| 摘要 |
Method for processing digital images, such as magnetic resonance images of a brain. The images contain a first object, for example the cerebrum white matter, a second object, for example the cerebral cortex, and a third object, for example the cerebrospinal fluid. The method involves providing a digital dataset representing a deformable curve or surface that in an initial state approximates the boundary between the first object and the second object. Further, a vector force field is provided that for each point on the deformable curve or surface defines a direction from the point approximately towards the second boundary between the second and the third object, and this vector force field is applied for iteratively deforming the deformable curve or surface convergently towards the second boundary. In order to reconstruct a folded structure of the second object, for example the cortex, the vector force field also involves a normal vector field with a vector normal or approximately normal to the deformable curve or surface for each point on the deformable curve or surface.
|
