Method and system for improved image segmentation

• US 20090226060A1
• Filed: 03/04/2009
• Published: 09/10/2009
• Est. Priority Date: 03/04/2008
• Status: Active Grant

• Current Assignee: TomoTherapy Incorporated (Accuray Incorporated)
• Original Assignee: TomoTherapy Incorporated (Accuray Incorporated)
• Inventors: Ruchala, Kenneth J., Gering, David T.
• Granted Patent: US 8,577,115 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 2207/20128 Atlas-based segmentation
• CPC Class Codes: G06T 2207/30004 Biomedical image processing
• 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 7/174 involving the use of two or more images

Abstract

A system and method of identifying anatomical structures in a patient. The method includes the acts of acquiring an image of the patient, the image including a set of image elements; segmenting the image to categorize each image elements according to its substance; computing the probability that the categorization of each image element is correct; resegmenting the image starting with image elements that have a high probability and progressing to image elements with lower probabilities; aligning at least one of the image elements with an anatomical atlas; and fitting the anatomical atlas to the segmented image.

Claims

1. A method of identifying anatomical structures in a patient, the method comprising:

acquiring an image of the patient, the image comprising a set of image elements;

segmenting the image to categorize at least some image elements according to their substance;

aligning the segmented image to an anatomical atlas;

computing the probability that the categorization of the image elements is correct; and

resegmenting the image emphasizing image elements based on their probabilities.

2. A method as set forth in claim 1 wherein the anatomical atlas is a probabilistic atlas.

3. A method as set forth in claim 1 and further comprising modifying the atlas based on data relating to the patient.

4. A method as set forth in claim 1 wherein the computing the probability that the categorization is correct further comprises using rules based logic to determine the probability of each image element being correctly categorized.

5. A method as set forth in claim 4 and further comprising modifying the probabilities based on data relating to the patient.

6. A method as set forth in claim 1 and further comprising developing an anatomical mesh representing an anatomical structure, the anatomical mesh having a surface, and fitting the anatomical mesh to the segmented image.

7. A method as set forth in claim 6 wherein the anatomical mesh has vertices, and wherein the fitting the anatomical mesh to the segmented image is based on the probabilities of the image elements near the vertices.

8. A method as set forth in claim 6 wherein the developing an anatomical mesh representing an anatomical structure, the anatomical mesh having a surface, and fitting the anatomical mesh to the segmented image further comprises developing a virtual surface representing an anatomical structure, the surface being governed by control points, and fitting the surface to the segmented image by adjusting the control points based on the image elements.

9. A method as set forth in claim 8 wherein the fitting the surface to the segmented image by adjusting the control points based on the image elements is based on the probability of the image elements.

10. A method as set forth in claim 1 and further comprising manually editing the segmentation, computing a difference between the initial segmentation and the manually edited segmentation;

and incorporating the difference into the initial segmentation to create a revised segmentation.

11. A method as set forth in claim 1 wherein the image element comprises a voxel.

12. A method as set forth in claim 1 and further comprising using rules-based analysis to infer the position of non-anatomical structures based on the known position of anatomical structures.

13. In a radiation therapy system, a method of automatically identifying anatomical structures in a patient, the method comprising:

acquiring an image of the patient;

segmenting the image using a hierarchical series of image processing steps to identify landmarks in the image; and

identifying the anatomical structures using the landmarks.

14. A method as set for the in claim 13 wherein the identifying the anatomical structures using the landmarks further comprises fitting a shape model to the image using at least one landmark.

15. A method as set forth in claim 14 wherein the shape model is an ellipse.

16. A method as set forth in claim 14 wherein the shape model is an anatomical mesh.

17. A method as set forth in claim 13 wherein the landmarks represent substantially the entire structure of a muscle or bone.

18. A method as set forth in claim 13 wherein the image comprises a set of three-dimensional image data divided into a series of two-dimensional image slices.

19. A method as set forth in claim 18 wherein each image slice comprises a group of image elements, and wherein the segmenting the image using a hierarchical series of image processing steps further comprises analyzing each image element to determine a probable tissue class, and analyzing each image element relative to neighboring image elements to confirm or modify the probability.

20. A method as set forth in claim 19 wherein the segmenting the image using a hierarchical series of image processing steps further comprises classifying image elements into tissue groups.

21. A method as set forth in claim 20 wherein the segmenting the image using a hierarchical series of image processing steps further comprises identifying organs, glands, muscles or nodal regions based on the tissue groups.

22. A method as set forth in claim 21 wherein the segmenting the image using a hierarchical series of image processing steps further comprises identifying systems based on the organs, glands, muscles or nodal regions and refining the image in response to the systems.

23. A method as set forth in claim 13 wherein the identifying the anatomical structures using the landmarks further comprises establishing a region of interest known to contain an anatomical structure by drawing lines between landmark structures, recognizing candidate tissues within the region of interest, segmenting a rough draft of the region of interest, fitting a shape model to the draft, smoothing the shape model parameters and reconciling the draft with the shape model to determine the borders of the anatomical structure.

24. A method as set forth in claim 23 wherein the shape model is at least one geometric primitive.

25. A method of segmenting an image of a patient to identify the tissue class of each image element in the image, the method comprising:

analyzing each image element using a set of rules to determine a probability of each image element being a given tissue class;

aligning an anatomical atlas to the image;

refining the tissue probabilities of each image element based on the atlas; and

fitting a mesh to the image elements based on the probabilities, the mesh representing a structural model of an organ and establishing an expected shape of an organ.

26. A method as set forth in claim 25 wherein the anatomical atlas is a probabilistic atlas.

27. A method as set forth in claim 25 wherein the anatomical atlas is derived from a large volume of scanned images.

28. A method as set forth in claim 25 wherein the anatomical atlas comprises a single template scan.

29. A method as set forth in claim 25 wherein the image element is a voxel.

30. A method as set forth in claim 25 and further comprising modifying the rules based on patient data.

31. A method as set forth in claim 25 and further comprising modifying the mesh based on patient data.

32. A method as set forth in claim 25 and further comprising modifying the anatomical atlas based on patient data.