Using GVF Snake to Segment Liver from CT Images

1
Using GVF Snake to Segment Liver from CT Images

• Shaohui Huang, Boliang Wang, Xiaoyang Huang

2
outline

• Traditional Active Contour (Snake)
• Gradient Vector Flow Snake (GVF Snake)
• SEGMENT CT IMAGES
• Edge map generation
• Contour initialization
• Maximum force angle map

3
Traditional Active Contour (Snake)
4
Snake
5
Gradient Vector Flow Snake (GVF Snack)

• The GVF field is defined to be a vector field
  V(x,y) (u(x,y),v(x,y)) , it comes from the
  force balance condition by replace the potential
  force
• with V(x,y), then we rewrite
• as

6
GVF Snack

• V(x,y) is defined such that it minimizes the
  energy functional
• where f(x,y) is the edge map of the image. GVF
  field can be obtained by solving following
  equations

7
GVF Snack
8
SEGMENT CT IMAGES

• Edge map generation
• Contour initialization
• Maximum force angle map

9
Edge map generation

• Gaussian function with sigma 1

10
Edge map generation
11
Edge map generation

• Two errors with this contour
• Over-segment of the liver
• Caused by the blur boundary.
• Error convergence inside the liver
• Caused by the fake boundary.

12
Edge map generation

• Canny edge detector

13
Edge map generation
14
Contour initialization

• The contour cannot cross the bottleneck of the
  liver.

15
Maximum force angle map

• Let u, v stand for two GVF forces. Note that u,
  v are vectors. The angle ? formed by u, v can be
  calculated as
• A maximum force angle MFA at the coordinate (m,n)
  is defined as

16
Maximum force angle map
17
Result