Method for detection and visional enhancement of blood vessels and pulmonary emboli

1. A method for detecting a substantially cylindrical internal structure from within a medical image in a medical image processing device, comprising:

acquiring a medical image;

calculating a gradient of the medical image;

calculating local shape index information for the calculated gradient of the medical image;

removing gradient information having a local shape index not indicative of a ridge shape;

calculating a diverging gradient field response (DGFR) based on the remaining gradient information;

using the DGFR as a discriminative feature in detecting the substantially cylindrical internal structure; and

displaying the desired internal structure as a visual enhancement of the medical image.

2. The method of claim 1 , wherein the substantially cylindrical internal structure is a blood vessel.

3. The method of claim 1 , wherein the medical image is a computed tomography (CT) image, an magnetic resonance (MR) image, an ultrasound image, or a positron emission tomography (PET) image.

4. The method of claim 1 , wherein calculating the gradient of the medical image includes calculating vectors of second-order partial derivatives along each special dimension.

5. The method of claim 1 , wherein the calculated gradient of the medical image is normalized to a set of three-dimensional unit vectors prior to the calculation of the local shape index information.

6. The method of claim 1 , wherein the local shape index information is calculated from a Hessian matrix of second-order partial derivatives of the medical image.

7. The method of claim 6 , wherein the local shape index information si is calculated according to the equation:

si

=

2

π

⁢

arctan

⁢

⁢

κ

2

+

κ

1

κ

2

-

κ

1

,

where K 2 represents an eigenvalue of the Hessian matrix corresponding to a direction of minimum curvature and K 1 represents an eigenvalue of the Hessian matrix corresponding to a direction of maximum curvature.

8. The method of claim 1 , wherein the calculated local shape index information is normalized within the range of [−1,+1].

9. The method of claim 8 , wherein the local shape index information not indicative of the ridge shape is local shape index information not within the range of [+⅜,+⅝].

10. The method of claim 1 , wherein the DGFR is calculated by vector convolution.

11. The method of claim 1 , additionally comprising the step of displaying the identified substantially cylindrical internal structure as a visual enhancement of the medical image.

12. The method of claim 1 , additionally comprising:

further removing gradient information having local shape index information not indicative of a rut shape;

inverting the remaining gradient information;

calculating a diverging gradient field response (DGFR) based on the inverted remaining gradient information; and

using the DGFR as a discriminative feature to detect an occlusion of the substantially cylindrical internal structure.

13. A method for detecting an occlusion of a substantially cylindrical internal structure from within a medical image in a medical image processing device, comprising:

acquiring a medical image;

calculating a gradient of the medical image;

calculating local shape index information based on the calculated gradient of the medical image;

removing gradient information having local shape index information not indicative of a rut shape;

inverting the remaining gradient information;

calculating a diverging gradient field response (DGFR) based on the inverted remaining gradient information;

using the DGFR as a discriminative feature in detecting the substantially cylindrical internal structure; and

displaying the desired internal structure as a visual enhancement of the medical image.

14. The method of claim 13 , wherein the substantially cylindrical internal structure is a blood vessel and the occlusion is an embolism.

15. The method of claim 13 , wherein the medical image is a computed tomography (CT) image, an magnetic resonance (MR) image, an ultrasound image, or a positron emission tomography (PET) image.

16. The method of claim 13 , wherein calculating the gradient of the medical image includes calculating vectors of second-order partial derivatives along each special dimension.

17. The method of claim 13 , wherein the calculated gradient of the medical image is normalized to a set of three-dimensional unit vectors prior to the calculation of the local shape index information.

18. The method of claim 13 , wherein the local shape index information is calculated from a Hessian matrix of second-order partial derivatives of the medical image.

19. The method of claim 18 , wherein wherein the local shape index information si is calculated according to the equation:

si

=

2

π

⁢

arctan

⁢

⁢

κ

2

+

κ

1

κ

2

-

κ

1

,

where K 2 represents an eigenvalue of the Hessian matrix corresponding to a direction of minimum curvature and K 1 represents an eigenvalue of the Hessian matrix corresponding to a direction of maximum curvature.

20. The method of claim 13 , wherein the calculated local shape index information is normalized within the range of [−1,+1].

21. The method of claim 20 , wherein the local shape index information not indicative of the rut shape is local shape index information not within the range of [−⅝,−⅜].

22. The method of claim 13 , wherein the DGFR is calculated by vector convolution.

23. The method of claim 13 , additionally comprising the step of displaying the occlusion as a visual enhancement of the medical image.

24. A method for detecting a desired internal structure from within a medical image in a medical image processing device, comprising:

acquiring a medical image;

calculating a gradient of the medical image;

calculating local shape index information for the calculated gradient of the medical image;

normalizing the calculated local shape index information within the range of [−1,+1];

removing gradient information having a local shape index not indicative of a predetermined shape;

calculating a diverging gradient field response (DGFR) based on the remaining local shape index information;

using the DGFR as a discriminative feature in detecting the desired internal structure; and

displaying the desired internal structure as a visual enhancement of the medical image.

25. The method of claim 24 , wherein the predetermined shape is a ridge shape, the local shape index information not indicative of the ridge shape is local shape index information not within the range of [+⅜,+⅝], and the desired internal structure has a substantially cylindrical shape.

26. The method of claim 24 , wherein:

the calculated gradient information is inverted prior to calculating the DGFR;

the predetermined shape is a rut shape;

the local shape index information not indicative of the rut shape is local shape index information not within the range of [−⅝,−⅜]; and

the desired internal structure is an occlusion within a substantially cylindrical shaped structure.