Autonomous boundary detection system for echocardiographic images

• US 6,346,124 B1
• Filed: 06/09/2000
• Issued: 02/12/2002
• 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): Kamm, William E.
• Application Number:
  US09/590,799
• Time in Patent Office: 613 Days
• Field of Search: 600/425, 600/437, 600/443, 600/450
• 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

A method for generating a synthesis echocardiographic image 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. Other features include a computing system programmed to perform the autonomous portions of the method and a program storage medium encoded with the instructions that perform the autonomous portions of the method when executed by a computer.

Claims

1. A method for autonomously detecting boundaries in a subject echocardiographic image, the method comprising:

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

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

(c) locating a plurality of predetermined features in each reference echocardiographic images;

(d) modeling an epicardial/endocardial border pair from the identified epicardial and endocardial borders;

(e) locating the predetermined features in the subject echocardiographic image from the location of the predetermined features in the reference echocardiographic images; and

(f) 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, 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.

3. The method of claim 1, wherein obtaining the echocardiographic image of each reference heart includes obtaining an echocardiographic image for a plurality of pathologically similar reference hearts manifesting at least one of normalcy, pericardial effusion, apical aneurysm, or dilated cardiomyopathy.

4. The method of claim 1, 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.

5. The method of claim 1, wherein identifying the epicardial and endocardial borders in each reference echocardiographic image includes manually tracing the borders on the echocardiographic images.

6. The method of claim 5, further comprising manually adjusting the autonomously detected borders.

7. The method of claim 1, wherein identifying the coupled epicardial and endocardial borders in each reference echocardiographic image includes invoking an autonomous border detection capability.

8. The method of claim 1, wherein modeling the coupled epicardial and endocardial border pair includes comprises a Procrustean method.

9. The method of claim 8, wherein the Procrustean method comprises averaging a collection of identified borders traced on a plurality of echocardiographic images of reference hearts.

10. The method of claim 9, 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.

11. The method of claim 10, wherein mapping the modeled epicardial and endocardial border pair includes:

(a) performing a first thin-plate spline transformation to embed the modeled epicardial border in the subject image relative to the located predetermined features in the subject image;

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

(c) performing a second thin-plate spline transformation to embed the modeled endocardial border in the subject image relative to the embedded epicardial border; and

(d) warping the septal and lateral walls of the embedded endocardial border.

12. The method of claim 1, wherein identifying the predetermined features in the echocardiographic images includes manually identifying the predetermined features.

13. The method of claim 1, 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.

14. The method of claim 1, wherein identifying the predetermined features in the echocardiographic images includes identifying the features in a stepwise, autonomous, hierarchical manner.

15. 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.

16. The method of claim 15, generating the synthetic echocardiographic images 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.

17. 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.

18. 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.

19. The method of claim 1, further comprising manually adjusting the mapped epicardial/endocardial border pair on the subject image.

20. The method of claim 1, further comprising utilizing the subject image in the autonomous detection of boundaries in a subsequent subject image.