Click here to


Are you sure ?

Yes, do it No, cancel

A Classification Algorithm to Reconstruct Voxel-By-Voxel Motion Trajectory From 4D-CT Images Using Displacement-Vector-Fields Calculated by Deformable Image Registration

N Alsbou1*, S Ahmad2, I Ali2 , (1) Department of Engineering and Physics, University of Central Oklahoma, Edmond, OK, (2) University of Oklahoma Health Sciences, Oklahoma City, OK


(Monday, 7/15/2019) 1:15 PM - 1:45 PM

Room: Exhibit Hall | Forum 5

Purpose: To develop a classification algorithm to reconstruct motion trajectory on a voxel-by-voxel basis using deformable-image-registration (DIR) of 4D-CT images. The displacement-vector-fields (DVF) were used to extract patient inter-phase motion and the cumulative motion from the different motion phases was used to reconstruct the motion trajectory for all voxels of the tumor and surrounding organs.

Methods: 4D-CT images of lung tumors were registered with different DIR-algorithms and DVF were calculated. CT image sets were reconstructed from 10-phases over the motion-cycle. The different image sets from the different phases were registered to the CT image set from end-of-exhale phase which was used as reference image. A motion algorithm was developed to extract and classify the DVF and reconstruct the motion trajectory for each voxel in the CT-images. The performance of this algorithm was verified with 4D-CT images of a mobile thorax-phantom.

Results: This algorithm used the DVF to reconstruct the motion trajectory of all the voxels in the CT-images. While the external-marker motion correlated only with respiratory motion, the DVF were affected by the different motions. The DVF for some voxels changed linearly with the motion of the external marker which were classified as respiratory motion. However, the motion of other voxels did not change linearly and was out of phase with the marker motion which was classified to result from other patient motions that included relaxations and variations in the filling of organs such as stomach close to the tumor. The performance of this algorithm was verified with DIR of CT-images of the mobile thorax phantom.

Conclusion: A novel algorithm was developed to classify and extract the motion trajectory from the DVF obtained from DIR-algorithms of 4D-CT images. This algorithm has potential clinical applications for correction of motion artifacts and anatomical variations in CT-images and for treatment with radiation therapy.


Registration, Respiration, Motion Artifacts


Not Applicable / None Entered.

Contact Email