A noise reduction enhancement method for CT positioning film

Abstract

The application provides a CT positioning sheet noise reduction enhancement method, which comprises the following steps: firstly, extracting and processing strip-shaped artifacts, i.e., removing the processed strip-shaped artifacts from the image to obtain an image with reduced noise; and then extracting and processing high-frequency information, i.e., adding the processed high-frequency information to the image to obtain an image with enhanced high-frequency information. The application effectively reduces the noise of the positioning sheet image, enhances the contrast of the image, and has the advantages of simple algorithm, small calculation amount, small memory occupation and good effect.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method for noise reduction and enhancement of CT localization images. Background Technology

[0002] During a CT scan, a preliminary localization scan is typically performed to determine the location, angle, and reconstructed field of view of the area to be scanned. Beyond these basic functions, the localization scan can also be used to assist in selecting the scanning protocol and setting scanning parameters. Furthermore, based on the image information provided by the localization scan, some simple diagnostics can be performed, such as quantifying fat. All of these applications require a certain level of image quality from the localization scan.

[0003] However, in practical applications, due to the low dose and limited information, the image of the localization film obtained directly from the reconstruction has low contrast and the edge structure of bones and tissues is not clear enough. When scanning some high-absorption areas, there is often a lot of noise in the image.

[0004] Existing noise reduction methods generally employ low-pass filtering, wavelet transform, nonlocal means, or even neural networks. These algorithms either produce unsatisfactory results, causing image distortion, or are computationally intensive and have poor real-time performance. Furthermore, different types of images require different enhancement algorithms; simple histogram equalization and grayscale stretching algorithms are insufficient to achieve the desired effects. Unique and systematic enhancement algorithms are needed to meet the processing requirements of localization patches. Summary of the Invention

[0005] In order to overcome the above-mentioned technical defects, the purpose of this invention is to provide a simple and effective method for noise reduction and enhancement of CT localization images.

[0006] This invention discloses a method for noise reduction and enhancement of CT localization images, comprising the following steps: adjusting the localization image I based on artifact distribution. ori The image I is processed to obtain the preliminary artifact removal result. sv ; Positioning image I ori Image I after initial artifact removal sv By subtracting the two images, we obtain the initial artifact image A. ori For the initial artifact image A ori Weighted processing yields the weighted artifact image A. w The weighted artifact image A w From positioning patch image I ori Image I after artifact removal is obtained. corr ; Image I after artifact correction corr The high-frequency edge image E is obtained by performing filtering processing. ori Generate high-frequency edge image E oriWeighted image W E ; weight the image W E Applied to high-frequency edge images E ori The weighted high-frequency image E is obtained. W The image after artifact correction I corr With the weighted high-frequency image E W The images are fused together to obtain the enhanced image I. corr-E .

[0007] Preferably, the positioning image is processed according to the artifact distribution to obtain image I after preliminary artifact removal. sv Includes: along the positioning patch image I ori A Gaussian filter is applied along the axis to obtain the image I after preliminary artifact removal, where the stripe artifacts are smoothed. sv .

[0008] Preferably, the image I along the positioning patch ori Before performing Gaussian filtering on the axis, the process also includes: processing the input original positioning patch image I in The positioning patch image I is obtained through projection reconstruction. ori I ori =-log(I in ).

[0009] Preferably, the initial artifact image A ori Weighted processing yields the weighted artifact image A. w Includes: the original positioning image I in Gaussian filtering and normalization are performed to obtain the weighted image W of the artifact distribution. A Use the atan function to reduce the initial artifacts in image A. ori Thus, the weakened artifact image A is obtained. com ; weight the image W A Applied to weakened artifact image A com The weighted artifact image A is obtained. w .

[0010] Preferably, the step of positioning the original positioning image I in Gaussian filtering and normalization are performed to obtain the weighted image W of the artifact distribution. A Includes: the original positioning image I in The filtered image I is obtained by performing maximum value filtering and Gaussian filtering. mg ; the filtered image I mg Image I is obtained by normalization. nor , Where C1 > 0 and is an adjustable parameter; if image I nor If W < 0, then the weighted image WA =0; if image I nor If the weighted image W is greater than 1, then the weighted image W is... A =1; otherwise, the weighted image W A =I nor .

[0011] Preferably, the use of the atan function to reduce the initial artifact image A ori Thus, the weakened artifact image A is obtained. com Including: A com =atan(C2*A ori ) / C2; where C1>0 and is an adjustable parameter.

[0012] Preferably, the initial artifact image A ori Weighted processing yields the weighted artifact image A. w This also includes: weighting the artifact image A w Median filtering and Gaussian filtering are performed along the direction of the artifact to obtain the final artifact A. f The weighted artifact image A w From positioning patch image I ori Image I after artifact removal is obtained. corr Including: Final Artifact A f From positioning patch image I ori Image I after artifact removal is obtained. corr .

[0013] Preferably, the high-frequency edge image E is generated ori Weighted image W E Includes: Image I after artifact correction corr Image E is obtained by performing Gaussian filtering. S If image E S If W < 1, then the weighted image W E =1; otherwise, the weighted image

[0014] Preferably, the weighted image W E Applied to high-frequency edge images E ori The weighted high-frequency image E is obtained. W This also includes: using the atan function to reduce the weighted high-frequency image E W Thus, the weakened high-frequency image E is obtained. com ,

[0015] E com =atan(C3*E W ) / C3; where C3 > 0 and is an adjustable parameter; the image I after artifact correction corr With the weighted high-frequency image EW The images are fused together to obtain the enhanced image I. corr-E Includes: Image I after artifact correction corr With weakened high-frequency image E com The images are fused together to obtain the enhanced image I. corr-E .

[0016] Preferably, the image I after artifact correction corr With the weighted high-frequency image E W The images are fused together to obtain the enhanced image I. corr-E This also includes: enhancing the image I corr-E The positioning patch output image I is obtained by performing filtering and grayscale transformation processing. out .

[0017] Compared with existing technologies, the above technical solution has the following advantages:

[0018] 1. This invention reduces the impact of noise on the positioning patch by removing strip artifacts; and enhances the positioning patch information through the extraction and processing of high-frequency information. This invention effectively reduces noise in the positioning patch image, enhances image contrast, has a simple algorithm, low computational load, low memory usage, and good results. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the noise reduction and enhancement method for CT positioning images provided by the present invention.

[0020] Figure 2 This is a schematic diagram of the noise reduction process provided by the present invention;

[0021] Figure 3 This is a schematic diagram of the enhanced processing flow provided by the present invention;

[0022] Figure 4 shows an example of a CT localization image, in which: Figure 4a The image is a CT localization film before processing according to the present invention; Figure 4b This is a CT positioning film processed by the present invention. Detailed Implementation

[0023] The advantages of the present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments.

[0024] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0025] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0026] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0027] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0028] In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two components. They can be direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0029] In the following description, suffixes such as "module," "part," or "unit" used to denote elements are used only for the convenience of the description of the invention and have no specific meaning in themselves. Therefore, "module" and "part" can be used interchangeably.

[0030] See appendix Figure 1Based on the noise distribution on the positioning patch, noise removal can be achieved by removing stripe artifacts. This invention first extracts and processes the stripe artifacts: subtracting the processed stripe artifacts from the image to obtain a noise-reduced image; then, it extracts and processes high-frequency information: adding the processed high-frequency information to the image to obtain a high-frequency enhanced image. Therefore, this invention includes three steps: image denoising, image enhancement, and image post-processing.

[0031] The first major step, image noise reduction, is detailed in the appendix. Figure 2 The noise on the positioning plate manifests as horizontal stripes perpendicular to the axis. Noise removal can be achieved by removing these horizontal stripes.

[0032] 1. Input the original positioning image I in The positioning patch image I is obtained through a series of preprocessing projection reconstructions. ori I ori =-log(I in ).

[0033] 2. Along the positioning patch image I ori A Gaussian filter is applied along the axis to obtain the image I after preliminary artifact removal, where the stripe artifacts are smoothed. sv .

[0034] 3. Position the image I ori Image I after initial artifact removal sv By subtracting the two images, we obtain the initial artifact image A. ori .

[0035] 4. Generate a weighted image W of the artifact distribution. A Specifically, this refers to the original positioning image I. in Gaussian filtering and normalization are performed to obtain the weighted image W of the artifact distribution. A :

[0036] 1) Striped artifacts are mostly distributed in high-absorption regions, i.e., locations with lower pixel values ​​in the image. This applies to the original localization image I. in The filtered image I is obtained by performing maximum value filtering and Gaussian filtering. mg ;

[0037] 2) Filter the image I mg Image I is obtained by normalization. nor , Where C1 > 0 and is an adjustable parameter, the pixel value in the region exceeding this threshold is high and no artifact removal processing is performed;

[0038] 3) Only regions within a preset threshold are processed; otherwise, regions exceeding the preset threshold with high pixel values ​​are not processed for artifact removal. Specifically: if image I... nor If W < 0, then the weighted image W A =0; if image I nor If the weighted image W is greater than 1, then the weighted image W is... A =1; otherwise, the weighted image W A =I nor .

[0039] 5. Strong stripe artifacts may originate from the edges of objects. The atan function can be used to reduce changes in this information. Specifically, the atan function is used to reduce artifacts in the initial image A. ori Thus, the weakened artifact image A is obtained. com A com =atan(C2*A ori ) / C2; where C2>0 and is an adjustable parameter used to adjust the degree of information attenuation. The larger the value of C2, the greater the attenuation.

[0040] 6. Weight the image W A Applied to weakened artifact image A com The weighted artifact image A is obtained. w A w =A com W A .

[0041] 7. Weighted artifact image A w Median filtering and Gaussian filtering are performed along the direction of the artifact to obtain the final artifact A. f .

[0042] 8. Final artifact A f From positioning patch image I ori Image I after artifact removal is obtained. corr That is, the image after noise reduction; I corr =I ori -A f .

[0043] Gaussian filtering is a commonly used image processing filtering method for smoothing, reducing noise, and blurring images. It's based on the weight distribution of a Gaussian function, applying a weighted average to each pixel in the image. The principle of Gaussian filtering is to apply a Gaussian kernel (also called a Gaussian template or Gaussian matrix) to each pixel of the image. The Gaussian kernel is a two-dimensional matrix whose size and shape can be adjusted as needed. In the Gaussian kernel, pixels farther from the center pixel have lower weights, while pixels closer to the center pixel have higher weights. This weight distribution makes the filtered pixel value more influenced by its surrounding pixels. During Gaussian filtering, for each pixel in the image, it is weighted and averaged with its surrounding pixels according to the weights in the Gaussian kernel. In this way, Gaussian filtering can effectively blur details and noise in an image, making the image smoother. Higher weights make the filtered result more influenced by neighboring pixels, while lower weights make the filtered result less influenced by pixels farther away.

[0044] Maxfilter is a commonly used non-linear image filtering method used to smooth images and reduce noise. Its principle is to replace the value of the center pixel with the maximum pixel value in a local neighborhood, thereby removing minor noise and preserving edges. A preferred step of maxfiltering is as follows:

[0045] 1. Define a sliding window or neighborhood that moves across the image, centered on the center pixel. The size of the sliding window is typically a square or rectangle, and can be adjusted as needed.

[0046] 2. For each position in the sliding window, compare the pixel values ​​within the window and select the maximum value.

[0047] 3. Assign the selected maximum value to the pixel value at the center of the window.

[0048] 4. Repeat steps 2 and 3 until the sliding window has moved to all positions of the image and processed all pixels.

[0049] Median filtering is a commonly used non-linear image filtering method used to remove noise and smooth images. Its principle is to replace the value of the center pixel with the median value (i.e., the sorted median value) within a local neighborhood, thereby eliminating the influence of noise. A preferred method for median filtering involves the following steps:

[0050] 1. Define a sliding window or neighborhood that moves across the image, centered on the center pixel. The size of the sliding window is typically a square or rectangle, and can be adjusted as needed.

[0051] 2. For each position in the sliding window, sort the pixel values ​​within the window, and then select the middle value of the sorted values ​​as the new value of the center pixel.

[0052] 3. Assign the selected median value to the pixel value at the center of the window.

[0053] 4. Repeat steps 2 and 3 until the sliding window has moved to all positions of the image and processed all pixels.

[0054] Median filtering is suitable for removing salt-and-pepper noise or impulse noise from images, which manifests as isolated black and white dots. By selecting the median within a sliding window, the value of the noise point is replaced with the median of the surrounding area, thus removing the noise. Compared to other filtering methods, median filtering has the advantage of preserving edges and details. Because median filtering selects the median of neighboring pixels, rather than selecting extreme values ​​or averages like maximum or mean filtering, it does not blur edges and details. This allows median filtering to effectively remove noise while preserving image details.

[0055] The second major step, image enhancement, is detailed in the appendix. Figure 3 The enhancement of the positioning patch is achieved by extracting and enhancing the high-frequency edge information of the image.

[0056] 1. Image I after artifact correction as described above corr The image is filtered using the Laplacian operator to obtain the high-frequency edge image E. ori .

[0057] 2. Generate high-frequency edge image E ori Weighted image W E :

[0058] 1) Image I after artifact correction corr Image E is obtained by performing Gaussian filtering. S ;

[0059] 2) In regions of high absorption, the enhancement of high-frequency information needs to be greater; in regions of low absorption, the enhancement needs to be smaller. Therefore, it is determined that if image E... S If W < 1, then the weighted image W E =1; otherwise, the weighted image

[0060] 3. Weight the image W E Applied to high-frequency edge images E ori The weighted high-frequency image E is obtained. W .

[0061] 4. Use the atan function to reduce the weighted high-frequency image EW Thus, the weakened high-frequency image E is obtained. com ,

[0062] E com =atan(C3*E W ) / C3; where C3>0 and is an adjustable parameter used to adjust the degree of information attenuation. The larger the value of C3, the greater the attenuation.

[0063] 5. Image I after artifact correction corr With weakened high-frequency image E com The images are fused together to obtain the enhanced image I. corr-E I corr-E =I corr +E com .

[0064] Laplacian filtering is a commonly used image processing filtering method to enhance the edges and details of an image. It is based on the Laplacian operator to calculate the second derivative of pixel values ​​in an image, thereby highlighting edges and variations. The principle of Laplacian filtering is to calculate the second derivative of the pixel value by applying the Laplacian operator to each pixel in the image. The Laplacian operator is a linear filter used to detect the rate of change of pixel values ​​in an image. In a two-dimensional image, the Laplacian operator can be represented as a difference operator of a 4-neighborhood or 8-neighborhood. When applying Laplacian filtering, the Laplacian operator is applied to each pixel of the image, and the calculated second derivative is used as the filtered pixel value. This highlights the edges and details in the image, making them more apparent after filtering. A preferred method for Laplacian filtering is through the following steps:

[0065] 1. Convert the image to grayscale (if it is not already grayscale).

[0066] 2. The Laplacian operator is used to calculate the second derivative value of each pixel.

[0067] 3. Use the calculated second derivative value as the filtered pixel value to form the filtered image.

[0068] The third major step, post-processing, involves denoising and enhancing the image, then converting the enhanced image I... corr-E The positioning patch output image I is obtained by performing filtering and grayscale transformation processing. outThis is done to further improve the visual effect of the image. Filtering processes include Gaussian filtering, median filtering, etc. Image grayscale changes refer to adjusting the brightness level of image pixel values, thereby changing the overall brightness of the image. By adjusting the grayscale of an image, the contrast, brightness, and visual effect of the image can be changed, such as through linear transformation, histogram equalization, logarithmic transformation, and phase inversion transformation.

[0069] See Figure 4, compared to CT localization films not treated with this invention ( Figure 4a The positioning piece after being processed by the present invention ( Figure 4b The noise-induced stripe artifacts were eliminated, the skeletal structure was clear, and the image contrast was significantly improved.

[0070] It should be noted that the embodiments of the present invention have better implementability and are not intended to limit the present invention in any way. Any person skilled in the art may use the above-disclosed technical content to change or modify it into equivalent effective embodiments. However, any modifications or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A method for noise reduction and enhancement of CT localization images, characterized in that, Includes the following steps: Based on the artifact distribution, the localization patch image Processing is performed to obtain an image after preliminary artifact removal. ; Positioning image Image after initial artifact removal The difference is calculated to obtain the initial artifact image. For the initial artifact image Weighted processing is performed to obtain the weighted artifact image. The weighted artifact image From positioning patch image By removing the artifacts, we obtain the image after artifact correction. ; Image after artifact correction High-frequency edge images are obtained by performing filtering processing. Generate high-frequency edge images Weighted image Weighted image Applied to high-frequency edge images The weighted high-frequency image is obtained. ; Image after artifact correction With weighted high-frequency images The images are then fused to obtain the enhanced image. ; The positioning image is processed according to the artifact distribution to obtain an image after preliminary artifact removal. Includes: along the positioning patch image Gaussian filtering is applied along the axis to obtain the image after preliminary artifact removal, where the stripe artifacts are smoothed. ; The image along the positioning patch Before performing Gaussian filtering on the axial direction, the following steps are also included: Input the original positioning image The positioning image is obtained through projection reconstruction. , ; The initial artifact image Weighted processing is performed to obtain the weighted artifact image. include: For the original positioning image Gaussian filtering and normalization are performed to obtain the weighted image of the artifact distribution. ; use Function to reduce initial artifact image Thus, the weakened artifact image is obtained. ; Weighted image Applied to weakened artifact images The weighted artifact image is obtained. ; The use Function to reduce initial artifact image Thus, the weakened artifact image is obtained. include: ;in, And these are adjustable parameters; The initial artifact image Weighted processing is performed to obtain the weighted artifact image. This also includes: Weighted artifact image Median filtering and Gaussian filtering are performed along the direction of the artifact to obtain the final artifact. ; The weighted artifact image From positioning patch image By removing the artifacts, we obtain the image after artifact correction. include: The final illusion From positioning patch image By removing the artifacts, we obtain the image after artifact correction. ; The generation of high-frequency edge images Weighted image include: Image after artifact correction Gaussian filtering is performed to obtain the image. ; If the image Then the weighted image Otherwise, the weighted image ; The weighted image Applied to high-frequency edge images The weighted high-frequency image is obtained. This also includes: use High-frequency image after function reduction weighting Thus, the weakened high-frequency image is obtained. , ;in And these are adjustable parameters; The image after artifact correction With weighted high-frequency images The images are then fused to obtain the enhanced image. include: Image after artifact correction Compared with weakened high-frequency images The images are then fused to obtain the enhanced image. .

2. The noise reduction and enhancement method for CT localization images according to claim 1, characterized in that, The original positioning image Gaussian filtering and normalization are performed to obtain the weighted image of the artifact distribution. include: For the original positioning image The filtered image is obtained by performing maximum value filtering and Gaussian filtering. ; The filtered image Image obtained by normalization , ;in, And these are adjustable parameters; If the image Then the weighted image If the image Then the weighted image Otherwise, the weighted image .

3. The noise reduction and enhancement method for CT localization images according to claim 1, characterized in that, The image after artifact correction With weighted high-frequency images The images are then fused to obtain the enhanced image. This also includes: Enhanced image The positioning patch output image is obtained by performing filtering and grayscale transformation processing. .

Images