Author: Kin Yin Cheung, Chen-Yu Huang, Chi Wa Kong, Pei-Xiong Li, Pak Hang Nam, Bin Yang, Siu Ki Yu, Chi To Yung 👨🔬
Affiliation: Medical Physics Department, Hong Kong Sanatorium and Hospital 🌍
Purpose:
The Elekta Unity system facilitates daily adaptive radiotherapy using MRI-based treatment planning. However, MR images are prone to motion artefacts caused by respiratory motion, potentially leading to inaccuracies in target volume delineation. This study investigates variations in tumour size determined under different MRI sequences and motion patterns.
Methods:
Different MRI sequences (T1 3DVaneXD [FFE], b3DVaneXD [FFE], T2 3D Tra [SE], and T1 3D Tra [FFE]) were employed to image a CIRS MRgRT Motion Management QA Phantom under static and simulated respiratory motion conditions. Motion patterns varied in superior-inferior amplitude and cycle time. The resulting MR images were registered to planning CT images in the Monaco planning system to minimize setup uncertainties. Target volumes were delineated, inferior edge extension and target centroid positions in MR were quantified relative to CT. Standardized window settings were applied to ensure consistent evaluation across all sequences.
Results:
Static tumour contour volumes (17.59±0.02cc) and target centroid position (0.04±0.09cm) showed no significant differences in images acquired by different MRI sequences as compared with CT image. Under identical imaging and motion condition, variations in cycle time (4 vs. 6sec) did not significantly affect target volume (0.19cc difference) and inferior edge position (0.0cm difference), suggesting the impact of cycle time in comparisons is negligible. 3DVaneXD image consistently demonstrated smaller increases in target volume and inferior edge extension compared to other MRI sequences, regardless of motion amplitude. For ±1 cm motion amplitude, target volumes (cc) and inferior edge extensions (cm) relative to CT were: T1 3DVaneXD: +6.87, +0.8; b3DVaneXD: +6.58, +0.8; T2 3DTra: +10.59, +1.1. For ±0.5 cm motion amplitude, they were: T1 3DVaneXD: +1.72, +0.3; b3DVaneXD: +1.36, +0.3; T2 3DTra: +6.4, +0.6.
Conclusion:
The 3DVaneXD sequence consistently reduced motion artefacts, which exhibited smaller target volume increases and inferior edge extensions during motion conditions, outperforming T1 and T2 3DTra.