Efficient features for shape analysis of lesions in breast MR

1. A method for analyzing a shape of a region of interest in a medical image of a body part, comprising:

collecting, by a scanner, the medical image of the body part;

finding, by a processor, a region of interest in the medical image;

calculating, by the processor, a Reeb graph of the region of interest; and

determining, by the processor, whether the region of interest is a malignant lesion candidate based on a shape characteristic of the Reeb graph.

2. The method of claim 1 , wherein the shape characteristic comprises a number of branches in the Reeb graph, and when the number of branches is above a threshold the region of interest is a malignant lesion candidate.

3. The method of claim 1 , wherein the shape characteristic comprises a number of branches in the Reeb graph that are above a length, and when the number of branches is above a threshold the region of interest is a malignant lesion candidate.

4. The method of claim 1 , further comprising:

choosing a function from which to calculate the Reeb graph.

5. The method of claim 4 , wherein the function is an isolation measure that is computed for each voxel in the region of interest.

6. The method of claim 5 , wherein the shape characteristic comprises a value that is obtained by dividing a value obtained from the isolation measures by a power of a volume of the region of interest, and when the value of the shape characteristic is above a threshold the region of interest is a malignant lesion candidate.

7. The method of claim 6 , wherein the value obtained from the isolation measures is an average of the isolation measures, a median of the isolation measures, a maximum of the isolation measures or a combination of the average, median or maximum of the isolation measures.

8. The method of claim 6 , wherein the power of the volume of the region of interest is a square root or a cubic root of the volume of the region of interest.

9. The method of claim 5 , wherein the isolation measure is represented by:

μ( v )=∫ pεS g ( v,p ) dS,

where v is a point of S at which to calculate μ, p is a point of S, g(v,p) is a function that returns a distance between v and p of S, S is a portion of the region of interest and μ(v) is a mean of a distance from v to all points of S.

10. The method of claim 9 , wherein S is a surface of the region of interest.

11. The method of claim 9 , wherein the distance is a geodesic distance or a Euclidean distance.

12. The method of claim 4 , wherein the function is a radial function that is computed for each voxel in the region of interest.

13. The method of claim 12 , wherein the radial function is represented by:

μ( v )= d ( v,c ),

where d(v,c) is a Euclidean distance between v and c, c is a point of the region of interest and v is a point of the region of interest at which to calculate μ.

14. The method of claim 13 , wherein c is a centroid or a center of the region of interest.

15. The method of claim 1 , wherein the medical image comprises a magnetic resonance image or a computed tomography image.

16. The method of claim 1 , wherein the body part comprises a breast, a lung or liver.

17. A system for analyzing a shape of a region of interest in a medical image of a body part, comprising:

a memory device for storing a program;

a processor in communication with the memory device, the processor operative with the program to:

find a region of interest in the medical image;

calculate a Reeb graph of the region of interest; and

determine whether the region of interest is a malignant lesion candidate based on a shape characteristic of the Reeb graph.

18. A method for analyzing a shape of a region of interest in a medical image of a body part, comprising:

collecting, by a scanner, the medical image of the body part;

finding, by a processor, a region of interest in the medical image;

calculating, by processor, an isolation measure for each voxel in the region of interest; and

determining, by the processor, whether the region of interest is a malignant lesion candidate by using the isolation measures.

19. A system for analyzing a shape of a region of interest in a medical image of a body part, comprising:

a memory device for storing a program;

a processor in communication with the memory device, the processor operative with the program to:

find a region of interest in the medical image;

calculate an isolation measure for each voxel in the region of interest; and

determine whether the region of interest is a malignant lesion candidate by using the isolation measures.