Pericardial adipose tissue segmentation method, device and equipment based on MR image and medium

Abstract

The invention discloses a pericardium adipose tissue segmentation method, device and equipment based on an MR image and a medium. The method comprises the following steps: acquiring a target heart magnetic resonance image containing pericardium adipose tissue of a target object; segmenting pericardium adipose tissue in the target heart magnetic resonance image based on a tissue segmentation model to obtain target pericardium adipose tissue of the target object; the tissue segmentation model comprises a three-branch cross-domain feature collaborative encoder, a double attention feature fusion module and a dynamic boundary perception decoder, and the three-branch cross-domain feature collaborative encoder performs feature extraction from global, local and frequency domains based on a three-branch structure; the double attention feature fusion module is established based on a space attention mechanism and a parallel channel attention mechanism, and the dynamic boundary perception decoder is used for enhancing the global and boundary perception capability. According to the scheme, the pericardium adipose tissue in the heart MR image can be accurately and efficiently segmented by using the tissue segmentation model based on multi-feature collaboration.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cardiac medical image segmentation, and in particular to a pericardial adipose tissue segmentation method and device based on MR images, equipment and media. BACKGROUND

[0002] Pericardial adipose tissue (PEAT) is an important pathological marker for the occurrence and development of cardiovascular diseases. Its accurate segmentation is of great significance for early diagnosis, risk assessment and individualized treatment plan. In the field of imaging technology, magnetic resonance (MR) imaging has become an important means for PEAT segmentation due to its high soft tissue contrast.

[0003] In recent years, deep learning technology has made significant progress in medical image segmentation. The fully supervised learning paradigm has shown good performance in various medical image segmentation tasks by training models using labeled data. However, when applied to pericardial adipose tissue segmentation, there are still obvious technical bottlenecks.

[0004] Existing deep learning-based fully supervised segmentation methods rely on convolutional neural networks to capture spatial features, making it difficult to effectively integrate global context information and local detail features, resulting in insufficient ability to locate the fuzzy boundary of PEAT. In addition, in the face of complex noise and motion artifacts of cardiac MR images, traditional network structures lack targeted optimization, and the model robustness is poor. SUMMARY

[0005] The present application provides a pericardial adipose tissue segmentation method, device, equipment and medium based on MR images, which can use a tissue segmentation model based on multi-feature collaboration to accurately and efficiently segment the pericardial adipose tissue in the cardiac MR image, improving the robustness and reliability of tissue segmentation.

[0006] According to an aspect of the present application, a pericardial adipose tissue segmentation method based on MR images is provided, the method comprising:

[0007] Obtaining a target cardiac magnetic resonance image of a target object, the target cardiac magnetic resonance image containing pericardial adipose tissue;

[0008] Segmenting the pericardial adipose tissue in the target cardiac magnetic resonance image based on a pre-trained tissue segmentation model to obtain the target pericardial adipose tissue of the target object;

[0009] The tissue segmentation model comprises a three-branch cross-domain feature collaborative encoder, a double attention feature fusion module and a dynamic boundary perception decoder.

[0010] According to another aspect of the present application, there is provided a pericardial fat tissue segmentation device based on MR images, the device comprising:

[0011] An image acquisition module is configured to acquire a target cardiac magnetic resonance image of a target object, the target cardiac magnetic resonance image containing pericardial fat tissue;

[0012] An image segmentation module is configured to segment the pericardial fat tissue in the target cardiac magnetic resonance image based on a pre-trained tissue segmentation model, to obtain target pericardial fat tissue of the target object.

[0013] The tissue segmentation model comprises a three-branch cross-domain feature collaborative encoder, a double attention feature fusion module and a dynamic boundary perception decoder.

[0014] According to another aspect of the present application, there is provided an electronic device, the electronic device comprising:

[0015] At least one processor; and,

[0016] A memory connected in communication with the at least one processor; wherein,

[0017] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the pericardial fat tissue segmentation method based on MR images according to any one of the embodiments of the present application.

[0018] According to another aspect of the present application, there is provided a computer readable storage medium storing computer instructions for causing a processor to perform the pericardial fat tissue segmentation method based on MR images according to any one of the embodiments of the present application.

[0019] The technical scheme of the embodiment of the present application firstly acquires a target cardiac magnetic resonance image of a target object, and the target cardiac magnetic resonance image contains pericardial adipose tissue; then the pericardial adipose tissue in the target cardiac magnetic resonance image is segmented based on a pre-trained tissue segmentation model to obtain the target pericardial adipose tissue of the target object; wherein the tissue segmentation model comprises a three-branch cross-domain feature collaborative encoder, a double-attention feature fusion module and a dynamic boundary perception decoder, the three-branch cross-domain feature collaborative encoder extracts features from global, local and frequency domains based on a three-branch structure, the double-attention feature fusion module is established based on a spatial attention mechanism and a parallel channel attention mechanism, and the dynamic boundary perception decoder is used to enhance the perception ability of the global and the boundary. The technical scheme can precisely and efficiently segment the pericardial adipose tissue in the cardiac MR image by using the tissue segmentation model based on multi-feature collaboration, and improve the robustness and reliability of the pericardial adipose tissue segmentation.

[0020] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0022] Figure 1 is a flow chart of a pericardial adipose tissue segmentation method based on MR image according to the embodiment one of the present application;

[0023] Figure 2 is a whole structure diagram of a tissue segmentation model according to the embodiment one of the present application;

[0024] Figure 3 is a structure diagram of a Mamba branch according to the embodiment one of the present application;

[0025] Figure 4 is a structure diagram of a double-attention feature fusion module according to the embodiment one of the present application;

[0026] Figure 5 is a structure diagram of a dynamic boundary perception decoder according to the embodiment one of the present application;

[0027] Figure 6 is a flow chart of a pericardial fat tissue segmentation method based on MR images according to Embodiment Two of the present application;

[0028] Figure 7 is a structural schematic diagram of a pericardial fat tissue segmentation device based on MR images according to Embodiment Three of the present application;

[0029] Figure 8 is a structural schematic diagram of an electronic device for implementing a pericardial fat tissue segmentation method based on MR images according to the present application. DETAILED DESCRIPTION

[0030] In order to make the person skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor should belong to the scope of protection of the present application.

[0031] It should be noted that the terms "first", "second", "target" and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0032] Embodiment One

[0033] Figure 1 A flow chart of a pericardial fat tissue segmentation method based on MR images is provided for Embodiment One of the present application. The present embodiment can be applicable to the case of accurately and efficiently segmenting pericardial fat tissue in a cardiac MR image. The method can be executed by a pericardial fat tissue segmentation device based on MR images. The pericardial fat tissue segmentation device based on MR images can be realized in the form of hardware and / or software, and can be configured in an electronic device with data processing capability. As shown in the figure, the method comprises: Figure 1

[0034] ​S110, obtain a target cardiac magnetic resonance image of a target object, the target cardiac magnetic resonance image containing pericardial adipose tissue.

[0035] The cardiac magnetic resonance image can refer to an image generated by imaging the heart using a magnetic resonance imaging technique. The target cardiac magnetic resonance image can refer to a cardiac magnetic resonance image that needs to be segmented for pericardial adipose tissue. The target object can refer to the object to which the target cardiac magnetic resonance image belongs, i.e., the target cardiac magnetic resonance image comes from which object.

[0036] In this embodiment, first, one or more cardiac magnetic resonance images of the target object are obtained as the target cardiac magnetic resonance image, and it is necessary to ensure that the target cardiac magnetic resonance image contains pericardial adipose tissue, so that the corresponding pericardial adipose tissue can be identified from the target cardiac magnetic resonance image subsequently.

[0037] S120, segmenting the pericardial adipose tissue in the target cardiac magnetic resonance image based on a pre-trained tissue segmentation model to obtain the target pericardial adipose tissue of the target object; wherein the tissue segmentation model comprises a three-branch cross-domain feature collaborative encoder, a double attention feature fusion module and a dynamic boundary perception decoder, the three-branch cross-domain feature collaborative encoder extracts features from global, local and frequency domains based on a three-branch structure, the double attention feature fusion module is established based on a spatial attention mechanism and a parallel channel attention mechanism, and the dynamic boundary perception decoder is used to enhance the perception ability of the global and boundary.

[0038] The tissue segmentation model can be a deep learning model that can be used to identify and segment pericardial adipose tissue from a cardiac magnetic resonance image. The target pericardial adipose tissue can refer to the pericardial adipose tissue region identified from the target cardiac magnetic resonance image using the tissue segmentation model.

[0039] The three-branch cross-domain feature collaborative encoder adopts a three-branch structure, and the three branches are respectively used to extract global features, local features and frequency domain features. Through the comprehensive multi-scale information, the diversity features and complex structure of the pericardial adipose tissue can be more effectively captured. Optionally, the three-branch structure includes a Mamba branch, a CNN branch and a frequency domain branch, the Mamba branch is used for global feature extraction, the CNN branch is used for local feature extraction, and the frequency domain branch is used for frequency feature extraction. Through the collaborative work of the three branches, the model can comprehensively consider the global and local information at different scales and spatial levels, effectively overcoming the shortcomings of existing methods in dealing with irregular boundaries, multi-scale features and similar gray value regions, so as to more accurately identify the boundary and morphology of the pericardial adipose tissue.

[0040] Wherein, CNN (Convolutional Neural Network) is a convolutional neural network; the English full name of Mamba is State Space Model with Selective Scan Mechanism, which is often referred to as Mamba or Mamba Architecture in the field of deep learning, and is a new type of sequence modeling architecture based on a state space model (State Space Model, SSM) and introducing a selective scanning mechanism.

[0041] In the embodiment, optionally, the Mamba branch includes a Mamba module, a scanning module and a residual connection structure, the Mamba module is used for processing global spatial dependence, the scanning module is used for local feature extraction, and the residual connection structure is used for fusing the outputs of the Mamba module and the scanning module.

[0042] Figure 2 A whole structure diagram of a tissue segmentation model provided for the first embodiment of the application is shown in FIG. 1. Figure 2 As shown in FIG. 1, the Mamba branch adopts a dynamic scanning Mamba mode, and integrates a lightweight convolution scanning operation into the Mamba module, thereby enhancing the modeling capability of spatial features while maintaining efficient calculation. Specifically, the Mamba branch adopts a double-branch structure, and the core components include a Mamba module for processing global spatial dependence, a scanning module for lightweight local feature extraction, and a residual connection structure for fusing the outputs of the two branch modules.

[0043] Figure 3 A structure diagram of a Mamba branch provided for the first embodiment of the application is shown in FIG. 2. Figure 3 As shown in FIG. 2, the input image enters the Mamba module and the scanning module respectively. The Mamba module transposes the image tensor to adapt to the processing paradigm of the model, and after normalization processing, the image tensor is sent to the Mamba model for feature extraction, and finally the output is restored to the original image dimension through inverse transposition. The scanning module is processed through a depth separable convolution, extracts more detailed local features, and applies a GELU activation function to enhance the non-linear expression capability. Finally, the features extracted by the two modules of the Mamba branch are fused to fully utilize the advantages of both to improve the overall performance and precision of the model.

[0044] In the embodiment, optionally, the CNN branch includes two convolution layers, a deformable convolution layer and a maximum pooling layer, and the frequency domain branch includes a frequency domain filter.

[0045] As shown in FIG. 3, Figure 2As shown, the deformable CNN branch is a sequential single-branch structure. It utilizes deformable convolutions to achieve adaptive receptive field adjustment, enhancing the feature extraction capability for irregular fat regions. Specifically, it first undergoes a 3×3 convolution operation through a regular convolutional layer, maintaining the same number of channels for both input and output, with padding set to 1 to ensure spatial alignment between the output and input. Next, the input passes through a deformable convolutional layer, as shown in the following formula: ,in, It is the output feature map at the location The value at that location, These are the convolution kernel weights. It is the input feature map at the location The value at that location, It is a learnable offset. This refers to the bias term. Deformable convolution is an extension of the standard convolution operation, allowing the kernel position to be dynamically adjusted based on the input feature map. Specifically, it is implemented by first calculating the offset through an additional offset convolution. The number of output channels for the offset is set to twice the square of the convolution kernel size. This way, the positional information of each convolution kernel is passed through the offset, guiding the weighted computation of the convolution operation in the image. This dynamic adjustment of the offset allows the convolution kernels to adaptively adjust their position, thus better capturing deformations and irregular structures in the image. The output feature map is then generated through standard convolutional layers and non-linearly transformed using the GELU (Gaussian Error Linear Unit) activation function. Finally, max pooling is used to halve the spatial size to reduce computation, and the number of channels is doubled through convolutional layers to continue extracting high-level features from the image, thereby enhancing the model's expressive power.

[0046] like Figure 2 As shown, the frequency domain branch includes convolutional layers and learnable frequency domain filters. For example, the frequency domain filter can be a Fourier filter. The frequency domain branch introduces a learnable Fourier filter, which helps the model better focus on important features of pericardial adipose tissue by enhancing specific frequency components in the image. Furthermore, the Fourier filter can effectively reduce noise and artifact interference without compromising important image structures, thereby improving the segmentation quality of pericardial adipose tissue.

[0047] Specifically, taking a Fourier filter as an example, the processing flow of a learnable frequency domain filter is as follows: First, perform a Fast Fourier Transform on the input image to convert the two-dimensional image from the time domain to the frequency domain. Then, resize the learnable filter to match the frequency components of the input image. Multiply the resized filter by the corresponding frequency components of the input image in the frequency domain to perform frequency domain filtering. Perform an Inverse Fourier Transform on the filtered frequency domain data to restore it to the time domain, obtaining the processed image.

[0048] where the detailed process of Fourier transform and filtering operation is as follows: assuming the input image is a multi-channel two-dimensional image, where , and represent the channel number, height and width of the input image respectively. After two-dimensional Fourier transform, the frequency domain representation is obtained, which can be seen from the formula . Wherein is the frequency domain index, is the pixel value of the input image in the spatial domain. For the learnable Fourier filter , it aims to enhance the features of the input image through frequency domain filtering. The filter will be adjusted by interpolation to adapt to the frequency domain size of the input image. The filter after interpolation adjustment is multiplied element by element with the input frequency domain representation to obtain the frequency domain filtering result, as shown in the following formula: . Wherein is the frequency domain filtering result, represents element-wise multiplication, is the filter after size adjustment, which has the same size as the input image frequency domain representation. Then it is converted back to the time domain through inverse Fourier transform, as shown in the formula . Wherein represents the pixel of the filtered time domain image. Finally, it has a similar structure as the input image in the time domain, but the specific frequency components are further strengthened through frequency domain operation, and the model's perception ability of details is improved.

[0049] The three-branch cross-domain feature collaborative encoder proposed in the application breaks through the limitation of existing single-domain modeling, and innovatively uses Mamba, CNN and frequency domain three-branch structure to extract features in parallel. Unlike existing methods that rely only on a single feature extraction method, this encoder can more comprehensively capture information of pericardial fat tissue at different scales and spatial levels through collaborative extraction of global, local and frequency domain features, effectively solving the problem that existing methods are difficult to handle irregular boundaries, multi-scale features and similar gray value regions, and significantly improving the accuracy and comprehensiveness of feature extraction.

[0050] In this embodiment, the dual attention feature fusion module independently evaluates the importance of each branch feature in the three-branch structure through parallel channel attention mechanism, dynamically adjusts the contribution of each branch to ensure that the features of key channels are paid more attention. At the same time, the spatial attention mechanism is used to generate a pixel-level weight map to enhance the model's ability to capture key areas and optimize feature fusion. In addition, the residual connection can effectively alleviate the problem of gradient disappearance, ensure stable gradient propagation, and thus improve the convergence speed and overall performance of the network. The dual attention feature fusion module is essentially different from the existing feature fusion method, which can effectively overcome the defects of the existing fusion method, such as insufficient attention to key features and easy gradient disappearance, achieve efficient feature fusion, and greatly improve the network convergence speed and overall performance.

[0051] Specifically, as shown in Figure 4 , each branch in the three-branch structure calculates the channel attention weight and the spatial attention weight, respectively, as shown in the following formula:

[0052] ;

[0053] .

[0054] wherein, and represent the channel attention weight and the spatial attention weight, is the input feature, is the Sigmoid activation function, represents the average pooling, represents the concatenation represents the convolution operation using a convolution kernel of size . Specifically, the channel attention mechanism can be used to independently process each input feature to calculate the importance weight of each channel. The spatial attention mechanism can be used to process the concatenated multi-source features to calculate the importance weight of each spatial position. Applying the channel attention weight and the spatial attention weight to the input feature, the weighted feature can be obtained, as shown in the following formula: .

[0055] In order to preserve the information of the original feature, the dual attention feature fusion module fuses the weighted feature with the concatenated original feature through the residual connection, as shown in the following formula:

[0056] .

[0057] wherein, represents the output feature of the dual attention feature fusion module.

[0058] In this embodiment, on the basis of the existing decoder, a dynamic boundary perception decoder is proposed to enhance the global perception ability and boundary perception ability of the decoder. Optionally, the dynamic boundary perception decoder includes a plurality of identical decoding modules, the decoding module includes a dynamic scanning Mamba module and a boundary refinement module, the dynamic scanning Mamba module includes a Mamba module and a scanning module, and the boundary refinement module includes an inflation layer, an erosion layer and an attention layer.

[0059] Figure 5 A structural diagram of a dynamic boundary perception decoder provided for embodiment one of the present application is shown in FIG. 1. Figure 5 As shown in FIG. 1, the dynamic boundary perception decoder introduces a dynamic scanning Mamba module on the basis of a traditional up-sampling convolution to enhance the global perception ability of the decoder; and through a boundary refinement module combining inflation, erosion and attention mechanisms, the boundary perception ability of the decoder is effectively improved. This design changes the current situation of insufficient boundary processing ability of the traditional decoder, makes the decoder more accurate when processing the boundary of pericardial fat tissue, significantly improves the boundary definition and integrity of the segmentation result, and effectively improves the overall segmentation accuracy and reliability.

[0060] Specifically, as shown in FIG. 2, the dynamic boundary perception decoder is composed of four identical decoding modules stacked together, and the input features are processed and the image segmentation result is gradually refined each time. Figure 5 In each decoding module, the input features are first up-sampled by a transpose convolution layer to restore the low-resolution feature map to a higher spatial resolution to enhance the details of the image. The up-sampled feature map then passes through the same Mamba module and scanning module as the encoder to achieve fine use of the encoder features when processing irregular areas and complex boundaries, and improve the boundary optimization ability of the model.

[0061] Meanwhile, the dynamic boundary-aware decoder further enhances the boundary precision of image features through a boundary refinement module. The boundary refinement module is composed of an expansion layer, an erosion layer, and an attention layer, which jointly act on the boundary part of the image. The expansion layer helps capture the outline and external structure of the object by expanding the object boundary in the feature map, enhancing the visibility of the boundary. The erosion layer further refines the boundary by shrinking the boundary in the feature map and removing noise. When the outputs of the expansion layer and the erosion layer are subtracted, the result highlights the boundary differences of the object, emphasizing the boundary part between the object and the background, while weakening the small noise and irrelevant information in the background. This operation helps to eliminate irrelevant small details, making the boundary clearer and more accurate. The attention layer generates an attention map to accurately control the boundary refinement process, enabling the network to selectively focus on important boundary parts, refine these areas, and suppress irrelevant parts, thereby improving the quality of the segmentation result. The attention layer adopts an attention mechanism and is constructed based on convolution, ReLU (Rectified Linear Unit) function, and Sigmoid function (i.e., S-shaped curve function). In each decoding module, residual connection is used to preserve the original features, avoiding feature loss and preventing gradient disappearance, ensuring effective information transmission in multiple levels.

[0062] In this embodiment, the tissue segmentation model needs to be pre-trained to segment the pericardial fat tissue in the target cardiac magnetic resonance image based on the trained tissue segmentation model. Optionally, the training process of the tissue segmentation model includes: obtaining multiple candidate cardiac magnetic resonance images containing pericardial fat tissue, determining training data according to the candidate cardiac magnetic resonance images; building an initial segmentation model using a semi-supervised training framework based on cross-teaching; and performing full-supervised training on the initial segmentation model to obtain the tissue segmentation model according to the hybrid loss function and the training data.

[0063] Specifically, when training the tissue segmentation model, first, multiple candidate cardiac magnetic resonance images containing pericardial fat tissue are obtained, and the pericardial fat tissue in each candidate cardiac magnetic resonance image is labeled accordingly to construct training data. The candidate cardiac magnetic resonance image can refer to the cardiac magnetic resonance image participating in the model training. In addition, existing public data sets can also be directly used as training data. For example, the MRPEAT data set can be used as training data, which contains three groups of short-axis cardiac magnetic resonance images of patients. Each patient has 8-10 consecutive dynamic image views at end-diastole and end-systole, and each image is provided with corresponding labels.

[0064] Then a semi-supervised training framework based on cross-teaching is built as an initial segmentation model, and the pre-labeled pericardial fat tissue label is used as a supervised target. The initial segmentation model is trained and optimized based on the training data using a full-supervised learning method using a hybrid loss function. Finally, a tissue segmentation model for segmenting pericardial fat tissue in a cardiac magnetic resonance image is obtained. The hybrid loss function can be a loss function composed of at least two losses. Optionally, the hybrid loss function includes a Dice loss, an edge loss, and an uncertainty loss. The Dice loss, the edge loss, and the uncertainty loss correspond to different optimization objectives, respectively, and refer to the following formulas:

[0065] ;

[0066] ;

[0067] .

[0068] wherein, 、 and represent the Dice loss, the edge loss, and the uncertainty loss, respectively, is a probability map output by the model, is a binary map of a real label, is a smoothing factor, is a Sigmoid activation function, is the total number of pixels participating in the calculation.

[0069] It should be noted that the Dice loss solves the class imbalance problem, is suitable for scenarios with large differences between foreground and background, emphasizes the overlap of predicted and real regions, and improves the segmentation coverage. The edge loss amplifies the edge difference through a high-slope Sigmoid function, aligns the predicted and real edges using a mean square error, and improves the boundary accuracy. The uncertainty loss minimizes the cross-entropy of the predicted probability and the real label, provides a global optimization signal, and enhances the pixel classification confidence.

[0070] The hybrid loss function is shown in the following formula: . Wherein, is a hybrid loss function; 、 and represent the weight coefficients of the Dice loss, the edge loss, and the uncertainty loss, respectively, and are used to control the contribution of each part of the loss. The hybrid loss function combines loss functions with different properties, which can optimize the region overlap, edge alignment, and pixel classification at the same time, thereby taking advantage of each loss function.

[0071] The hybrid loss function proposed in the scheme fuses the Dice loss, the edge loss and the uncertainty loss, sets optimization targets respectively for the volume overlap of the segmentation region, the boundary positioning accuracy and the prediction confidence, and effectively improves the overall segmentation performance of the model through the multi-task joint learning mechanism.

[0072] The technical scheme of the embodiment of the application first acquires a target cardiac magnetic resonance image of a target object, the target cardiac magnetic resonance image containing pericardial adipose tissue; then the pericardial adipose tissue in the target cardiac magnetic resonance image is segmented based on a pre-trained tissue segmentation model to obtain the target pericardial adipose tissue of the target object; wherein the tissue segmentation model comprises a three-branch cross-domain feature collaborative encoder, a double-attention feature fusion module and a dynamic boundary perception decoder, the three-branch cross-domain feature collaborative encoder performing feature extraction from the global, local and frequency domains based on a three-branch structure, the double-attention feature fusion module being established based on a spatial attention mechanism and a parallel channel attention mechanism, and the dynamic boundary perception decoder being used to enhance the perception ability of the global and the boundary. The technical scheme can perform parallel feature extraction through the three-branch cross-domain feature collaborative encoder, realize efficient fusion of the features through the double-attention feature fusion module, improve the boundary definition and integrity of the segmentation result through the dynamic boundary perception decoder, and can use the full-supervision tissue segmentation model based on multi-feature collaboration to accurately and efficiently segment the pericardial adipose tissue in the cardiac MR image, thereby improving the reliability and robustness of the pericardial adipose tissue segmentation.

[0073] In the embodiment, after the plurality of candidate cardiac magnetic resonance images containing pericardial adipose tissue are acquired, the candidate cardiac magnetic resonance images are optionally preprocessed, and the preprocessing includes image size adjustment; accordingly, the training data are determined according to the candidate cardiac magnetic resonance images, including determining the training data according to the preprocessed candidate cardiac magnetic resonance images.

[0074] Specifically, to meet the input requirements of model training, all original image data (i.e., candidate cardiac magnetic resonance images) participating in the training are subjected to standardization processing. Through image scaling and size adjustment technology, the original images can be uniformly adjusted to a standard size of 256x256, which not only effectively eliminates the differences in resolution of the data, but also lays a foundation for efficient training and stable performance of the model.

[0075] Embodiment two

[0076] Figure 6A flowchart of a pericardial fat tissue segmentation method based on an MR image is provided for the second embodiment of the present application, which is optimized on the basis of the above-mentioned embodiment. The specific optimization is that after obtaining the target cardiac magnetic resonance image of the target object, the method further includes: pre-processing the target cardiac magnetic resonance image, the pre-processing including image size adjustment; and correspondingly, segmenting the pericardial fat tissue in the target cardiac magnetic resonance image based on the pre-trained tissue segmentation model, including: segmenting the pericardial fat tissue in the pre-processed target cardiac magnetic resonance image based on the pre-trained tissue segmentation model.

[0077] As shown in Figure 6 , the method of the present embodiment specifically includes the following steps:

[0078] S210, obtaining a target cardiac magnetic resonance image of a target object, the target cardiac magnetic resonance image containing pericardial fat tissue.

[0079] S220, pre-processing the target cardiac magnetic resonance image, the pre-processing including image size adjustment.

[0080] Similarly, referring to the pre-processing operation in the model training process described above, in order to meet the input requirements of the model, the target cardiac magnetic resonance image obtained also needs to be standardized. Through image scaling and size adjustment technology, the target cardiac magnetic resonance image can be adjusted to a standard size of 256x256, which helps to further improve the prediction accuracy of the tissue segmentation model. In addition, if other pre-processing operations are also used during model training, such as image resolution adjustment and image intensity normalization, corresponding pre-processing operations need to be performed when the model is used.

[0081] S230, segmenting the pericardial fat tissue in the pre-processed target cardiac magnetic resonance image based on the pre-trained tissue segmentation model, to obtain the target pericardial fat tissue of the target object.

[0082] The tissue segmentation model includes a three-branch cross-domain feature collaborative encoder, a double attention feature fusion module, and a dynamic boundary perception decoder. The three-branch cross-domain feature collaborative encoder extracts features from global, local, and frequency domains based on a three-branch structure. The double attention feature fusion module is established based on a spatial attention mechanism and a parallel channel attention mechanism. The dynamic boundary perception decoder is used to enhance the perception ability of the global and boundary.

[0083] The technical scheme of the embodiment of the present application, after obtaining the target cardiac magnetic resonance image of the target object, further pre-processes the target cardiac magnetic resonance image, wherein the pre-processing includes image size adjustment, and the pericardial fat tissue in the pre-processed target cardiac magnetic resonance image is segmented based on a pre-trained tissue segmentation model. The technical scheme can meet the input requirements of the model by increasing the image pre-processing operation, and helps to further improve the prediction accuracy of the tissue segmentation model.

[0084] Embodiment three

[0085] Figure 7 A structure schematic diagram of a pericardial fat tissue segmentation device based on an MR image is provided for the third embodiment of the present application. The device can perform the pericardial fat tissue segmentation method based on the MR image provided by any embodiment of the present application, and has the corresponding function modules and beneficial effects of the execution method. As shown in the figure, Figure 7 The device comprises:

[0086] The image acquisition module 310 is configured to acquire a target cardiac magnetic resonance image of a target object, and the target cardiac magnetic resonance image contains pericardial fat tissue.

[0087] The image segmentation module 320 is configured to segment the pericardial fat tissue in the target cardiac magnetic resonance image based on a pre-trained tissue segmentation model, and obtain the target pericardial fat tissue of the target object.

[0088] The tissue segmentation model comprises a three-branch cross-domain feature collaborative encoder, a double attention feature fusion module and a dynamic boundary perception decoder. The three-branch cross-domain feature collaborative encoder extracts features from the global, local and frequency domains based on a three-branch structure. The double attention feature fusion module is established based on a spatial attention mechanism and a parallel channel attention mechanism. The dynamic boundary perception decoder is used to enhance the perception ability of the global and boundary.

[0089] Optionally, the three-branch structure comprises a Mamba branch, a CNN branch and a frequency domain branch. The Mamba branch is used for global feature extraction. The CNN branch is used for local feature extraction. The frequency domain branch is used for frequency feature extraction.

[0090] Optionally, the Mamba branch comprises a Mamba module, a scanning module and a residual connection structure. The Mamba module is used to process global spatial dependence. The scanning module is used to extract local features. The residual connection structure is used to fuse the outputs of the Mamba module and the scanning module.

[0091] Optionally, the CNN branch comprises two convolution layers, one deformable convolution layer and one max-pooling layer. The frequency domain branch comprises a frequency domain filter.

[0092] Optionally, the dynamic boundary-aware decoder comprises a plurality of identical decoding modules, the decoding module comprising a dynamic scanning Mamba module and a boundary refinement module, the dynamic scanning Mamba module comprising a Mamba module and a scanning module, and the boundary refinement module comprising an expansion layer, an erosion layer and an attention layer.

[0093] Optionally, the device further comprises a model training module configured to:

[0094] Obtain a plurality of candidate cardiac magnetic resonance images containing pericardial adipose tissue, and determine training data according to the candidate cardiac magnetic resonance images.

[0095] An initial segmentation model is built using a semi-supervised training framework based on cross-teaching.

[0096] The initial segmentation model is fully supervised trained according to a hybrid loss function and the training data to obtain a tissue segmentation model.

[0097] Optionally, the hybrid loss function comprises a Dice loss, an edge loss and an uncertainty loss.

[0098] Optionally, the model training module is further configured to:

[0099] After obtaining a plurality of candidate cardiac magnetic resonance images containing pericardial adipose tissue, the candidate cardiac magnetic resonance images are preprocessed, and the preprocessing comprises image size adjustment.

[0100] Correspondingly, the model training module is configured to:

[0101] The training data is determined according to the preprocessed candidate cardiac magnetic resonance images.

[0102] The pericardial adipose tissue segmentation device based on MR images provided by the embodiments of the present application can execute the pericardial adipose tissue segmentation method based on MR images provided by any embodiment of the present application, and has the corresponding functional modules and beneficial effects of the execution method.

[0103] Embodiment Four

[0104] Figure 8A structural diagram of an electronic device 10 that can be used to implement embodiments of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular telephones, smartphones, wearable devices (e.g., headsets, glasses, watches, etc.), and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present application described and / or claimed in this document.

[0105] As shown, the electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., communicatively connected to the at least one processor 11, where the memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer programs stored in the read-only memory (ROM) 12 or loaded into the random access memory (RAM) 13 from the storage unit 18. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14. Figure 8

[0106] Various components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc., an output unit 17, such as various types of displays, speakers, etc., a storage unit 18, such as a magnetic disk, an optical disk, etc., and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0107] The processor 11 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as the pericardial adipose tissue segmentation method based on MR images.

[0108] ​In some embodiments, the MR image-based pericardial adipose tissue segmentation method can be implemented as a computer program tangibly embodied in a computer readable storage medium, e.g., storage unit 18. In some embodiments, parts or all of the computer program can be loaded and / or installed onto electronic device 10 via, e.g., ROM 12 and / or communication unit 19. When the computer program is loaded onto RAM 13 and executed by processor 11, one or more steps of the MR image-based pericardial adipose tissue segmentation method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the MR image-based pericardial adipose tissue segmentation method by way of other any suitable means, e.g., by way of firmware.

[0109] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (PLD), a computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0110] Computer programs used to implement the processes of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program, when executed, implements the functions / acts specified in the flowcharts and / or block diagrams. The computer program can be executed entirely on a machine, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0111] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0112] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0113] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0114] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0115] It should be understood that the various forms of flow shown above can be used to reorder, add or delete steps. For example, each step described in the present application can be executed in parallel, sequentially or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, which is not limited herein.

[0116] The above detailed description does not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for segmenting pericardial adipose tissue based on MR images, characterized in that, The method includes: Acquire a target cardiac magnetic resonance image of the target object, wherein the target cardiac magnetic resonance image contains pericardial adipose tissue; The pericardial adipose tissue in the target cardiac magnetic resonance image is segmented based on a pre-trained tissue segmentation model to obtain the target pericardial adipose tissue of the target object. The tissue segmentation model includes a three-branch cross-domain feature co-encoder, a dual-attention feature fusion module, and a dynamic boundary-aware decoder. The three-branch cross-domain feature co-encoder extracts features from the global, local, and frequency domains based on a three-branch structure. The dual-attention feature fusion module is established based on spatial attention and parallel channel attention mechanisms. The dynamic boundary-aware decoder is used to enhance the perception of the global and boundary information.

2. The method according to claim 1, characterized in that, The three-branch structure includes a Mamba branch, a CNN branch, and a frequency domain branch. The Mamba branch is used for global feature extraction, the CNN branch is used for local feature extraction, and the frequency domain branch is used for frequency feature extraction.

3. The method according to claim 2, characterized in that, The Mamba branch includes a Mamba module, a scanning module, and a residual connection structure. The Mamba module is used to handle global spatial dependencies, the scanning module is used to extract local features, and the residual connection structure is used to fuse the outputs of the Mamba module and the scanning module.

4. The method according to claim 2, characterized in that, The CNN branch includes two convolutional layers, one deformable convolutional layer, and one max pooling layer, and the frequency domain branch includes a frequency domain filter.

5. The method according to claim 1, characterized in that, The dynamic boundary-aware decoder includes multiple identical decoding modules. Each decoding module includes a dynamic scanning Mamba module and a boundary refinement module. The dynamic scanning Mamba module includes a Mamba module and a scanning module. The boundary refinement module includes an expansion layer, an erosion layer, and an attention layer.

6. The method according to any one of claims 1-5, characterized in that, The training process of the tissue segmentation model includes: Acquire multiple candidate cardiac magnetic resonance images containing pericardial adipose tissue, and determine training data based on the candidate cardiac magnetic resonance images; An initial segmentation model was built using a semi-supervised training framework based on cross-teaching. The tissue segmentation model is obtained by performing fully supervised training on the initial segmentation model based on the hybrid loss function and the training data.

7. The method according to claim 6, characterized in that, The hybrid loss function includes Dice loss, edge loss, and uncertainty loss.

8. The method according to claim 6, characterized in that, After acquiring multiple candidate cardiac magnetic resonance images containing pericardial adipose tissue, the method further includes: The candidate cardiac magnetic resonance images are preprocessed, including image resizing. Accordingly, training data is determined based on the candidate cardiac magnetic resonance images, including: Training data is determined based on preprocessed candidate cardiac magnetic resonance images.

9. A pericardial adipose tissue segmentation device based on MR images, characterized in that, The device includes: The image acquisition module is used to acquire a target cardiac magnetic resonance image of the target object, wherein the target cardiac magnetic resonance image contains pericardial adipose tissue; The image segmentation module is used to segment the pericardial adipose tissue in the target cardiac magnetic resonance image based on a pre-trained tissue segmentation model, so as to obtain the target pericardial adipose tissue of the target object. The tissue segmentation model includes a three-branch cross-domain feature co-encoder, a dual-attention feature fusion module, and a dynamic boundary-aware decoder. The three-branch cross-domain feature co-encoder extracts features from the global, local, and frequency domains based on a three-branch structure. The dual-attention feature fusion module is established based on spatial attention and parallel channel attention mechanisms. The dynamic boundary-aware decoder is used to enhance the perception of the global and boundary information.

10. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, which enables the at least one processor to perform the pericardial fat tissue segmentation method based on MR images according to any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the pericardial adipose tissue segmentation method based on MR images as described in any one of claims 1-8.

Images