Method and system for the computerized radiographic analysis of bone

• US 5,931,780 A
• Filed: 11/29/1993
• Issued: 08/03/1999
• 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., Doi, Kunio
• Primary Examiner(s): CASLER, BRIAN L
• Application Number:
  US08/158,388
• Time in Patent Office: 2,073 Days
• Field of Search: 128/653.1, 128/653.2, 128/660.01, 128/920, 128/916, 382/6, 364/413.13, 364/413.14, 600/407, 600/410, 600/437
• US Class Current: 600/407
• CPC Class Codes: A61B 6/4042 K-edge filters
• CPC Class Codes: A61B 6/482 involving multiple energy imaging
• CPC Class Codes: A61B 6/505 for diagnosis of bone
• CPC Class Codes: A61B 6/583 using calibration phantoms
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/30008 Bone
• CPC Class Codes: G06T 2207/30012 Spine; Backbone
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

A computerized method and system for the radiographic analysis of bone structure and risk of future fracture with or without the measurement of bone mass. Techniques including texture analysis for use in quantitating the bone structure and risk of future fracture. The texture analysis of the bone structure incorporates directionality information, for example in terms of the angular dependence of the RMS variation and first moment of the power spectrum of a ROI in the bony region of interest. The system also includes using dual energy imaging in order to obtain measures of both bone mass and bone structure with one exam. Specific applications are given for the analysis of regions within the vertebral bodies on conventional spine radiographs. Techniques include novel features that characterize the power spectrum of the bone structure and allow extraction of directionality features with which to characterize the spatial distribution and thickness of the bone trabeculae. These features are then merged using artificial neural networks in order to yield a likelihood of risk of future fracture. In addition, a method and system is presented in which dual-energy imaging techniques are used to yield measures of both bone mass and bone structure with one low-dose radiographic examination; thus, making the system desirable for screening (for osteoporosis and risk of future fracture).

Claims

1. A method for analyzing bone, comprising:

obtaining an image containing said bone;

selecting a region of interest of said bone;

determining at least one texture measure of said region of interest of said bone; and

analyzing said bone using said at least one texture measure.

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

background-trend correcting said region of interest to produce a background-trend corrected region of interest; and

determining said at least one texture measure of said background-trend corrected region of interest.

3. A method as recited in claim 1, wherein selecting said region of interest comprises:

selecting said region of interest on at least one vertebral body in a spine.

4. A method as recited in claim 1, wherein determining said at least one texture measure comprises:

determining a power spectrum of said region of interest; and

determining said at least one texture measure using said power spectrum.

5. A method as recited in claim 4, comprising:

dividing said power spectrum into a number of sectors; and

determining said at least one texture measure for each of said sectors.

6. A method as recited in claim 4, wherein determining said at least one texture measure comprises:

determining a root mean square variation and a first moment of said power spectrum.

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

filtering said power spectra;

determining said root mean square variation as R;

##EQU4## determining said first moment as M;

##EQU5## where;

V(u,v) is a filter function; and

F(u,v) is said power spectrum.

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

determining an angular dependence Rθ

of said variation R;

##EQU6## determining an angular dependence Mθ

of said first moment M;

##EQU7## for Θ

₁ ≦

tan^-1 (n/m)<

Θ

₂.

9. A method as recited in claim 8, comprising:

dividing said power spectrum into a number of sectors; and

determining said angular dependence of said variation and said angular dependence of M for each of said sectors.

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

determining a minimum of said angular dependence of M.

11. A method as recited in claim 4, wherein determining said at least one texture measure comprises:

determining a root mean square variation of said power spectrum;

determining an angular dependence of said variation;

determining a maximum of said angular dependence;

determining a minimum of said angular dependence;

determining a standard deviation of said variation; and

determining a relative standard deviation of said variation.

12. A method as recited in claim 11, comprising:

inputting selected of said variation, said angular dependence of said variation, said maximum of said angular dependence, said minimum of said angular dependence, said standard deviation of said variation and said relative standard deviation of said variation into a discriminator; and

determining a likelihood of risk of fracture using said discriminator.

13. A method as recited in claim 4, wherein determining said at least one texture measure comprises:

determining a first moment of said power spectrum;

determining an angular dependence of said first moment;

determining a maximum of said angular dependence;

determining a minimum of said angular dependence;

determining a standard deviation of said first moment; and

determining a relative standard deviation of said first moment.

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

inputting selected of said first moment, said angular dependence of said first moment, said maximum of said angular dependence, said minimum of said angular dependence, said standard deviation of said first moment and said relative standard deviation of said first moment into a discriminator; and

determining a likelihood of risk of fracture using said discriminator.

15. A method as recited in claim 4, comprising:

determining a second power spectrum of a uniform tissue region in said image; and

normalizing said power spectrum using said second power spectrum.

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

performing edge gradient analysis on said region of interest; and

determining a maximum cumulative edge gradient.

17. A method as recited in claim 4, comprising:

inputting said at least one texture measure into a discriminator; and

determining a likelihood of risk of fracture using said discriminator.

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

obtaining a spine image;

selecting said region of interest in a vertebral body; and

determining a trabecular pattern in said vertebral body.

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

background-trend correcting said region of interest.

20. A method as recited in claim 18, wherein determining said trabecular pattern comprises:

determining a power spectrum of said region of interest; and

determining said at least one texture measure using said power spectrum.

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

dividing said power spectrum into a number of sectors; and

determining said at least one texture measure for each of said sectors.

22. A method as recited in claim 21, wherein determining said at least one texture measure comprises:

determining a root mean square variation and a first moment of said power spectrum.

23. A method as recited in claim 18, wherein determining said at least one texture measure comprises:

determining a root mean square variation and a first moment of said power spectrum.

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

performing edge gradient analysis on said region of interest; and

determining a maximum cumulative edge gradient.

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

inputting said at least one features into a discriminator; and

determining a likelihood of risk of fracture of using said discriminator.

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

obtaining a second image of said bone;

selecting a second region of interest in said second image;

determining bone mass based on said region of interest and said second region of interest; and

analyzing said bone based on said bone mass and said texture measure.

27. A method as recited in claim 26, wherein determining said bone mass comprises:

integrating said region of interest to obtain a first integration value;

integrating said second region of interest to obtain a second integration value; and

determining a weighted sum of said first and second integration values.

28. A method as recited in claim 26, comprising:

obtaining said image and said second image such that each has a plurality of pixels with a gray level value; and

determining a linear fit from said gray level values of said image and said second image.

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

determining a first sum of gray levels of first pixels in said image over said region of interest;

determining a second sum of gray levels of second pixels in said second image over said second region of interest; and

determining said bone mass based on said first and second sums and said linear fit.

30. A method as recited in claim 26, comprising:

obtaining said image at a first energy level and having a plurality of first gray-level pixels;

obtaining said second image at a second energy level higher than said first energy level and having a plurality of second gray-level pixels;

integrating said image using said region of interest to obtain a first integration value;

integrating said second image using said second region of interest to obtain a second integration value;

determining a linear fit from said first and second gray-level pixels;

determining first and second weighted sums of said first and second integration values; and

determining said bone mass using said first and second weighted sums and said linear fit.

31. A method as recited in claim 26, comprising:

inputting said at least one texture measure and said bone mass into a discriminator; and

determining a likelihood of risk of fracture using said discriminator.

32. A method as recited in claim 31, wherein:

determining said at least one texture measure comprises;

determining a power spectrum of said region of interest; and

determining a root mean square variation and a first moment of said power spectrum; and

determining said bone mass comprises;

integrating said region of interest to obtain a first integration value;

integrating said second region of interest to obtain a second integration value; and

determining a weighted sum of said first and second integration values.

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

selecting said region of interest having a plurality of gray level pixels; and

determining a surface area of said region of interest at each of a plurality of levels of resolution.

34. A method as recited in claim 33, wherein determining said surface area comprises:

determining a slope using said surface areas and said levels of resolution;

determining a fractal dimension using said slope.

35. A method as recited in claim 34, comprising:

determining said fractal dimension D=2-H, where H is said slope.

36. A method as recited in claim 33, comprising:

determining a first surface area of said region of interest based upon said pixels;

selectively combining said pixels to obtain combined pixels;

determining a second surface area based upon said combined pixels; and

determining a fractal dimension based upon said first and second surface areas.

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

determining a slope using said first and second surface areas and said levels of resolution; and

determining said fractal dimension D=2-H, where H is said slope.