Method for segmenting image data for detecting a liver

• US 20090003672A1
• Filed: 06/13/2008
• Published: 01/01/2009
• Est. Priority Date: 06/15/2007
• Status: Active Grant

• Current Assignee: Siemens Healthcare GMBH (Siemens AG)
• Original Assignee: Siemens AG
• Inventors: Wimmer, Andreas, Maier, Florian, Soza, Grzegorz
• Granted Patent: US 8,090,180 B2
• Time in Patent Office: Days
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/20156 Automatic seed setting
• CPC Class Codes: G06T 2207/30056 Liver; Hepatic
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling
• CPC Class Codes: G06V 10/457 by analysing connectivity, e.g. edge linking, connected component analysis or slices
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

A method is disclosed for segmenting image data for detecting a liver of a subject under examination. In at least one embodiment, the method includes determining as rib pixels which represent a rib, determining pixels which delimit an area inside the rib cage, determining pixels which represent a pixel of the liver and determining a probability, for each pixel inside the rib cage, as to whether the pixel belongs to the liver or not, with the aid of a Random Walker method.

Claims

1. A method for segmenting image data for detecting a liver of a subject under examination, wherein the image data include a number of pixels, each pixel including one signal value allocated thereto, the method comprising:

determining, as rib pixels, pixels which represent a rib of the subject under examination;

determining pixels which delimit an area inside a rib cage of the subject under examination, with the aid of the determined rib pixels;

determining at least one pixel which represents a pixel of the liver of the subject under examination as a liver pixel, within pixels which are located in an area inside the rib cage;

defining, as background seed points, the determined rib pixels and the determined pixels which delimit the area inside the rib cage;

defining, as liver seed points, the at least one determined liver pixel; and

determining a probability, for each pixel of an area inside the rib cage, as to whether the pixel belongs to the liver or not, with the aid of a Random Walker method, with the aid of the defined liver seed points and with the aid of the defined background seed points.

2. The method as claimed in claim 1, wherein the image data include a number of image data layers which are arranged in anatomical order of the subject under examination,

wherein the determining of pixels which represent a rib is carried out separately for various layers of the image data.

3. The method as claimed in claim 1, wherein the determining of pixels which represent a rib also includes:

determining pixels which are located inside the subject under examination and pixels which are located outside the subject under examination in order to determine a first boundary which represents a skin surface of the subject under examination;

determining pixels which are located inside a fatty tissue and the skin surface of the subject under examination and pixels which are located inside a muscle tissue under the fatty tissue of the subject under examination in order to determine a second boundary between the fatty tissue and the muscle tissue on the basis of the first boundary; and

determining pixels which represent a rib in only one area which extends from the second boundary towards the center of the subject under examination.

4. The method as claimed in claim 3, wherein the extent of the area which extends from the second boundary towards the center of the subject under examination is selected in dependence on the location of the pixels.

5. The method as claimed in claim 1, wherein the determining of pixels which delimit the area inside the rib cage includes a joining of the rib pixels by way of a continuous line, wherein pixels which are located on the continuous line are defined as background seed points.

6. The method as claimed in claim 1, further comprising:

determining individual ribs by combining coherent pixels which belong to precisely one rib, and determining a center of each individual rib and determining a pixel for each individual rib which corresponds to the center of the rib, as rib center.

7. The method as claimed in claim 6, wherein the determining of pixels which delimit an area inside the rib cage includes a joining of the rib centers by way of a spline interpolation, wherein pixels which are located on the spline interpolation are defined as background seed points.

8. The method as claimed in claim 1, wherein a signal value histogram of the signal values of the pixels which are located inside the rib cage is determined and

wherein pixels with a signal value within a first variance of a first maximum of the signal value histogram are defined as background seed points and pixels with a signal value within a second variance of a second maximum of the signal value histogram are defined as liver seed points.

9. The method as claimed in claim 1, wherein the determination with the aid of the Random Walker method includes a first coarse determination on the basis of image data of a first resolution which has a coarser resolution than the original resolution and a second fine determination on the basis of the results of the first coarse determination and image data in the original resolution.

10. The method as claimed in claim 9, wherein, for the second fine determination, pixels for which a minimum probability for belonging to the liver was determined in the first determination, are used as additional liver seed points.

11. The method as claimed in claim 9, wherein, for the second fine determination, pixels for which a minimum probability for belonging to the background was determined in the first determination, are used as additional background seed points.

12. The method as claimed in claim 1, wherein the image data include a number of image data layers which are arranged in anatomical order of the subject under examination,

wherein the pixels of an image data layer are addressed in a system of polar coordinates, and wherein the origin of the system of polar coordinates is arranged approximately in a centroid of the layer of the subject under examination.

13. The method as claimed in claim 1, wherein an edge value function w_i,j of the Random Walker method between two adjacent pixels i and j is determined in accordance with the equation

w_ij=e⁻

β

·

(Δ

d)<sup>2·

c·

n·

f(gi-gj)</sup>,

where n is a scaling value which is selected in dependence on a selected function f and a signal value range of the pixels,

Δ

d is a distance between the pixels i and j,

c is a scaling value which is selected in dependence on a movement of the subject under examination between determinations of the pixels i and j which are temporally offset,

β

is a normalization factor which is selected in dependence on a contrast range of the image data, and

f is a function for determining the difference between the signal values g_i and g_j.

14. The method as claimed in claim 13, wherein the following applies to the function f:

f(g_i-g_j)=|g_i-g_j|.

15. The method as claimed in claim 13, wherein the following applies to the function f:

f(g_i-g_j)=(g_i-g_j)².

16. The method as claimed in claim 1, wherein the image data include computer tomography data and wherein the signal values include Hounsfield values.

17. The method as claimed in claim 1, wherein pixels which represent at least one of air and other gases are defined as background seed points.

18. The method as claimed in claim 1, wherein pixels which represent bones are defined as background seed points.

19. The method as claimed in claim 1, wherein pixels which represent fat are defined as background seed points.

20. The method as claimed in claim 1, wherein pixels which represent an aorta are defined as background seed points.

21. The method as claimed in claim 1, wherein the image data for each pixel include a number of signal values, wherein the number of signal values are detected successively in time, the method further comprising:

defining pixel blocks which include a number of pixels which are located in the area inside the rib cage,

performing a perfusion of the subject under examination during the detection of the number of temporally successive signal values,

determining a pixel block as a starting pixel block in which the number of pixels which exceed a threshold value rises during the temporally successive signal value detection, and

defining as liver seed points pixels of the starting pixel block which exceed the threshold value.

22. The method as claimed in claim 21, wherein a pixel block is defined as starting pixel block if the number of pixels which exceed a threshold value rises during the temporally successive signal value detection, and if the number of pixels which exceed a threshold value includes a proportion of the total number of pixels of the pixel block.

23. The method as claimed in claim 22, wherein the proportion includes at least 40%.

24. The method as claimed in claim 21, wherein the method further comprises:

applying a region growing method to the pixel blocks, wherein the starting pixel block is used as starting point for the region growing method, and

determining pixel blocks which, according to the region growing method, belong to the starting pixel block and defining as liver seed points pixels which belong to the determined pixel blocks and exceed the threshold value.

25. The method as claimed in claim 1, wherein the liver seed points and the background seed points also have probability information of belonging to the liver or to the background, respectively, wherein the membership probability information is used by the Random Walker method for determining the probability for each pixel of an area inside the rib cage whether the pixel belongs to the liver or not.

27. A device for segmenting image data for detecting a liver of a subject under examination, wherein the image data include a number of pixels, each pixel being allocated one signal value, the device comprising:

an element for determining as rib pixels, pixels which represent a rib of the subject under examination;

an element for determining pixels which delimit an area inside a rib cage of the subject under examination, with the aid of the determined rib pixels;

an element for determining at least one pixel which represents a pixel of the liver as a liver pixel within pixels which are located in the area inside the rib cage;

an element for defining, as background seed points, the rib pixels and the pixels which delimit the area inside the rib cage;

an element for defining as liver seed points the determined at least one liver pixel; and

an element for determining a probability for each pixel of an area inside the rib cage, of whether the pixel belongs to the liver or not, with the aid of a Random Walker method, with the aid of the defined liver seed points and with the aid of defined the background seed points.

wherein the device is designed for carrying out the method as claimed in claim 1.

28. A computer program product which is loadable directly into a memory of a programmable processing unit of a device for segmenting image data, with program segments for carrying out all steps of a method as claimed in claim 1 when the program is executed in the processing unit.

29. An electronically readable data medium with electronically readable control information stored thereon, designed in such a manner that, when the data medium is used in a processing unit of a device for segmenting image data, it carries out the method as claimed in claim 1.

30. The method as claimed in claim 2, wherein the determining of pixels which represent a rib also includes:

determining pixels which are located inside the subject under examination and pixels which are located outside the subject under examination in order to determine a first boundary which represents a skin surface of the subject under examination;

determining pixels which are located inside a fatty tissue and the skin surface of the subject under examination and pixels which are located inside a muscle tissue under the fatty tissue of the subject under examination in order to determine a second boundary between the fatty tissue and the muscle tissue on the basis of the first boundary; and

determining pixels which represent a rib in only one area which extends from the second boundary towards the center of the subject under examination.

31. The method as claimed in claim 10, wherein, for the second fine determination, pixels for which a minimum probability for belonging to the background was determined in the first determination, are used as additional background seed points.

26. A device for segmenting image data for detecting a liver of a subject under examination, wherein the image data include a number of pixels, each pixel being allocated one signal value, the device comprising:

an element for determining as rib pixels, pixels which represent a rib of the subject under examination;

an element for determining pixels which delimit an area inside a rib cage of the subject under examination, with the aid of the determined rib pixels;

an element for determining at least one pixel which represents a pixel of the liver as a liver pixel within pixels which are located in the area inside the rib cage;

an element for defining, as background seed-points, the rib pixels and the pixels which delimit the area inside the rib cage;

an element for defining as liver seed points the determined at least one liver pixel; and

an element for determining a probability for each pixel of an area inside the rib cage, of whether the pixel belongs to the liver or not, with the aid of a Random Walker method, with the aid of the defined liver seed points and with the aid of defined the background seed points.