Automatic coronary isolation using a n-MIP ray casting technique
View Patent ↗A novel method is presented for detecting coronary arteries as well as other peripheral vessels of the heart. After finding the location of the myocardium through a segmentation method, such as a graph theoretic segmentation method, the method models the heart with a biaxial ellipsoid. For each point of the ellipsoid, a collection of intensities are computed that are normal to the surface. This collection is then filtered to detect the cardiovascular structures. Ultimately, vessel centerline points are detected using a vessel tracking method, and linked together to form a complete coronary artery tree.
1. A method for analyzing a coronary image, comprising:
retrieving, automatically, centerlines of peripheral vessels of a heart in the coronary image, wherein retrieving the centerlines comprises,
segmenting the heart in the coronary image to produce a three-dimensional (“3D”) segmented myocardium;
computing a 3D distance map of distances from a surface of the three-dimensional segmented myocardium to each pixel corresponding to the heart in the coronary image, modeling the heart from the coronary image;
casting normal maximum intensity projection rays for each voxel on the surface of the heart in the 3D distance map;
filtering the rays to obtain vessel candidate points; and
refining the vessel candidates points to obtain the centerlines.
2. The method of claim 1 , wherein the step of segmenting the heart comprises segmenting the heart using a graph cut method.
3. The method of claim 1 , wherein the step of segmenting the heart comprises segmenting the heart using model-based segmentation.
4. The method of claim 1 , wherein the step of segmenting the heart comprises segmenting the heart using level set methods.
5. The method of claim 1 , wherein each of the rays is a collection of voxel intensities that are crossed by a straight line.
6. The method of claim 1 , wherein the step of filtering the rays to obtain vessel candidate points comprises detecting peaks of high intensity as vessel candidates.
7. The method of claim 6 , wherein the step of detecting peaks of high intensity as vessel candidates comprises detecting peaks of high intensity as vessel candidates if the peak is within a predetermined intensity range.
8. The method of claim 6 , wherein the step of detecting peaks of high intensity as vessel candidates comprises detecting peaks of high intensity as vessel candidates if the peak is a predetermined distance from the heart wall.
9. The method of claim 6 , wherein the step of detecting peaks of high intensity as vessel candidates comprises detecting peaks of high intensity as vessel candidates if the peak is in a predetermined range.
10. The method of claim 9 , wherein the step of detecting peaks of high intensity as vessel candidates if the peak is in a predetermined range comprises detecting peaks of high intensity as vessel candidates if the peak is in a predetermined range of 1 mm to 10 mm.
11. The method of claim 6 , wherein the step of detecting peaks of high intensity as vessel candidates comprises using a local maximum detection method.
12. The method of claim 1 , wherein the step of refining the vessel candidates points to obtain centerlines comprises computing a Hessian matrix and Eigen values for each of the vessel candidate points.
13. The method of claim 1 , wherein the image is one of a CT image and MR image.
14. The method of claim 1 , further comprising:
linking the centerlines to construct a vessel tree; and
outputting a visualization of the vessel tree.
15. A method for analyzing a coronary image, comprising:
retrieving, automatically, centerlines of peripheral vessels of a heart in the coronary image, wherein retrieving the centerlines comprises,
segmenting the heart in the coronary image to produce a three-dimensional (“3D”) segmented myocardium;
computing a 3D distance map of distances from a surface of the three-dimensional segmented myocardium to each pixel corresponding to the heart in the coronary image;
modeling a heart wall by a spheroid using the 3D distance map;
casting normal maximum intensity projection rays for each voxel on the spheroid;
filtering the rays to obtain vessel candidate points; and
refining the vessel candidates points to obtain the centerlines.
16. The method of claim 15 , further comprising:
linking the centerlines to construct a vessel tree; and
outputting a visualization of the vessel tree.
17. A machine-readable medium having instructions stored thereon for execution by a processor to perform method for analyzing a coronary image, the method comprising:
retrieving, automatically, centerlines of peripheral vessels of a heart in the coronary image, wherein retrieving the centerlines comprises,
segmenting the heart in the coronary image to produce a three-dimensional (“3D”) segmented myocardium;
computing a 3D distance map of distances from a surface of the three-dimensional segmented myocardium to each pixel corresponding to the heart in the coronary image, modeling the heart from the coronary image;
casting normal maximum intensity projection rays for each voxel on the surface of the heart in the 3D distance map;
filtering the rays to obtain vessel candidate points; and
refining the vessel candidates points to obtain the centerlines.
18. The method of claim 17 , further comprising:
linking the centerlines to construct a vessel tree; and
outputting a visualization of the vessel tree.
19. A machine-readable medium having instructions stored thereon for execution by a processor to perform method of automatically retrieving centerlines of peripheral vessels of a heart in an image, the method comprising:
retrieving, automatically, centerlines of peripheral vessels of a heart in the coronary image, wherein retrieving the centerlines comprises,
segmenting the heart in the coronary image to produce a three-dimensional (“3D”) segmented myocardium;
computing a 3D distance map of distances from a surface of the three-dimensional segmented myocardium to each pixel corresponding to the heart in the coronary image;
modeling a heart wall by a spheroid using the 3D distance map;
casting normal maximum intensity projection rays for each voxel on the spheroid;
filtering the rays to obtain vessel candidate points; and
refining the vessel candidates points to obtain the centerlines.
20. The method of claim 19 , further comprising:
linking the centerlines to construct a vessel tree; and
outputting a visualization of the vessel tree.