Method and system for selecting the number of gaussian modes for the segmentation of cardiac magnetic resonance (MR) images

1. A method for automatically selecting a number of Gaussian modes for segmentation of a cardiac magnetic resonance (MR) image, comprising:

identifying a left ventricle (LV) in a cardiac MR image slice;

quantifying the LV blood pool;

obtaining a mask for the LV blood pool;

generating a ring mask for a myocardium of the LV from the LV blood pool mask;

fitting three Gaussian modes to a histogram of the image slice to obtain a corresponding homogeneity image for the myocardium;

computing a quality of fitting (QOF) measure for the three Gaussian modes based on the corresponding homogeneity image;

repeating the fitting and computing steps for four and five Gaussian modes; and

selecting the homogeneity image of the number of Gaussian modes with the largest QOF measure as the homogeneity image for processing.

2. The method of claim 1 , further comprising:

determining inner and outer contours of the myocardium using the selected homogeneity image.

3. The method of claim 1 , wherein the QOF measure is computed by the following equation:

Q

⁢

⁢

O

⁢

⁢

F

=

(

1

N

⁢

∑

(

i

,

j

)

∈

S

⁢

H

ij

)

⁢

log

⁡

(

N

)

,

where S is a collection of pixels that meet two constraints: H ij and R y , N is the total number of pixels in S, H ij is a probability of a pixel's inclusion in the LV blood pool and R y indicates that a pixel was identified as part of the LV blood pool by the ring mask.

4. The method of claim 3 , wherein H ij >about 0.25 and R y =about 1.

5. A system for automatically selecting a number of Gaussian modes for segmentation of a cardiac magnetic resonance (MR) image, comprising:

a memory device for storing a program: and

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

identify a left ventricle (LV) in a cardiac MR image slice;

quantify the LV blood pool;

obtain a mask for the LV blood pool;

generate a ring mask for a myocardium of the LV from the LV blood pool mask;

fit three Gaussian modes to a histogram of the image slice to obtain a corresponding homogeneity image for the myocardium;

compute a quality of fitting (QOF) measure for the three Gaussian modes based on the corresponding homogeneity image:

repeat the fitting and computing steps for four and five Gaussian modes; and

select the homogeneity image of the number of Gaussian modes with the largest QOF measure as the homogeneity image for processing.

6. The system of claim 5 , wherein the processor is further operative with the program to:

determine inner and outer contours of the myocardium using the selected homogeneity image.

7. The system of claim 5 , wherein the QOF measure is computed by the following equation:

Q

⁢

⁢

O

⁢

⁢

F

=

(

1

N

⁢

∑

(

i

,

j

)

∈

S

⁢

H

ij

)

⁢

log

⁡

(

N

)

,

where S is a collection of pixels that meet two constraints: H ij and R y , N is the total number of pixels in S, H ij is a probability of a pixel's inclusion in the LV blood pool and R y indicates that a pixel was identified as part of the LV blood pool by the ring mask.

8. The system of claim 7 , wherein H ij >about 0.25 and R y =about 1.

9. A non-transitory computer readable medium tangibly embodying a program of instructions executable by a processor to perform method steps for automatically selecting a number of Gaussian modes for segmentation of a cardiac magnetic resonance (MR) image, the method steps comprising:

identifying a left ventricle (LV) in a cardiac MR image slice;

quantifying the LV blood pool;

obtaining a mask for the LV blood pool;

generating a ring mask for a myocardium of the LV from the LV blood pool mask;

fitting three Gaussian modes to a histogram of the image slice to obtain a corresponding homogeneity image for the myocardium;

computing a quality of fitting (QOF) measure for the three Gaussian modes based on the corresponding homogeneity image;

repeating the fitting and computing steps for four and five Gaussian modes; and

selecting the homogeneity image of the number of Gaussian modes with the largest QOF measure as the homogeneity image for processing.

10. The computer readable medium of claim 9 , the method steps further comprising:

determining inner and outer contours of the myocardium using the selected homogeneity image.

11. The computer readable medium of claim 9 , wherein the QOF measure is computed by the following equation:

Q

⁢

⁢

O

⁢

⁢

F

=

(

1

N

⁢

∑

(

i

,

j

)

∈

S

⁢

H

ij

)

⁢

log

⁡

(

N

)

,

where S is a collection of pixels that meet two constraints: H ij and R y , N is the total number of pixels in S, H ij is a probability of a pixel s inclusion in the LV blood pool and R y indicates that a pixel was identified as part of the LV blood pool by the ring mask.

12. The computer readable medium of claim 11 , wherein H ij >about 0.25 and R y =about 1.

13. A method for automatically selecting a number of Gaussian modes for image segmentation, comprising:

coarsely identifying a region of interest in an image based on its intensity; and

optimizing the region of interest by:

fitting a first number of Gaussian modes to a histogram of the image to obtain a corresponding homogeneity image for the region of interest;

computing a quality of fitting (QOF) measure for the first number of Gaussian modes based on the corresponding homogeneity image; and

repeating the fitting and computing steps for at least one other number of Gaussian modes;

selecting the homogeneity image of the number of Gaussian modes with the largest QOF measure as the homogeneity image for processing.

14. The method of claim 13 , wherein the image comprises an anatomical part.

15. The method of claim 13 , wherein the image comprises a computed tomography (CT) or magnetic resonance (MR) image.