Method of segmenting anatomic entities in 3D digital medical images
First Claim
1. A method of segmenting an anatomic entity in a digital medical image, the method comprising the steps of:
- defining, for each of a number of landmark points in the digital medical image, an initial position of each of the landmark points;
connecting a plurality of pairs of the landmark points to each other to form polygons defining a mesh;
constructing a rectangular grid around each of the landmark points in the mesh;
sampling a neighborhood in each rectangular grid around the initial position for each of the connected landmark points in the mesh, the neighborhood including a number of candidate locations of the connected landmark points in the mesh;
calculating a graph data structure for all of the connected landmark points in the mesh, each node (vertex) of the graph data structure representing a connected landmark point of the connected landmark points in the mesh and each edge (arc) of the graph data structure representing the connection between a pair of the landmark points in the mesh;
calculating a path through the graph data structure wherein an order of visiting the connected landmark points in the mesh is determined by the path through the graph data structure, the path being a minimum spanning tree;
optimizing a cost function expressing a weighted sum of an overall gray level cost and an overall shape cost for all of the candidate locations of the connected landmark points in the mesh based on the path; and
calculating a segmented anatomic entity as a surface or a volume through a selected combination of the candidate locations of each of the connected landmark points in the mesh for which combination the cost function is optimized.
2 Assignments
0 Petitions
Accused Products
Abstract
For each of a number of landmarks in an image an initial position of the landmark is defined. Next a neighborhood around the initial position comprising a number of candidate locations of the landmark, is sampled and a cost is associated with each of the candidate locations. A cost function expressing a weighted sum of overall gray level cost and overall shape cost for all candidate locations is optimized. A segmented anatomic entity is defined as a path through a selected combination of candidate locations for which combination the cost function is optimized. During optimization towards the optimal segmented surface/volume graph traversal methods are exploited.
7 Citations
17 Claims
-
1. A method of segmenting an anatomic entity in a digital medical image, the method comprising the steps of:
-
defining, for each of a number of landmark points in the digital medical image, an initial position of each of the landmark points; connecting a plurality of pairs of the landmark points to each other to form polygons defining a mesh; constructing a rectangular grid around each of the landmark points in the mesh; sampling a neighborhood in each rectangular grid around the initial position for each of the connected landmark points in the mesh, the neighborhood including a number of candidate locations of the connected landmark points in the mesh; calculating a graph data structure for all of the connected landmark points in the mesh, each node (vertex) of the graph data structure representing a connected landmark point of the connected landmark points in the mesh and each edge (arc) of the graph data structure representing the connection between a pair of the landmark points in the mesh; calculating a path through the graph data structure wherein an order of visiting the connected landmark points in the mesh is determined by the path through the graph data structure, the path being a minimum spanning tree; optimizing a cost function expressing a weighted sum of an overall gray level cost and an overall shape cost for all of the candidate locations of the connected landmark points in the mesh based on the path; and calculating a segmented anatomic entity as a surface or a volume through a selected combination of the candidate locations of each of the connected landmark points in the mesh for which combination the cost function is optimized. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
Specification