithm was used for target tracking during the respiratory cycle. The central coronal image was selected for comparison.
Results: The sphere target was tracked with delineation contours among 10 phases of the 4D MR and CT scans. The average area of the coronal slice of the target for the 10 phases was 700.1Â±55.1mmÂ² using MR. The average area of the CT was 826.2Â±20.5mmÂ². The calculated area of the central slice of the sphere target was 855.3 mmÂ².
Conclusion: Both MR and CT scans showed reduced area compared to the calculated target dimension. This could have resulted from approximating the central slice position. The standard deviation of MR images was higher than that of CT, agreeing with the apparent distortion of the sphere in the 4D MR images.