Raj Varadhan PhD, Abstract Co-Author: Nothing to Disclose

Susanta K. Hui PhD, Presenter: Nothing to Disclose

To establish the fundamental relationship between deformation and its causative physical force using a deformable bladder phantom. To ascertain if a threshold limit exists for DIR accuracy, beyond which its applicability in deformed anatomy may be clinically inappropriate.

A tissue equivalent deformable bladder phantom with 21 implanted aluminum markers was developed using a viscoelastic polymer with Young’s modulus and physical density properties comparable to human bladder. Applied force on the organ was incrementally varied from 10N to 70N which in turn deforms the organ. DIR accuracy was studied for commercially available algorithms (MIM and Velocity AI) by comparing the centroid (3D vector) of the 21 marker locations at the undeformed CT (ground truth) with the synthetically derived marker positions from each target image obtained from DIR

The relationship between applied force and both 1D deformation along the axis of applied force (R2 =0.99) and 3D deformation as quantified by the 95% hausdorff distance (R2 =0.96) showed a linear response. The maximum 1D deformation (compression) ranged from 3mm to 28 mm. A threshold limit at 30N (maximum13mm 1D and 6.8 mm 3D deformation) was established beyond which average registration error of 21 markers was ≥ 2mm for all algorithms studied. Utilizing the “Reg Reveal” feature of the MIM software, the marker accuracy beyond 50N improved by 2mm while the average marker error using Velocity multipass deformation was ≤3mm up to 50N.  

The accuracy of DIR was evaluated using a tissue equivalent mass and density conserving bladder phantom in the best case scenario using 21 implanted aluminum high contrast markers to improve the accuracy. The limits of applicability of DIR are strongly dependent on the magnitude of deformation. There is a threshold limit beyond which the accuracy of DIR fails for the range of mass and density conserving deformation studied.  

There is a threshold limit in the accuracy of deformable image registration beyond which its applicability in deformed anatomy may be clinically inappropriate

Varadhan, R, Hui, S, Is there a Threshold Limit in the Accuracy of Deformable Image Registration (DIR)? A Validation Study Using Deformable Bladder Phantom with Implanted Markers.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14019067.html