Method and system for vessel enhancement and artifact reduction in TOF MR angiography of brain

1. A method for vessel enhancement and artifact reduction in a 3D time-of-flight (TOF) magnetic resonance (MR) angiography brain image, comprising:

extracting structures in the brain image based on an intensity threshold;

isolating vessels in said structures by filtering said structures using a vesselness measure, resulting in a filtered image;

multiplying the filtered image by a coefficient map based on intensities of the brain image to enhance a vascular structure by eliminating surrounding structures, resulting in an enhanced image of the vascular structure;

detecting a scalp in the enhanced image; and

generating a noise-reduced enhanced image by subtracting the scalp from the enhanced image.

2. The method of claim 1 , wherein said step of extracting structures in the brain image based on an intensity threshold comprises:

calculating the intensity threshold based on a histogram of the brain image; and

extracting voxels in the brain image having intensities greater than the intensity threshold.

3. The method of claim 2 , wherein said step of calculating the intensity threshold based on a histogram of the brain image comprises:

calculating the intensity threshold as an intensity corresponding to a last point of a second positive section in a 1 st order derivative of the histogram.

4. The method of claim 1 , wherein said vesselness measure is one of a Frangi, Shikata, and Li vesselness measure.

5. The method of claim 1 , wherein said step of isolating vessels in said structures by filtering said structures using a vesselness measure comprises:

filtering said structures using the vesselness measure at multiple scales.

6. The method of claim 5 , wherein said step of filtering said structures using the vesselness measure at multiple scales comprises:

calculating first and second Hessian matrices for each voxel of said structures art first and second scales;

calculating eigenvalues and eigenvectors of the first and second Hessian matrices for each voxel of said structures; and

calculating first and second Frangi vesselness measures at each voxel of said structures based on the eigenvalues and eigenvectors of the first and second Hessian matrices, resulting in first and second filtered images.

7. The method of claim 6 , wherein said step of multiplying the filtered image by a coefficient map based on intensities of the brain image to enhance a vascular structure by eliminating surrounding structures comprises:

generating the coefficient map by dividing an intensity of each voxel in the brain image by a maximum intensity of the brain image; and

multiplying the coefficient map by the first and second filtered images to generate first and second enhanced images corresponding to the first and second scales, respectively; and

generating the enhanced image of the vascular structure as a union of the first and second enhanced images.

8. The method of claim 1 , wherein said step of detecting a scalp in the enhanced image comprises:

obtaining curves representing a head contour;

smoothing the curves to remove ears and background intrusions into a head region;

segmenting the head region based on the smoothed curves;

eroding the segmented head region using morphological operators to generate an eroded sub-region; and

detecting the scalp as a difference between the segmented head region and the eroded sub-region.

9. The method of claim 8 , wherein said step of obtaining a curve representing a head contour comprises:

applying a threshold and mathematical morphology to make the background of the enhanced image uniform; and

tracking a scalp boundary from left, right, top, and bottom boundaries of each of a plurality of slices of the enhanced image to generated curves representing the scalp in four directions.

10. An apparatus for vessel enhancement and artifact reduction in a 3D time-of-flight (TOF) magnetic resonance (MR) angiography brain image, comprising:

means for extracting structures in the brain image based on an intensity threshold;

means for isolating vessels in said structures by filtering said structures using a vesselness measure to generate a filtered image;

means for enhancing the vessels by multiplying the filtered image by a coefficient map based on intensities of the brain image to generate an enhanced image;

means for detecting a scalp in the enhanced image; and

means for generating a noise-reduced enhanced image by subtracting the scalp from the enhanced image.

11. The apparatus of claim 10 , wherein said means for extracting structures in the brain image based on an intensity threshold comprises:

means for calculating the intensity threshold based on a histogram of the brain image; and

means for extracting voxels in the brain image having intensities greater than the intensity threshold.

12. The apparatus of claim 10 , wherein said vesselness measure is one of a Frangi, Shikata, and Li vesselness measure.

13. The apparatus of claim 10 , wherein said means for isolating vessels in said structures by filtering said structures using a vesselness measure comprises:

means for filtering said structures using the vesselness measure at multiple scales.

14. The apparatus of claim 13 , wherein said means for filtering said structures using the vesselness measure at multiple scales comprises:

means for calculating first and second Hessian matrices for each voxel of said structures art first and second scales;

means for calculating eigenvalues and eigenvectors of the first and second Hessian matrices for each voxel of said structures; and

means for calculating first and second Frangi vesselness measures at each voxel of said structures based on the eigenvalues and eigenvectors of the first and second Hessian matrices, resulting in first and second filtered images.

15. The apparatus of claim 10 , wherein said means for enhancing the vessels by multiplying the filtered image by a coefficient map based on intensities of the brain image to generate an enhanced image comprises:

means for generating the coefficient map by dividing an intensity of each voxel in the brain image by a maximum intensity of the brain image; and

means for multiplying the coefficient map by the first and second filtered images to generate first and second enhanced images corresponding to the first and second scales, respectively; and

means for generating the enhanced image as a union of the first and second enhanced images.

16. The apparatus of claim 10 , wherein said means for detecting a scalp in the enhanced image comprises:

means for obtaining curves representing a head contour;

means for smoothing the curves to remove ears and background intrusions into a head region;

means for segmenting the head region based on the smoothed curves;

means for eroding the segmented head region using morphological operators to generate an eroded sub-region; and

means for defining the scalp as a difference between the segmented head region and the eroded sub-region.

17. The apparatus of claim 16 , wherein said means for obtaining a curve representing a head contour comprises:

means for applying a threshold and mathematical morphology to make the background of the enhanced image uniform; and

means for tracking a scalp boundary from left, right, top, and bottom boundaries of each of a plurality of slices of the enhanced image to generated curves representing the scalp along four directions.

18. A computer readable medium encoded with computer executable instructions for vessel enhancement and artifact reduction in a 3D time-of-flight (TOF) magnetic resonance (MR) angiography brain image, the computer executable instructions defining steps comprising:

extracting structures in the brain image based on an intensity threshold;

isolating vessels in said structures by filtering said structures using a vesselness measure, resulting in a filtered image;

enhancing the vessels by multiplying the filtered image by a coefficient map based on intensities of the brain image, resulting in an enhanced image;

detecting a scalp in the enhanced image; and

generating a noise-reduced enhanced image by subtracting the scalp from the enhanced image.

19. The computer readable medium of claim 18 , wherein the computer executable instructions defining the step of extracting structures in the brain image based on an intensity threshold comprise computer executable instructions defining the steps of:

calculating the intensity threshold based on a histogram of the brain image; and

extracting voxels in the brain image having intensities greater than the intensity threshold.

20. The computer readable medium of claim 18 , wherein said vesselness measure is one of a Frangi, Shikata, and Li vesselness measure.

21. The computer readable medium of claim 18 , wherein the computer executable instructions defining the step of isolating vessels in said structures by filtering said structures using a vesselness measure comprise computer executable instructions defining the step of:

filtering said structures using the vesselness measure at multiple scales.

22. The computer readable medium of claim 21 , wherein the computer executable instructions defining the step of filtering said structures using the vesselness measure at multiple scales comprise computer executable instructions defining the steps of:

calculating first and second Hessian matrices for each voxel of said structures art first and second scales;

calculating eigenvalues and eigenvectors of the first and second Hessian matrices for each voxel of said structures; and

calculating first and second Frangi vesselness measures at each voxel of said structures based on the eigenvalues and eigenvectors of the first and second Hessian matrices, resulting in first and second filtered images.

23. The computer readable medium of claim 18 , wherein the computer executable instructions defining the step of enhancing the vessels by multiplying the filtered image by a coefficient map based on intensities of the brain image comprise computer executable instructions defining the steps of:

generating the coefficient map by dividing an intensity of each voxel in the brain image by a maximum intensity of the brain image; and

multiplying the coefficient map by the first and second filtered images to generate first and second enhanced images corresponding to the first and second scales, respectively; and

generating the enhanced image of the vascular structure as a union of the first and second enhanced images.

24. The computer readable medium of claim 18 , wherein the computer executable instructions defining the step of detecting a scalp in the enhanced image comprise computer executable instructions defining the steps of:

obtaining a curve representing a head contour;

smoothing the curve to remove ears and background intrusions into a head region;

segmenting the head region based on the smoothed curve;

eroding the segmented head region using morphological operators to generate an eroded sub-region; and

detecting the scalp as a difference between the segmented head region and the eroded sub-region.

25. The computer readable medium of claim 24 , wherein the computer executable instructions defining the step of obtaining a curve representing a head contour comprise computer executable instructions defining the steps of:

applying a threshold and mathematical morphology to make the background of the enhanced image uniform; and

tracking a scalp boundary from left, right, top, and bottom boundaries of each of a plurality of slices of the enhanced image to generated curves representing the scalp in four directions.