ROBUST ARTIFACT REDUCTION IN IMAGE RECONSTRUCTION

• US 20150029178A1
• Filed: 07/26/2013
• Published: 01/29/2015
• Est. Priority Date: 07/26/2013
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Claus, Bernhard Erich Hermann, Langan, David Allen, Braun-Jones, Taylor Robert
• Granted Patent: US 9,349,198 B2
• Time in Patent Office: Days
• US Class Current: 345/419
• CPC Class Codes: A61B 6/032 Transmission computed tomography [CT]
• CPC Class Codes: A61B 6/5258 involving detection or reduction of artifacts or noise
• CPC Class Codes: G06T 11/005 Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
• CPC Class Codes: G06T 2200/04 involving 3D image data

Abstract

Approaches are disclosed for removing or reducing metal artifacts in reconstructed images. The approaches include creating a metal mask in the projection domain, interpolating data within the metal mask, and perform a reconstruction using the interpolated data. In certain embodiments the metal structure is separately reconstructed and combined with the reconstructed volume.

Claims

1. A computer-implemented method for processing projection image data, comprising:

generating a set of tentative two-dimensional (2D) metal masks, wherein each tentative 2D metal mask is generated based on a comparison between a respective projection image and a corresponding background image;

backprojecting each tentative 2D metal mask to generate a respective set of backprojected three-dimensional (3D) data in image space for each tentative 2D metal mask;

combining the sets of backprojected three-dimensional (3D) data in image space to generate a 3D metal mask; and

reprojecting the 3D metal mask to generate one or more metal masks in projection space.

2. The computer-implemented method of claim 1, wherein each corresponding background image is generated by performing a weight-based smoothing of the respective projection image using a respective weight image.

3. The computer-implemented method of claim 2, wherein each respective weight image is derived from a label image corresponding to the respective projection image.

4. The computer-implemented method of claim 3, wherein each corresponding background image is iteratively generated or updated based on updates to the respective label image after each iteration.

5. The computer-implemented method of claim 1, wherein generating the 3D metal mask comprises, at each voxel within the combined backprojected 3D data, determining how many tentative 2D metal masks show non-metal data at each voxel location and how many tentative 2D metal masks show the presence of a metal object at each voxel location, wherein the tentative 2D metal masks may be weighted or unweighted.

7. The computer-implemented method of claim 1, wherein generating the 3D metal mask comprises determining, using a set of confidence weights, a confidence-weighted sum of the respective backprojected tentative metal masks at each voxel location and determining the sum of backprojected confidence-weights at each voxel location.

8. The computer-implemented method of claim 7, wherein the confidence weights used in determining the confidence weighted sums are binary.

9. The computer-implemented method of claim 7, wherein the confidence weights for each respective projection are derived from the local penetration or signal level in the corresponding background image.

10. The computer-implemented method of claim 7, wherein the confidence weights for each respective projection are derived from a measure of confidence in a tentative metal label.

11. The computer-implemented method of claim 7, wherein the confidence weights for each respective projection are derived from prior knowledge of a collimator position with the respective projection image.

12. The computer-implemented method of claim 7, wherein the confidence weights for each respective projection are derived from prior knowledge of data points being invalid in portions of the projection image.

13. The computer-implemented method of claim 7, wherein a voxel within the 3D metal mask is characterized as metal if the ratio of confidence weighted sums of the backprojected tentative 2D metal masks to the sum of the confidence weights exceeds a threshold that is less than 1.

14. The computer-implemented method of claim 1, comprising expanding or dilating the tentative 2D metal masks prior to backprojection.

15. The computer-implemented method of claim 1, comprising refining the one or more metal masks in projection space.

16. A computer-implemented method for processing projection image data, comprising:

accessing a mathematically incomplete set of acquired projection images;

reconstructing an initial volume from the mathematically incomplete set of acquired projection images;

generating a three-dimensional (3D) metal mask from the initial volume;

refining the 3D metal mask; and

reprojecting the 3D metal mask to generate one or more metal masks in projection space.

17. The computer-implemented method of claim 16, wherein the set of acquired projection images is acquired over a limited angular range less than 160 degrees, or wherein the angular separation between projection images is at about 1 degree or less.

18. The computer-implemented method of claim 16, wherein the 3D metal mask is generated by applying one or more thresholds to the initial volume.

19. The computer-implemented method of claim 16, comprising refining the one or more metal masks generated from the 3D metal mask.

20. An image processing system, comprising:

a memory storing one or more routines; and

a processing component configured to execute the one or more routines stored in the memory, wherein the one or more routines, when executed by the processing component, cause acts to be performed comprising;

accessing or acquiring projection image data;

generating a set of tentative two-dimensional (2D) metal masks, wherein each tentative 2D metal mask is generated based on a comparison between a respective projection image and a corresponding background image;

backprojecting each tentative 2D metal mask to generate a respective set of backprojected three-dimensional (3D) data in image space for each tentative 2D metal mask;

combining the sets of backprojected three-dimensional (3D) data in image space to generate a 3D metal mask; and

reprojecting the 3D metal mask to generate one or more metal masks in projection space.