Conference Paper/Proceeding/Abstract 1556 views
Computing 3D Mesh Correspondence for Aortic Root Shape Modelling
19th Conference on Medical Image Understanding and Analysis, Pages: 1 - 6
Swansea University Authors: Gary Tam , Xianghua Xie
Abstract
Aortic valve disorder is one of the common diseases affecting elderly people. To provide visual assessment and improve success of surgical treatment, a segmentation technique equipped with a reliable statistical shape model is required. This in turn requires reliable dense correspondences establishm...
Published in: | 19th Conference on Medical Image Understanding and Analysis |
---|---|
Published: |
19th Conference on Medical Image Understanding and Analysis
2015
|
Online Access: |
http://csvision.swan.ac.uk/uploads/Site/PublicationCategorisedVersion/miua2015_XX.pdf |
URI: | https://cronfa.swan.ac.uk/Record/cronfa22234 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Abstract: |
Aortic valve disorder is one of the common diseases affecting elderly people. To provide visual assessment and improve success of surgical treatment, a segmentation technique equipped with a reliable statistical shape model is required. This in turn requires reliable dense correspondences establishment. This paper develops a reliable 3D registration technique targeting aortic region. Given a few easily identifiable land- mark correspondences, our technique obtains a much denser set of point correspondences across a set of 3D aortic sources meshes to the target mesh. We proposes to use geodesic interpolation, a new mesh based similarity metric, and a two-stage local transformation to develop a better registration technique for 3D aortic meshes. It results in better corre- spondences compared to existing work, shows an average Hausdorff distance of 3.65mm and point-to-mesh distance of 0.41mm. Visual comparison is also provided to assess the quality of the point correspondences. |
---|---|
Keywords: |
Medical image analysis, mesh processing |
College: |
Faculty of Science and Engineering |
Start Page: |
1 |
End Page: |
6 |