Method and apparatus for fully automated segmentation of joint using patient-specific optimal thresholding and watershed algorithm

1. A method of fully automatically segmenting an image, the method comprising steps of:

extracting region information corresponding to a target object to be segmented from a medical image of the target object by associating a thresholding method and a load path algorithm, wherein the medical image indicates a quantitative computed tomography (QCT), computed tomography (CT) or magnetic resonance imaging (MRI), and wherein the target object includes one or more bones;

generating a first mask based on the extracted region information;

generating a morphological patch by performing morphological subdivision on the medical image based on a watershed algorithm;

generating a second mask based on the generated morphological patch; and

segmenting an image corresponding to the target object from the medical image based on the first mask and the second mask, wherein the step of segmenting the image corresponding to the target object comprises segmenting the image corresponding to the target object from the medical image by merging the first mask corresponding to a regional patch and the second mask corresponding to the morphological patch.

2. The method of claim 1 , wherein the step of extracting the region information corresponding to the target object comprises steps of:

performing thresholding on the medical image based on a predetermined threshold;

extracting a flag array related to the target object from an image on which the thresholding has been performed based on the load path algorithm; and

generating a regional patch corresponding to the target object based on the extracted flag array.

3. The method of claim 2 , wherein the step of performing the thresholding comprises steps of:

setting the threshold used upon performing the thresholding as a design variable of a golden section method; and

changing the threshold so that the threshold is relatively reduced according to the golden section method until predetermined convergence conditions are satisfied.

4. The method of claim 1 , wherein the step of generating the morphological patch comprises generating the morphological patch on the medical image by performing preprocessing based on a middle value filter, dilation operation, and erosion operation.

5. The method of claim 1 , wherein the step of generating the morphological patch comprises steps of:

removing noise from the medical image based on a middle value filter;

reinforcing a boundary of the image corresponding to the target object by performing dilation operation and erosion operation on an image from which the noise has been removed; and

performing morphological subdivision on the image whose boundary has been enhanced based on the watershed algorithm.

6. The method of claim 1 , wherein the step of segmenting the image comprises automatically segmenting the image corresponding to the target object by selectively extracting a patch corresponding to the first mask from the second mask.

7. The method of claim 1 , wherein:

the step of segmenting the image comprises segmenting an image corresponding to a hip joint from the QCT, CT or MRI image.

8. The method of claim 7 , wherein the step of segmenting the image comprises segmenting images corresponding to different two bones forming the hip joint from the QCT, CT or MRI image.

9. A method of fully automatically segmenting an image, the method comprising steps of:

extracting region information corresponding to a target object to be segmented from a medical image of the target object by associating a thresholding method and a load path algorithm, wherein the medical image indicates a quantitative computed tomography (QCT), computed tomography (CT) or magnetic resonance imaging (MRI), wherein the target object includes one or more bones, and wherein the step of extracting the region information corresponding to the target object comprises steps of:

performing thresholding on the medical image based on a predetermined threshold;

extracting a flag array related to the target object from an image on which the thresholding has been performed based on the load path algorithm; and

generating a regional patch corresponding to the target object based on the extracted flag array;

generating a first mask based on the extracted region information;

generating a morphological patch by performing morphological subdivision on the medical image based on a watershed algorithm;

generating a second mask based on the generated morphological patch; and

segmenting an image corresponding to the target object from the medical image based on the first mask and the second mask.

10. The method of claim 9 , wherein the step of segmenting the image corresponding to the target object comprises segmenting the image corresponding to the target object from the medical image by merging the first mask corresponding to a regional patch and the second mask corresponding to the morphological patch.

11. The method of claim 10 , wherein the step of performing the thresholding comprises steps of:

setting the threshold used upon performing the thresholding as a design variable of a golden section method; and

changing the threshold so that the threshold is relatively reduced according to the golden section method until predetermined convergence conditions are satisfied.

12. The method of claim 9 , wherein the step of generating the morphological patch comprises generating the morphological patch on the medical image by performing preprocessing based on a middle value filter, dilation operation, and erosion operation.

13. The method of claim 9 , wherein the step of generating the morphological patch comprises steps of:

removing noise from the medical image based on a middle value filter;

reinforcing a boundary of the image corresponding to the target object by performing dilation operation and erosion operation on an image from which the noise has been removed; and

performing morphological subdivision on the image whose boundary has been enhanced based on the watershed algorithm.

14. The method of claim 9 , wherein the step of segmenting the image comprises automatically segmenting the image corresponding to the target object by selectively extracting a patch corresponding to the first mask from the second mask.

15. The method of claim 9 , wherein:

the step of segmenting the image comprises segmenting an image corresponding to a hip joint from the QCT, CT or MRI image.

16. The method of claim 15 , wherein the step of segmenting the image comprises segmenting images corresponding to different two bones forming the hip joint from the QCT, CT or MRI image.

17. A method of fully automatically segmenting an image, the method comprising steps of:

extracting region information corresponding to a target object to be segmented from a medical image of the target object by associating a thresholding method and a load path algorithm, wherein the medical image indicates a quantitative computed tomography (QCT), computed tomography (CT) or magnetic resonance imaging (MRI), and wherein the target object includes one or more bones;

generating a first mask based on the extracted region information;

generating a morphological patch by performing morphological subdivision on the medical image based on a watershed algorithm, wherein the step of generating the morphological patch comprises steps of:

removing noise from the medical image based on a middle value filter;

reinforcing a boundary of the image corresponding to the target object by performing dilation operation and erosion operation on an image from which the noise has been removed; and

performing morphological subdivision on the image whose boundary has been enhanced based on the watershed algorithm;

generating a second mask based on the generated morphological patch; and

segmenting an image corresponding to the target object from the medical image based on the first mask and the second mask.

18. The method of claim 17 , wherein the step of segmenting the image corresponding to the target object comprises segmenting the image corresponding to the target object from the medical image by merging the first mask corresponding to a regional patch and the second mask corresponding to the morphological patch.

19. The method of claim 17 , wherein the step of extracting the region information corresponding to the target object comprises steps of:

performing thresholding on the medical image based on a predetermined threshold;

extracting a flag array related to the target object from an image on which the thresholding has been performed based on the load path algorithm; and

generating a regional patch corresponding to the target object based on the extracted flag array.

20. The method of claim 19 , wherein the step of performing the thresholding comprises steps of:

setting the threshold used upon performing the thresholding as a design variable of a golden section method; and

changing the threshold so that the threshold is relatively reduced according to the golden section method until predetermined convergence conditions are satisfied.

21. The method of claim 17 , wherein the step of generating the morphological patch comprises generating the morphological patch on the medical image by performing preprocessing based on a middle value filter, dilation operation, and erosion operation.

22. The method of claim 17 , wherein the step of segmenting the image comprises automatically segmenting the image corresponding to the target object by selectively extracting a patch corresponding to the first mask from the second mask.

23. The method of claim 17 , wherein:

the step of segmenting the image comprises segmenting an image corresponding to a hip joint from the QCT, CT or MRI image.

24. The method of claim 23 , wherein the step of segmenting the image comprises segmenting images corresponding to different two bones forming the hip joint from the QCT, CT or MRI image.