Methods and apparatus for virtual coronary mapping

1. A method for mapping vessels comprising:

determining a plurality of positions of a coronary guidewire in a plurality of non-contrast enhanced fluoroscopic images;

determining, in a plurality of contrast enhanced fluoroscopic images, a plurality of reference points of a vessel segment corresponding to the plurality of positions of the coronary guidewire in a plurality of contrast enhanced fluoroscopic images; and

modifying a non-contrast enhanced fluoroscopic image based at least in part on the determined plurality of reference points of a vessel segment.

2. The method of claim 1 further comprising:

determining a reference system relative to the coronary guidewire;

detecting a contrast agent contrasting the vessels; and

extracting the locations of a plurality of vessel boundaries relative to the determined reference system based on the contrasted vessels.

3. The method of claim 2 wherein the reference system is a relative coordinate system wherein a first dimension of the relative coordinate system is a length along the coronary guidewire relative to a reference point and a second dimension of the relative coordinate system is a distance between the coronary guidewire and a vessel boundary.

4. The method of claim 3 wherein the reference point is a location of a catheter along the coronary guidewire.

5. The method of claim 1 further comprising:

determining a position of a coronary guidewire in a first of the plurality of non-contrast enhanced fluoroscopic images by applying an image-based guidewire detector, and image-based guidewire tip detector, and an image-based catheter end detector.

6. The method of claim 5 further comprising:

learning the image-based guidewire detector from a set of guidewire images using marginal space learning and a probabilistic boosting tree;

learning the image-based guidewire tip detector from a set of guidewire tip images using a probabilistic boosting tree and steerable features; and

learning the image-based catheter end detector from a set of catheter end images using a probabilistic boosting tree.

7. The method of claim 1 further comprising:

determining a position of the coronary guidewire by tracking a two-dimensional guidewire shape in a plurality of images.

8. The method of claim 7 further comprising:

tracking the coronary guidewire with a robust estimation of translational motion of a two-dimensional guidewire shape; and

tracking the coronary guidewire with a robust estimation of non-rigid deformation of a two-dimensional guidewire shape.

9. The method of claim 8 wherein the position of the coronary guidewire is further refined by integrating results of a guidewire tip detection and a catheter end detection.

10. The method of claim 1 further comprising:

determining a centerline and a plurality of vessel boundaries of a coronary vessel segment by applying an image-based vessel detector learned from a set of vessel segment images using marginal space learning and a probabilistic boosting tree.

11. The method of 10 further comprising:

determining the plurality of vessel boundaries of the coronary vessel segment using graph cuts and a learned vessel boundary detector.

12. The method of claim 1 further comprising:

modifying a contrast free image by mapping pixel intensities around a guidewire.

13. A method of coronary mapping:

determining a position of a coronary guidewire in a coronary vessel segment;

determining a centerline of the coronary vessel segment; and

determining a plurality of boundaries of the coronary vessel segment;

wherein determining a position of a coronary guidewire comprises:

determining an estimation of translational motion of a two-dimensional coronary guidewire shape; and

determining an estimation of non-rigid deformation of the two-dimensional guidewire shape;

wherein determining an estimation of translational motion of a two-dimensional guidewire shape comprises:

determining a first set of guidewire features using an image-based guidewire detector;

determining a second set of guidewire features using an image-based guidewire detector; and

optimizing a kernel-based matching function between the first set of guidewire features and the second set of guidewire features.

14. The method of claim 13 wherein optimizing a kernel-based matching function between the first set of guidewire features and the second set of guidewire features comprises determining

T

opt

=

argmax

⁢

∑

i

⁢

⁢

∑

j

⁢

⁢

exp

⁡

(

1

σ

2

⁢



y

i

-

(

z

j

+

T

)



2

)

wherein {y i } is the first set of guidewire features, {z i } is the second set of guidewire features, T is a two-dimensional translation vector, and

exp

⁡

(

1

σ

2

⁢



y

i

-

(

z

j

+

T

)



2

)

is an exponential function that defines a robust kernel with a bandwidth σhu 2 .

15. The method of claim 13 wherein determining an estimation of non-rigid deformation of the two-dimensional coronary guidewire shape comprises:

determining a vessel shape;

matching the two-dimensional coronary guidewire shape with the vessel shape; and

optimizing a probability of a vessel appearance with an image-based vessel detector.

16. The method of claim 15 wherein determining a vessel shape comprises:

detecting a location of the coronary guidewire in an image frame; and

tracking the coronary guidewire in the image frame.

17. The method of claim 13 wherein determining a centerline of the coronary vessel segment comprises:

determining a two-dimensional vessel shape in a plurality of image frames;

determining a estimation of translational motion of the two-dimensional shape; and

determining an estimation of non-rigid deformation of the two-dimensional vessel shape.