Encrypted login | home

Program Information

CT Number and Size Variations of Well-Known Mobile Targets in Deformable Image Registration Algorithms in CT Images with Motion Artifacts Induced by Controlled Motion Patterns

no image available
J Jaskowiak

J Jaskowiak1*, N Alsbou2 , O Algan1 , S Ahmad1 , I Ali1 , (1) University of Oklahoma Health Sciences Center, Oklahoma City, OK, (2) Ohio Northern University, Ada, OH,


TH-EF-BRA-9 (Thursday, July 16, 2015) 1:00 PM - 2:50 PM Room: Ballroom A

Purpose: To evaluate quantitatively the performance of different deformable image registration algorithms using a mobile thorax phantom with well-defined targets and controlled motion-patterns imaged with helical, axial and cone-beam CT.
Materials and Methods: A thorax phantom was scanned using helical, axial and cone-beam CT imaging while it is moving and stationary. The phantom system was moved using cyclic motion with adjustable motion amplitude (0-40 mm) and frequency (0.125-0.5Hz). Several water equivalent targets (small, medium and large) with known sizes were inserted in the middle of low-density foam. The CT images of the mobile phantom were registered with the images of the stationary phantom with different deformable image registration algorithms using the DIRART software. The performance of different deformable image registration algorithms were evaluated in terms of reducing image artifact and reproducing the shape, size and CT-number level of well-defined targets.
Results: The image artifacts induced by phantom motion which included deformation and spread-out of the targets increased with motion amplitude. The deformable image registration improved image-quality from artifacts induced by motion in the deformed CT images. The CT-number profiles showed that the deformable image registration algorithms were able to reproduce the shape and size of the mobile targets where the spread-out artifacts were eliminated. However, the CT-numbers deviated from the corresponding values for the stationary targets. The discrepancy between the CT-number level from the deformed CT images and that from the stationary CT images increased with motion amplitude.
Conclusions: The performance of the deformable image registration depends on several parameters that include: motion artifacts, deformable image registration algorithm, and imaging modes. The deformable image registration algorithms reproduced size and shape of mobile targets with different motion-patterns. However, they did not reproduce the CT-number value which deviated strongly from the actual value of the stationary CT images with increasing motion amplitude.

Contact Email: