Autonomous boundary detection system for echocardiographic images

• US 6,716,175 B2
• Filed: 01/11/2002
• Issued: 04/06/2004
• Est. Priority Date: 08/25/1998
• Status: Expired due to Fees

• Current Assignee: University of Florida (State University System of Florida)
• Original Assignee: University of Florida (State University System of Florida)
• Inventors: Wilson, David C., Geiser, Edward A.
• Primary Examiner(s): Getzow, Scott M.
• Application Number:
  US10/042,303
• Publication Number:
  US 20020151793A1
• Time in Patent Office: 816 Days
• Field of Search: 600/450, 600/425, 600/437, 600/443
• US Class Current: 600/450
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/12 Edge-based segmentation

Abstract

The invention comprises a method and apparatus for generating a synthetic echocardiographic image. The method comprises first obtaining, for a plurality of pathologically similar reference hearts, a reference echocardiographic image of each reference heart at end-systole and at end-diastole. Next, the coupled epicardial and endocardial borders are identified in each echocardiographic image. An epicardial/endocardial border pair is then modeled from the identified borders. The method then locates a plurality of predetermined features in the reference echocardiographic images. The predetermined features are then located in the subject echocardiographic image from the location of the predetermined features in the reference echocardiographic images. The modeled epicardial/endocardial border pair is then mapped onto the subject echocardiographic image relative to the location of the predetermined features in the subject echocardiographic image. The apparatus generally comprises an echocardiographic machine for obtaining the echocardiographic images that are then processed by a computing system. In one aspect of the invention, the invention comprises such a computing system programmed to perform the autonomous portions of the method. In another aspect, the invention comprises a program storage medium encoded with instructions that perform the autonomous portions of the method when executed by a computer.

Claims

1. A method for generating a synthetic echocardiographic image, comprising:

(a) retrieving a modeled epicardial/endocardial border pair;

(b) retrieving the location of a plurality of predetermined features identified in a plurality of modeled echocardiographic images;

(c) locating the predetermined features in an echocardiographic image of subject heart; and

(d) mapping the modeled epicardial/endocardial border pair onto the subject echocardiographic image relative to the location of the predetermined features in the subject echocardiographic image.

2. The method of claim 1, further comprising:

(a) obtaining, for a plurality of pathologically similar reference hearts, an echocardiographic image of each reference heart at end-systole and at end-diastole;

(b) identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image; and

(c) storing the identified border pairs in a data structure.

3. The method of claim 2, wherein obtaining the echocardiographic image of each reference heart includes obtaining an echocardiographic image of each reference heart in an apical four chamber view, an apical two-chamber view, or a parasternal short-axis view.

4. The method of claim 2, wherein obtaining the echocardiographic image of each reference heart includes obtaining an echocardiographic image for a plurality of pathologically similar reference hearts, manifesting normalcy, pericardial effusion, apical aneurysm, or dilated cardiomyopathy.

5. The method of claim 2, wherein obtaining the echocardiographic image of each reference heart includes:

(a) obtaining a video comprising a sequence of digitized echocardiographic images;

(b) displaying the sequence of digitized echocardiographic images;

(c) selecting two images from the displayed sequence; and

(d) capturing the selected images.

6. The method of claim 2, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image includes manually tracing the borders on the echocardiographic images.

7. The method of claim 2, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image includes invoking an autonomous border detection capability.

8. The method of claim 7, further comprising manually adjusting the autonomously detected borders.

9. The method of claim 1, wherein generating the synthetic image includes:

(a) modeling a coupled epicardial and endocardial border pair from the identified epicardial and endocardial borders; and

(b) mapping the modeled epicardial and endocardial border pair into the synthetic image.

10. The method of claim 9, wherein modeling the coupled epicardial and endocardial border pair includes comprises a Procrustean method.

11. The method of claim 10, wherein the Procrustean method comprises averaging a collection of identified borders traced on a plurality of echocardiographic images of reference hearts.

12. The method of claim 11, wherein averaging the collection of expert defined borders includes:

(a) translating a plurality of curves representing the expert defined borders to an origin;

(b) reparameterizing each translated curve such that each reparamaterized curve is defined by the same number of points;

(c) scale each curve to a vector length of one;

(d) rotate each curve to align the matrices they define using a singular value decomposition; and

(e) averaging the translated, reparameterized, scaled, and rotated curves.

13. The method of claim 9, wherein mapping the modeled epicardial and endocardial border pair includes:

(a) performing a thin-plate spline transformation to embed the modeled epicardial border in the synthetic image; and

(b) warping the septal and lateral walls of the embedded epicardial border.

14. The method of claim 1, further comprising:

(a) identifying a plurality of predetermined features in the echocardiographic images; and

(b) storing the identified predetermined features in a data structure.

15. The method of claim 14, wherein identifying the predetermined features in the echocardiographic images includes manually identifying the predetermined features.

16. The method of claim 15, wherein identifying the predetermined features in the echocardiographic images includes identifying at least one of the medial mitral annulus position, the lateral mitral annulus position, and the epicardial apex of the left ventricle.

17. The method of claim 15, wherein identifying the predetermined features in the echocardiographic images includes identifying the features in a stepwise, autonomous, hierarchical manner.

18. The method of claim 1, wherein locating the predetermined features in the synthetic image from the identified predetermined features includes:

(a) generating a synthetic echocardiographic image at end-systole and at end-diastole from the reference echocardiographic images at end-systole and the reference echocardiographic images at end-diastole, respectively;

(b) creating a convolution filter for each of the predetermined features from the synthetic image; and

(c) applying the convolution filters in a hierarchical manner to locate the predetermined features in serial.

19. The method of claim 1, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image and identifying the predetermined features in the echocardiographic images includes identifying the borders and the predetermined features in the same copies of the echocardiographic images.

20. The method of claim 1, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image and identifying the predetermined features in the echocardiographic images includes identifying the borders and the predetermined features in different copies of the echocardiographic images.

21. A program storage medium encoded with instructions that, when executed by a computer, perform a method for generating a synthetic echocardiographic image, the method comprising:

(a) retrieving a modeled epicardial/endocardial border pair;

(b) retrieving the location of a plurality of predetermined features identified in a plurality of modeled echocardiographic images;

(c) locating the predetermined features in an echocardiographic image of subject heart; and

(d) mapping the modeled epicardial/endocardial border pair onto the subject echocardiographic image relative to the location of the predetermined features in the subject echocardiographic image.

22. The program storage medium of claim 21, wherein the encoded method further comprises:

(a) identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image in the encoded method includes invoking an autonomous border detection capability; and

(b) storing the identified border pair for subsequent retrieval.

23. The program storage medium of claim 21, wherein generating the synthetic image in the encoded method includes:

(a) modeling a coupled epicardial and endocardial border pair from the identified epicardial and endocardial borders; and

(b) mapping the modeled epicardial and endocardial border pair into the synthetic image.

24. The program storage medium of claim 23, wherein modeling the coupled epicardial and endocardial border pair in the encoded method comprises a Procrustean method.

25. The program storage medium of claim 23, wherein the Procrustean method in the encoded method comprises averaging a collection of identified borders traced on a plurality of echocardiographic images of reference hearts.

26. The program storage medium of claim 25, wherein averaging the collection of expert defined borders in the encoded method includes:

(a) translating a plurality of curves representing the expert defined borders to an origin;

(b) reparameterizing each translated curve such that each reparamaterized curve is defined by the same number of points;

(c) scale each curve to a vector length of one;

(d) rotate each curve to align the matrices they define using a singular value decomposition; and

(e) averaging the translated. reparameterized, scaled, and rotated curves.

27. The program storage medium of claim 21, wherein mapping the modeled epicardial and endocardial border pair in the encoded method includes:

(a) performing a thin-plate spline transformation to embed the modeled epicardial border in the synthetic image; and

(b) warping the septal and lateral walls of the embedded epicardial border.

28. The program storage medium of claim 21, further comprising:

(a) identifying a plurality of predetermined features in the echocardiographic images; and

(b) storing the identified predetermined features in a data structure.

29. The program storage medium of claim 28, wherein identifying the predetermined features in the echocardiographic images in the encoded method includes manually identifying the predetermined features.

30. The program storage medium of claim 28, wherein identifying the predetermined features in the echocardiographic images in the encoded method includes identifying at least one of the medial mitral annulus position, the lateral mitral annulus position, and the epicardial apex of the left ventricle.

31. The program storage medium of claim 28, wherein identifying the predetermined features in the echocardiographic images in the encoded method includes identifying the features in a stepwise, autonomous, hierarchical manner.

32. The program storage medium of claim 21, wherein locating the predetermined features in the synthetic image from the identified predetermined features in the encoded method includes:

(a) generating a synthetic echocardiographic image at end-systole and at end-diastole from the reference echocardiographic images at end-systole and the reference echocardiographic images at end-diastole, respectively;

(b) creating a convolution filter for each of the predetermined features from the synthetic image; and

(c) applying the convolution filters in a hierarchical manner to locate the predetermined features in serial.

33. The program storage medium of claim 21, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image and identifying the predetermined features in the echocardiographic images in the encoded method includes identifying the borders and the predetermined features in the same copies of the echocardiographic images.

34. The program storage medium of claim 21, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image and identifying the predetermined features in the echocardiographic images in the encoded method includes identifying the borders and the predetermined features in different copies of the echocardiographic images.

35. An apparatus, comprising a computing system programmed to perform a method for generating a synthetic echocardiographic image, the method comprising:

(a) retrieving a modeled epicardial/endocardial border pair;

(b) retrieving the location of a plurality of predetermined features identified in a plurality of modeled echocardiographic images;

(c) locating the predetermined features in an echocardiographic image of subject heart; and

(d) mapping the modeled epicardial/endocardial border pair onto the subject echocardiographic image relative to the location of the predetermined features in the subject echocardiographic image.

36. The apparatus of claim 35, further comprising an echocardiograph machine capable of generating the echocardiographic images.

37. The apparatus of claim 35, wherein the programmed method further comprises:

(a) identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiogoraphic image in the programmed method includes invoking an autonomous border detection capability; and

(b) storing the identified border pair for subsequent retrieval.

38. The apparatus of claim 35, wherein generating the synthetic image in the programmed method includes:

(a) modeling a coupled epicardial and endocardial border pair from the identified epicardial and endocardial borders; and

b) mapping the modeled epicardial and endocardial border pair into the synthetic image.

39. The apparatus of claim 38, wherein modeling the coupled epicardial and endocardial border pair in the programmed method comprises a Procrustean method.

40. The apparatus of claim 39, wherein the Procrustean method in the programmed method comprises averaging a collection of identified borders traced on a plurality of echocardiographic images of reference hearts.

41. The apparatus of claim 40, wherein averaging the collection of expert defined borders in the programmed method includes:

(a) translating a plurality of curves representing the expert defined borders to an origin;

(b) reparameterizing each translated curve such that each reparamaterized curve is defined by the same number of points;

(c) scale each curve to a vector length of one;

(d) rotate each curve to align the matrices they define using a singular value decomposition; and

(e) averaging the translated, reparameterized, scaled, and rotated curves.

42. The apparatus of claim 35, wherein mapping the modeled epicardial and endocardial border pair in the programmed method includes:

(a) performing a thin-plate spline transformation to embed the modeled epicardial border in the synthetic image; and

(b) warping the septal and lateral walls of the embedded epicardial border.

47. The apparatus of claim 35, wherein locating the predetermined features in the synthetic image from the identified predetermined features in the programmed method includes;

(a) generating a synthetic echocardiographic image at end-systole and at end-diastole from the reference echocardiographic images at end-systole and the reference echocardiographic images at end-diastole, respectively;

(b) creating a convolution filter for each of the predetermined features from the synthetic image; and

(c) applying the convolution filters in a hierarchical manner to locate the predetermined features in serial.

48. The apparatus of claim 35, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image and identifying the predetermined features in the echocardiographic images in the programmed method includes identifying the borders and the predetermined features in the same copies of the echocardiographic images.

49. The apparatus of claim 35, wherein identifying the coupled epicardial and endocardial borders of each reference heart in each echocardiographic image and identifying the predetermined features in the echocardiographic images in the programmed method includes identifying the borders and the predetermined features in different copies of the echocardiographic images.

43. The apparatus of claim 55, further comprising:

(a) identifying a plurality of predetermined features in the echocardiographic images; and

(b) storing the identified predetermined features in a data structure.

44. The apparatus of claim 43, wherein identifying the predetermined features in the echocardiographic images in the programmed method includes manually identifying the predetermined features.

45. The apparatus of claim 43, wherein identifying the predetermined features in the echocardiographic images in the programmed method includes identifying at least one of the medial mitral annulus position, the lateral mitral annulus position, and the epicardial apex of the left ventricle.

46. The apparatus of claim 43, wherein identifying the predetermined features in the echocardiographic images in the programmed method includes identifying the features in a stepwise, autonomous, hierarchical manner.