Multiscale Segmentation of the Aorta in 3D Ultrasound Images

K. Krissian, J. Ellsmere, K. Vosburgh, R. Kikinis, C.-F. Westin
EMBS, Cancun, Mexico
25th Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society (EMBS)
Pages 638-641
September, 2003

Download full paper

Abstract

Fast, reliable segmentation of the abdominal aorta from three dimensional ultrasound remains a difficult problem. Standard methods based on local information like thresholding, region growing or active contours fail in separating the arteries from the veins and suffer from the lack of homogeneity of the vessel intensity and from the partial contour information. We propose to use a model-based multiscale detection of the vessels centerlines based on a cylindrical model with circular cross-section, and adapted from previous work. Our method provides a set of centerlines and an estimate of the vessel radii along each line. After an interactive selection of the desired lines, a model of the aorta is generated using the radii information and compared to a manual segmentation. This model can also be locally improved using a level set technique in order to stick to the contours of the image and to allow non-circular cross-section.

Figure 5 compares the model generated from a manual segmentation with the reconstructed model from the centerlines and the radii estimates. Left, model generated from a manual segmentation. Right, model generated from the multiscale approach.

Reference

Krissian K, Ellsmere J, Vosburgh K, Kikinis R, Westin CF. Multiscale segmentation of the aorta in 3d ultrasound images. In 25th Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society (EMBS). Cancun, Mexico, 2003;638-641.

Bibtex entry

@INPROCEEDINGS{krissianEMBS03,
  author         = {K. Krissian and J. Ellsmere and K. Vosburgh and R. Kikinis 
                   and C.-F. Westin},                                          
  title          = {Multiscale Segmentation of the Aorta in 3D Ultrasound      
                   Images},                                                    
  booktitle      = {25th Annual Int. Conf. of the IEEE Engineering in Medicine 
                   and Biology Society (EMBS)},                                
  pages          = {638-641},                                                  
  month          = {September},                                                
  year           = {2003},                                                     
  address        = {Cancun, Mexico}
}

Grants

CIMIT, NIH P41-RR13218 (NAC)

Research areas

Level Sets, Segmentation, Vessles