Click here to

Session: Multi-Disciplinary General ePoster Viewing [Return to Session]

A Voxel-By-Voxel Motion Modeling Algorithm Using Deformable Image Registration and 4D-CT Imaging

T Meyers1*, N Alsbou2, O Algan1, S Ahmad1, I Ali1, ((1) University of Oklahoma Health Sciences, Oklahoma City, OK, (2) Department of Engineering and Physics, University of Central Oklahoma, Edmond, OK

Presentations

PO-GePV-M-170 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: To develop a 4D-patient specific motion modeling that can accurately depict motion of the individual voxels using 4D-CT imaging and deformable image registration (DIR).

Methods: Respiratory gated 4D-CT patient datasets were exported into DIRART-software to apply DIR algorithms for each phase of a patient’s respiratory cycle. The 4D-CT image sets of five patients were investigated where 10 respiratory phases were collected (0-90%) with phase 0 representing the stationary reference CT-image set at the end of exhale. Subsequent phases represented the moving CT-image sets. Four deformable image registration algorithms (Demons, Horn-Schunck-Optical-Flow, Iterative-Optical-Flow, and Level-Set-Motion) were used to register the moving CT-image (10%-90%) sets to the stationary CT-image.

Results: The deformation vector fields (DVFs) were obtained in 3-dimensions for every voxel in the 4D-CT images. The DVFs for a single-voxel located along the patient’s diaphragm were plotted across each respiratory phase which depicted the motion of that voxel as it traveled through the respiratory cycle. The different DIR algorithms produced relatively similar motion-patterns of the voxel-of-interest traveling farther away from its point of origin as the patient inhaled and the lungs expanded, then returning close to the origin during exhaling. Although the DVFs for a certain voxel in the CT-image differed slightly comparing the four DIR-algorithms, the pattern of the DVFs remained consistent in the different directions and used to model the patient respiratory motion.

Conclusion: A 4D-motion model was developed to extract motion trajectories using DIR for all voxels in the 4D-CT images. These motion trajectories could be used to deform the dose distributions covering the planning target volumes and critical structures in treatment-planning-system to create 4D-dose-optimized treatment plans. This motion model may come up with a robust tool for adaptive radiation therapy that could potentially provide an alternative for the management of patient motion and anatomical variations.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email

Share: