Apparatus and method for forming 3D ultrasound image

1. An apparatus for forming a 3-dimensional (3D) ultrasound image, comprising:

a generation unit configured to generate a binarization image of a 2D ultrasound image to generate a surface curved line of a target object on the 2D ultrasound image and automatically generate a region of interest (ROI) box for the target object on the 2D ultrasound image by moving bounds of the ROI box to pixels closest to the surface curved line;

a first determination unit configured to calculate a mean square error value between a predetermined standard histogram for a 2D ultrasound image of the target object and a histogram for a 2D ultrasound image within the ROI box, and determine whether the ROI box is suitably generated for a display of the target object therewithin based on the calculation result;

a detection unit configured to detect a contour of the target object in the ROI box; and

a formation unit configured to form the 3D ultrasound image by rendering volume data existing in the detected contour.

2. The apparatus as recited in claim 1 , further comprising:

a second determination unit configured to determine whether the contour of the 2D ultrasound image of the target object is detectable.

3. The apparatus as recited in claim 1 , wherein the generation unit comprises:

a first unit configured to segment the 2D ultrasound image into an image of the target object and an image of a background;

a second unit configured to set the ROI box with a predetermined size according to types of the target object on the 2D ultrasound image; and

a third unit configured to adjust a size of the ROI box to be suitable for a size of the target object.

4. The apparatus as recited in claim 3 , wherein the third unit is configured to perform the following functions:

removing a noise region in a binarization region of the image of the target object, after performing binarization for the image of the target object in the ROI box;

generating the binarization image for the target object by providing predetermined pixel values to the binarization region and the surface curved line of the binarization image; and

adjusting the bounds of the ROI box by moving each bound to the pixels closest to the surface curved line.

5. The apparatus as recited in claim 2 , wherein the second determination unit is configured to perform the following functions:

extracting pixels forming a boundary of the target object by matching pixels of a mask of a predetermined type with pixels of the image of the target object in the ROI box one-to-one; and

determining a capability of contour detection of the target object by calculating a ratio of the number of boundary pixels to the number of all pixels matched with the mask one-to-one and a variance of pixels forming the boundary according to the following equation:

D i =αR 1i +βR 2i

where i is the number of a 2D ultrasound image of a same type of target object and previously stored in a memory, D i is a determination numerical value representing the capability of the contour detection, R 1i is a ratio of the number of the boundary pixels to the number of all pixels matched with the mask one-to-one, and R 2i is the variance of the pixels forming the boundary, and where α and β, which are coefficients obtained through a contour detection experiment of various 2D ultrasound images for the same type of target objects as the target object, are the values where a mean square error value defined in the following equation is minimized:

∈

=

∑

i

⁢

(

D

i

-

(

α

⁢

⁢

R

1

⁢

i

+

β

⁢

⁢

R

2

⁢

i

)

)

2

where, to minimize the mean square error value, a partial differentiation of the mean square error value is performed for each α and β, and where the α and β are determined as the coefficients when resulting values of the partial differentiation are zero.

6. The apparatus as recited in claim 1 , wherein the detection unit is configured to perform the following functions:

performing a brightness contrast stretching for the image of the target object for sufficient brightness contrast of the target object in the ROI box;

removing a noise region from a binarization region of the target object, after performing binarization of the target object in the ROI box;

extracting pixels existing in a surface of the binarization region which meets the bounds of the ROI box by adjusting the bounds of the ROI box; and

detecting the contour of the target object by connecting the extracted pixels.

7. A method for forming a 3-dimensional (3D) ultrasound image, comprising:

a) generating a binarization image of a 2D ultrasound image to generate a surface curved line of a target object on the 2D ultrasound image and automatically generating a region of interest (ROI) box for the target object on the 2D ultrasound image by moving bounds of the ROI box to pixels closest to the surface curved line;

b) calculating a mean square error value of a predetermined standard histogram for a 2D ultrasound image of the target object and a histogram for a 2D ultrasound image within the ROI box, and determining whether the ROI box is suitably generated for a display of the target object therewith based on the calculation result;

c) detecting a contour of the target object in the ROI box; and

d) forming the 3D ultrasound image by rendering volume data existing in the detected contour.

8. The method as recited in claim 7 , further comprising:

e) determining whether the contour of the 2D ultrasound image of the target object is detectable.

9. The method as recited in claim 7 , wherein a) further comprises:

a1) segmenting the 2D ultrasound image into an image of the target object and an image of a background neighboring the target object;

a2) setting a ROI box with a predetermined size on the 2D ultrasound image; and

a3) adjusting the size of the ROI box to be suitable for a size of the 2D ultrasound image of the target object.

10. The method as recited in claim 9 , wherein a3) further comprises:

a3-1) removing a noise region in a binarization region of the image of the target object, after performing binarization for the image of the target object in the ROI box;

a3-2) generating the binarization image for the target object by providing predetermined pixel values to the binarization region and the surface curved line of the binarization image; and

a3-3) adjusting the bounds of the ROI box by moving each bound to the pixels closest to the surface curved line.

11. The method as recited in claim 8 , wherein e) further comprises:

e1) extracting pixels forming a boundary of the target object by matching pixels of a mask of a predetermined type with pixels of the image of the target object in the ROI box one-to-one; and

e2) determining a capability of contour detection of the target object by calculating a ratio of the number of boundary pixels to the number of all pixels matched with the mask one-to-one and a variance of pixels forming the boundary according to the following equation:

D i =αR 1i +βR 2i

where i is the number of a 2D ultrasound image of a same type of target object and previously stored in a memory, D i is a determination numerical value representing the capability of the contour detection, R 1i is a ratio of the number of the boundary pixels to the number of all pixels matched with the mask one-to-one, and R 2i is the variance of the pixels forming the boundary, and where α and β, which are coefficients obtained through a contour detection experiment of various 2D ultrasound images for the same type of target objects as the target object, are the values where a mean square error value defined in the following equation is minimized:

∈

=

∑

i

⁢

(

D

i

-

(

α

⁢

⁢

R

1

⁢

i

+

β

⁢

⁢

R

2

⁢

i

)

)

2

where, to minimize the mean square error value, a partial differentiation of the mean square error value is performed for each α and β, and where the α and β are determined as the coefficients when resulting values of the partial differentiation are zero.

12. The method as recited in claim 7 , wherein b) further comprises:

b1) performing a brightness contrast stretching for the image of the target object for sufficient brightness contrast of the target object in the ROI box;

b2) removing a noise region from a binarization region of the target object, after performing binarization of the target object in the ROI box;

b3) extracting pixels existing in a surface of the binarization region which meets the bounds of the ROI box by adjusting the bounds of the ROI box; and

b4) detecting the contour of the target object by connecting the extracted pixels.