A method of rib suppression for chest dr images
By combining grayscale compensation and gradient correction with coordinate transformation, the problem of overlapping interference between ribs and lung fields in chest X-rays was solved. This method effectively suppresses ribs in a single X-ray image, adapts to different equipment and body shapes, reduces radiation dose, and preserves lung field texture information.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- LIAONING KAMPO MEDICAL SYST
- Filing Date
- 2021-07-09
- Publication Date
- 2026-04-10
AI Technical Summary
Existing technologies cause diagnostic interference due to the overlap of rib and lung field images in chest X-rays. Dual-energy subtraction angiography increases radiation dose and is costly, while single-image X-ray images do not provide ideal rib suppression, especially for curved ribs.
The rib region was suppressed using grayscale compensation and gradient correction methods. Combined with contour-based coordinate transformation, soft tissue images were obtained through gradient image reconstruction. The rib region mask and gradient correction techniques were used to process the rib edges and reconstruct lung soft tissue images.
It effectively suppresses ribs in a single X-ray image, adapts to different equipment and body shapes, preserves lung field texture information, avoids unnatural transitions at rib edges, and reduces radiation dose and cost.
Smart Images
Figure CN115601246B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of X-ray digital image processing, and particularly relates to a rib suppression method for chest DR images. BACKGROUND
[0002] X-ray chest radiography is widely used for screening and diagnosis of lung diseases due to its fast and simple, economical and practical, and low radiation dose. However, due to the two-dimensional characteristics of X-ray chest radiography, the rib and lung field image overlap, which will interfere with the diagnosis of lung diseases. Therefore, suppressing the rib structure in the image can improve the above situation to a certain extent.
[0003] At present, dual-energy subtraction technology can separate bone tissue and soft tissue images. This technology uses two different energy X-rays to take images, and uses the different attenuation coefficients of human tissues to different energy X-rays to obtain bone and soft tissue images by weighted subtraction processing. However, compared with ordinary X-rays, this technology has a larger radiation dose, which increases the cost and also affects the health of patients.
[0004] In recent years, machine learning and pattern recognition methods have become a research hotspot in the field of computer technology. Researchers use artificial neural network technology to obtain corresponding bone tissue images and soft tissue images. However, this technology requires a large number of bone tissue image and soft tissue image training sets, so it is not easy to apply.
[0005] Therefore, rib suppression for single X-ray image has always been the focus of people's research. A method for suppressing clavicle proposed by Simkó et al. (Elimination of clavicle shadows to help automatic lung nodule detection on chest radiographs) has good suppression effect on clavicle located in the horizontal direction, but the suppression effect on the curved rib is not ideal. The rib suppression method proposed in the present application does not have this problem. SUMMARY
[0006] In view of the above problems, the purpose of the present application is to provide a rib suppression method for chest DR images, which can effectively suppress the rib region and obtain lung soft tissue images.
[0007] According to the rib region mask, the present application uses gray scale compensation to suppress the inside of the rib. The suppression of the rib edge uses gradient correction to perform gradient suppression, and the soft tissue image is reconstructed from the gradient image. In view of the problem that the suppression effect on the curved rib is not ideal, the present application uses a contour-based coordinate transformation method to solve this problem.
[0008] The technical scheme adopted by the present application to achieve the above object is as follows: a chest DR image rib suppression method, comprising the following steps:
[0009] 1) obtaining a DR original image and a rib region mask through a DR device;
[0010] 2) performing gray suppression on the rib region mask to obtain a gray value after rib internal suppression and a rib edge region after gray suppression;
[0011] 3) extracting the DR original image to obtain a rib edge image, and performing coordinate transformation on the rib edge image to obtain a rib edge image after coordinate transformation;
[0012] 4) performing gradient calculation in two directions on the rib edge image after coordinate transformation to obtain gradient images of the rib edge in two directions;
[0013] 5) performing rib edge gradient suppression on the rib edge gradient image to obtain gradient images Ix and Iy after rib edge suppression;
[0014] 6) reconstructing the gradient image after rib edge suppression back to a gray image and replacing the rib edge region after gray suppression, and the gray image is a soft tissue image after rib suppression.
[0015] The step 2) is specifically as follows:
[0016] The gray average values of each rib region in the rib region mask and the vertical up and down regions outside the rib region are calculated at a distance d, respectively, to obtain a rib gray average value I in , a rib outer upper region gray average value I up and a rib outer lower region I down , wherein the rib region includes the edge of the rib and the inside of the rib.
[0017] The gray value after rib internal suppression is calculated according to the following formula:
[0018]
[0019] wherein, is the gray value after rib internal suppression, and I(i,j) is the original rib region gray value.
[0020] The distance d is set to 1 / 2 of the average width of the rib.
[0021] In the step 3), the rib edge image is obtained, specifically as follows:
[0022] The up and down boundaries of each rib in the original image are extracted in the vertical direction at a distance of 1 / 4 of the average width of the current rib to obtain the rib edge image.
[0023] In step 3), the coordinate transformation of the rib edge image comprises:
[0024] The Cartesian coordinates (x, y) of a single rib edge image are transformed into contour-based coordinates (r, c), and the transformed rib edge remains horizontal.
[0025] The step 5) is specifically:
[0026] Step 1: r-direction and c-direction gradient images are respectively subjected to one-dimensional Gaussian smoothing in r-direction and c-direction, the r-direction Gaussian kernel size is selected according to 1 / 4 of the r-direction length, and the c-direction Gaussian kernel size is selected according to 1 / 2 of the c-direction length.
[0027] Wherein, the r, c direction length is the length and width of the rib edge image after coordinate transformation respectively;
[0028] Step 2: the r, c direction gradient images are respectively subtracted from the r, c direction Gaussian smoothed images to obtain the r, c direction contour coordinates (r, c).
[0029] Step 3: the contour coordinates (r, c) are transformed into the original Cartesian coordinates (x, y) again, thereby obtaining the rib edge suppressed gradient image Ix and Iy.
[0030] The step 6) is specifically:
[0031] The divergence of the rib edge suppressed gradient image is calculated
[0032] The rib edge suppressed gradient image is reconstructed back to a grayscale image by establishing a Poisson equation and replacing The rib edge region, to obtain the rib suppressed soft tissue image:
[0033] Delta G = DIV (I),
[0034] Wherein, delta is the Laplace operator, G is the output image, is the rib inside suppressed image, is the boundary of the rib edge region.
[0035] The present application has the following beneficial effects and advantages:
[0036] 1. The method of the present application does not require dual-energy radiography, and can utilize a single X-ray image for rib suppression.
[0037] 2. The present application can be used for X-ray images taken by different devices, and is suitable for different body shapes of the photographer.
[0038] 3. The method of the present application retains the texture information of the lung field area and does not cause unnatural rib edge transition. BRIEF DESCRIPTION OF DRAWINGS
[0039] Figure 1 A flow chart of the present application;
[0040] Figure 2 A rib edge image extracted by the present application;
[0041] Figure 3 An image after rib edge coordinate transformation of the present application;
[0042] Figure 4 An original image of the present application;
[0043] Figure 5 An image after rib suppression of the present application. DETAILED DESCRIPTION
[0044] In order to make the objectives, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described below in connection with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.
[0045] The present application will be further described in detail below in connection with the drawings and embodiments.
[0046] A specific flow chart of the present application is shown in Figure 1
[0047] Step 101: Obtain a DR original image and a rib region mask.
[0048] Step 102: Rib internal suppression. The average gray value is calculated for each rib region and the vertical direction up and down outside the rib with a distance d, respectively, I in , I up and I down The distance d is set to 1 / 2 of the average width of the rib, which can be adjusted according to the actual effect. The gray value calculation formula after rib internal suppression is as follows:
[0049]
[0050] Among them, is the gray value after rib internal suppression, and I(i,j) is the original rib region gray value.
[0051] Step 103: rib edge image extraction. The upper and lower boundaries of each rib in the original image are extracted in the vertical direction with a distance of ds to obtain a rib edge image. The distance ds is set to 1 / 4 of the average width of the rib, which can be adjusted according to the actual effect, Figure 2 The extracted rib edge image is shown.
[0052] Step 104: rib edge coordinate transformation. The single rib edge image is rearranged as a rectangular image according to the column pixels, the height of the rectangle is the vertical width of the original edge image, and the width of the rectangle is the horizontal width of the original edge image. The Cartesian coordinates (x, y) of the single rib edge image are transformed into contour-based coordinates (r, c), and the transformed rib edge remains substantially horizontal. Figure 3 The image after the rib edge coordinate transformation is shown.
[0053] Step 105: calculate the rib edge gradient image. The r, c direction gradient calculation is performed on the rib edge image after the coordinate transformation.
[0054] Step 106: rib edge gradient suppression. One-dimensional Gaussian smoothing is performed on the r, c gradient images in the r, c direction, the r direction Gaussian kernel size is selected according to 1 / 4 of the r direction length, and the c direction Gaussian kernel size is selected according to 1 / 2 of the c direction length. The r, c gradient images are respectively subtracted by the Gaussian smoothed images, the processed images are rearranged as the original edge image position according to the column pixels, the contour-based coordinates (r, c) are transformed into the original Cartesian coordinates (x, y), and thus the suppressed gradient image is obtained.
[0055] Step 107: reconstruct the soft tissue image. The divergence of the suppressed gradient image is calculated Thus, the Poisson equation ΔG = DIV(I), s.t. where G is the output image, is the rib interior suppressed image, is the boundary of the rib edge region. The Poisson equation is solved, the gradient image is reconstructed back to a grayscale image to obtain the rib suppressed soft tissue image. Figure 4 、 Figure 5 The original image and the rib suppressed image are shown respectively, it can be seen that the lung texture after suppression is not damaged, and the rib edge transition is natural.
[0056] According to the above steps of the present application, those skilled in the art can understand and implement without creative labor.
[0057] The above description is only an embodiment of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, extension, etc. made within the spirit and principle of the present application is included in the protection scope of the present application.
Claims
1. A method of rib suppression for chest DR images, characterized in that, The method comprises the following steps: 1) obtaining a DR original image and a rib region mask through a DR device; 2) performing gray scale suppression on the rib region mask to obtain an internal rib gray scale value after suppression and a rib edge region after gray scale suppression; 3) extracting the DR original image again to obtain a rib edge image; performing coordinate transformation on the rib edge image to obtain a rib edge image after coordinate transformation; 4) performing gradient calculation in two directions based on the rib edge image after coordinate transformation to obtain gradient images of the rib edge in two directions; 5) performing rib edge gradient suppression on the gradient images of the rib edge to obtain gradient images Ix and Iy after rib edge suppression; 6) reconstructing the gradient images after rib edge suppression into a gray scale image and replacing the rib edge region after gray scale suppression, and the gray scale image is a soft tissue image after rib suppression.
2. The method of rib suppression for chest DR images of claim 1, wherein, The step 2) is specifically: The average value of the rib grayscale is obtained by calculating the average value of the grayscale of each rib region in the rib region mask and the vertical direction above and below the rib region with a distance d The average value of the rib grayscale is obtained by calculating the average value of the grayscale of each rib region in the rib region mask and the vertical direction above and below the rib region with a distance d The average value of the rib grayscale is obtained by calculating the average value of the grayscale of each rib region in the rib region mask and the vertical direction above and below the rib region with a distance d The rib region includes the edge of the rib and the inside of the rib. The internal rib gray scale value after suppression is calculated according to the following formula: ； wherein, is the gray value of the rib interior after suppression, is the gray value of the original rib region.
3. The method of rib suppression for chest DR images of claim 2, wherein, The distance d is set as 1 / 2 of the average width of the rib.
4. The method of rib suppression for chest DR images of claim 1, wherein, In the step 3), the rib edge image is obtained by: Extracting the rib edge image by taking 1 / 4 of the average width of the rib as the distance in the vertical direction above and below the upper and lower boundaries of each rib in the original image.
5. The method of rib suppression for chest DR images of claim 1, wherein, In the step 3), the coordinate transformation on the rib edge image comprises: Transforming the Cartesian coordinates (x, y) of a single rib edge image into contour-based coordinates (r, c), and the rib edge after transformation remains horizontal.
6. The method of rib suppression for chest DR images of claim 5, wherein, The step 5) is specifically: Step 1: performing one-dimensional Gaussian smoothing in the r direction and the c direction on the gradient images in the r direction and the c direction respectively, and the Gaussian kernel size in the r direction is selected according to 1 / 4 of the length in the r direction, and the Gaussian kernel size in the c direction is selected according to 1 / 2 of the length in the c direction; Wherein, the length in the r direction and the length in the c direction are the length and the width of the rib edge image after coordinate transformation; the r direction is the horizontal direction in the contour-based coordinates, and the c direction is the vertical direction in the contour-based coordinates; Step 2: subtracting the images after Gaussian smoothing in the r direction and the c direction from the gradient images in the r direction and the c direction respectively to obtain processed images; Step 3: transforming the processed images from the contour-based coordinates (r, c) into the original Cartesian coordinates (x, y) again to obtain the gradient images Ix and Iy after rib edge suppression.
7. The method of rib suppression for chest DR images of claim 6, wherein, The step 6) is specifically: divergence of the gradient image after rib edge suppression ; Wherein, Ix and Iy are gradient image components in the Cartesian coordinate system, and x and y are Cartesian coordinate system coordinates. The Poisson equation is established to reconstruct the gradient image after rib edge suppression back to the gray image and replace The rib edge region is obtained, and the soft tissue image after rib suppression is obtained: , ； where Δ is the Laplacian operator, G is the output image, is the inside of the rib, is the boundary of the rib edge region.
Citation Information
Patent Citations
Method and system for inhibiting bone shadows in digital X ray chest radiograph
CN105469365A
Poisson model-based X-ray chest image rib inhibition method
CN106023200A