System and method for boundary estimation using CT metrology

• US 7,203,267 B2
• Filed: 06/30/2004
• Issued: 04/10/2007
• Est. Priority Date: 06/30/2004
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Basu, Samit Kumar, De Man, Bruno Kristiaan Bernard, Ye, Jong Chul
• Primary Examiner(s): Thomas; Courtney
• Assistant Examiner(s): TANINGCO, ALEXANDER H
• Application Number:
  US10/880,875
• Publication Number:
  US 20060002504A1
• Time in Patent Office: 1,014 Days
• Field of Search: 378/4, 378/382
• US Class Current: 378/4
• CPC Class Codes: G06T 11/006 Inverse problem, transformation from projection-space into object-space, e.g. transform methods, back-projection, algebraic methods
• CPC Class Codes: G06T 2211/424 Iterative

Abstract

A technique is provided for CT reconstruction for use in CT metrology. The boundary based CT reconstruction method includes the steps of initializing a boundary of an object to obtain a boundary estimate, defining a forward model based on the boundary estimate, linearizing the forward model to obtain a system matrix and implementing an iterative image reconstruction process using the system matrix to update the boundary estimate.

Claims

1. A method for computed tomography (CT) reconstruction, the method comprising:

initializing a boundary of an object to obtain a boundary estimate;

defining a forward model based on the boundary estimate;

linearizing the forward model to obtain a system matrix;

implementing an iterative image reconstruction process using the system matrix to update the boundary estimate;

generating an image of the object based on a final boundary estimate; and

displaying or storing the image for use by a user.

2. The method of claim 1, wherein the boundary corresponds to an edge of a region of substantially uniform attenuation.

3. The method of claim 1, wherein the boundary estimate is represented directly.

4. The method of claim 3, wherein the boundary estimate is defined by at least one spline.

5. The method of claim 4, wherein each spline is a piecewise linear contour represented by a plurality of corner points.

6. The method of claim 4, wherein the at least one spline is of second or higher order.

7. The method of claim 3, wherein the boundary estimate is defined parametrically.

8. The method of claim 7, wherein the boundary estimate comprises a weighted sum of a plurality of basis functions.

9. The method of claim 8, wherein the plurality of basis functions comprise at least one of a harmonic decomposition, a wavelet decomposition and a polynomial expansion.

10. The method of claim 3, wherein the boundary estimate is represented by a plurality of coordinates, wherein the coordinates are modeled by a distance from at least one origin as a function of an angle relative to an axis.

11. The method of claim 3, wherein the boundary estimate is represented by a plurality of Cartesian coordinates.

12. The method of claim 3, wherein the boundary estimate is represented by a plurality of coordinates, wherein the coordinates are modeled by their respective perpendicular distances to a reference boundary.

13. The method of claim 1, wherein the boundary estimate is represented indirectly.

14. The method of claim 13, wherein the boundary estimate is represented using the method of level sets.

15. The method of claim 1, wherein the boundary is initialized based on at least one of a pixel-based reconstruction, at least one contour from a neighboring slice, at least one contour from a previous reconstruction, a CAD model, and an image obtained from at least one non-CT imaging modality.

16. The method of claim 15, wherein the pixel-based reconstruction comprises at least one of a filtered back projection (FBP), an iterative filtered back projection (IFBP), an iterative reconstruction and a statistical reconstruction.

17. The method of claim 15, wherein the at least one non-CT imaging modality comprises at least one of a coordinate measuring machine (CMM), a micrometer, and a laser-based measurement system.

18. The method of claim 1, wherein linearizing the forward model comprises:

incrementing a plurality of corner points defining the boundary in the respective normal direction by an incremental amount; and

calculating a response in a measurement domain to generate the system matrix.

19. The method of claim 1, wherein implementing the iterative image reconstruction process comprises at least one of an iterative filtered back projection (IFBP) algorithm, a maximum a posteriori probability (MAP) algorithm, a maximum likelihood (ML) algorithm, an algebraic reconstruction technique (ART) algorithm, an entropy-based algorithm, a least squares (LS) algorithm, a penalized weighted least squares (PWLS) algorithm, a statistical reconstruction algorithm, and an iterative reconstruction algorithm.

20. The method of claim 1, wherein implementing the iterative image reconstruction process comprises directly solving a linearized system of equations to find a boundary perturbation.

21. The method of claim 1, wherein the forward model based on the boundary estimate incorporates a model of the acquisition physics associated with the acquisition of CT image data.

22. The method of claim 21, wherein the model of the acquisition physics models at least one of a focal spot size, a detector cell size, a detector point-spread function, azimuthal blur, a polychromatic source x-ray spectrum, a detector sensitivity, hysteresis, afterglow, lag, detector cross-talk, off-focal radiation, beam hardening, scattered radiation, measurement noise, and detector memory effects.

23. The method of claim 1, wherein the iterative image reconstruction process incorporates constraints on at least a part of the boundary.

24. The method of claim 23, wherein the constraints on at least a part of the boundary are based on data obtained with at least one of a coordinate measuring machine (CMM), a micrometer, and a laser-based measurement system.

25. The method of claim 1, wherein the iterative image reconstruction process is based on applying a rigid transformation to a known boundary shape for at least a part of the boundary.

26. The method of claim 1, wherein the iterative image reconstruction process is based on applying a non-rigid transformation to a known boundary shape for at least a part of the boundary.

27. The method of claim 1, wherein the iterative image reconstruction process incorporates regularization to enforce smoothness of the boundary.

28. A computed tomography (CT) metrology method, the method comprising:

generating a boundary estimate of an object from a measured sinogram;

generating a partially calculated sinogram based upon the boundary estimate;

updating a system matrix based upon the boundary estimate;

calculating at least one sinogram perturbation from at least one of the system matrix and a set of boundary perturbations;

adding the partially calculated sinogram and the at least one sinogram perturbation to derive a total calculated sinogram;

updating the set of boundary perturbations based upon at least one of the measured sinogram, the total calculated sinogram, and the system matrix;

updating the boundary estimate based upon the set of boundary perturbations;

generating an image of the object based on a final boundary estimate; and

displaying or storing the image for use by a user.

29. The method of claim 28, further comprising initializing the system matrix to an initial value.

30. The method of claim 28, further comprising initializing the set of boundary perturbations to an initial value.

31. The method of claim 28, wherein updating the system matrix comprises linearizing a forward model with respect to the boundary estimate.

32. The method of claim 28, further comprising iterating the steps of updating the system matrix, generating the partially calculated sinogram, calculating the at least one sinogram perturbation, adding the partially calculated sinogram and the at least one sinogram perturbation, updating the set of boundary perturbations, and updating the boundary estimate.

33. The method of claim 28, wherein updating the set of boundary perturbations is based upon the total calculated sinogram and the measured sinogram.

34. The method of claim 28, wherein updating the set of boundary perturbations comprises applying at least one of a maximum likelihood algorithm, a maximum a posteriori algorithm, a statistical reconstruction algorithm, a conjugate gradient algorithm, a gradient ascent algorithm, an algebraic reconstruction algorithm, a weighted least squares algorithm, an iterative filtered back projection, an entropy based algorithm, and an iterative reconstruction algorithm.

35. The method of claim 28, wherein generating the partially calculated sinogram and updating the system matrix comprise applying a model of the acquisition physics used to acquire the measured sinogram.

36. The method of claim 35, wherein the model of the acquisition physics accounts for at least one of a focal spot size, a detector cell size, a detector point-spread function, azimuthal blur, a polychromatic source x-ray spectrum, a detector sensitivity, hysteresis, afterglow, lag, detector cross-talk, off-focal radiation, beam hardening, scattered radiation, measurement noise, and detector memory effects.

37. The method of claim 28, wherein updating the set of boundary perturbations is constrained for at least a part of a boundary.

38. The method of claim 37, wherein updating the set of boundary perturbations is constrained for at least a part of the boundary based on information from a non-CT modality.

39. The method of claim 28, wherein updating the set of boundary perturbations accounts for prior information on smoothness of a boundary.

40. The method of claim 28, wherein updating the set of boundary perturbations comprises applying a rigid or non-rigid transformation to a known boundary shape for at least a part of a boundary.

41. The method of claim 28, wherein the boundary estimate is represented indirectly.

42. The method of claim 41, wherein the boundary estimate is represented using the method of level sets.

43. The method of claim 28, wherein the boundary estimate is represented directly.

44. The method of claim 43, wherein the boundary estimate is defined parametrically.

45. The method of claim 44, wherein the boundary estimate comprises a weighted sum of a plurality of basis functions.

46. The method of claim 45, wherein the plurality of basis functions comprise at least one of a harmonic decomposition, a wavelet decomposition and a polynomial expansion.

47. The method of claim 43, wherein the boundary estimate is defined by at least one spline.

48. The method of claim 47, wherein each spline is a piecewise linear contour represented by a plurality of corner points.

49. The method of claim 47, wherein the at least one spline is of second or higher order.

50. A computed tomography (CT) metrology system comprising:

a source for emitting radiation;

a controller configured to operate the source;

an array of detector units configured to detect radiation transmitted through the object and to generate a detector output signal in response to the detected radiation;

data acquisition circuitry for acquiring the detector output signal;

an image reconstructor coupled to the data acquisition circuitry for generating an image signal in response to the detector output signal, wherein the image reconstructor is configured to initialize a boundary of an object to obtain a boundary estimate, to define a forward model based on the boundary estimate, to linearize the forward model to obtain a system matrix, and to implement an iterative image reconstruction process using the system matrix to update the boundary estimate; and

a computer for controlling the operation of at least one of the controller, the data acquisition circuitry and the image reconstructor.

51. The computed tomography (CT) metrology system of claim 50, wherein the image reconstructor linearizes the forward model by:

incrementing a plurality of corner points defining the boundary in the respective normal direction by an incremental amount; and

calculating a response in a measurement domain to generate the system matrix.

52. A computed tomography (CT) metrology system comprising:

a source for emitting radiation;

a controller configured to operate the source;

an array of detector units configured to detect radiation transmitted through the object and to generate a detector output signal in response to the detected radiation;

data acquisition circuitry for acquiring the detector output signal;

an image reconstructor coupled to the data acquisition circuitry for generating an image signal in response to the detector output signal, wherein the image reconstructor is configured to generate a boundary estimate of an object from a measured sinogram, to update a system matrix based upon the boundary estimate, to generate a partially calculated sinogram based on the boundary estimate, to calculate at least one sinogram perturbation from at least one of the system matrix and a set of boundary perturbations, to add the partially calculated sinogram and the at least one sinogram perturbation to derive a total calculated sinogram, to update the set of boundary perturbations based upon at least one of the measured sinogram, the total calculated sinogram, and the system matrix, and to update the boundary estimate based upon the set of boundary perturbations; and

a computer for controlling the operation of at least one of the controller, the data acquisition circuitry and the image reconstructor.

53. The computed tomography (CT) metrology system of claim 52, wherein the image reconstructor is configured to initialize the system matrix to an initial value.

54. The computed tomography (CT) metrology system of claim 52, wherein the image reconstructor is configured to initialize the set of boundary perturbations to an initial value.

55. The computed tomography (CT) metrology system of claim 52, wherein the image reconstructor updates the system matrix by linearizing a forward model with respect to the boundary estimate.

56. The computed tomography (CT) metrology system of claim 52, wherein the image reconstructor is configured to iterate the steps of updating the system matrix, generating the partially calculated sinogram, calculating the at least one sinogram perturbation, adding the partially calculated sinogram and the at least one sinogram perturbation, updating the set of boundary perturbations, and updating the boundary estimate.

57. A, tangible, machine readable media, comprising:

code adapted to initialize a boundary of an object to obtain a boundary estimate;

code adapted to define a forward model based on the boundary estimate

code adapted to linearize the forward model to obtain a system matrix; and

code adapted to implement an iterative image reconstruction process using the system matrix to update the boundary estimate.

58. A tangible, machine readable media, comprising:

code adapted to generate a boundary estimate of an object from a measured sinogram;

code adapted to update a system matrix based upon the boundary estimate;

code adapted to generate a partially calculated sinogram based on the boundary estimate;

code adapted to calculate at least one sinogram perturbation from at least one of the system matrix and a set of boundary perturbations;

code adapted to add the partially calculated sinogram and the at least one sinogram perturbation to derive a total calculated sinogram;

code adapted to update the set of boundary perturbations based upon at least one of the measured sinogram, the total calculated sinogram, and the system matrix; and

code adapted to update the boundary estimate based upon the set of boundary perturbations.

59. The tangible, machine readable media of claim 58, further comprising code adapted to initialize the system matrix to an initial value.

60. The tangible, machine readable media of claim 58, further comprising code adapted to initialize the set of boundary perturbations to an initial value.

61. A computed tomography (CT) metrology system comprising:

means for initializing a boundary of an object to obtain a boundary estimate;

means for defining a forward model based on the boundary estimate

means for linearizing the forward model to obtain a system matrix; and

means for implementing an iterative image reconstruction process using the system matrix to update the boundary estimate.

62. A computed tomography (CT) metrology system comprising:

means for generating a boundary estimate of an object from a measured sinogram;

means for updating a system matrix based upon the boundary estimate;

means for generating a partially calculated sinogram based on the boundary estimate;

means for calculating at least one sinogram perturbation from at least one of the system matrix and a set of boundary perturbations;

means for adding the partially calculated sinogram and the at least one sinogram perturbation to derive a total calculated sinogram;

means for updating the set of boundary perturbations based upon at least one of the measured sinogram, the total calculated sinogram, and the system matrix; and

means for updating the boundary estimate based upon the set of boundary perturbations.