Neutron CT image denoising method, model training method, device and equipment

The feature fusion of neutron CT images and X-ray CT images through the image denoising model solves the noise problem of neutron CT images, improves the detailed performance and clarity of the image, and improves the quality and reliability of neutron CT images.

CN118941458AActive Publication Date: 2024-11-12CHINA SPALLATION NEUTRON SOURCE SCI CENT +1

Patent Information

Application Number
CN202410980567.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-12
Estimated Expiration
2044-07-22

AI Technical Summary

Technical Problem

Noise problems in the reconstruction process of neutron CT images lead to reduced image clarity and accuracy. Existing filtering and denoising methods sacrifice details and resolution, while iterative reconstruction algorithms are computationally expensive and converging speed is slow.

Method used

By acquiring neutron CT images and X-ray CT images of the same lesion area, the image denoising model is used to perform feature fusion, including encoder, feature fusion and decoder, to reduce noise in neutron CT images and improve the detailed performance of the image and overall clarity.

Benefits of technology

It effectively reduces random noise and system noise in neutron CT images, improves the detailed performance and overall clarity of the image, and improves the quality and reliability of neutron CT images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118941458A_ABST
    Figure CN118941458A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a neutron CT image denoising method, a neutron CT image model training method, a neutron CT image model training device and neutron CT image model training equipment. A neutron CT image and an X-ray CT image corresponding to the same focus area are obtained; processing the neutron CT image and the X-ray CT image based on an image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image; wherein the image denoising model comprises an encoder, a feature fusion device for performing feature fusion on a first feature vector which is output by the encoder and corresponds to the neutron CT image and a second feature vector which is output by the encoder and corresponds to the X-ray CT image, and a decoder for decoding the fused features to obtain a denoised neutron CT image; according to the method, the complementary characteristics of the neutron and the X-ray CT image are fully utilized, the noise in the neutron CT image is effectively reduced, the detail expression and the overall definition of the image are improved, and the quality and the reliability of the neutron CT image are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of image processing technology, and in particular to a neutron CT image denoising method and a model training method, device and equipment. Background Art

[0002] In the field of modern science and technology, computed tomography (CT) technology is widely used in research and diagnosis scenarios, for example, neutron computed tomography (NCT). Unlike X-rays, neutrons have unique penetrating properties. They can effectively penetrate certain metals and heavy elements, while showing high sensitivity to light elements (especially hydrogen), which makes NCT show great potential and value. However, in the process of neutron CT image reconstruction, the noise problem is a recognized challenge. The presence of noise can significantly reduce the clarity and accuracy of the image, affecting the results of diagnosis and analysis.

[0003] At present, the commonly used neutron CT image denoising methods are filtering denoising methods; and methods using iterative reconstruction algorithms, such as algebraic reconstruction technique (ART) and statistical iterative reconstruction technique (SIRT).

[0004] However, the filtering denoising method will sacrifice the details and resolution in the neutron CT sample image, and there is a problem of transition smoothness, which leads to blurred image edges and affects the clarity of the image. The method using iterative reconstruction algorithm usually has a large amount of calculation, requires a long processing time, and may have a slow convergence speed during the iteration process. Summary of the invention

[0005] The embodiments of the present invention provide a neutron CT image denoising method and a model training method, device and equipment to reduce the noise in the neutron CT image, improve the detail expression and overall clarity of the neutron CT image, and improve the quality and reliability of the neutron CT image.

[0006] In a first aspect, an embodiment of the present invention provides a neutron CT image denoising method, the method comprising:

[0007] Obtaining neutron CT images and X-ray CT images corresponding to the same lesion area;

[0008] The neutron CT image and the X-ray CT image are processed based on the image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image; wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features to obtain the denoised neutron CT image.

[0009] In a second aspect, an embodiment of the present invention further provides a method for training an image denoising model, wherein the image denoising model includes an encoder to be trained, a feature fuser to be trained, and a decoder to be trained, and the method includes:

[0010] Acquire multiple training samples; wherein each training sample includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image;

[0011] For each training sample, the neutron CT sample image and the X-ray CT sample image in the current training sample are input into the encoder to be trained to obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image;

[0012] Inputting the first sample feature vector and the second sample feature vector into the feature fuser to be trained to obtain a target sample feature vector;

[0013] Input the target sample feature vector into the decoder to be trained to obtain the target sample output image;

[0014] Determine a target loss value based on each first pixel value in the target sample output image and each second pixel value at a corresponding pixel position in the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value;

[0015] The convergence of the loss function in the image denoising model is taken as the training goal to obtain the trained image denoising model.

[0016] In a third aspect, an embodiment of the present invention further provides a neutron CT image denoising device, the device comprising:

[0017] A dual-modality image acquisition module, used to acquire neutron CT images and X-ray CT images corresponding to the same lesion area;

[0018] The denoising image output module is used to process the neutron CT image and the X-ray CT image based on the image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image; wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on the first feature vector corresponding to the neutron CT image and the second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features.

[0019] In a fourth aspect, an embodiment of the present invention further provides a training device for an image denoising model, wherein the image denoising model includes an encoder to be trained, a feature fuser to be trained, and a decoder to be trained, and the device includes:

[0020] A sample acquisition module, used to acquire multiple training samples; wherein each training sample includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image;

[0021] A sample feature extraction module is used to input the neutron CT sample image and the X-ray CT sample image in the current training sample into the encoder to be trained for each training sample, and obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image;

[0022] A sample feature fusion module, used for inputting the first sample feature vector and the second sample feature vector into a feature fuser to be trained to obtain a target sample feature vector;

[0023] An image output module is used to input the target sample feature vector into the decoder to be trained to obtain the target sample output image;

[0024] A model correction module, used to determine a target loss value based on a target sample output image and a denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value;

[0025] The denoising model determination module is used to take the convergence of the loss function in the image denoising model as a training target to obtain a trained image denoising model.

[0026] In a fifth aspect, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0027] one or more processors;

[0028] a storage device for storing one or more programs,

[0029] When the one or more programs are executed by the one or more processors, the one or more processors implement the neutron CT image denoising method or the image denoising model training method as described in any one of the embodiments of the present invention.

[0030] In a sixth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the neutron CT image denoising method or the image denoising model training method as described in any one of the embodiments of the present invention.

[0031] The technical solution of the embodiment of the present invention obtains a neutron CT image and an X-ray CT image corresponding to the same lesion area, and then processes the neutron CT image and the X-ray CT image based on an image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image, wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features to obtain a denoised neutron CT image, which fully utilizes the complementary characteristics of neutron and X-ray CT images, enhances the expression ability of the fused image, effectively reduces random noise and system noise in the neutron CT image, improves the detail expression and overall clarity of the image, and improves the quality and reliability of the neutron CT image. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required to describe the embodiments. Obviously, the drawings introduced are only drawings of a part of the embodiments to be described in the present invention, rather than all the drawings. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 A schematic diagram of a process for neutron CT image denoising provided by an embodiment of the present invention;

[0034] Figure 2 A schematic diagram of an image denoising model involved in an embodiment of the present invention;

[0035] Figure 3 A schematic diagram of an image denoising model with a Swin-Transformer network structure according to an embodiment of the present invention;

[0036] Figure 4 It is a schematic diagram of the structure of a feature fuser involved in an embodiment of the present invention;

[0037] Figure 5A schematic diagram of a flow chart of a method for training an image denoising model provided by an embodiment of the present invention;

[0038] Figure 6 A schematic diagram of the structure of a neutron CT image denoising device provided by an embodiment of the present invention;

[0039] Figure 7 A schematic diagram of the structure of a training device for an image denoising model provided by an embodiment of the present invention;

[0040] Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0042] Embodiment 1

[0043] Before introducing the technical solution, an exemplary description of the application scenario may be given. The neutron CT image denoising method provided in this embodiment may be applied in any scenario where neutron CT images need to be denoised.

[0044] In the field of modern science and technology, imaging technology has become an important tool for research and diagnosis. Traditional imaging technologies, such as X-ray Computed Tomography (X-ray CT), are widely known for their wide applications in medical imaging and industrial non-destructive testing. X-ray CT technology uses X-rays to penetrate samples, captures the rays after passing through the sample through the detector, and then reconstructs the internal structure image of the sample. However, X-ray CT technology has limitations in certain application scenarios. X-rays have weak penetration ability for light elements, which limits their application in analyzing samples containing a large amount of light element materials (such as biological tissues). In addition, X-rays also have limited penetration ability for metals and heavy elements, which poses a challenge in materials science and archaeological research.

[0045] To overcome these limitations, scientists have developed a new imaging technology, Neutron Computed Tomography (NCT). Unlike X-rays, neutrons have unique penetrating properties. They can effectively penetrate certain metals and heavy elements, while showing high sensitivity to light elements (especially hydrogen). This feature makes NCT technology show great potential and value in many fields.

[0046] In the image reconstruction process of computed tomography (CT), the noise problem is a recognized challenge. The presence of noise can significantly reduce the clarity and accuracy of the image, affecting the results of diagnosis and analysis. Although the existing technology has taken a series of measures to reduce the impact of noise, there are still some shortcomings, which are summarized as follows:

[0047] 1. Limitations of filtering technology: Traditional filtering technologies, such as Gaussian filtering, mean filtering, and median filtering, can reduce random noise in images to a certain extent, but these methods often sacrifice image details and resolution. In particular, Gaussian filtering, due to its smoothing effect, may cause image edges to be blurred, affecting image clarity.

[0048] 2. Optimization space of reconstruction algorithms: Iterative reconstruction algorithms, including algebraic reconstruction technology (ART) and statistical iterative reconstruction technology (SIRT), can optimize image quality and reduce noise through multiple iterations, but these algorithms are usually computationally intensive, require a long processing time, and may have slow convergence problems during the iteration process.

[0049] 3. Challenges of multimodal fusion: Although multimodal fusion technology can take advantage of the advantages of different imaging modalities and improve imaging quality, in actual operation, the accuracy of image registration is crucial to the fusion effect. The differences in physical properties between images of different modalities may lead to difficulties in registration and affect the quality of the fused image.

[0050] 4. Insufficient preprocessing of raw projection data: In the preprocessing stage before image reconstruction, although noise filtering and correction techniques are used, these methods may not be able to completely eliminate noise, especially for images with complex structures or high noise levels, the preprocessing effect may not be satisfactory.

[0051] In summary, although the existing noise reduction technology has solved the noise problem in CT image reconstruction to a certain extent, there are still many challenges to overcome. Based on this, the present application aims to provide a neutron CT noise reduction method in a neutron and X-ray dual-modality system to solve the problem that the image quality in the existing neutron CT technology is seriously affected by noise.

[0052] Figure 1 A flow chart of a neutron CT image denoising method provided in an embodiment of the present invention is provided. This embodiment is applicable to situations where neutron CT images need to be denoised. The method can be executed by a neutron CT image denoising device, which can be implemented in the form of software and / or hardware. The hardware can be an electronic device such as a server. The electronic device can execute the neutron CT image denoising method provided in the technical solution to display target display items for users.

[0053] like Figure 1 As shown, the method includes:

[0054] S110 , obtaining a neutron CT image and an X-ray CT image corresponding to the same lesion area.

[0055] The neutron CT image is an image of the lesion region acquired using neutron computed tomography technology, and the X-ray CT image is an image of the lesion region acquired using X-ray computed tomography.

[0056] In this embodiment, a dual-modality imaging system can be used to simultaneously acquire neutron CT images and X-ray CT images. The dual-modality imaging system combines the advantages of neutron CT and X-ray CT, and can simultaneously acquire neutron images and X-ray images of the same sample, thereby providing a more comprehensive and in-depth analysis. The key components of the dual-modality imaging system include:

[0057] Neutron source and X-ray source: two independent radiation sources, used to generate neutrons and X-rays respectively to meet different imaging requirements;

[0058] Sample stage: A versatile device that can stably position and rotate samples in both imaging modes, ensuring consistency and accuracy during imaging;

[0059] Dual-mode detector: An innovative detector design that can detect neutrons and X-rays simultaneously, or detect neutrons and X-rays separately through two independent detectors to accommodate different imaging needs.

[0060] For example, a medical image of the abdominal area of ​​a patient may be taken based on the dual-modality imaging system to obtain a neutron CT image and an X-ray CT image corresponding to the abdominal area of ​​the patient.

[0061] S120 , processing the neutron CT image and the X-ray CT image based on the image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image.

[0062] The denoised neutron CT image is the image content obtained after denoising the neutron CT image. The image denoising model includes an encoder, a feature fusion device for performing feature fusion on the first feature vector corresponding to the neutron CT image and the second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features to obtain the denoised neutron CT image. In this embodiment, the schematic diagram of the image denoising model is shown in FIG. Figure 2 ,like Figure 2 As shown, the image denoising model includes an encoder, a feature fusion device and a decoder. For example, the encoder and the decoder are Swin-Transformer network structures.

[0063] In practical applications, the neutron CT image and the X-ray CT image are input into the image denoising model, and the image denoising model can process the input image to obtain a denoised neutron CT image.

[0064] More specifically, based on the image denoising model, the neutron CT image and the X-ray CT image are processed to obtain a denoised neutron CT image corresponding to the neutron CT image. The specific implementation method includes:

[0065] S1201. Input a neutron CT image and an X-ray CT image into an encoder to obtain a first eigenvector corresponding to the neutron CT image and a second eigenvector corresponding to the X-ray CT image.

[0066] In this embodiment, the neutron CT image and the X-ray CT image are input into the encoder in parallel, and the encoder encodes the neutron CT image to obtain a first feature vector corresponding to the neutron CT image; the encoder encodes the X-ray CT image to obtain a second feature vector corresponding to the X-ray CT image.

[0067] For an exemplary image denoising model with a Swin-Transformer network structure, see Figure 3 , the neutron CT image (i.e. Figure 3 Noise in neutron CT images) and X-ray CT images (i.e. Figure 3 The raw data of these two modalities are input into the Swin-Transformer network in parallel. The Swin-Transformer network consists of a Transformer encoder and a decoder. The encoder uses a multi-head self-attention mechanism to extract features from the input neutron CT images and X-ray CT images to capture the long-range dependencies between features of different modalities. Specifically, the encoder extracts multi-scale features from the input data through a stacked multi-head self-attention and feedforward neural network to ensure that the details and overall information of each modality data can be fully captured.

[0068] S1202: Input the first feature vector and the second feature vector into a feature fuser to obtain a target fused feature vector.

[0069] The target fused feature vector is a feature vector obtained by fusing the first feature vector and the second feature vector.

[0070] In this embodiment, the first eigenvector and the second eigenvector can be spliced ​​to obtain a target fused eigenvector, and this feature fuser is a feature splicing fuser. In addition, the first eigenvector and the second eigenvector can be fused by bilinear pooling fusion, and this feature fuser is a bilinear pooling feature fuser. The bilinear fusion method generates a fused feature representation by calculating the bilinear pooling of two modal features. Bilinear pooling is an effective multimodal feature fusion method that can capture the complex interactive relationship between different modal features. This fusion method can make full use of the complementarity of neutron and X-ray data, enhance the expressiveness of features, and make the fused features richer and more accurate.

[0071] Specifically, for the bilinear pooling fusion method, the feature fusion structure diagram is as follows: Figure 4 As shown, the feature fusion device may include: a feature element normalization unit, a low-rank bilinear mapping unit, a self-attention unit and a fusion feature output unit. The first feature vector and the second feature vector are input into the feature fusion device, and a specific implementation method of obtaining a target fusion feature vector may include:

[0072] S1. Based on a feature element standardization unit, the first feature vector and the second feature vector are respectively standardized to obtain a first standard feature vector and a second standard feature vector.

[0073] Specifically, the first eigenvector and the second eigenvector can be expressed as: the first eigenvector X∈R^(b×d1); the second eigenvector Y∈R^(b×d2), where b is the batch size, and d1 and d2 are the respective feature dimensions. Each feature element in the first eigenvector and the second eigenvector is normalized by the feature element normalization unit, that is, L2 normalization is performed using the following formula:

[0074]

[0075] Wherein, x is any characteristic element in the first characteristic vector, and y is any characteristic element in the second characteristic vector.

[0076] S2. Perform low-rank bilinear mapping on the first standard feature vector and the second standard feature vector based on a low-rank bilinear mapping unit to obtain a fused feature vector to be processed.

[0077] Specifically, based on the low-rank bilinear mapping unit, the first standard eigenvector and the second standard eigenvector are subjected to low-rank bilinear mapping processing to obtain the fused feature vector to be processed. The specific implementation method may include: performing element-by-element multiplication on the first standard eigenvector and the second standard eigenvector to obtain the fused feature vector to be transformed; performing regularization processing and linear transformation processing on the fused feature vector to be transformed to obtain the fused feature vector to be processed.

[0078] In this embodiment, the specific processing process in the low-rank bilinear mapping unit includes:

[0079] First, a low-rank bilinear mapping is implemented by matrix multiplication and element-wise multiplication:

[0080] X proj =XU

[0081] Y proj =YH

[0082] Z=X proj ⊙Y proj

[0083] Among them, U and H are coefficient matrices, ⊙ represents the element-wise multiplication operator.

[0084] Subsequently, Z is subjected to dropout and linear transformation by the following formula:

[0085] Z=Dropout(Z)

[0086] Z out =ZW+b

[0087] Where W is the coefficient matrix, Z out is the fusion feature vector to be processed.

[0088] S3. Input the fused feature vector to be processed into the self-attention unit to obtain the attention weight vector corresponding to the fused feature vector to be processed.

[0089] In this embodiment, the specific processing process in the self-attention unit includes:

[0090] After obtaining the fusion feature vector Z to be processed out Based on this, first, Z is calculated by the following formula out Use the self-attention mechanism to process:

[0091] Q=Z out W q

[0092] K=Z out W k

[0093] V=Z out W v

[0094] in, is the attention weight matrix.

[0095] Furthermore, the attention weight is calculated as:

[0096]

[0097] Thus, the attention weight vector corresponding to the fused feature vector to be processed can be expressed as:

[0098] Z attn =AV

[0099] S4. Based on the fusion feature output unit, the attention weight vector is nonlinearly activated and Euclidean normalized to obtain the target feature vector.

[0100] In this embodiment, the specific processing process of the fusion feature output unit includes:

[0101] The attention weight vector is nonlinearly activated and L2 normalized using the following formula:

[0102] F = ReLU(Z attn )

[0103]

[0104] Among them, Fo is the final target feature vector, and ∈ is a small constant used to prevent division by zero.

[0105] On the basis of the above examples, Figure 3 As shown, this feature fusion ( Figure 3 The bilinear fusion module shown in Figure 1 is embedded into the Swin-Transformer network to construct an end-to-end network structure, and then the weight matrices U, V, and W of the feature fuser are optimized by back propagation. This bilinear pooling method can effectively capture the high-order correlation between neutron and X-ray data, thereby generating richer and more discriminative fusion features. Its advantage is that it can learn fine-grained interactions between modalities, rather than just simple feature splicing.

[0106] It should be noted that the computational complexity of bilinear pooling is relatively high, and in practical applications, it may be necessary to consider computational efficiency and memory consumption. To address this, the above-mentioned low-rank approximation or random projection techniques can be used for optimization.

[0107] S1203 , input the target fusion feature vector into the decoder to obtain a denoised neutron CT image corresponding to the neutron CT image.

[0108] In this embodiment, the fused target fused feature vector is input into the decoder to obtain a denoised neutron CT image corresponding to the neutron CT image.

[0109] On the basis of the above examples, Figure 3 As shown in the figure, the fused target fusion feature vector is input into the Transformer decoder of Swin-Transformer, and the output denoised CT image obtained is the denoised neutron CT image corresponding to the neutron CT image. The decoder processes the fused features through stacked multi-head self-attention and feedforward neural networks to generate a reconstructed neutron image. In this process, the decoder module not only reconstructs the original neutron data, but also applies denoising processing, which effectively reduces the impact of noise and improves the quality of the neutron CT image. The decoder module generates high-quality neutron CT images by decoding the fused features layer by layer, and retains the complementary information of neutron and X-ray data.

[0110] The technical solution of the embodiment of the present invention obtains a neutron CT image and an X-ray CT image corresponding to the same lesion area, and then processes the neutron CT image and the X-ray CT image based on an image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image, wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features to obtain a denoised neutron CT image, which fully utilizes the complementary characteristics of neutron and X-ray CT images, enhances the expression ability of the fused image, effectively reduces random noise and system noise in the neutron CT image, improves the detail expression and overall clarity of the image, and improves the quality and reliability of the neutron CT image.

[0111] Embodiment 2

[0112] Figure 5 A flow chart of a method for training an image denoising model provided in an embodiment of the present invention. This embodiment is applicable to the case of training an image denoising model for removing noise in neutron CT images. The method can be executed by a training device for the image denoising model. The device can be implemented in the form of software and / or hardware. The hardware can be an electronic device such as a server. The electronic device can execute the training method for the image denoising model provided in the present technical solution to obtain an image denoising model.

[0113] like Figure 5 As shown, the method includes:

[0114] S210: Obtain multiple training samples.

[0115] Each training sample includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, as well as a denoised sample image corresponding to the neutron CT sample image. The denoised sample image is a high-quality image that has been denoised in advance using conventional denoising technology.

[0116] S220. For each training sample, the neutron CT sample image and the X-ray CT sample image in the current training sample are input into the encoder to be trained to obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image.

[0117] The image denoising model includes an encoder to be trained, a feature fuser to be trained, and a decoder to be trained.

[0118] In this embodiment, for any one of the multiple training samples, the training process is the same, and any one of them is taken as an example for description. Specifically, the neutron CT sample image and the X-ray CT sample image in the current training sample are input into the encoder to be trained in parallel, and the encoder to be trained can output a first sample feature vector corresponding to the neutron CT sample image, and a second sample feature vector corresponding to the X-ray CT sample image.

[0119] S230: Input the first sample feature vector and the second sample feature vector into the feature fuser to be trained to obtain a target sample feature vector.

[0120] In this embodiment, if the feature fuser to be trained is a feature splicing fuser, the first sample feature vector and the second sample feature vector are spliced ​​based on the feature splicing fuser to obtain a target sample feature vector.

[0121] Optionally, if the feature fuser to be trained is a bilinear pooling feature fuser, the bilinear pooling feature fuser may include: a feature element normalization unit, a low-rank bilinear mapping unit, a self-attention unit, and a fusion feature output unit, then the specific implementation method of determining the target sample feature vector may include:

[0122] Based on the feature element standardization unit, the first sample feature vector and the second sample feature vector are standardized respectively to obtain the first standard sample feature vector and the second standard sample feature vector; based on the low-rank bilinear mapping unit, the first standard sample feature vector and the second standard sample feature vector are low-rank bilinear mapped to obtain the fused sample feature vector to be processed; the fused sample feature vector to be processed is input into the self-attention unit to obtain the attention sample weight vector corresponding to the fused feature vector to be processed; based on the fused feature output unit, the attention sample weight vector is nonlinearly activated and Euclidean normalized to obtain the target sample feature vector.

[0123] S240: Input the target sample feature vector into the decoder to be trained to obtain the target sample output image.

[0124] In this embodiment, the target sample feature vector is input into the decoder to be trained, and the decoder to be trained outputs the target sample output image.

[0125] S250. Determine a target loss value based on each first pixel value in the target sample output image and each second pixel value at the corresponding pixel position in the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value.

[0126] In this embodiment, based on each first pixel value in the target sample output image and each second pixel value at the corresponding pixel position in the denoised sample image, a specific implementation method for determining the target loss value may include: determining a mean square error loss value and a structural similarity loss value based on each first pixel value and each corresponding second pixel value; determining a target loss value based on a preset weight value, a mean square error loss value and a structural similarity loss value.

[0127] In this embodiment, in order to optimize the model parameters, a loss function can be used to measure the difference between the neutron denoised image output during image denoising model training (i.e., the target sample output image) and the given high-quality neutron CT image (i.e., the denoised sample image).

[0128] In this embodiment, the target loss function is a composite loss function consisting of a mean square error (MSE) loss function and a SSIM (structural similarity index) loss function. The MSE loss function measures the square difference between the predicted value and the true value, and the SSIM is a structural similarity index ranging from [0,1]. The larger the value, the more similar the two images are.

[0129] Among them, the formula of the objective loss function is as follows:

[0130]

[0131] In the formula, y i is the i-th first pixel value in the target sample output image, is the second pixel value of the i-th pixel in the sample image after denoising; It is the structural similarity index between the target sample output image and the denoised sample image. α and β are preset weight values. In practical applications, the values ​​of α and β can be adjusted according to the actual effect to balance the influence of MSE and SSIM.

[0132] Each first pixel value and each corresponding second pixel value are brought into the target loss function to obtain a target loss value. Thus, the model parameters in the encoder, feature fuser and decoder in the image denoising model are modified according to the loss value.

[0133] S260: Taking the convergence of the loss function in the image denoising model as a training goal, and obtaining a trained image denoising model.

[0134] In this embodiment, the image denoising model is trained based on each training sample, and the model parameters in the encoder, feature fuser and decoder in the image denoising model are corrected according to the loss value until the target loss function converges to obtain a trained image denoising model.

[0135] The technical solution of this embodiment is that the image denoising model includes an encoder to be trained, a feature fusion device to be trained and a decoder to be trained. The training method includes: obtaining multiple training samples, wherein each training sample includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image; for each training sample, the neutron CT sample image and the X-ray CT sample image in the current training sample are input into the encoder to be trained to obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image; then, the first sample feature vector and the second sample feature vector are input into the feature fusion device to be trained to obtain a target sample feature vector; and then, the target sample feature vector is input into the decoder to be trained. A target sample output image is obtained. Further, a target loss value is determined based on each first pixel value in the target sample output image and each second pixel value at the corresponding pixel position in the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value. Finally, the convergence of the loss function in the image denoising model is used as a training target to obtain a trained image denoising model. The image denoising model is trained by continuously reducing the difference between the neutron denoised image output by the image denoising model and a given high-quality denoised neutron CT image. During this training process, the decoder not only reconstructs the original neutron data, but also applies denoising processing to effectively reduce the influence of noise. The image denoising model finally obtained can improve the detail expression and overall clarity of the image, and improve the quality and reliability of the neutron CT image.

[0136] Embodiment 3

[0137] Figure 6 A schematic diagram of the structure of a neutron CT image denoising device provided by an embodiment of the present invention is shown in FIG. Figure 6 As shown, the device includes: a dual-modality image acquisition module 310 and a denoised image output module 320.

[0138] Wherein, the dual-modality image acquisition module 310 is used to acquire a neutron CT image and an X-ray CT image corresponding to the same lesion area;

[0139] The denoising image output module 320 is used to process the neutron CT image and the X-ray CT image based on the image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image; wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on the first feature vector corresponding to the neutron CT image and the second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features.

[0140] Based on the above embodiment, the denoised image output module 320 includes:

[0141] a feature vector determination module, used for inputting the neutron CT image and the X-ray CT image into the encoder to obtain a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image;

[0142] A feature fusion module, used for inputting the first feature vector and the second feature vector into a feature fuser to obtain a target fused feature vector;

[0143] The denoised image output module is used to input the target fusion feature vector into the decoder to obtain a denoised neutron CT image corresponding to the neutron CT image.

[0144] On the basis of the above embodiment, the feature fusion unit includes: a feature element normalization unit, a low-rank bilinear mapping unit, a self-attention unit and a fusion feature output unit, and the feature fusion module includes:

[0145] A feature vector standardization unit, used to perform standardization processing on the first feature vector and the second feature vector based on the feature element standardization unit, to obtain a first standard feature vector and a second standard feature vector;

[0146] A feature bilinear mapping unit, used for performing low-rank bilinear mapping processing on the first standard feature vector and the second standard feature vector based on the low-rank bilinear mapping unit to obtain a fused feature vector to be processed;

[0147] An attention weight determination unit, used for inputting the fused feature vector to be processed into the self-attention unit to obtain an attention weight vector corresponding to the fused feature vector to be processed;

[0148] The target feature vector determination unit is used to perform nonlinear activation processing and Euclidean normalization processing on the attention weight vector based on the fusion feature output unit to obtain the target feature vector.

[0149] On the basis of the above embodiment, the feature bilinear mapping unit is specifically used to perform element-by-element multiplication on the first standard feature vector and the second standard feature vector to obtain the fused feature vector to be transformed; and perform regularization processing and linear transformation processing on the fused feature vector to be transformed to obtain the fused feature vector to be processed.

[0150] The technical solution of the embodiment of the present invention obtains a neutron CT image and an X-ray CT image corresponding to the same lesion area, and then processes the neutron CT image and the X-ray CT image based on an image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image, wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features to obtain a denoised neutron CT image, which fully utilizes the complementary characteristics of neutron and X-ray CT images, enhances the expression ability of the fused image, effectively reduces random noise and system noise in the neutron CT image, improves the detail expression and overall clarity of the image, and improves the quality and reliability of the neutron CT image.

[0151] The neutron CT image denoising device provided in the embodiment of the present invention can execute the neutron CT image denoising method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0152] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the embodiments of the present invention.

[0153] Embodiment 4

[0154] Figure 7 A structural diagram of a training device for an image denoising model provided in an embodiment of the present invention, wherein the image denoising model includes an encoder to be trained, a feature fuser to be trained, and a decoder to be trained, such as Figure 7 As shown, the device includes: a sample acquisition module 410, a sample feature extraction module 420, a sample feature fusion module 430, an image output module 440, a model correction module 450 and a denoising model determination module 460.

[0155] The sample acquisition module 410 is used to acquire a plurality of training samples; wherein each of the training samples includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image;

[0156] A sample feature extraction module 420 is used for inputting the neutron CT sample image and the X-ray CT sample image in the current training sample into the encoder to be trained for each of the training samples, so as to obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image;

[0157] A sample feature fusion module 430 is used to input the first sample feature vector and the second sample feature vector into the feature fuser to be trained to obtain a target sample feature vector;

[0158] An image output module 440 is used to input the target sample feature vector into the decoder to be trained to obtain a target sample output image;

[0159] A model correction module 450, configured to determine a target loss value based on the target sample output image and the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value;

[0160] The denoising model determination module 460 is used to take the convergence of the loss function in the image denoising model as a training target to obtain a trained image denoising model.

[0161] On the basis of the above embodiment, the model correction module 450 further includes: a loss value determination unit;

[0162] A loss value determination unit is used to determine a mean square error loss value and a structural similarity loss value based on each of the first pixel values ​​and the corresponding each of the second pixel values; and to determine a target loss value based on a preset weight value, the mean square error loss value and the structural similarity loss value.

[0163] The technical solution of this embodiment is that the image denoising model includes an encoder to be trained, a feature fusion device to be trained and a decoder to be trained. The training method includes: obtaining multiple training samples, wherein each training sample includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image; for each training sample, the neutron CT sample image and the X-ray CT sample image in the current training sample are input into the encoder to be trained to obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image; then, the first sample feature vector and the second sample feature vector are input into the feature fusion device to be trained to obtain a target sample feature vector; and then, the target sample feature vector is input into the decoder to be trained. A target sample output image is obtained. Further, a target loss value is determined based on each first pixel value in the target sample output image and each second pixel value at the corresponding pixel position in the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value. Finally, the convergence of the loss function in the image denoising model is used as a training target to obtain a trained image denoising model. The image denoising model is trained by continuously reducing the difference between the neutron denoised image output by the image denoising model and a given high-quality denoised neutron CT image. During this training process, the decoder not only reconstructs the original neutron data, but also applies denoising processing to effectively reduce the influence of noise. The image denoising model finally obtained can improve the detail expression and overall clarity of the image, and improve the quality and reliability of the neutron CT image.

[0164] Embodiment 5

[0165] Figure 8 The present invention is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 8 A block diagram of an exemplary electronic device 50 suitable for implementing exemplary embodiments of the present invention is shown. Figure 8 The electronic device 50 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0166] like Figure 8 As shown, the electronic device 50 is in the form of a general computing device. The components of the electronic device 50 may include but are not limited to: one or more processors or processing units 501, a system memory 502, and a bus 503 connecting different system components (including the system memory 502 and the processing unit 501).

[0167] Bus 503 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus structures. For example, these architectures include but are not limited to Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus and Peripheral Component Interconnect (PCI) bus.

[0168] The electronic device 50 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 50, including volatile and non-volatile media, removable and non-removable media.

[0169] System memory 502 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 504 and / or cache memory 505. Electronic device 50 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 506 may be used to read and write non-removable, non-volatile magnetic media ( Figure 8 not shown, usually called a "hard drive"). Although Figure 8 Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 503 via one or more data medium interfaces. The memory 502 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the embodiments of the present invention.

[0170] A program / utility 508 having a set (at least one) of program modules 507 may be stored, for example, in the memory 502, such program modules 507 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 507 generally perform the functions and / or methods of the embodiments described herein.

[0171] The electronic device 50 may also communicate with one or more external devices 509 (e.g., keyboards, pointing devices, displays 510, etc.), may communicate with one or more devices that enable a user to interact with the electronic device 50, and / or communicate with any device that enables the electronic device 50 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). Such communication may be performed through an input / output (I / O) interface 511. Furthermore, the electronic device 50 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 512. As shown, the network adapter 512 communicates with other modules of the electronic device 50 through the bus 503. It should be understood that although Figure 8 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 50, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0172] The processing unit 501 executes various functional applications and page processing by running the programs stored in the system memory 502, such as implementing the neutron CT image denoising method or the image denoising model training method provided in the embodiment of the present invention.

[0173] Embodiment 6

[0174] An embodiment of the present invention further provides a storage medium containing computer executable instructions, wherein the computer executable instructions are used to perform a neutron CT image denoising method when executed by a computer processor, the method comprising:

[0175] Obtaining neutron CT images and X-ray CT images corresponding to the same lesion area;

[0176] The neutron CT image and the X-ray CT image are processed based on an image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image; wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features to obtain the denoised neutron CT image.

[0177] Alternatively, the computer executable instructions are used to execute a training method for an image denoising model when executed by a computer processor, the image denoising model including an encoder to be trained, a feature fuser to be trained, and a decoder to be trained, the method comprising:

[0178] Acquire multiple training samples; wherein each of the training samples includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image;

[0179] For each of the training samples, the neutron CT sample image and the X-ray CT sample image in the current training sample are input into the encoder to be trained to obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image;

[0180] Inputting the first sample feature vector and the second sample feature vector into the feature fuser to be trained to obtain a target sample feature vector;

[0181] Inputting the target sample feature vector into the decoder to be trained to obtain a target sample output image;

[0182] Determine a target loss value based on each first pixel value in the target sample output image and each second pixel value at a corresponding pixel position in the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value;

[0183] The convergence of the loss function in the image denoising model is taken as a training goal to obtain a trained image denoising model.

[0184] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0185] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0186] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0187] Computer program code for performing the operation of embodiments of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0188] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A neutron CT image denoising method, It is characterized in that include: Obtaining neutron CT images and X-ray CT images corresponding to the same lesion area; The neutron CT image and the X-ray CT image are processed based on an image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image; wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features to obtain the denoised neutron CT image.

2. The method according to claim 1, It is characterized in that The step of processing the neutron CT image and the X-ray CT image based on the image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image includes: Inputting the neutron CT image and the X-ray CT image into the encoder to obtain a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image; Inputting the first feature vector and the second feature vector into the feature fuser to obtain a target fused feature vector; The target fusion feature vector is input into the decoder to obtain a denoised neutron CT image corresponding to the neutron CT image.

3. The method according to claim 2, It is characterized in that The feature fusion unit includes: a feature element normalization unit, a low-rank bilinear mapping unit, a self-attention unit and a fusion feature output unit. The first feature vector and the second feature vector are input into the feature fusion unit to obtain a target fusion feature vector, including: Based on the feature element standardization unit, the first feature vector and the second feature vector are respectively standardized to obtain a first standard feature vector and a second standard feature vector; Performing low-rank bilinear mapping processing on the first standard feature vector and the second standard feature vector based on the low-rank bilinear mapping unit to obtain a fused feature vector to be processed; Inputting the to-be-processed fused feature vector into a self-attention unit to obtain an attention weight vector corresponding to the to-be-processed fused feature vector; Based on the fusion feature output unit, nonlinear activation processing and Euclidean normalization processing are performed on the attention weight vector to obtain a target feature vector.

4. The method according to claim 3, It is characterized in that The performing low-rank bilinear mapping processing on the first standard feature vector and the second standard feature vector based on the low-rank bilinear mapping unit to obtain a fused feature vector to be processed includes: Multiplying the first standard feature vector and the second standard feature vector element by element to obtain a fused feature vector to be transformed; Regularization processing and linear transformation processing are performed on the to-be-transformed fused feature vector to obtain the to-be-processed fused feature vector.

5. A training method for an image denoising model, It is characterized in that The image denoising model includes an encoder to be trained, a feature fuser to be trained, and a decoder to be trained. The method includes: Acquire multiple training samples; wherein each of the training samples includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image; For each of the training samples, the neutron CT sample image and the X-ray CT sample image in the current training sample are input into the encoder to be trained to obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image; Inputting the first sample feature vector and the second sample feature vector into the feature fuser to be trained to obtain a target sample feature vector; Inputting the target sample feature vector into the decoder to be trained to obtain a target sample output image; Determine a target loss value based on each first pixel value in the target sample output image and each second pixel value at a corresponding pixel position in the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value; The convergence of the loss function in the image denoising model is taken as a training goal to obtain a trained image denoising model.

6. The method according to claim 5, It is characterized in that The determining of the target loss value based on each first pixel value in the target sample output image and each second pixel value at a corresponding pixel position in the denoised sample image comprises: Determining a mean square error loss value and a structural similarity loss value based on each of the first pixel values ​​and each of the corresponding second pixel values; A target loss value is determined based on a preset weight value, the mean square error loss value, and the structural similarity loss value.

7. A neutron CT image denoising device, It is characterized in that include: A dual-modality image acquisition module, used to acquire neutron CT images and X-ray CT images corresponding to the same lesion area; A denoising image output module is used to process the neutron CT image and the X-ray CT image based on an image denoising model to obtain a denoised neutron CT image corresponding to the neutron CT image; wherein the image denoising model includes an encoder, a feature fusion device for performing feature fusion on a first feature vector corresponding to the neutron CT image and a second feature vector corresponding to the X-ray CT image output by the encoder, and a decoder for decoding the fused features.

8. A training device for an image denoising model, It is characterized in that The image denoising model includes an encoder to be trained, a feature aggregator to be trained, and a decoder to be trained. The device includes: A sample acquisition module, used to acquire a plurality of training samples; wherein each of the training samples includes a neutron CT sample image and an X-ray CT sample image corresponding to the same lesion area, and a denoised sample image corresponding to the neutron CT sample image; a sample feature extraction module, configured to input the neutron CT sample image and the X-ray CT sample image in the current training sample into the encoder to be trained for each of the training samples, and obtain a first sample feature vector corresponding to the neutron CT sample image and a second sample feature vector corresponding to the X-ray CT sample image; A sample feature fusion module, used for inputting the first sample feature vector and the second sample feature vector into the feature fuser to be trained to obtain a target sample feature vector; An image output module, used for inputting the target sample feature vector into the decoder to be trained to obtain a target sample output image; A model correction module, used to determine a target loss value based on the target sample output image and the denoised sample image, so as to perform parameter correction on each processor in the image denoising model based on the target loss value; The denoising model determination module is used to take the convergence of the loss function in the image denoising model as a training target to obtain a trained image denoising model.

9. An electronic device, It is characterized in that Electronic equipment includes: one or more processors; a storage device for storing one or more programs, When one or more programs are executed by one or more processors, the one or more processors implement the neutron CT image denoising method described in any one of claims 1-4 or the image denoising model training method described in any one of claims 5-6.

10. A storage medium containing computer executable instructions, It is characterized in that When the computer executable instructions are executed by a computer processor, they are used to execute the neutron CT image denoising method as described in any one of claims 1 to 4 or the image denoising model training method as described in any one of claims 5 to 6.

Citation Information

Patent Citations

  • CT image blind denoising method based on multiple scales and attention mechanism

    CN114782408A

  • Multi-modal image registration method based on deep learning

    CN116523981A

Cited By

  • Cylindrical multilayer structure neutron / X-ray heterogenous image three-dimensional registration method

    CN120747181A