Room: Exhibit Hall | Forum 2
Purpose: To develop and validate a 3-dimensional (3-D) motion assessment framework for rectal tumor during radiation treatment using dynamic contrast enhanced (DCE) MRI.
Methods: A motion correction tool (MCFLIRT) originally designed for use on fMRI time series is adapted in this study to assess the 3D intra-fractional motion of rectal cancer during radiotherapy. This algorithm is optimized for sub-pixel volumetric motion tracking of MRI images. A DCE-MRI image dataset with 31 frames was acquired with an imaging time of 5.92 minutes. The extent of the tumor motion during this period was assessed. To validate the algorithm, 4D (3D + time) sequences with pre-defined small (<0.5 mm) and large (mm level) motions along A-P, L-R, and S-I directions were generated using 3D volume of a selected timestamp. MCFLIRT tracking is then performed to inspect gross body motion (Whole 3-D volume), as well as region-of-interest (ROI) motion (3-D volume with 3-D binary mask) and compared with the ground truth. Finally, motions of the DCE scans from the same patient were analyzed.
Results: For both small and large gross body motion, the tracking error is less than 0.1 mm; the tracking error is less than 0.2 mm for small ROI motion, and less than 11% for total large ROI motion. Patient result demonstrates less than 0.25 mm motion in A-P and L-R direction, and 0.6 mm in S-I direction for gross body motion; ROI tracking shows less than 2mm total motion during the 5.92 minutes acquisition time.
Conclusion: The feasibility of sub-pixel 3D rigid motion tracking using MCFLIRT is validated in DCE pelvis MRI scans to be within Â±0.2 mm accuracy for both gross body and ROI volume. This study constructs a framework to quantify small, but significant motions of tumor volume and organ-at-risk using DCE-MRI, which has comparable time as EBRT treatment.