MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Evaluation of An Intensity Based Deformable Registration Method for the Generation of Liver Volumes On Post Y90 PET/CT

N Lamba*, A Kruzer , S Pirozzi , A Nelson , MIM Software Inc., Cleveland, OH

Presentations

(Monday, 7/30/2018) 1:15 PM - 1:45 PM

Room: Exhibit Hall | Forum 9

Purpose: In Yttrium-90 (Y90) microsphere therapy, pre-treatment liver volumes are necessary for determining the amount of activity to order for therapy. Now that methods are available for calculating absorbed dose on post-Y90 PET or SPECT images, liver volumes are important for generating DVHs and other dose statistics on post-treatment exams. In this study, our goal was to evaluate the accuracy of an automatic intensity based deformable image registration (IDIR) method for transferring liver contours from pre-treatment CT scans to post-treatment Y90 PET/CT scans.

Methods: 24 non-contrast CTs were selected from 12 subjects with liver cancer from a single institution. Each study had two time points. Liver contours were manually defined on each CT scan by a radiation oncologist or an experienced anatomist. The first CT was deformed to the second CT using the IDIR method. Mean Dice similarity coefficients (DSC) and the mean distance to agreement (MDA) were calculated by comparing the liver contours generated by the deformable method to those generated by manual contouring.

Results: Based on liver contours from all 12 subjects, the mean DSC obtained from the IDIR method was 0.93 ± 0.03, and the mean MDA was 2.08 ± 1.12 mm.

Conclusion: This intensity based deformable image registration method demonstrated accurate contour transfer with a mean DSC of 0.93. This method could be used in conjunction with new dose quantification methods on post-treatment Y90 exams by providing fast and accurate liver contour transfer from pre to post Y90 PET/CT and SPECT/CTs. We plan to continue to investigate this and other deformation methods for the purpose of improving contour transfer through deformable fusion of liver Y90 images.

Funding Support, Disclosures, and Conflict of Interest: All authors are employees of MIM Software Inc.

Keywords

Deformation, Registration, Nuclear Medicine

Taxonomy

IM/TH- image segmentation: CT

Contact Email