KR100849163B1 - Image processing method of automatically matching levels of nipples in digital left and right mammographic images so as to symmetrically arrange said images - Google Patents

Abstract

Provided are a breast image processing method of automatically matching left and right breast images obtained by digital mammography to a symmetrical state by automatically adjusting the height of a nipple position (tip). The method includes finding a leading position from each of the left breast image and the right breast image, and based on the found leading positions, the leading position of the left breast image and the leading position of the right breast image are collinear in the left and right directions. And a breast image placement step of placing left and right breast images so as to lie in. Therefore, since the position of the tip of the breast can be the same height and unnecessary parts are removed by using a pretreatment or the like, there is an advantage of eliminating the inconvenience that the doctors have to manually adjust the height of the teat in a symmetrical state at the time of reading. The method also includes a step of combining two left and right symmetrical images into one image by stitching the left and right symmetrical images together. Therefore, there is also an advantage in that the film can be saved because it can be printed by including the left and right mammography images in the symmetrical state on a single film.

Digital Breast Imaging, Nipple Height, Fit

Description

Image processing method of automatically matching levels of nipples in digital left and right mammographic images so as to symmetrically arrange said images}

1 is a flowchart illustrating a digital left and right breast image processing method according to an exemplary embodiment of the present invention;

2A and 2B are views showing left and right breast images obtained by digital mammography, FIG. 2A shows left and right breast images taken in a vertical direction, and FIG. 2B shows left and right breast images taken in an internal and external direction.

3 is a diagram for explaining preprocessing of breast images;

4A is a view in which the left breast image is rotated 180 degrees among the left and right breast images of FIG. 2A, and FIG. 4B is a view in which the left breast image is rotated 180 degrees among the left and right breast images of FIG. 2B;

5 is a view for explaining an automatic papillary positioning algorithm of left and right breast images;

6 is a view for explaining the operation of manually matching the nipple position,

7A is a view in which the nipple positions of the left and right breast images of FIG. 4A are matched in left and right directions, and FIG. 7B is a view in which the nipple positions of the left and right breast images of FIG. 4B are matched in left and right directions,

8 is a view for explaining stitching of left and right breast images;

9 is a view showing a single breast image obtained from left and right breast images, Figure 9a shows a single breast image obtained from the left and right breast images taken in the vertical direction, Figure 9b is a left and right breast image taken in the internal and external oblique direction Single breast images obtained from these are shown.

The present invention relates to an image processing method of digital left and right breast images.

In film mammography, left and right breasts are photographed in craniocaudal (CC) and mediolateral oblique (MLO), respectively, to obtain four films. Doctors hang these films on a fluorescent plate viewbox for reading or diagnosis.

Recently, computed radiography (CR) and digital mammography have also been used to obtain digital images of left and right breasts. Currently, doctors make four films of left and right breasts for doctors to read or use digital high resolution monitors to read digital images.

This conventional method was inconvenient for the doctor and the like to manually adjust the height of the left and right breast images to the left and right symmetrical state.

Accordingly, an object of the present invention is to provide a method for processing digital left and right breast images that can automatically match the height of the left and right breast images.

In order to achieve the above object of the present invention, the present invention provides a breast image processing method for left and right breast images obtained by digital mammography, which method,
Masking the left breast image and the right breast image to obtain masked left and right breast images;
Finding a breast boundary point from each of the masked left and right breast images in each of the lines parallel to the direction line crossing the background and breast areas;
Finding a leading position of the breast from each of the left breast image and the right breast image based on the boundary points found by the breast boundary finding step; And
Based on the found tip positions, a breast image disposing step of arranging left and right breast images such that the leading position of the left breast and the leading position of the right breast lie in the same line in left and right directions;
The breast boundary finding step,
Detecting a change in the pixel value of the masked breast image for each line while traveling from the edge of the background region located opposite the breast toward the breast region; And
When the change of the pixel value is detected in the first step, the pixel position at which the change in the pixel value is detected is set as the boundary point, and the operation of detecting the change in the pixel value for the line where the change in the pixel value is detected is stopped. It has a second step.

The breast image placement step may include extracting a boundary of the breast from each of the left and right breast images; Determining tip positions of the boundaries of the extracted breasts; And aligning the left and right breast images so that the positions of the left and right breast images corresponding to the determined leading positions are located on the same line in the left and right directions.

The tip position is the nipple position.

The breast image arranging step may further include arranging the left and right breast images so that the tips of the left and right breasts are far from each other based on the symmetry axis of the left and right breast images.

After the breast image arranging step, the method may further include generating a breast image including the left breast image and the right breast image by combining the left breast image and the right breast image.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart of a digital left and right breast image processing method according to an exemplary embodiment of the present invention. The flowchart of FIG. 1 is described in detail with reference to FIGS. 2A-9B.

First, breast images shown in FIGS. 2A and 2B are obtained by computer or digital mammography (step S11). In this case, the acquired left and right breast images are determined as a left breast image or a right breast image based on the header information included in the image, and are disposed on the left or right side of the screen. The breast images shown in FIG. 2A are images obtained by vertical imaging, and the breast images shown in FIG. 2B are images obtained by internal and external oblique imaging. 2A illustrates an example in which a left breast image is disposed on the left side and a right breast image is disposed on the right side. 2B shows an example in which a right breast image is disposed on the left side and a left breast image is disposed on the right side.

The left and right breast images shown in Figs. 2A and 2B are preprocessed to remove unnecessary portions from the images (step S13). Referring to the images shown in FIG. 2B in step S13, the image on the right side of FIG. 2B is rotated 180 degrees. The images shown in FIG. 2B are then arranged similarly to those shown in FIG. 4B. Thereafter, the regions 33 indicated by dotted lines in FIG. 3 are cut off from the right breast image 31 and the left breast image 32. At this time, the left part on the ground, that is, the right part is cut off and removed from the right breast image in the right breast image 31, and the right part on the ground, that is, the left part is cut off and removed in the left breast image 32. Thereafter, the upper left and lower left corner portions 34 of the right breast image 31 are filled with black, and the upper right and lower right corner portions 35 of the left breast image 32 are also black. Filled with color. Step S13 may be performed automatically. As a result, left and right breast images as shown in FIG. 4B are obtained. Similar preprocessing is performed on the images shown in FIG. 2A, and as a result, left and right breast images as shown in FIG. 4A are obtained. As described above, left and right breast images of the pre-processed breasts are not located at the same height as the nipples corresponding to the tip of the breast.

The images shown in Figs. 4A and 4B are processed so that the tips of the breasts are located at the same height (step S15). 5A is a diagram for explaining an automatic nipple search algorithm used in step S15. An automatic nipple finding algorithm includes masking a digital breast image, finding boundary points of the breast region and the background region, and determining a papillary position (exactly, a tip position) at a defined boundary.

In the image masking step, for example, the same breast image (FIG. 5A) as the breast image (right breast image) shown on the left side of FIG. 4B is image masked. When W is the width of the image of FIG. 5A, H is its height, and Min is the minimum pixel value, the algorithm used for image masking is approximately:

       For (y = 0; y <H; y ++)

         For (x = 0; x <W; x ++)

            if (P (x, y)> Min) then

               M (x, y) = 1

            Else

               M (x, y) = 0

            End if

         End for

       End for

By such an image masking algorithm, an image as shown in Fig. 5B is obtained. P (x, y) means a pixel value on the coordinates of x and y. M (x, y) is a masking value of x, y coordinates. If the pixel value of P (x, y) is larger than Min, the corresponding masking value is 1, otherwise it is classified as 0. That is, the area where M (x, y) = 1 is the white area where the breast is located, and M (x, y) = 0 The phosphorus region becomes a black region which is a background region.

Then, in the boundary point search step, boundary points of the black region 51 and the white region 53 are found in the masked image shown in FIG. 5B. An example of an algorithm used for this purpose, where E (y) is the x value of the boundary at y, is:

       For (y = 0; y <H; y ++)

         For (x = 0; x <W; x ++)

           if (M (x, y) == 1) then

                E (y) = x

                Break;

           End if

         End for

       End for

Using this algorithm, each boundary point is found as it progresses from the top line to the bottom line constituting the image of FIG. 5B. Scanning from the top to the bottom of the image finds the boundary point and stores the x position at this time in the variable E (y). E (y) is the x position of the boundary where the masking value changes. x is from 0 Up to W, the y value is From 0 to H.

Thereafter, a teat positioning step is performed. In the case of the image shown in Fig. 5C, the nipple position NP is the y position corresponding to the smallest x value among the boundary points, so an example of the algorithm used is as follows:

       NP = 0;

       MinEY = 65535;

       For (y = 0; y <H; y ++)

         If (MinEY> E (y))

            MinEY = E (y)

            NP = y

          End if

       End for

NP = 0 is a variable that is taken from the top y value (0) down to the y axis of the image. In the case of the right breast, MinEY is a variable to find the minimum value of E (y) in the process of finding the tip of the breast. Since this must exceed the maximum number of pixels on the x-axis of the image, it is set here as 65535 arbitrarily. In Figure 5 (c), it is a variable used to store the minimum value of E (y) in the process of finding the smallest value among the x position (E (y)) values of the boundary point.
On the other hand, in the case of the breast which protrudes to the right as shown in the left breast image shown in Figure 4a, the nipple position is the y position corresponding to the largest x value in the boundary line of the breast. Using the automatic nipple finding algorithm, you can find the most protruding part of the breast in the left direction. Therefore, even when the nipple is recessed, it is possible to find the tip of the breast that can be regarded as the nipple.

When the nipple position is not found by the automatic nipple search algorithm, a screen such as the example shown in FIG. 6 is provided to the user. The user viewing this screen designates the nipple locations using a double click of the mouse, etc. Then, based on the specified nipple locations, the nipples of the left breast image and the right breast image are automatically adjusted to be the same height. Can be. Alternatively, the user may adjust the nipple height of both breast images by adjusting the nipple position on the guideline by dragging the image using the mouse. In this manner, when step S15 is completed, as shown in FIGS. 7A and 7B, both breast images are positioned to have the same height of the nipple.

After step S15 is completed, marks indicating how the image is taken and whether it is a left breast image or a right breast image are inserted into each of the left and right breast images where the nipple height is adjusted (step S17). For example, in the right image of FIG. 7A, a mark including an English letter R indicating a right breast image and an English letter CC indicating an image captured in a craniocaudal direction is displayed. In the left image of FIG. 7B, a mark including an English letter L indicating a left breast image and an English letter MLO indicating an image taken in a mediolateral oblique direction is displayed.

Meanwhile, in the images shown in FIG. 7A or 7B, the left breast image and the right breast image are simultaneously displayed on the same screen, but these are separate images. Therefore, it is necessary to make the images that are aligned with each other as a single image. For this purpose, a process of combining the left and right breast images into one image, for example, a stitching process, is performed (step S19). When the stitching process is performed in step S19, the left breast image and the right breast image are made into one image. The image may be an unnecessary margin, that is, an image in which a background image of a breast is excluded to some extent. 8 shows that the stitching process for the breast images is in progress when the user presses a corresponding button to perform the stitching process.

9A and 9B show images of left and right breasts when the stitching process is completed, in particular, FIG. 9A shows one left and right breast image taken up and down and FIG. 9B shows one left and right breast image taken in and out of four directions. . Each of the images shown in FIGS. 9A and 9B may be stored or output to a film as needed.

As described above, according to the present invention, the tip position of the breast can be automatically the same height, so that the amount of work required for arranging the image, aligning the nipple, etc. until the doctor reads the image can be reduced. In addition, since the left and right breast images are combined into one image, the amount of film used can be reduced by half. Furthermore, there is an advantage in that the position of the tip of the breast can be the same height and it is convenient at the time of reading because unnecessary parts are removed using pretreatment or the like.

Claims (5)

In the breast image processing method for left and right breast images obtained by digital mammography, Masking the left breast image and the right breast image to obtain masked left and right breast images; Finding a breast boundary point from each of the masked left and right breast images in each of the lines parallel to the direction line crossing the background and breast areas; Finding a leading position of the breast from each of the left breast image and the right breast image based on the boundary points found by the breast boundary finding step; And Based on the found tip positions, a breast image disposing step of arranging left and right breast images such that the leading position of the left breast and the leading position of the right breast lie in the same line in left and right directions; The breast boundary finding step, Detecting a change in the pixel value of the masked breast image for each line while traveling from the edge of the background region located opposite the breast toward the breast region; And When the change of the pixel value is detected in the first step, the pixel position at which the change of the pixel value is detected is set as the boundary point, and the operation of detecting the change of the pixel value for the line where the change of the pixel value is detected is finished. Breast image processing method comprising two steps. delete The method of claim 1, wherein the leading position is a papillary position. The method of claim 1, wherein the arranging of the breast image comprises arranging the left and right breast images so that the ends of the left breast and the right breast are far from each other based on the symmetry axis of the left breast and right breast images. Breast image processing method further comprising. The breast image according to any one of claims 1, 3, and 4, wherein after the breast image placement step, the left breast image and the right breast image are combined to form one breast image containing the left breast image and the right breast image. Breast image processing method further comprising the step of generating.

Images