Border detection for medical imaging

• US 7,022,073 B2
• Filed: 04/02/2003
• Issued: 04/04/2006
• Est. Priority Date: 04/02/2003
• Status: Active Grant

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Fan, Liexiang, Sumanaweera, Thilaka S., Von Behren, Patrick L.
• Primary Examiner(s): IMAM, ALI M
• Application Number:
  US10/405,136
• Publication Number:
  US 20040197015A1
• Time in Patent Office: 1,098 Days
• Field of Search: 600407-472, 382/199, 736/25, 736/26, 367/7, 367/11, 367/130, 367/138, 128/916
• US Class Current: 600/437
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/20104 Interactive definition of region of interest [ROI]
• CPC Class Codes: G06T 2207/20116 Active contour; Active surface; Snakes
• CPC Class Codes: G06T 2207/30101 Blood vessel; Artery; Vein; Vascular
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/149 involving deformable models, e.g. active contour models
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

Methods and systems for detecting a border in a medical image are provided. A boundary is chosen as a connected curve whose tangent is substantially perpendicular to the gradient of the image everywhere along the curve. As an alternative to a tangent, a normal or other border direction may be used. At a given point within the image, the tangent to the boundary and the image gradient direction are orthogonal. Using an initial boundary detection, the boundary associated with the minimum cost or associated with the closest boundary where the boundary tangent and the image gradient directional are orthogonal for locations along the boundary is identified. By refining an initial border location to minimize divergence from the boundary tangent being orthogonal to the image gradient direction or by identifying a border based on the orthogonal relationship, accurate border detection may be provided in ultrasound images as well as other medical images.

Claims

1. A method for detecting a border in a medical image, the method comprising:

(a) determining a gradient direction for each of plurality of locations in the medical image;

(b) identifying potential border pixels;

(c) determining a set of potential borders as a function of connecting locations of the potential border pixels; and

(d) detecting the border, the detecting being a function of the gradient direction and the set of potential borders.

2. The method of claim 1 wherein the function of (d) is a minimization of a cost function dependent on the gradient and a normal to the border.

3. The method of claim 2 wherein the cost function is computed only over selected ones of the potential borders that are predicted to have decreasing costs.

4. The method of claim 1 wherein (a) comprises determining the gradient direction for each of the plurality of locations in a three-dimensional volume, the image associated with one of a plane within the volume and a rendering of the volume, (d) comprises determining a normal to a surface at each of the plurality of locations, and (e) comprises detecting the border as connected locations based on the normal and gradient direction at each of the plurality of locations.

5. A method for detecting a border in a medical image, the method comprising:

(a) determining a gradient direction for each of a plurality of locations in the image;

(b) detecting an initial boundary;

(c) identifying a border direction on the initial boundary, the border direction relative to the gradient direction for the plurality of locations; and

(d) detecting the border as connected locations of the plurality of locations, the detecting being a function of the gradient direction and the border direction.

6. The method of claim 5 wherein (c) comprises identifying a tangent to the gradient direction at each of the plurality of locations.

7. The method of claim 6 wherein (d) comprises detecting the border as the connected locations where the tangents are about perpendicular to the gradient directions along the border.

8. The method of claim 5 further comprising:

(e) receiving an indication of a region of interest;

wherein (b) comprises determining the initial boundary as a function of image intensity and the indication; and

wherein (d) is performed as a function of the initial boundary.

9. The method of claim 5 further comprising:

(e) minimizing a cost of a cost function as a function of the gradient direction and the border direction for each of the plurality of locations;

wherein (d) comprises selecting the connected locations corresponding to the minimum cost.

10. The method of claim 9 wherein (e) comprises calculating cost multiple times for each of the plurality of locations, the calculating for each of the multiple times being a function of different possible border directions;

wherein the cost for each of the plurality of locations is selected as the minimum of the multiple calculations for a given location.

11. The method of claim 5 wherein the medical image comprises a longitudinal vessel view, (a) comprises determining the gradient direction for each of a plurality of locations in a region of interest, and (c) comprises identifying a border direction for each of the plurality of locations in the region of interest, the border direction being selected as a function of calculations for a plurality of possible border directions for each location.

12. The method of claim 5 wherein the medical image comprises a cross-sectional vessel view, (a) comprises determining the gradient direction for each of a plurality of locations along an edge of a geometric shape approximating the cross-section of the vessel, (c) comprises identifying the border direction as the tangent to the geometric shape at each of the plurality of locations.

13. The method of claim 12 further comprising:

(e) minimizing a cost of a cost function in response to different parameters defining the geometric shape, the cost function being a function of the gradient direction and the tangent for each of the plurality of locations; and

(f) determining a different parameter of the geometric shape for the minimization of (e) as a function of the image intensity;

wherein (d) comprises selecting a final geometric shape as a function of the parameters corresponding to a minimum cost.

14. The method of claim 12 wherein (d) comprises:

(d1) altering the geometric shape as a function of the tangent and the gradient direction; and

(d2) selecting a final geometric shape from the altered geometric shapes;

further comprising;

(e) determining a vessel diameter as a function of the final geometric shape.

15. A method for detecting a border in a medical image, the method comprising:

(a) determining a gradient for a plurality of locations in the image; and

(b) identifying at least a line segment in response to a tangent to the line segment, the tangent substantially perpendicular to a direction of the gradient for each location of the plurality of locations along the line segment.

16. The method of claim 15 wherein (b) comprises:

(b1) calculating a minimum cost for each of the plurality of locations, the minimum cost for each location being a function of the tangent at the location and the direction of the gradient for the location; and

(b2) selecting the line segment as adjacent locations with a minimum sum of the minimum costs.

17. The method of claim 15 wherein (b) comprises identifying a closed geometric form.

18. The method of claim 15 wherein (b) comprises identifying a line without connected ends.

19. A system for detecting a border in a medical image, the system comprising:

a memory operable to store the medical image; and

a processor operable to determine a gradient direction for each of a plurality of locations in the image, identify a border direction relative to the gradient direction for the plurality of locations, and detect the border as connected locations of the plurality of locations, the detecting being a function of the gradient direction and the border direction.

20. The system of claim 19 further comprising:

a user input device, wherein the processor is operable to determine an initial boundary in response to user input with the user input device.

21. The system of claim 19 wherein the processor is operable to minimize a cost of a cost function as a function of the gradient direction and the border direction for each of the plurality of locations and select the connected locations corresponding to the minimum cost functions.

22. The system of claim 19 wherein the processor is operable to identify a closed geometric form as the border.

23. The system of claim 19 wherein the processor is operable to identify a line without connected ends as the border.