METHOD AND APPARATUS FOR REDUCTION OF METAL ARTIFACTS IN CT IMAGES

• US 20110081071A1
• Filed: 10/06/2009
• Published: 04/07/2011
• Est. Priority Date: 10/06/2009
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Langan, David Allen, Chandra, Naveen, Benson, Thomas Matthew
• Granted Patent: US 8,503,750 B2
• Time in Patent Office: Days
• US Class Current: 382/154
• 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

Abstract

A method and apparatus include acquisition of a view dataset based on x-rays received by a detector corresponding to a energy level, reconstruction of an initial image using the view dataset, the initial image comprising a plurality of metal voxels at respective metal voxel locations, and generation of a metal mask corresponding to the plurality of metal voxels within the initial image. The method and apparatus also include forward projection of the metal mask onto the view dataset to identify metal dexels in the view dataset, performance of a weighted interpolation based on the identified metal dexels to generate a completed view dataset, reconstruction of a final image using the completed view dataset, the final image comprising a plurality of image voxels corresponding to the metal voxel locations, and replacement of a portion of the plurality of image voxels corresponding to the metal voxel locations with smoothed metal values.

Claims

1. A computer readable storage medium having stored thereon a computer program comprising instructions, which, when executed by a computer, cause the computer to:

acquire a first view dataset based on x-rays received by a detector corresponding to a first energy level;

reconstruct an initial image using the first view dataset, the initial image comprising a plurality of metal voxels at respective metal voxel locations;

generate a metal mask corresponding to the plurality of metal voxels within the initial image;

forward project the metal mask onto the first view dataset to identify metal dexels in the first view dataset;

perform a weighted interpolation based on the identified metal dexels to generate a completed first view dataset;

reconstruct a first final image using the completed first view dataset, the first final image comprising a plurality of image voxels corresponding to the metal voxel locations; and

replace a portion of the plurality of image voxels corresponding to the metal voxel locations with smoothed metal values.

2. The computer readable storage medium of claim 1 wherein the instructions cause the computer to generate the metal mask by applying a metal threshold to the initial image.

3. The computer readable storage medium of claim 2 wherein the instructions further cause the computer to perform dilation and erosion operations on the metal mask.

4. The computer readable storage medium of claim 1 wherein the instructions further cause the computer to replace the plurality of metal voxels in the reconstructed initial image with at least one of a constant value and interpolated values corresponding to nearby non-metal voxels.

5. The computer readable storage medium of claim 1 wherein the instructions further cause the computer to apply a full forward projection to the reconstructed initial image.

6. The computer readable storage medium of claim 1 wherein the instructions further cause the computer to apply a forward projection-guided interpolation to generate a completion patch corresponding to the metal dexel locations.

7. The computer readable storage medium of claim 1 wherein the instructions further cause the computer to forward project the metal mask onto the first view dataset based on a spread of a respective metal voxel onto the detector.

8. The computer readable storage medium of claim 1 wherein the instructions cause the computer to perform a weighted three-dimensional interpolation using row, channel, and view data in a neighborhood of respective metal dexels of the first view dataset being interpolated.

9. The computer readable storage medium of claim 1 wherein the instructions cause the computer to perform the weighted interpolation using historical completed data.

10. The computer readable storage medium of claim 1 wherein the instructions that cause the computer to acquire a first view dataset cause the computer to acquire the first view dataset in an axial scanning mode;

and

wherein the instructions that cause the computer to reconstruct an initial image cause the computer to;

reconstruct a first plurality of image slices corresponding to a primary reconstruction volume for an axial scan; and

reconstruct a second plurality of image slices corresponding to a non-primary reconstruction volume in a direction orthogonal to an axial scan plane of the first view dataset, the non-primary reconstruction volume outside the primary reconstruction volume.

11. The computer readable storage medium of claim 1 wherein the instructions further cause the computer to:

acquire a second view dataset based on x-rays received by the detector corresponding to a second energy level;

forward project the metal mask onto the second view dataset to identify metal dexels in the second view dataset;

perform the weighted interpolation based on the metal dexels identified in the second view dataset to generate a completed second view dataset; and

reconstruct a second final image, the second final image comprising a plurality of image voxels corresponding to the metal voxel locations.

12. The computer readable storage medium of claim 11 wherein the instructions that cause the computer to perform the weighted interpolation to generate the completed first view dataset and the completed second view dataset cause the computer to:

determine a set of weighting values; and

apply the set of weighting values to corresponding metal dexels in the first and second view datasets.

13. The computer readable storage medium of claim 11 wherein the instructions that cause the computer to reconstruct an initial image cause the computer to reconstruct one of a monoenergetic image and a material decomposed image using the first view dataset and the second view dataset.

14. The computer readable storage medium of claim 11 wherein the instructions that cause the computer to reconstruct a second final image cause the computer to reconstruct one of a monoenergetic image and a material decomposed image.

15. The computer readable storage medium of claim 11 wherein the instructions cause the computer to reconstruct a second final image using the completed second view dataset.

16. The computer readable storage medium of claim 11 wherein the instructions cause the computer to:

modify the plurality of image voxels corresponding to the metal voxel locations of the first final image to have values corresponding to expected metal values for the first energy level;

modify the plurality of image voxels corresponding to the metal voxel locations of the second final image to have values corresponding to expected metal values for the second energy level; and

apply a smoothing procedure on the modified plurality of image voxels.

17. The computer readable storage medium of claim 11 wherein the instructions further cause the computer to:

modify the plurality of image voxels corresponding to the metal voxel locations of the first final image according to a weighted least squares fit;

modify the plurality of image voxels corresponding to the metal voxel locations of the second final image according to the weighted least squares fit;

scale the modified plurality of image voxels of the first and second final images to match one of the metal threshold and an expected value; and

apply a smoothing procedure on the modified plurality of image voxels.

18. The computer readable storage medium of claim 17 wherein the instructions further cause the computer to combine the first final image and the second final image to generate a monoenergetic image.

19. A method comprising:

reconstructing an image using an imaging dataset based on x-rays received by a detector corresponding to a first energy level;

generating a metal mask from the image corresponding to metal voxel locations within the imaging dataset;

forward projecting the metal mask onto the imaging dataset;

identifying a plurality of metal dexels and a plurality of non-metal dexels within the imaging dataset based on the metal mask;

removing the plurality of metal dexels from the imaging dataset;

generating a plurality of interpolated dexels via a weighted interpolation algorithm;

replacing each of the plurality of metal dexels in the imaging dataset with a respective interpolated dexel of the plurality of interpolated dexels;

reconstructing a final image using the plurality of non-metal dexels and the plurality of interpolated dexels; and

smoothing image voxels in the final image corresponding to the metal voxel locations.

20. The method of claim 19 wherein generating the plurality of interpolated dexels comprises generating the plurality of interpolated dexels using row, channel, and view data in a neighborhood of respective metal dexels of the imaging dataset.

21. The method of claim 19 further comprising:

forward projecting the metal mask onto a second imaging dataset corresponding to a second energy level different from the first energy level;

identifying a second plurality of metal dexels and a second plurality of non-metal dexels within the second imaging dataset based on the metal mask;

removing the second plurality of metal dexels from the second imaging dataset;

generating a second plurality of interpolated dexels via the weighted interpolation algorithm;

replacing each of the second plurality of metal dexels in the second imaging dataset with a respective interpolated dexel of the second plurality of interpolated dexels;

reconstructing a second final image; and

smoothing image voxels in the second final image corresponding to the metal voxel locations.

22. The method of claim 21 further comprising:

modifying the plurality of image voxels corresponding to the metal voxel locations of the final image to have values corresponding to expected metal values for the first energy level; and

modifying the plurality of image voxels corresponding to the metal voxel locations of the second final image to have values corresponding to expected metal values for the second energy level.

23. The method of claim 19 further comprising smoothing the plurality of interpolated dexels using at least one neighboring dexel.

24. The method of claim 19 further comprising synthesizing a plurality of replacement voxels corresponding to the metal voxels prior to forward projecting the metal mask onto the imaging dataset, wherein the plurality of replacement voxels correspond to one of a constant value and interpolated values corresponding to nearby non-metal voxels.

25. The method of claim 19 comprising generating the plurality of interpolated dexels using row, channel, and view data corresponding to at least one of historical completed data and neighboring non-metal dexels.

26. An imaging system comprising:

a rotatable gantry having an opening for receiving an object to be scanned;

an x-ray source coupled to the gantry and configured to project x-rays through the opening;

a generator configured to energize the x-ray source to a first energy level to generate x-rays corresponding to the first energy level;

a detector having pixels therein, the detector attached to the gantry and positioned to receive x-rays projected from the x-ray source; and

a computer programmed to;

access a first projection dataset corresponding to the first energy level;

reconstruct a first image from the first projection dataset;

segment the first image to identify metal locations;

forward project the segmentation onto the first projection dataset to identify a plurality of metal detector pixels in the first projection dataset;

remove the plurality of metal detector pixels from the first projection dataset;

use a weighted interpolation to replace the removed plurality of metal detector pixels and to complete the first projection dataset;

reconstruct a completed first image using the completed first projection dataset; and

smooth portions of the completed first image corresponding to the metal locations of the first energy image.

27. The imaging system of claim 26 wherein the computer is further programmed to:

access a second projection dataset corresponding to a second energy level;

forward project the segmentation onto the second projection dataset to identify a plurality of metal detector pixels in the second projection dataset;

remove the plurality of metal dexels from the second projection dataset;

use a weighted interpolation to replace the removed plurality of metal detector pixels and to complete the second projection dataset;

reconstruct a completed second image using the completed second projection dataset; and

smooth portions of the completed second image corresponding to metal locations of the second projection dataset.

28. The imaging system of claim 27 wherein the computer is further programmed to:

modify portions of the completed first image corresponding to metal locations such that the completed first image depicts expected metal values for the metal locations at the first energy level; and

modify portions of the completed second image corresponding to metal locations such that the completed second image depicts expected metal values for the metal locations at the second energy level.

29. The imaging system of claim 26 wherein the computer is programmed to apply a weighted three-dimensional interpolation using at least one of historical completed data and row, channel, and view data to replace the removed plurality of metal detector pixels.