Methods and systems for segmentation using boundary reparameterization

• US 8,199,981 B2
• Filed: 05/18/2007
• Issued: 06/12/2012
• Est. Priority Date: 05/18/2006
• Status: Active Grant

• Current Assignee: Elekta Ltd. (Elekta AB)
• Original Assignee: Elekta Ltd. (Elekta AB)
• Inventors: Koptenko, Sergei, Lachaine, Martin, Huang, Xing, Falco, Tony
• Primary Examiner(s): Wang, Claire X
• Application Number:
  US11/804,594
• Publication Number:
  US 20080008369A1
• Time in Patent Office: 1,852 Days
• Field of Search: 382/128, 382/152, 382/224, 382/190, 382/173, 382/131, 382/132
• US Class Current: 382/128
• 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 2207/30096 Tumor; Lesion
• CPC Class Codes: G06T 3/073 Transforming surfaces of revolution to planar images, e.g. cylindrical surfaces to planar images
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06V 10/44 Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

Representations of a segmented, contoured organ or lesion are obtained from two-dimensional or three-dimensional images. A contour within the image of the lesion or organ of interest is used to identify a region around the initial contour and transform it into a boundary image comprising sampling lines that contain points identifying the organ boundary.

Claims

1. A computer-implemented method of generating a segmented image of an anatomical structure, the method comprising:

a) providing, from a register to a mapping module, an initial boundary estimate of the anatomical structure;

b) determining, by the mapping module, a boundary band region that includes the initial boundary estimate of the anatomical structure by sampling image data along a series of lines normal to the initial border estimate;

c) converting the boundary band region to a rectangular boundary image and the boundary estimate to a curve within the rectangular boundary image such that the boundary estimate is defined relative to a coordinate grid of the rectangular boundary image;

d) using a processor, (i) segmenting the boundary estimate within the rectangular boundary image and (ii) mapping the segmented boundary estimate from the rectangular boundary image onto the boundary band region to obtain a segmented boundary estimate;

e) repeating steps b) through d) using the segmented boundary estimate from the rectangular boundary image as the initial boundary estimate within an updated boundary band, until the tolerance difference is below a threshold; and

(f) generating, using the processor, an improved segmented image of the anatomical structure from the segmented boundary estimate.

2. The method of claim 1 further comprising obtaining a tolerance difference between the segmented boundary estimate and the initial boundary estimate.

3. The method of claim 2 wherein the threshold is modified for at least one repetition of steps b) through d).

4. The method of claim 2 wherein a width of the boundary band is adjusted for at least one iteration of steps b) through d).

5. The method of claim 1 wherein determining the boundary band comprises the mapping module expanding the initial boundary estimate in multiple directions by a predefined amount.

6. The method of claim 1 wherein the step of determining the boundary band comprises the mapping module sampling image data along a series of lines normal to the initial boundary estimate and mapping the sampled data into a rectangular array.

7. The method of claim 1 wherein the step of determining the boundary band comprises the mapping module sampling image data using a plurality of co-radial lines, each of the lines passing through the initial boundary estimate, and mapping the sampled data into a rectangular array.

8. The method of claim 1 wherein the step of determining the boundary band comprises the mapping module sampling image data with one or more curved lines passing through the initial boundary estimate and mapping the sampled data into a rectangular array.

9. The method of claim 1 wherein the step of segmenting the boundary band comprises the processor:

i) calculating a weighted-sum image comprising global and local statistics of the image;

ii) applying one or more thresholds to the weighted-sum image to form a binary image;

iii) detecting an edge curve based on the binary image; and

iv) modifying one or more data points along the edge curve.

10. The method of claim 1 wherein the step of segmenting the boundary band comprises the processor using one or more of level sets, active contours, active shapes, deformable templates, graph-based techniques, statistical clustering, Markov random field-based techniques or active-appearances methods.

11. The method of claim 1 wherein the step of segmenting the boundary band comprises the processor implementing one or more of thresholding of the boundary band, gradient edge-detection or Laplacian-edge detection to detect an edge in the boundary band.

12. The method of claim 1 wherein the step of segmenting the boundary band comprises the processor using texture information to detect an edge in the boundary band.

13. The method of claim 9 further comprising the processor deriving the thresholds based on statistical measurements of the image.

14. The method of claim 13 wherein the statistical measurements comprise one or more of a mean of the weighted sum of the image or an estimate of the mean of the image.

15. The method of claim 9 wherein the thresholds are determined using one or more of gradient edge-detection methods, Laplacian edge-detection methods or second-derivative gradient-direction edge-filter methods.

16. The method of claim 9 further comprising the processor removing false binary islands.

17. The method of claim 9 wherein modifying one or more data points along the edge curve comprises removing false local concavities.

18. The method of claim 9 wherein modifying one or more data points along the edge curve comprises removing erroneous edge points, and fitting a spline curve through remaining edge points.

19. The method of claim 1 wherein the image is a three-dimensional image, and further comprising the processor dividing the three-dimensional image into a plurality of two-dimensional images prior to performing steps b) through d) thereon.

20. The method of claim 19 further comprising the processor forming a three-dimensional mesh from the two-dimensional segmented boundary estimates generated by performing steps b) through d) on the plurality of two-dimensional images.

21. The method of claim 19 further comprising maintaining continuity between the two-dimensional segmented boundary estimates.

22. The method of claim 19 wherein the new segmented boundary band is based, at least in part, on data sampled from points in proximity to the initial boundary estimate.

23. The method of claim 1 wherein the initial boundary estimate is arbitrarily defined.

24. The method of claim 1 wherein the initial boundary estimate user-provided.

25. The method of claim 1 where the initial boundary estimate is based, at least in part, on a limited number of user-provided points.

26. A system for generating a segmented image of an anatomical structure, the system comprising:

a) a register for receiving an image and an initial boundary estimate of an anatomical structure represented in the image; and

b) a mapping module for;

i) determining a boundary band region that includes the initial boundary estimate of the anatomical structure by sampling image data along a series of lines normal to the initial border estimate; and

ii) converting the boundary band region to a rectangular boundary image and the boundary estimate to a curve within the rectangular boundary image such that the boundary estimate is defined relative to a coordinate grid of the boundary image; and

c) a processor for;

ii) segmenting the boundary estimate within the rectangular boundary image;

iii) mapping the segmented boundary estimate from the rectangular boundary image onto the boundary band region to obtain a segmented boundary estimate; and

iv) generating a segmented image of the anatomical structure from the segmented boundary estimate;

wherein the mapping module and processor iteratively segment and map the segmented boundary band using the new boundary estimate generated based on the rectangular boundary image as the initial boundary estimate within the boundary band.

27. The system of claim 26 wherein the image is obtained using one of a CT scanner, a three-dimensional ultrasound device, a PET scanner, a SPECT scanner and an MRI.

28. An article of manufacture having computer-readable program portions embodied thereon for segmenting representations of an anatomical structure in an image, the article comprising non-transitory computer-readable instructions for:

a) providing an initial boundary estimate of the anatomical structure;

b) determining a boundary band that includes the initial boundary estimate of the anatomical structure by sampling image data along a series of lines normal to the initial border estimate;

c) converting the boundary band region to a rectangular boundary image and the boundary estimate to a curve within the rectangular boundary image such that the boundary estimate is defined relative to a coordinate grid of the rectangular boundary image;

d) segmenting the boundary estimate within the rectangular boundary image;

e) mapping the segmented boundary estimate from the rectangular boundary image onto the boundary band region to obtain a segmented boundary estimate;

f) repeating steps b) through e) using the segmented boundary estimate from the rectangular boundary image as the initial boundary estimate within the boundary band, until the tolerance difference is below a threshold; and

g) mapping the segmented boundary band onto the image, thus providing a segmented boundary estimate.