Encrypted login | home

Program Information

Novel Validation Paradigm of MRI to CT Deformation of Prostate

no image available
K Padgett

K Padgett1,2*, S Pirozzi3 , A Nelson3 , R Stoyanova1 , J Piper3 , M Horvat3 , N Dogan1 , A Pollack1 , (1) University of Miami School of Medicine - Radiation Oncology, Miami, FL. (2) University of Miami School of Medicine - Radiology, Miami, FL.(3) MIM Software Inc., Cleveland, Ohio.


SU-E-J-218 (Sunday, July 12, 2015) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose: Deformable registration algorithms are inherently difficult to characterize in the multi-modality setting due to a significant differences in the characteristics of the different modalities (CT and MRI) as well as tissue deformations. We present a unique paradigm where this is overcome by utilizing a planning-MRI acquired within an hour of the planning-CT serving as a surrogate for quantifying MRI to CT deformation by eliminating the issues of multi-modality comparisons.

Methods: For nine subjects, T2 fast-spin-echo images were acquired at two different time points, the first several weeks prior to planning (diagnostic-MRI) and the second on the same day as the planning-CT (planning-MRI). Significant effort in patient positioning and bowel/bladder preparation was undertaken to minimize distortion of the prostate in all datasets. The diagnostic-MRI was rigidly and deformably aligned to the planning-CT utilizing a commercially available deformable registration algorithm synthesized from local registrations. Additionally, the quality of rigid alignment was ranked by an imaging physicist. The distances between corresponding anatomical landmarks on rigid and deformed registrations (diagnostic-MR to planning-CT) were evaluated.

Results: It was discovered that in cases where the rigid registration was of acceptable quality the deformable registration didn’t improve the alignment, this was true of all metrics employed. If the analysis is separated into cases where the rigid alignment was ranked as unacceptable the deformable registration significantly improved the alignment, 4.62mm residual error in landmarks as compared to 5.72mm residual error in rigid alignments with a p-value of 0.0008.

Conclusion: This paradigm provides an ideal testing ground for MR to CT deformable registration algorithms by allowing for inter-modality comparisons of multi-modality registrations. Consistent positioning, bowel and bladder preparation may result in higher quality rigid registrations than typically achieved which limits the impact of deformable registrations. In this study cases where significant differences exist, deformable registrations provide significant value.

Funding Support, Disclosures, and Conflict of Interest: This work was funded by an NIH grant# 1R01CA189295-01. PI: Alan Pollack, MD PhD. "MRI IMAGING AND GENETIC SIGNATURES TO MANAGE PROSTATE CANCER OVERDIAGNOSIS"

Contact Email: