Oral cone beam CT 3D image denoising method and system based on anatomical priors

Abstract

The present application relates to the field of medical imaging technology, and specifically to an oral cone beam CT three-dimensional image denoising method and system based on anatomical priors, comprising: obtaining a CBCT slice sequence image to be denoised, and performing enhancement sharpening and Gaussian filtering on the denoised image f to obtain a first denoised image I; calculating and extracting an anatomical prior weight image based on the first denoised image I, and further calculating and generating a standard deviation image; performing bilateral filtering on the first denoised image I based on the standard deviation image to obtain a second denoised image I b ; The second denoised image I b Perform median filtering to obtain the third denoised image I m , and the second denoised image I b Perform weighted fusion processing to obtain the final denoised image I dn The present invention can perform smoothing and denoising on oral CBCT images containing a large amount of noise to a large extent, and at the same time can better protect important anatomical detail information, so that doctors can perform more accurate clinical examinations and diagnoses of the oral cavity.

Description

Technical Field

[0001] The present application relates to the field of medical imaging technology, and more specifically, to an oral cone beam CT three-dimensional image denoising method and system based on anatomical priors. Background Art

[0002] Cone beam CT has the advantages of fast scanning, high spatial resolution, and low radiation dose, and is widely used in oral medicine. However, due to factors such as detector electronic noise and X-ray scattering, the reconstructed CBCT images often contain a lot of noise, which seriously affects the image quality, and thus causes certain difficulties and interference for doctors in clinical examination and diagnosis of patients.

[0003] At present, commonly used traditional image denoising methods include mean filtering, median filtering, Gaussian filtering and other methods, but these methods are very likely to cause image blur and lose many image details with clinical value, and cannot meet the needs of clinical diagnosis; some commonly used edge-preserving denoising methods such as bilateral filtering have a certain preservation effect on the edges of some strong-contrast structures, but while performing a large degree of smoothing, the preservation effect on some weak-contrast detail edges is still limited; there are also some image denoising methods such as non-local mean filtering, BM3D filtering and other methods, which have better detail preservation effects, but the calculation is large and time-consuming, and it is difficult to apply in actual clinical practice.

[0004] Therefore, it is a challenging and critical issue to minimize noise while maintaining important details. In oral medicine, details such as alveolar bone and tooth margins are very important for doctors to examine and diagnose. How to provide a denoising method and system for oral cone beam CT, which can reduce noise as much as possible during the image denoising process while maintaining clinically important details such as alveolar bone and tooth margins, has become a technical problem that needs to be solved urgently. To this end, we propose an oral cone beam CT three-dimensional image denoising method and system based on anatomical priors to solve the above-mentioned problems. Summary of the invention

[0005] In view of the above problems, this application proposes an oral cone-beam CT three-dimensional image denoising method and system based on anatomical priors, which can reduce noise as much as possible during the image denoising process and simultaneously maintain clinically important detail information such as alveolar bone and tooth edges, so that doctors can perform more accurate clinical examinations and diagnoses of the oral cavity, thereby improving the above problems.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an oral cone beam CT three-dimensional image denoising method based on anatomical priors, comprising the following steps: step 1: obtaining a set of CBCT slice sequence images to be denoised;

[0007] Step 2: Perform enhancement sharpening and Gaussian filtering on each layer of the denoised image f in turn to obtain a first denoised image I;

[0008] Step 3: Calculate and extract the anatomical prior weight image w of the current image I according to the image edge coefficient and image grayscale distribution of the first denoised image I, and perform median filtering on it;

[0009] Step 4: Calculate and generate the standard deviation image based on the anatomical prior weight image w obtained in step 3, including the spatial standard deviation image σ s And grayscale standard deviation image σ g ;

[0010] Step 5: Based on the spatial standard deviation image σ obtained in step 4 s And gray standard deviation image σ g , a three-dimensional bilateral filtering process is performed on the first denoised image I to obtain a second denoised image I b ;

[0011] Step 6: The second denoised image I obtained in step 5 b A median filter can be further performed to obtain a third denoised image I m , and then with the second denoised image I b Perform weighted fusion processing to obtain the final denoised image I dn .

[0012] Preferably, the step of obtaining a set of CBCT slice sequence images to be denoised comprises: scanning the patient at different angles using oral cone beam CT to acquire projection data, and reconstructing the projection data to obtain a set of slice sequence images to be denoised.

[0013] Preferably, the step of sequentially performing enhancement sharpening and Gaussian filtering on each layer of the to-be-denoised image f to obtain the first denoised image I is specifically:

[0014] Select any voxel point with coordinates (x, y, z) in the image to be denoised f and perform three-dimensional edge enhancement using the Laplacian operator to obtain an edge-sharpened image h;

[0015] In order to suppress the enhancement of noise interference, a three-dimensional Gaussian filter is performed on the sharpened image h to obtain a first denoised image I.

[0016] Preferably, the anatomical prior weight image w of the current image I is obtained by calculating and extracting the image edge coefficient and the image grayscale distribution of the first denoised image I, and performing median filtering processing on it, specifically:

[0017] Calculate the edge coefficient D(x,y,z) of the current pixel based on the neighborhood of the pixel to be processed;

[0018] The anatomical prior weighted image w can be calculated based on the image grayscale distribution and the obtained edge coefficient D(x, y, z); the obtained anatomical prior weighted image w is subjected to median filtering.

[0019] Preferably, the calculation method of the anatomical prior weight image w in step 3 is specifically as follows:

[0020]

[0021] in,

[0022] Preferably, the anatomical prior weight image w obtained in step 3 is calculated to generate a standard deviation image, including a spatial standard deviation image σ s And gray standard deviation image σ g , the specific calculation method is:

[0023] σ s (x,y,z)=σ smax -w(x,y,z)·(σ smax -σ smin )

[0024] σ g (x,y,z)=σ gmax -w(x,y,z)·(σ gmax -σ gmin ).

[0025] Preferably, the spatial standard deviation image σ obtained according to step 4 s And gray standard deviation image σ g , a three-dimensional bilateral filtering process is performed on the first denoised image I to obtain a second denoised image I b , the specific calculation is as follows:

[0026]

[0027] Preferably, the second denoised image I obtained in step 5 b A median filter can be further performed to obtain a third denoised image I m , and then with the second denoised image I b Perform weighted fusion processing to obtain the final denoised image I dn :

[0028] I dn (x,y,z)=w m (x,y,z)·I m (x,y,z)+[1-w m (x,y,z)]·I b (x,y,z).

[0029] The oral cone beam CT three-dimensional image denoising system based on anatomical priors is characterized by comprising: an acquisition module: mainly used to acquire CBCT slice sequence image data to be denoised;

[0030] Enhancement module: mainly uses the Laplacian operator to enhance and sharpen the image edges;

[0031] Extraction module: Calculate and extract the anatomical prior weight image based on the image edge coefficient and image grayscale distribution;

[0032] Generation module: Calculate and generate the standard deviation image based on the anatomical prior weight image;

[0033] Filtering module: Use Gaussian filtering, median filtering or bilateral filtering to filter and smooth the image;

[0034] Fusion module: mainly performs weighted fusion of images obtained using different processing methods.

[0035] Preferably, the method comprises the following steps:

[0036] S1, firstly obtain a set of CBCT slice sequence image data to be denoised from the CBCT system through an acquisition module;

[0037] S2, enhancing and sharpening the edge of the denoised image through an enhancement module to obtain an edge-sharpened image;

[0038] S3, then using a filtering module to perform Gaussian filtering on the edge-sharpened image to obtain a first denoised image I;

[0039] S4, calculating and extracting the first denoised image I through the extraction module to obtain an anatomical priori weighted image, and performing median filtering on the anatomical priori weighted image through the filtering module;

[0040] S5, then generate a standard deviation image by using a generation module and calculating the anatomical prior weight image;

[0041] S6, then use the filtering module to perform bilateral filtering on the first denoised image I according to the standard deviation image to obtain a second denoised image I b ;

[0042] S7, then the second denoised image I is filtered by a filtering module b Perform median filtering to obtain the third denoised image I m ;

[0043] S8, finally, the second denoised image I is fused through the fusion module b and the third denoised image I m Perform weighted fusion to obtain the final denoised image Idn .

[0044] The oral cone beam CT three-dimensional image denoising method and system based on anatomical priors provided in this application have the following beneficial effects:

[0045] 1. The oral cone beam CT three-dimensional image denoising method and system based on anatomical prior enhances the image through step 2 to improve the contrast of details such as alveolar bone and tooth edge, and then calculates the anatomical prior weight image through step 3, and then generates the standard deviation image according to the weight calculation through step 4, and performs filtering and denoising through step 5. This process can protect important anatomical details such as alveolar bone and tooth edge, and finally smoothes the tooth hard tissue through step 6 to further eliminate residual noise. In the denoising process, important anatomical details such as alveolar bone and tooth edge can be effectively maintained, so that doctors can make more accurate clinical examinations and diagnoses of the oral cavity.

[0046] 2. The oral cone beam CT three-dimensional image denoising method and system based on anatomical priors has low computational complexity and reduces the demand for hardware, software and other infrastructure, and can be quickly applied in clinical practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0048] Figure 1 It is a flow chart of the oral cone beam CT three-dimensional image denoising method based on anatomical priors of the present invention;

[0049] Figure 2 It is a flow chart of the oral cone beam CT three-dimensional image denoising system based on anatomical priors of the present invention;

[0050] Figure 3 The image to be denoised in the oral cone beam CT three-dimensional image denoising method based on anatomical priors of the present invention;

[0051] Figure 4 This is the final denoised image of the oral cone beam CT three-dimensional image denoising method based on anatomical priors of the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0053] See also Figures 1 to 4 The present invention provides a technical solution: an oral cone beam CT three-dimensional image denoising method based on anatomical priors, comprising the following steps: Step 1: acquiring a set of CBCT slice sequence images to be denoised;

[0054] Step 2: Perform enhancement sharpening and Gaussian filtering on each layer of the denoised image f in turn to obtain a first denoised image I;

[0055] Step 3: Calculate and extract the anatomical prior weight image w of the current image I according to the image edge coefficient and image grayscale distribution of the first denoised image I, and perform median filtering on it;

[0056] Step 4: Calculate and generate the standard deviation image based on the anatomical prior weight image w obtained in step 3, including the spatial standard deviation image σ s And gray standard deviation image σ g ;

[0057] Step 5: Based on the spatial standard deviation image σ obtained in step 4 s And gray standard deviation image σ g , a three-dimensional bilateral filtering process is performed on the first denoised image I to obtain a second denoised image I b ;

[0058] Step 6: The second denoised image I obtained in step 5 b A median filter can be further performed to obtain a third denoised image I m , and then with the second denoised image I b Perform weighted fusion processing to obtain the final denoised image I dn .

[0059] The step of obtaining a set of CBCT slice sequence images to be denoised is as follows: scanning the patient at different angles using oral cone beam CT to acquire projection data, and reconstructing the projection data to obtain a set of slice sequence images to be denoised.

[0060] The first denoised image I obtained by sequentially performing enhancement sharpening and Gaussian filtering on each layer of the denoised image f is specifically:

[0061] Select any voxel point with coordinates (x, y, z) in the denoised image f and use the Laplace operator to perform three-dimensional edge enhancement to obtain an edge-sharpened image h; this step is mainly for edge enhancement, which can enhance the contrast of details such as alveolar bone and tooth edge, which is conducive to the calculation and extraction of its edge coefficient;

[0062] The calculation formula is: Wherein, c is the enhancement factor parameter, and in the present invention, c=0.04;

[0063] In order to suppress the enhancement of noise interference, the sharpened image h is subjected to three-dimensional Gaussian filtering to obtain the first denoised image I, as follows: Among them, point q is the voxel point to be calculated in image h, point p is any point in the neighborhood Q of point q, and w g (x p ,y p ,z p ) is the weight of point p, which is calculated as follows: Wherein, σ is a preset Gaussian filter smoothing parameter (the value of the present invention is σ=0.5).

[0064] The anatomical prior weight image w of the current image I is obtained by calculating and extracting the image edge coefficient and image grayscale distribution of the first denoised image I, and performing median filtering on the image. Specifically:

[0065] Calculate the edge coefficient D(x,y,z) of the current pixel based on the neighborhood of the pixel to be processed;

[0066] The calculation formula is: D(x,y,z)=ex(x,y,z)+ey(x,y,z)+ez(x,y,z),

[0067] Where ex, ey, ez are the edge coefficients dx, dy, dz obtained along different directions of x, y, z respectively, which are normalized as follows:

[0068] in, The coefficient of the edge coefficient dx, dy or dz at the point (x, y, z) can be expressed as d(x, y, z), and after normalization, it can correspond to e(x, y, z). The edge coefficient here mainly reflects the degree of change between layers of the anatomical structure in different directions. The smaller edge coefficient is mainly relatively uniform tissue such as soft tissue, and the larger edge coefficient is mainly the edge details of hard tissues such as tooth edges and alveolar bones. The edge coefficients dx, dy, and dz are obtained as follows:

[0069]

[0070]

[0071]

[0072] According to the image grayscale distribution and the obtained edge coefficient D(x, y, z), the anatomical prior weight image w can be calculated. The calculation formula is:

[0073] in, α is a preset fixed parameter (the value of the present invention is α=50), t min and t max is the preset grayscale threshold (the value of the present invention is t min =0,t max =800). The anatomical prior weight mainly reflects the degree of detail preservation during the denoising process. Relatively large weights are assigned to alveolar bones and tooth edges with larger anatomical edge coefficients and tooth hard tissues with higher grayscale, which can better protect the weaker edge details on the tooth hard tissues.

[0074] The obtained anatomical prior weight image w is subjected to median filtering.

[0075] The anatomical prior weight image w obtained in step 3 is calculated to generate a standard deviation image, including a spatial standard deviation image σ s And gray standard deviation image σ g , the specific calculation method is:

[0076] σ s (x,y,z)=σ smax -w(x,y,z)·(σ smax -σ smin )

[0077] σ g (x,y,z)=σ gmax -w(x,y,z)·(σ gmax -σ gmin )

[0078] Among them, σ smax , σ smin and σ gmax , σ gmin is the preset threshold parameter (the value of the present invention is σ smax =2.1,σ smin =1.4,σ gmax =700,σ gmin =250). Through the calculation in this step, a smaller smoothing parameter is obtained for a voxel with a larger anatomical prior weight, and a larger smoothing parameter is obtained for a voxel with a smaller anatomical prior weight.

[0079] The spatial standard deviation image σ obtained according to step 4 s And gray standard deviation image σg , a three-dimensional bilateral filtering process is performed on the first denoised image I to obtain a second denoised image I b , the specific calculation is as follows:

[0080]

[0081] in, Point q is any pixel point in the image I, point p is any point in the neighborhood Q of point q with a size of r×r×r (the value of r in the present invention is 5), and w s (x p ,y p ,z p ) and w g (x p ,y p ,z p ) represent the spatial weight and grayscale weight respectively, which are calculated as follows:

[0082]

[0083]

[0084] The second denoised image I obtained in step 5 b A median filter can be further performed to obtain a third denoised image I m , and then with the second denoised image I b Perform weighted fusion processing to obtain the final denoised image I dn :

[0085] I dn (x,y,z)=w m (x,y,z)·I m (x,y,z)+[1-w m (x,y,z)]·I b (x,y,z)

[0086] Among them, the weight w m It can be calculated as follows:

[0087]

[0088] Among them, s min and max is the preset grayscale threshold (the value of this invention is s min =700,s max =1000). In order to avoid the residual noise in the uniform hard tissue caused by the low smoothing filtering of the tooth hard tissue in step 5, this step mainly allocates a larger smoothing weight to the tooth hard tissue, so that the smoothing of the tooth hard tissue is further enhanced and the residual noise is eliminated.

[0089] The oral cone beam CT three-dimensional image denoising system based on anatomical priors is characterized by comprising: an acquisition module: mainly used to acquire CBCT slice sequence image data to be denoised;

[0090] Enhancement module: mainly uses the Laplacian operator to enhance and sharpen the image edges;

[0091] Extraction module: Calculate and extract the anatomical prior weight image based on the image edge coefficient and image grayscale distribution;

[0092] Generation module: Calculate and generate the standard deviation image based on the anatomical prior weight image;

[0093] Filtering module: Use Gaussian filtering, median filtering or bilateral filtering to filter and smooth the image;

[0094] Fusion module: mainly performs weighted fusion of images obtained using different processing methods.

[0095] The oral cone beam CT three-dimensional image denoising system based on anatomical prior includes the following steps:

[0096] S1, firstly obtain a set of CBCT slice sequence image data to be denoised from the CBCT system through an acquisition module;

[0097] S2, enhancing and sharpening the edge of the denoised image through an enhancement module to obtain an edge-sharpened image;

[0098] S3, then using a filtering module to perform Gaussian filtering on the edge-sharpened image to obtain a first denoised image I;

[0099] S4, calculating and extracting the first denoised image I through the extraction module to obtain an anatomical priori weighted image, and performing median filtering on the anatomical priori weighted image through the filtering module;

[0100] S5, then generate a standard deviation image by using a generation module and calculating the anatomical prior weight image;

[0101] S6, then use the filtering module to perform bilateral filtering on the first denoised image I according to the standard deviation image to obtain a second denoised image I b ;

[0102] S7, then the second denoised image I is filtered by a filtering module b Perform median filtering to obtain the third denoised image I m ;

[0103] S8, finally, the second denoised image I is fused through the fusion module b and the third denoised image I mPerform weighted fusion to obtain the final denoised image I dn .

[0104] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific position, be constructed and operated in a specific position, and therefore cannot be understood as limiting the present invention; the terms "first", "second", and "third" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance. In addition, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not expressly listed, or also includes elements inherent to such process, method, article, or apparatus.

[0105] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An oral cone beam CT three-dimensional image denoising method based on anatomical priors, characterized by: The method comprises the following steps: Step 1: obtaining a set of CBCT slice sequence images to be denoised; Step 2: Perform enhancement sharpening and Gaussian filtering on each layer of the denoised image f in turn to obtain a first denoised image I; Step 3: Calculate and extract the anatomical prior weight image w of the current image I according to the image edge coefficient and image grayscale distribution of the first denoised image I, and perform median filtering on it; The method of calculating and extracting the anatomical prior weight image w of the current image I based on the image edge coefficient and the image grayscale distribution of the first denoised image I and performing median filtering on the image w is as follows: Calculate the edge coefficient D(x,y,z) of the current pixel based on the neighborhood of the pixel to be processed; According to the image grayscale distribution and the obtained edge coefficient D(x, y, z), its anatomical prior weight image w can be calculated; The calculation formula is: in, α is a preset fixed parameter, t min and t max To preset the grayscale threshold, median filtering is performed on the obtained anatomical prior weight image w; Step 4: Calculate and generate the standard deviation image based on the anatomical prior weight image w obtained in step 3, including the spatial standard deviation image σ s And grayscale standard deviation image σ g ; Step 5: Based on the spatial standard deviation image σ obtained in step 4 s And grayscale standard deviation image σ g , a three-dimensional bilateral filtering process is performed on the first denoised image I to obtain a second denoised image I b ; Step 6: The second denoised image I obtained in step 5 b A median filter can be further performed to obtain a third denoised image I m , and then with the second denoised image I b Perform weighted fusion processing to obtain the final denoised image I dn .

2. The oral cone beam CT three-dimensional image denoising method based on anatomical priors according to claim 1, characterized in that: The step of obtaining a set of CBCT slice sequence images to be denoised is as follows: scanning the patient at different angles using oral cone beam CT to acquire projection data, and reconstructing the projection data to obtain a set of slice sequence images to be denoised.

3. The oral cone beam CT three-dimensional image denoising method based on anatomical priors according to claim 1, characterized in that: The first denoised image I obtained by sequentially performing enhancement sharpening and Gaussian filtering on each layer of the denoised image f is specifically: Select any voxel point with coordinates (x, y, z) in the image to be denoised f and perform three-dimensional edge enhancement using the Laplacian operator to obtain an edge-sharpened image h; In order to suppress the enhancement of noise interference, a three-dimensional Gaussian filter is performed on the sharpened image h to obtain a first denoised image I.

4. The oral cone beam CT three-dimensional image denoising method based on anatomical priors according to claim 1, characterized in that: The anatomical prior weight image w obtained in step 3 is calculated to generate a standard deviation image, including a spatial standard deviation image σ s And grayscale standard deviation image σ g The specific calculation method is: s s (x,y,z)=σ smax -w(x,y,z)·(σ smax -s smin ) s g (x,y,z)=σ gmax -w(x,y,z)·(σ gmax -s gmin ) Among them, σ smax , σ smin and σ gmax , σ gmin is the preset threshold parameter.

5. The oral cone beam CT three-dimensional image denoising method based on anatomical priors according to claim 1, characterized in that: The spatial standard deviation image σ obtained according to step 4 s And grayscale standard deviation image σ g , a three-dimensional bilateral filtering process is performed on the first denoised image I to obtain a second denoised image I b , the specific calculation is as follows: in, Point q is any pixel point in image I, point p is any point in the neighborhood Q of point q with a size of r×r×r, and w s (x p ,y p ,z p ) and w g (x p ,y p ,z p ) represent the spatial weight and grayscale weight respectively, which are calculated as follows:

6. The oral cone beam CT three-dimensional image denoising method based on anatomical priors according to claim 1, characterized in that: The second denoised image I obtained in step five b A median filter can be further performed to obtain a third denoised image I m , and then with the second denoised image I b Perform weighted fusion processing to obtain the final denoised image I dn : I dn (x,y,z)=w m (x,y,z)·I m (x,y,z)+[1-w m (x,y,z)]·I b (x,y,z) Among them, the weight w m It can be calculated as follows: Among them, s min and max is the preset grayscale threshold.

7. A denoising system for implementing the oral cone beam CT three-dimensional image denoising method based on anatomical priors as described in any one of claims 1 to 6, characterized in that: It includes: an acquisition module: used for acquiring CBCT slice sequence image data to be denoised; Enhancement module: Use the Laplacian operator to enhance and sharpen the image edges; Extraction module: Calculate and extract the anatomical prior weight image based on the image edge coefficient and image grayscale distribution; Generation module: Calculate and generate the standard deviation image based on the anatomical prior weight image; Filtering module: Use Gaussian filtering, median filtering or bilateral filtering to filter and smooth the image; Fusion module: weighted fusion of images obtained using different processing methods.

8. The oral cone beam CT three-dimensional image denoising system based on anatomical priors according to claim 7, characterized in that: S1, firstly obtain a set of CBCT slice sequence image data to be denoised from the CBCT system through an acquisition module; S2, enhancing and sharpening the edge of the denoised image through an enhancement module to obtain an edge-sharpened image; S3, then using a filtering module to perform Gaussian filtering on the edge-sharpened image to obtain a first denoised image I; S4, calculating and extracting the first denoised image I through the extraction module to obtain an anatomical priori weighted image, and performing median filtering on the anatomical priori weighted image through the filtering module; S5, then generate a standard deviation image by using a generation module and calculating the anatomical prior weight image; S6, then use the filtering module to perform bilateral filtering on the first denoised image I according to the standard deviation image to obtain a second denoised image I b ; S7, then the second denoised image I is filtered by a filtering module b Perform median filtering to obtain the third denoised image I m ; S8, finally, the second denoised image I is fused through the fusion module b and the third denoised image I m Perform weighted fusion to obtain the final denoised image I dn .

Images