Biomedical Engineering Reference
In-Depth Information
(a) Initialization contours
(b) Gray matter
(c) White matter
(d) Cerebrospinal fluid
Figure 7.
3D MRI segmentation using three-phase level set model: (a) initial surface; 27
small spheres are initialized for
φ
0
, 4 larger spheres are initialized for
φ
1
; (b) final GM
segmentation result; (c) final WM segmentation result; (d) final CSF segmentation result.
4.
CONCLUSIONS AND FUTURE RESEARCH
We proposed a variational segmentation algorithm that measures the dissim-
ilarity between local and global regions. Region boundaries were modeled as
a regular hypersurface implicitly represented by level set functions. The image
regions were assumed to follow Gaussian distributions. A geodesic curve-based
energy was imposed on the energy functional to reduce segmentation errors when
