Method for interactive segmentation of a structure contained in an object

• US 20050157925A1
• Filed: 03/17/2003
• Published: 07/21/2005
• Est. Priority Date: 03/23/2002
• Status: Active Grant

• Current Assignee: Koninklijke Philips N.V.
• Original Assignee: Koninklijke Philips N.V.
• Inventors: Lorenz, Cristian
• Granted Patent: US 7,142,208 B2
• Time in Patent Office: Days
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/20101 Interactive definition of point of interest, landmark or seed
• CPC Class Codes: G06T 2207/30008 Bone
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling

Abstract

The invention relates to a method for the interactive segmentation of a structure contained in an object from a three-dimensional image of the object. In an expansion mode a region expansion process is carried out, the results of said process simply being completely or partly undone again in a contraction mode. A quasi-continuous similarity measure can be used to determine whether or not a voxel belongs to the structure to be segmented. The expansion taking place in the region expansion process always involves the voxels having the largest similarity measure each time.

Claims

1. A method for the interactive segmentation of a structure contained in an object from a three-dimensional image of the object, which method involves a continuous visualization of the instantaneous segmentation status in a segmentation image and includes an expansion mode and a contraction mode, the following steps being carried out in the expansion mode

a) determining the similarity between the neighboring voxel of a defined voxel and the structure,

b) storing the order in which the similarity of the voxels belonging to the structure is determined,

c) repeating the steps a) and b) for voxels of the structure which have not yet been processed or have been newly determined,

while in the contraction mode the voxels are removed, on the basis of the instantaneous segmentation tatus, from the segmented structure in an order which is the reverse of that in which they have been determined as belonging to the structure in the expansion mode.

2. A method as claimed in claim 1, in which the expansion mode includes the following steps

a) determining a similarity measure, describing the similarity between a voxel and the structure, for the neighboring voxel of the set voxel of the structure,

b) storing the neighboring voxel in a list,

c) selecting the voxel having the largest similarity measure from the list,

d) determining the similarity measure for the neighboring voxel of the selected voxel,

e) storing neighboring voxels in the list,

f) repeating the steps f) to h) for voxels which have not been selected thus far.

3. A method as claimed in claim 2, in which the number of selection steps can be interactively set and the segmentation, or the reversal of the segmentation, is continued until a segmentation status corresponding to the number of selection steps is reached.

4. A method as claimed in claim 2, in which the list is subdivided into a number of sub-lists (300 . . . 304) in which voxels having the same value of the similarity measure are taken up, each of said sub-lists being associated with a different value of the similarity measure.

5. A method as claimed in claim 2, in which from among the voxels neighboring a voxel selected from the list exclusively those voxels whose similarity measure exceeds a minimum value are taken up in the list.

6. A method as claimed in claim 2, in which the similarity measure of the voxels is derived from their grey value.

7. An image processing device for carrying out the method claimed in claim 1, which device includes

a memory for storing a three-dimensional image of the object and for storing lists which are processed in the course of the segmentation,

an image display unit for displaying a segmented structure, and

image processing means for segmenting a three-dimensional structure which is contained in the object, said image processing means having an expansion mode and a contraction mode,

the following steps being carried out in the expansion mode:

a) determining the similarity between the neighboring voxel of a defined voxel and the structure,

b) storing the order in which the similarity of the voxels belonging to the structure is determined,

c) repeating the steps a) and b) for voxels of the structure which have not yet been processed or have been newly determined,

while in the contraction mode the voxels are removed, on the basis of the instantaneous segmentation status, from the segmented structure in an order which is the reverse of that in which they have been determined as belonging to the structure in the expansion mode.

8. A computer program for an image processing unit as claimed in claim 7 for the segmentation of a three-dimensional structure, contained in an object, from a three-dimensional image of the object, which program includes the steps of:

a) determining the similarity between the neighboring voxel of a defined voxel and the structure,

b) storing the order in which the similarity of the voxels belonging to the structure is determined,

c) repeating the steps a) and b) for voxels of the structure which have not yet been processed or have been newly determined,

while in the contraction mode the voxels are removed, on the basis of the instantaneous segmentation status, from the segmented structure in an order which is the reverse of that in which they have been determined as belonging to the structure in the expansion mode.