Automated method and system for the segmentation of medical images

• US 5,452,367 A
• Filed: 11/29/1993
• Issued: 09/19/1995
• Est. Priority Date: 11/29/1993
• Status: Expired due to Term

• Current Assignee: ARCH Development Corporation (University of Chicago)
• Original Assignee: ARCH Development Corporation (University of Chicago)
• Inventors: Giger, Maryellen L., Bick, Ulrich
• Primary Examiner(s): McElheny, Jr., Donald E.
• Application Number:
  US08/158,320
• Time in Patent Office: 659 Days
• Field of Search: 364/413.13, 364/413.14, 364/413.22, 364/413.16, 378/901, 378/37, 382/6, 382/9
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20036 Morphological image processing
• CPC Class Codes: G06T 2207/20041 Distance transform
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06T 7/155 involving morphological operators
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling
• CPC Class Codes: G06V 10/267 by performing operations on regions, e.g. growing, shrinking or watersheds
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

A method for the automated segmentation of medical images, including generating image data from radiographic images of the breast. The method is applicable to breast mammograms including the extraction of the skinline as well as correction for non-uniform exposure conditions, hand radiographs, and chest radiographs. Techniques for the segmentation include noise filtering, local gray value range determination, modified global histogram analysis, region growing and determination of object contour. The method also is applicable to skin detection and analysis of skin thickening in medical images, where image segmentation, local optimization of external skinline, creation of a gradient image, identification of the internal skinline and then skin thickness determination are carried out. The method for enhancement of medical images includes, after image segmentation and identification of the skinline, calculation of pixel distances from the skinline, determination of the enhancement parameter fit and then selective enhancement of the periphery.

Claims

1. A method of segmentation of an image of an anatomic feature of a subject, comprising:

obtaining said image of said anatomic feature;

obtaining a global histogram using said image;

modifying said global histogram to obtain a modified global histogram; and

segmenting said image using said modified global histogram.

2. A method as recited in claim 1, wherein modifying said global histogram comprises:

determining local minimum pixel values in said image;

determining local maximum pixel values in said image;

determining a gray value range using said local minimum and said local maximum pixel values; and

modifying said global histogram using said range.

3. A method as recited in claim 2, wherein segmenting said image comprises:

classifying a pixel in said image as a non-feature pixel based upon said gray value range, upon whether said pixel is in a vicinity of a gray value peak of said modified histogram, and upon whether said pixel is part of a connected region; and

obtaining a first segmented image containing first pixels having a gray value corresponding to a feature pixel and second pixels having at least one gray value corresponding to said non-feature pixel.

4. A method as recited in claim 3, wherein classifying said pixel in said image based upon whether said pixel is part of said connected region comprises region growing.

5. A method as recited in claim 3, comprising:

performing a morphological open operation on said first segmented image to obtain a second segmented image; and

performing a contouring routine on said second segmented region.

6. A method as recited in claim 3, comprising:

determining a distance map of said first segmented region.

7. A method as recited in claim 6, wherein determining said distance map comprises:

determining a connection path for each first pixel in said first segmented image; and

thresholding said distance map based upon said connection path.

8. A method as recited in claim 7, wherein determining said connection path comprises:

calculating a distance from each first pixel to an edge of said first segmented image if a corresponding connection path does not cross a region of said first segmented image corresponding to a directly exposed region of said image.

9. A method as recited in claim 7, comprising:

thresholding said distance map to generate a second segmented image having pixels corresponding to said feature pixels; and

determining a contour of said feature using said second segmented region.

10. A method as recited in claim 9, wherein determining said contour comprises:

analyzing points along said contour having a connection path to an edge of said second segmented region less than a predetermined length and a concave angle.

11. A method as recited in claim 9, wherein determining said contour comprises:

analyzing points along said contour having a connection path to an edge of said second segmented region crossing a transition zone corresponding to an area in said image between a directly-exposed area and a non-exposed area with a length less than a predetermined length and a concave angle.

12. A method as recited in claim 1, comprising:

obtaining a contour of said feature in said segmented image; and

analyzing points along said contour where said feature connects with a transition zone corresponding to an area in said image between a directly-exposed area and a non-exposed area.

13. A method as recited in claim 1, comprising:

digitizing said image of said feature to obtain a digitized image; and

performing noise filtering on said digitized image to obtain a filtered image;

wherein modifying said global histogram comprises;

determining local minimum pixel values in said filtered image;

determining local maximum pixel values in said filtered image;

determining a gray value range using said local minimum and said local maximum pixel values; and

modifying said global histogram using said range.

14. A method as recited in claim 2, wherein segmenting said image comprises:

determining a first segmented image by classifying first pixels in said image as non-feature pixels based upon said gray value range;

determining a second segmented image by classifying second pixels in said first segmented image as said non-feature pixels based upon whether said second pixels are in a vicinity of a gray value peak of said modified histogram; and

determining a third segmented region by classifying third pixels in said second segmented image as said non-feature pixels based upon whether said third pixels are part of a connected region.

15. A method as recited in claim 14, comprising:

obtaining a fourth segmented image containing fourth pixels having a gray value corresponding to a feature pixel and fifth pixels having at least one gray value corresponding to a non-feature pixel.

16. A method as recited in claim 1, comprising:

obtaining a radiographic breast image; and

segmenting said breast in said radiographic breast image.

17. A method as recited in claim 16, comprising:

obtaining a contour of said breast in said segmented image; and

analyzing points along said contour where said breast connects with a transition zone corresponding to an area in said image between a directly-exposed area and a non-exposed area.

18. A method as recited in claim 1, comprising:

obtaining a radiographic hand image; and

segmenting said hand from said image using local thresholding.

19. A method as recited in claim 18, wherein said local thresholding comprises:

placing a predetermined number of regions-of-interest each having a plurality of pixels on said hand image;

determining a distribution of said pixels in each of said regions-of-interest; and

setting selected pixels in each of said regions-of-interest to a predetermined value based upon corresponding distributions of said pixels.

20. A method as recited in claim 1, wherein:

said step of segmenting said image produces a segmented image; and

said method further comprises;

determining an external skinline of said feature using said image segmented image;

determining an internal skinline of said feature using said segmented image; and

measuring a skin thickness of said feature using said internal and external skinlines.

21. A method as recited in claim 20, comprising:

obtaining a gradient image of said segmented image; and

determining said internal and external skinlines using said gradient image.

22. A method as recited in claim 20, wherein determining said internal skinline comprises:

obtaining a gray value profile perpendicular to a border of said feature; and

determining a local gradient minimum in said gray value profile as said internal skinline.

23. A method as recited in claim 20, wherein determining said external skinline comprises:

determining an initial external skinline of said feature;

calculating a second derivative of a dark-side LaPlacian; and

determining said external skinline as a ridge in said second derivative.

24. A method as recited in claim 23, wherein determining said external skinline as said ridge comprises:

determining a gray scale skeleton of pixels in said gradient image; and

determining a maxima ridge in said skeleton as said external skinline.

25. A method as recited in claim 20, further comprising determining skin thickening using said internal and external skinlines.

26. A method as recited in claim 20, further comprising superimposing said internal and external skinlines on said image.

27. A method as recited in claim 1, wherein:

said step of segmenting said image produces a segmented image; and

said method further comprises;

determining an external skinline using said segmented image;

calculating a distance of pixels of said image being potential feature pixels to said external skinline;

determining an average gray value of said potential feature pixels as a function of said distance; and

enhancing a periphery of said image using said average gray value.

35. A method as recited in claim 1, wherein:

said step of segmenting said image produces a segmented image; and

said method further comprises;

determining an external skinline using said segmented image;

calculating a distance of pixels of said image being potential feature pixels to said external skinline;

determining an average gray value of said potential feature pixels as a function of said distance; and

enhancing a periphery of said image using said average gray value.

28. A method of segmenting and determining a skinline in an image of an anatomic feature of a subject, comprising:

obtaining said image of said anatomic feature;

segmenting said image to obtain a segmented image;

determining an external skinline of said feature using said segmented image;

determining an internal skinline of said feature using said segmented image; and

measuring a skin thickness of said feature using said internal and external skinlines.

29. A method as recited in claim 28, comprising:

obtaining a gradient image of said segmented image; and

determining said internal and external skinlines using said gradient image.

30. A method as recited in claim 28, wherein determining said internal skinline comprises:

obtaining a gray value profile perpendicular to a border of said feature; and

determining a local gradient minimum in said gray value profile as said internal skinline.

31. A method as recited in claim 28, wherein determining said external skinline comprises:

determining an initial external skinline of said feature;

calculating a second derivative of a dark-side LaPlacian; and

determining said external skinline as a ridge in said second derivative.

32. A method as recited in claim 31, wherein determining said external skinline as said ridge comprises:

determining a gray scale skeleton of pixels in said gradient image; and

determining a maxima ridge in said skeleton as said external skinline.

33. A method as recited in claim 28, further comprising determining skin thickening using said internal and external skinlines.

34. A method as recited in claim 28, further comprising superimposing said internal and external skinlines on said image.

36. A system for segmenting images, comprising:

an image acquisition device;

an image memory connected to said image acquisition device;

a local range circuit connected to said image memory;

a modified global thresholding circuit connected to said local range circuit;

a distance map circuit connected to said modified global thresholding circuit;

a contouring circuit connected to distance map circuit;

a contour connection circuit connected to said contouring circuit; and

a display.

37. A system as recited in claim 36, comprising:

a noise filtering circuit connected between said image memory and said local range circuit;

a region growing circuit connected to said modified global histogram circuit;

an erosion circuit connected between said region growing circuit and said distance map circuit; and

a superimposing circuit connected between said contour connection circuit and said display.

38. A system as recited in claim 36, comprising:

a skinline identification circuit.

39. A system for segmenting images, comprising:

an image acquisition device;

means for determining a global histogram of said image connected to said image acquisition device;

means for determining a local range of said global histogram connected to said means for determining a global histogram;

means for modifying said global histogram using said local range;

means for segmenting said image based upon said modified global histogram.

40. A system as recited in claim 39, further comprising:

means for contouring an image segmented by said means for segmenting; and

a display.

41. A system for segmenting and determining a skinline in an image, comprising:

an image acquisition device;

a segmentation circuit connected to said image acquisition device;

a gradient circuit connected to said segmentation circuit;

a local optimization circuit connected to said gradient circuit;

a skinline identification circuit; and

a skin analysis circuit.

42. A system as recited in claim 41, wherein said gradient circuit produces a gradient image;

and

wherein said local optimization circuit comprises;

means for calculating a second derivative of a dark-side LaPlacian of said gradient image; and

means for finding a ridge in said second derivative.

43. A system as recited in claim 42, wherein said means for finding a ridge comprises:

means for generating a gray scale skeleton; and

means for finding a maxima ridge in said gray scale skeleton.

44. A system for segmenting an image having an anatomic feature and determining a skinline in said image, comprising:

an image acquisition device;

a segmentation circuit connected to said image acquisition device;

a skinline identification circuit connected to said segmentation circuit;

a distance circuit connected to said skinline identification circuit; and

an image enhancement circuit connected to said distance circuit.

45. A system as recited in claim 44, wherein:

said segmentation circuit comprises means for obtaining a segmented image having a plurality of pixels;

said distance circuit comprises means for determining a distance of selected pixels of said segmented image from said skinline; and

said image enhancement circuit comprises means for determining average gray values of pixels in said segmented image as a function of said distance and means for enhancing a peripheral region of said feature.

46. A system as recited in claim 45, further comprising:

a curve fitting circuit connected to said enhancement circuit and having means for fitting a curve to said average gray values as a function of said distance.