Multi-Atlas based Segmentation of Head and Neck CT Images using Active Contour Framework

This paper presents the segmentation of bilateral parotid glands in the Head and Neck (H&N) CT images using an active contour based atlas registration. We compare segmentation results from three atlas selection strategies: (i) selection of “single-most-similar” atlas for each image to be segmented, (ii) fusion of segmentation results from multiple atlases using STAPLE, and (iii) fusion of segmentation results using majority voting. Among these three approaches, fusion using majority voting provided the best results. Finally, we present a detailed evaluation on a dataset of eight images (provided as a part of H&N auto segmentation challenge conducted in conjunction with MICCAI-2010 conference) using majority voting strategy.


Published in:
MICCAI workshop on 3D Segmentation Challenge for Clinical Applications
Presented at:
MICCAI workshop on 3D Segmentation Challenge for Clinical Applications, Beijing, China, September 20-24, 2010
Year:
2010
Keywords:
Laboratories:




 Record created 2010-08-18, last modified 2018-03-17

Postprint:
Download fulltextPDF
n/a:
Download fulltextJPG
External link:
Download fulltextURL
Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)