WO2024119337A1 - 脑网络统一表征计算方法、装置、电子设备及存储介质 - Google Patents

脑网络统一表征计算方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2024119337A1
WO2024119337A1 PCT/CN2022/136710 CN2022136710W WO2024119337A1 WO 2024119337 A1 WO2024119337 A1 WO 2024119337A1 CN 2022136710 W CN2022136710 W CN 2022136710W WO 2024119337 A1 WO2024119337 A1 WO 2024119337A1
Authority
WO
WIPO (PCT)
Prior art keywords
brain
image
network
brain network
reference image
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2022/136710
Other languages
English (en)
French (fr)
Inventor
王书强
宗永承
荆常宏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to PCT/CN2022/136710 priority Critical patent/WO2024119337A1/zh
Publication of WO2024119337A1 publication Critical patent/WO2024119337A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis

Definitions

  • the present invention relates to the field of medical image processing technology, and in particular to a brain network unified representation calculation method, device, electronic equipment and storage medium.
  • Brain imaging data of different modalities are usually used to construct three types of brain networks: morphological networks, structural networks, and functional networks.
  • morphological networks In 2007, the cortical thickness of each brain region was used as a morphological feature to construct the brain morphological network for the first time, and the small-world property of the brain network was discovered. Since then, a large number of complex network analysis methods based on graph theory have begun to study the connection patterns and graph property analysis between brain regions, and have been used to explore the diagnosis of neurodegenerative diseases from structure to function.
  • Graph neural networks are one of the current research hotspots in deep learning technology and have become the most widely used method for extracting information from graphs to perform neurodegenerative disease diagnosis.
  • the embodiments of the present invention provide a brain network unified representation calculation method, device, electronic device and storage medium to solve the problems of strong subjectivity, low efficiency, low accuracy and low degree of automation existing in the related art.
  • a method for calculating a unified representation of a brain network includes: obtaining a brain image of a sample, and based on a standard brain space provided by a reference image, guiding the brain image to align with the standard brain space to obtain a brain region structural deformation field of the brain image; learning the local deformation of the brain image through deformable convolution, adjusting the local deformation difference of the brain region structural deformation field to obtain an optimized brain region structural deformation field; applying the optimized brain region structural deformation field to the brain image to obtain a moving image for indicating the distribution of brain regions, and obtaining regional information of multiple brain regions related to the disease by constraining the similarity between the moving image and the reference image; constructing an initial brain network based on the regional information of each brain region, and optimizing the initial brain network by performing graph comparison learning between the subgraphs of the brain network and their diffusion graphs to obtain a reconstructed brain network, wherein the reconstructed brain network is used to indicate a unified representation of the brain network of the
  • a brain network unified representation calculation device includes: a global positioning module, which is used to obtain a brain image of a sample, and based on the standard brain space provided by a reference image, guide the brain image to align with the standard brain space to obtain a brain region structural deformation field of the brain image; a local perception module, which is used to learn the local deformation of the brain image through deformable convolution, adjust the local deformation difference of the brain region structural deformation field, and obtain the optimized brain region structural deformation field; a weighted coding module, which is used to apply the optimized brain region structural deformation field to the brain image to obtain a moving image for indicating the distribution of brain regions, and obtain regional information of multiple brain regions related to the disease by constraining the similarity between the moving image and the reference image; a brain network reconstruction module, which is used to perform an initial brain network construction based on the regional information of each brain region, and optimize the initial brain network by performing graph comparison learning between the subgraphs of the brain network and their diffusion graphs to
  • an electronic device includes a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the unified brain network representation calculation method as described above is implemented.
  • a storage medium stores a computer program thereon, and when the computer program is executed by a processor, the above-mentioned unified brain network representation calculation method is implemented.
  • a computer program product includes a computer program, the computer program is stored in a storage medium, a processor of a computer device reads the computer program from the storage medium, and the processor executes the computer program, so that when the computer device executes the computer program, the unified brain network representation calculation method as described above is implemented.
  • the present invention realizes the unified representation calculation of brain networks with strong subjectivity, high efficiency, high degree of automation and high accuracy.
  • the present invention obtains regional information of multiple brain regions related to the disease through a global positioning module guided by the spatial information of brain structure provided by a reference image and a weighted coding module based on brain abnormality representation learning, and constructs a relatively accurate initial brain network based on the regional information of each brain region.
  • the initial brain network is optimized by using the idea of graph contrast learning through a brain network reconstruction module based on the embedding of brain disease topological features, and then based on the brain network representation indicated by the reconstructed brain network, the disease category of the sample is predicted to obtain the disease prediction result of the sample, further accurately locates the disease-related brain regions and abnormal connection features, explores the potential pathogenesis of the disease, improves the accuracy of the unified representation calculation of the brain network, and thus improves the accuracy of disease prediction.
  • each module in the disease prediction model used for disease prediction uses deep learning technology to have corresponding prediction capabilities, improves the degree of automation and efficiency, enhances subjectivity, and thus solves the problems of strong objectivity, low efficiency, low accuracy and low degree of automation in the prior art.
  • FIG1 is a flow chart of a method for calculating a unified representation of a brain network according to an exemplary embodiment
  • FIG2 is a schematic diagram of step 110 in an embodiment of the embodiment corresponding to FIG1 ;
  • FIG3 is a schematic diagram of the flow of convolution and downsampling operations in the embodiment corresponding to FIG2 ;
  • FIG4 is a schematic diagram of optimizing the brain region structural deformation field in the embodiment corresponding to FIG2 ;
  • FIG5 is a schematic diagram of step 130 in an embodiment of the embodiment corresponding to FIG1 ;
  • FIG6 is a schematic diagram of step 150 in an embodiment of the embodiment corresponding to FIG1 ;
  • FIG7 is a schematic diagram showing the overall framework of a disease prediction model based on a brain network unified representation calculation method according to an exemplary embodiment
  • FIG8 is a schematic diagram showing a training process of a disease prediction model based on a brain network unified representation calculation method according to an exemplary embodiment
  • FIG9 is a block diagram of a brain network unified representation computing device according to an exemplary embodiment
  • FIG10 is an experimental result of a simulation experiment performed according to an embodiment of the present invention.
  • FIG11 is a hardware structure diagram of an electronic device according to an exemplary embodiment
  • Fig. 12 is a block diagram of an electronic device according to an exemplary embodiment.
  • the brain network unified representation calculation method takes the original brain image data as input, combines the two tasks of brain network construction and graph representation learning, and utilizes the ability of automatic feature extraction of deep learning and the idea of graph neural network optimization based on brain networks to reconstruct the brain network corresponding to each sample. It is not limited to the registration or alignment of images of different modalities, and further learns the knowledge of brain area boundaries, accurately locates the spatial position of the brain area divided by the template, and uses texture feature similarity and gradient smoothness constraints to improve the accuracy of regional positioning. A multi-classification cross entropy loss function is proposed, and multiple input samples are mapped to corresponding disease categories.
  • the brain network unified representation calculation method is applicable to a unified representation calculation device and a disease prediction device based on a unified representation calculation method.
  • the unified representation calculation device and the disease prediction device based on the unified representation calculation method can be deployed in an electronic device.
  • the electronic device can be a computer device configured with a von Neumann architecture, and the computer device includes but is not limited to a desktop computer, a laptop computer, a server, and the like.
  • An embodiment of the present application provides a brain network unified representation calculation method, which is applicable to electronic devices, for example, the electronic device can be a desktop computer, a laptop computer, a server, etc.
  • the method may include the following steps:
  • Step 110 obtaining a brain image of the sample, and guiding the brain image to align with the standard brain space based on the standard brain space provided by the reference image, to obtain a brain region structural deformation field of the brain image.
  • the brain image can be any one of structural T1-weighted image (T1WI), diffusion-weighted image (DTI) and resting-state functional MRI (rs-fMRI).
  • T1WI structural T1-weighted image
  • DTI diffusion-weighted image
  • rs-fMRI resting-state functional MRI
  • the reference image refers to the manually aligned MNI brain template that can provide a standard brain space.
  • the brain image and the reference image are downsampled respectively to obtain a low-dimensional representation of the brain image and a low-dimensional representation of the reference image.
  • the low-dimensional representation of the reference image guides the alignment of the low-dimensional representation of the brain image to obtain the brain area structural deformation field of the brain image.
  • the alignment process in the decoding stage may include the following steps: in the decoding stage, the low-dimensional features of the brain image and the reference image are upsampled respectively to obtain the high-dimensional features of the brain image and the reference image; through the jump connection between the encoding and decoding stages, the low-dimensional features of the brain image and the reference image are spliced and implicitly aligned with the corresponding high-dimensional features in the channel direction; the high-dimensional features of the brain image are fused with the high-dimensional features of the reference image, and the fused high-dimensional features of the brain image are transformed and adjusted to obtain the brain region structural deformation field.
  • the brain region voxel-level alignment effect of the brain image to the reference image is achieved, which is conducive to the subsequent construction of an accurate brain network.
  • FIG3 shows the specific process of downsampling the brain image and the reference image, which includes multiple convolutions, multiple batch normalizations, and multiple downsampling operations.
  • the brain image and the reference image are converted from a size of 128 ⁇ 128 ⁇ 128 to a feature map of 8 ⁇ 8 ⁇ 8 with 32 channels.
  • Step 130 adjusting the local deformation difference of the brain region structural deformation field by deformable convolution learning of the local deformation of the brain image, obtaining an optimized brain region structural deformation field, and constructing and optimizing the brain network based on the optimized brain region structural deformation field.
  • a deformable convolution kernel is used to sample the input feature map X, and the deformable convolution is used to learn the rotation invariance and scale invariance features of the image texture.
  • the input feature map X and the filter parameter W are matrix multiplied by the convolution kernel and the convolution bias is added to obtain the output feature map Y.
  • the input feature map X is used to characterize the structural deformation field of the brain area before optimization
  • the output feature map Y is used to characterize the structural deformation field of the brain area after optimization.
  • p n is the position of each grid point in the convolution kernel
  • ⁇ p n represents the learnable position offset parameter, which is usually a floating point number and needs to be resampled and mapped to an integer coordinate position.
  • bilinear interpolation is used to obtain the voxel value X(p) of the non-integer coordinate position on the feature map.
  • the specific calculation formula is as follows:
  • q is the integer coordinate on the input feature map X
  • p is the floating point coordinate on the input feature map X that needs to be calculated
  • the function F is a bilinear interpolation function.
  • the bilinear interpolation function F is selected as:
  • function g is constructed by the maximum value function, and the specific calculation formula of function g is as follows:
  • g(a,b) max(0,1-
  • X(p) can be further calculated by the following formula:
  • the local deformation of brain images is learned through deformable convolution, the local deformation difference of the brain region structure deformation field is adjusted, the brain region structure deformation field is optimized, and a more accurate brain region structure deformation field is obtained, which is beneficial to improve the accuracy of subsequent brain network construction.
  • Step 150 applying the brain region structural deformation field to the brain image to obtain a moving image for indicating the distribution of brain regions, and obtaining regional information of multiple brain regions related to the disease by constraining the similarity between the moving image and the reference image.
  • a weighted encoding method of a cross-attention mechanism is used to jointly constrain the similarity between the moving image and the reference image.
  • multiple transformation functions composed of a neural network perform multiple cross-attention and self-attention on the moving image and the reference image respectively, and the moving image and the reference image are mapped to corresponding feature spaces respectively. Then, the feature spaces corresponding to the moving image and the reference image are spliced to realize weighted encoding of the cross-attention mechanism, and finally the regional information related to the disease in each brain region is obtained.
  • M is a brain image
  • M′ refers to a moving image formed by the brain area structural deformation field acting on the brain image
  • ⁇ is the brain area structural deformation field
  • F is a reference image
  • q, k, v respectively represent multiple transformation functions composed of a neural network
  • the moving image and the reference image are mapped to feature spaces of corresponding dimensions by matrix multiplication with the moving image and the reference image, and then the feature space Mi obtained by mapping the moving image and the feature space obtained by mapping the reference image are feature spliced to obtain M' i .
  • the regional information Mregioni related to the disease in each brain region can be calculated according to the following calculation formula.
  • the weighted encoding method of the cross-attention mechanism is used to jointly constrain the similarity between the moving image and the reference image, so that the accurate disease-related brain area information of each brain region can be obtained.
  • Step 170 constructing an initial brain network based on the regional information of each brain region, and optimizing the initial brain network by performing graph comparison learning between the subgraphs of the brain network and their diffusion graphs to obtain a reconstructed brain network.
  • the reconstructed brain network is used to indicate the brain network characteristics of the sample.
  • a subgraph of the brain network is constructed by sampling the initial brain network, and a corresponding diffusion map is generated (Diffuse) based on the heat kernel features of the subgraph.
  • the node features of the subgraph of the brain network and each node in its diffusion map are updated and aggregated to obtain the node embedding of the subgraph and the node embedding of the diffusion map, respectively.
  • the corresponding graph embedding is extracted from the node embeddings of the subgraph and its diffusion map.
  • the subgraph is compared with the diffusion map, and finally the initial brain network is optimized to be a reconstructed brain network.
  • GCN graph neural network
  • A represents the adjacency matrix of the subgraph
  • D represents the degree matrix of the subgraph
  • MI represents the brain network mutual information function.
  • the calculation formula of the brain network mutual information function MI is as follows:
  • a graph contrast learning strategy was introduced.
  • the model was guided to learn the common connections between the same category and the differences between different categories through the feature similarity between the node embedding and graph embedding of the brain network graph and its diffusion graph, thereby improving the performance of the brain network, optimizing the initial brain network, and obtaining a reconstructed brain network.
  • the disease category of the sample is predicted based on the brain network representation indicated by the reconstructed brain network to obtain the disease prediction result of the sample.
  • disease category prediction is implemented by a disease prediction module based on a unified brain network representation calculation method.
  • the disease prediction module aggregates graph representations containing rich pathological information through a multi-layer perceptron, and connects it with the Softmax function to output the probability that the sample belongs to different disease categories, and then obtains the disease prediction result of the sample based on the probability.
  • the disease prediction module is composed of a three-layer BP neural network, a two-layer ReLu activation function, and a Softmax output layer; the Softmax output layer contains four neurons, which are used to map the brain network representation of the sample indicated by the reconstructed brain network to the probabilities of four different disease categories.
  • the disease prediction module can also include a Dropout layer, which is used to adopt a Dropout strategy to prevent model overfitting.
  • an embodiment of the present invention provides an end-to-end brain network unified representation calculation method, which directly uses a deep learning model to map image data to brain network connections, avoiding the cumbersome image data processing steps in the prior art, with a high degree of automation, simple and efficient operation, and learning the rotation invariance and scale invariance characteristics of standard brain space image textures, thereby improving the accuracy of brain network construction.
  • the stability of brain network construction is improved.
  • the high-precision prediction results on classification indicators are improved, thereby further improving the accuracy of brain network unified representation calculation.
  • Figures 7 and 8 show the overall framework and training process of the disease prediction model based on the unified characterization calculation method of the brain network.
  • Figure 7 shows a schematic diagram of the overall framework of the disease prediction model based on the unified characterization calculation method of the brain network in one embodiment
  • Figure 8 shows a schematic diagram of the training process of the disease prediction model based on the unified characterization calculation method of the brain network in one embodiment.
  • the disease prediction model based on the unified characterization calculation method of the brain network includes a global positioning module, a local perception module, a weighted coding module, a brain network reconstruction module and a disease diagnosis module.
  • each of the above modules in the disease prediction model is a machine learning model that has been trained and has corresponding prediction capabilities.
  • the method may include the following steps:
  • Step 310 the regional position information of the brain image is learned through the global positioning module, so that the brain image is aligned with the standard brain space to obtain the brain region structural deformation field.
  • brain images are used as moving images. Based on the knowledge of anatomical brain regions, a manually aligned standard brain space is used as a reference image.
  • the reference image is convolved and down-sampled by maximum pooling to obtain a low-dimensional image representation.
  • the alignment of the moving image is guided by the low-dimensional image representation of the reference image to narrow the gap in the semantic subspace, achieve voxel-level alignment of the brain regions in the moving image, and construct an accurate brain region structural deformation field.
  • Step 330 Use the local perception module to learn the local micro-deformation and rotation invariance of the brain image to optimize the structural deformation field of the brain area.
  • Step 350 applying the optimized brain region structural deformation field to the brain image, and updating the parameters of the global positioning module and the local perception module according to the transformed moving image and the reference image through the weighted coding module.
  • the optimized brain region structural deformation field is applied to brain images, and the transformed moving image and the reference image are input into a weighted coding module based on brain abnormality representation learning. Weighted coding is performed according to the brain abnormality representation knowledge, the feature spaces of the two are aligned, and the similarity loss is back-propagated to guide the global positioning module and the local perception module to update parameters.
  • the loss function is as follows:
  • the image texture feature similarity loss L similarity is calculated using mean square voxel difference, and the specific calculation formula is as follows:
  • the gradient regularization function is used to promote the smoothness of the smooth displacement field L smooth , which is calculated as follows:
  • the parameters of the global positioning module and the local perception module are updated.
  • Step 370 Use the brain network reconstruction module to optimize the brain network through graph comparison learning ideas.
  • the brain network reconstruction module is trained by the brain network mutual information contrast loss, where the calculation formula of the brain network mutual information contrast loss is as follows:
  • the contrast loss of the brain network mutual information is back-propagated to guide the brain network reconstruction module to update parameters.
  • Step 390 predict the disease category of the sample through the disease diagnosis module, and update the parameters according to the multi-classification cross entropy guidance.
  • the multi-class cross entropy loss is calculated as follows:
  • each element pi represents the probability that the sample belongs to the i-th disease category
  • yi represents the sample label, which indicates the actual disease category to which the sample belongs
  • C represents the number of sample labels, that is, the number of disease categories.
  • the multi-classification cross entropy loss is back-propagated to guide the parameter updates of the above modules in the disease prediction model.
  • the disease prediction model based on the unified representation calculation method of brain networks will have the ability to predict the disease category of the sample. Then, by inputting the brain image of the sample into the disease prediction model, the disease prediction result of the sample can be obtained. At the same time, combined with the brain network representation of the sample extracted from the brain network reconstruction module, the brain network connection differences of the sample can be analyzed, and compared with the group brain network to obtain the abnormal brain connection of the sample, thereby realizing intelligent assisted diagnosis and treatment.
  • a brain network unified representation computing device 900 is shown.
  • the device 900 includes but is not limited to: a global positioning module 910 , a local perception module 930 , a weighted coding module 950 and a brain network reconstruction module 970 .
  • the global positioning module 910 is used to obtain the brain image of the sample, and guide the brain image to align with the standard brain space based on the standard brain space provided by the reference image to obtain the brain area structural deformation field of the brain image.
  • the local perception module 930 is used to learn the local deformation of the brain image through deformable convolution, adjust the local deformation difference of the brain area structure deformation field, and obtain the optimized brain area structure deformation field.
  • the weighted coding module 950 is used to apply the optimized brain region structural deformation field to the brain image to obtain a moving image indicating the distribution of brain regions, and to obtain regional information of multiple brain regions related to the disease by constraining the similarity between the moving image and the reference image.
  • the brain network reconstruction module 970 is used to construct an initial brain network based on the regional information of each brain region, and optimize the initial brain network by performing graph comparison learning between the subgraphs of the brain network and its diffusion graph to obtain a reconstructed brain network.
  • the disease prediction device provided in the above embodiment only uses the division of the above functional modules as an example when predicting diseases.
  • the above functions can be assigned to different functional modules as needed, that is, the internal structure of the disease prediction device will be divided into different functional modules to complete all or part of the functions described above.
  • the embodiments of the brain network unified characterization calculation device and the brain network unified characterization calculation method provided in the above embodiments belong to the same concept, and the specific manner in which each module performs operations has been described in detail in the method embodiments and will not be repeated here.
  • the ADNI data set is used for experimental simulation, and AD, EMCI, SMCI, and NC are selected to contain 50, 54, 44, and 44 samples respectively. All data preprocessing is unified to 128 ⁇ 128 ⁇ 128 size and aligned by global positioning module. 5-fold cross validation is adopted, and the Python sklearn library is used to randomly divide 5 folds. The average value of each indicator is calculated as the final result for comparison.
  • the five evaluation indicators are: accuracy (ACC), sensitivity (SEN), specificity (SPE), and ROC curve area (AUC).
  • the other two methods are: empirical method (Empirical) and benchmark method (Benchmark), and each method uses the same classifier to calculate the classification performance.
  • the Empirical method uses the software toolbox to preprocess DTI to obtain the structural connection matrix, and the Empirical brain connection is obtained after taking the average value;
  • the Benchmark method uses a simple 2-layer GCN network to obtain the Benchmark brain connection by fusing the structural connection matrix.
  • the experimental results are shown in Figure 10. It has been experimentally proved that the brain connection constructed by the present model by fusing structural and functional images has a great advantage in classification performance over other similar methods.
  • Fig. 11 is a schematic diagram showing the structure of an electronic device according to an exemplary embodiment.
  • the electronic device is only an example adapted to the present application and cannot be considered to provide any limitation on the scope of use of the present application.
  • the electronic device cannot be interpreted as needing to rely on or necessarily having one or more components in the exemplary electronic device 2000 shown in FIG. 11.
  • the hardware structure of the electronic device 2000 may vary greatly due to different configurations or performances.
  • the electronic device 2000 includes: a power supply 210 , an interface 230 , at least one memory 250 , and at least one central processing unit (CPU, Central Processing Units) 270 .
  • CPU Central Processing Unit
  • the power supply 210 is used to provide operating voltage for each hardware device on the electronic device 2000 .
  • the interface 230 includes at least one wired or wireless network interface 231 for interacting with external devices.
  • the interface 230 may further include at least one serial-to-parallel conversion interface 233, at least one input-output interface 235, and at least one USB interface 237, as shown in FIG. 9, which is not specifically limited here.
  • the memory 250 is a carrier for storing resources, which may be a read-only memory, a random access memory, a disk or an optical disk, etc.
  • the resources stored thereon include an operating system 251, an application 253 and data 255, etc.
  • the storage method may be temporary storage or permanent storage.
  • the operating system 251 is used to manage and control various hardware devices and application programs 253 on the electronic device 2000, so as to enable the central processing unit 270 to calculate and process the massive data 255 in the memory 250. It can be Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.
  • the application 253 is a computer program that performs at least one specific task based on the operating system 251, and may include at least one module (not shown in FIG. 11 ), each of which may include a computer program for the electronic device 2000.
  • a disease prediction device may be considered as an application 253 deployed on the electronic device 2000.
  • the data 255 may be photos, pictures, etc. stored in a disk, or may be brain images, etc., stored in the memory 250 .
  • the central processor 270 may include one or more processors and is configured to communicate with the memory 250 through at least one communication bus to read the computer program stored in the memory 250, thereby realizing the operation and processing of the mass data 255 in the memory 250.
  • the brain network unified representation calculation method is completed in the form of the central processor 270 reading a series of computer programs stored in the memory 250.
  • present application can also be implemented through hardware circuits or hardware circuits combined with software. Therefore, the implementation of the present application is not limited to any specific hardware circuits, software, or a combination of the two.
  • An electronic device 4000 is provided in an embodiment of the present application.
  • the electronic device 4000 may include: a desktop computer, a laptop computer, a server, etc.
  • the electronic device 4000 includes at least one processor 4001, at least one communication bus 4002 and at least one memory 4003.
  • the processor 4001 and the memory 4003 are connected, such as through a communication bus 4002.
  • the electronic device 4000 may also include a transceiver 4004, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and/or data reception.
  • the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.
  • Processor 4001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 4001 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.
  • the communication bus 4002 may include a path for transmitting information between the above components.
  • the communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc.
  • the communication bus 4002 may be divided into an address bus, a data bus, a control bus, etc.
  • FIG12 only uses a thick line, but it does not mean that there is only one bus or one type of bus.
  • the memory 4003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.
  • ROM Read Only Memory
  • RAM Random Access Memory
  • EEPROM Electrically Erasable Programmable Read Only Memory
  • CD-ROM Compact Disc Read Only Memory
  • optical disk storage including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.
  • magnetic disk storage medium or other magnetic storage device or any other medium
  • the memory 4003 stores a computer program
  • the processor 4001 reads the computer program stored in the memory 4003 through the communication bus 4002 .
  • a storage medium is provided in an embodiment of the present application, on which a computer program is stored.
  • the computer program is executed by a processor, the brain network unified representation calculation method in the above embodiments is implemented.
  • a computer program product includes a computer program, and the computer program is stored in a storage medium.
  • a processor of a computer device reads the computer program from the storage medium, and the processor executes the computer program, so that the computer device executes the brain network unified representation calculation method in each of the above embodiments.
  • the present invention has the following beneficial effects:
  • the present invention proposes a new unified brain network representation calculation method. Compared with the prior art, the present invention is a completely end-to-end processing flow that directly uses a deep learning model to map brain image data into a brain network, avoiding the cumbersome image data preprocessing steps of traditional software, with a high degree of automation, simple and efficient operation.
  • the present invention uses the original brain image data as the input of the model, combines the two tasks of brain network construction and graph representation learning, and utilizes the ability of automatic feature extraction of deep learning and the idea of graph neural network optimization based on brain networks to reconstruct the brain network corresponding to each sample, which is conducive to the analysis and diagnosis of abnormal brain connections in patients with brain diseases.
  • the present invention proposes a new global positioning module guided by brain structure spatial information. Compared with the prior art, the present invention is not limited to the registration or alignment of images of different modalities, but further learns the knowledge of brain area boundaries, so that the model can accurately locate the spatial position of the brain area divided by the template.
  • the weighted coding module based on brain abnormality representation learning of the present invention uses texture feature similarity and gradient smoothness constraints. Compared with the existing method that only relies on image consistency, the extracted features have higher-level semantics in the representation space and contain richer spatial information. The accuracy of regional positioning is improved through local deformation fine-tuning of the module.
  • the brain network reconstruction module based on embedding of brain disease topological features proposed in the present invention utilizes graph diffusion operations to enhance data, and utilizes the differences between brain networks of different categories to strengthen the common connections between similar samples and highlight the differences between different categories, thereby improving the robustness of brain network reconstruction learning.
  • the disease diagnosis module based on brain pathology knowledge designed by the present invention is not only limited to the auxiliary diagnosis of Alzheimer's disease, but can also be applied to the medical diagnosis and treatment scenarios of other diseases: by modifying the corresponding disease sample data and model output labels during model training, it can be conveniently applied to the analysis and diagnosis of other brain diseases.
  • the brain network mutual information comparison loss function proposed in the present invention calculates the similarity between the graph embedding features and the node features of another view. Compared with the existing technology that compares the node features and the graph embedding features, it is more efficient and improves the stability of brain network representation learning.
  • the present invention proposes a multi-classification cross entropy loss function, which maps multiple input samples to corresponding categories and utilizes the powerful nonlinear fitting ability of the multilayer perceptron to improve the high-precision prediction results on the classification indicators.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

(图1)本发明实施例公开了一种脑网络统一表征计算方法及装置，所述方法包括：获取样本的脑影像，并基于参考图像所提供的标准脑空间，引导脑影像与标准脑空间进行对齐，得到脑影像的脑区结构形变场；通过可形变卷积学习脑影像的局部形变，对脑区结构形变场的局部形变差异进行调整，得到优化后的脑区结构形变场；将优化后的脑区结构形变场作用于脑影像得到移动图像，并通过约束移动图像与参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息；根据各脑区的区域信息进行脑网络构建，并通过在脑网络的子图及其扩散图之间进行的图对比学习，对脑网络进行优化，得到重构脑网络。本发明解决了现有技术主观性强，效率低，准确率和自动化程度不高的问题。

Description

脑网络统一表征计算方法、装置、电子设备及存储介质 技术领域
本发明涉及医学影像处理技术领域，尤其涉及一种脑网络统一表征计算方法、装置、电子设备及存储介质。
背景技术
不同的模态的脑影像数据通常被用来构建三种类型的脑网络：即形态学网络、结构网络和功能网络，2007年以各个脑区的皮层厚度作为形态学特征，首次构建了大脑形态学脑网络，发现了脑网络的小世界属性，此后，大量基于图论的复杂网络分析方法开始研究大脑区域之间的连接模式和图属性分析，从结构到功能上用于探索神经退行性疾病的诊断，图神经网络是当前的深度学习技术研究热点之一，已成为从图中提取信息以执行神经退行性疾病诊断的最广泛使用的方法。
然而，现有脑网络构建方法主要依赖于医学影像处理软件包，依赖经验知识人工设置大量的参数，自动化程度不高，主观性较强，导致脑网络构建不准确，从而导致疾病预测的结果不够准确。
除此之外，现有技术基于群体水平的统计分析构建得到脑网络，重建脑区之间的脑连接操作复杂，需要不断迭代优化找到全局最优的模型，参数量大，处理十分耗时，且无法构建针对个体的个性化脑网络，难以探索疾病的潜在发病机制，无法实现精准化诊断。
因此，急需一种客观性强、效率高、自动化程度高、准确率高的脑网络统一表征计算方法。
发明内容
本发明各实施例提供一种脑网络统一表征计算方法、装置、电子设备及存储介质，以解决相关技术中存在的主观性强、效率低、准确率和自动化程度不高的问题。
其中，本发明所采用的技术方案为：
根据本发明的一个方面，一种脑网络统一表征计算方法，所述方法包括： 获取样本的脑影像，并基于参考图像所提供的标准脑空间，引导所述脑影像与所述标准脑空间进行对齐，得到所述脑影像的脑区结构形变场；通过可形变卷积学习所述脑影像的局部形变，对所述脑区结构形变场的局部形变差异进行调整，得到优化后的所述脑区结构形变场；将优化后的所述脑区结构形变场作用于所述脑影像，得到用于指示脑区分布的移动图像，并通过约束所述移动图像与所述参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息；根据各所述脑区的区域信息进行初始的脑网络构建，并通过在所述脑网络的子图及其扩散图之间进行图对比学习，对初始的脑网络进行优化，得到重构脑网络，所述重构脑网络用于指示所述样本的脑网络统一表征。根据本发明的一个方面，一种脑网络统一表征计算装置，所述装置包括：全局定位模块，用于获取样本的脑影像，并基于参考图像所提供的标准脑空间，引导所述脑影像与所述标准脑空间进行对齐，得到所述脑影像的脑区结构形变场；局部感知模块，用于通过可形变卷积学习所述脑影像的局部形变，对所述脑区结构形变场的局部形变差异进行调整，得到优化后的所述脑区结构形变场；加权编码模块，用于将优化后的脑区结构形变场作用于脑影像，得到用于指示脑区分布的移动图像，并通过约束移动图像与参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息；脑网络重构模块，用于根据各脑区的区域信息进行初始的脑网络构建，并通过在脑网络的子图及其扩散图之间进行图对比学习，对初始的脑网络进行优化，得到重构脑网络。
根据本发明的一个方面，一种电子设备，包括处理器及存储器，所述存储器上存储有计算机可读指令，所述计算机可读指令被所述处理器执行时实现如上所述的脑网络统一表征计算方法。
根据本发明的一个方面，一种存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现如上所述的脑网络统一表征计算方法。
根据本发明的一个方面，一种计算机程序产品，计算机程序产品包括计算机程序，计算机程序存储在存储介质中，计算机设备的处理器从存储介质读取计算机程序，处理器执行计算机程序，使得计算机设备执行时实现如上所述的脑网络统一表征计算方法。
在上述技术方案中，本发明实现了主观性强、效率高自动化程度高、准确率高的脑网络统一表征计算。
具体而言，本发明通过基于参考图像提供的脑结构空间信息引导的全局定位模块和基于大脑异常表征学习的加权编码模块，得到与疾病相关的多个脑区的区域信息，并根据各脑区的区域信息构建了较为准确的初始的脑网络，通过基于脑疾病拓扑特征嵌入的脑网络重构模块利用图对比学习的思想优化初始的脑网络，进而基于重构脑网络所指示的脑网络表征，对样本进行疾病类别预测得到样本的疾病预测结果，进一步准确定位了疾病相关脑区和异常连接特征，探索疾病的潜在发病机制，提高了脑网络统一表征计算的准确度，进而提高了疾病预测的准确性，此外，用于疾病预测的疾病预测模型中的各模块利用深度学习技术而具有相应预测能力，提高了自动化程度和效率，增强了主观性从而解决了现有技术中客观性强，效率低，准确率和自动化程度不高的问题。
应当理解的是，以上的一般描述和后文的细节描述仅是示例性和解释性的，并不能限制本发明。
附图说明
此处的附图被并入说明书中并构成本说明书的一部分，示出了符合本发明的实施例，并于说明书一起用于解释本发明的原理。
图1是根据一示例性实施例示出的一种脑网络统一表征计算方法的流程图；
图2是图1对应实施例中步骤110在一个实施例的示意图；
图3是图2对应实施例中卷积和下采样操作的流程示意图；
图4是图2对应实施例中优化脑区结构形变场的示意图；
图5是图1对应实施例中步骤130在一个实施例的示意图；
图6是图1对应实施例中步骤150在一个实施例的示意图；
图7是根据一示例性实施例示出基于脑网络统一表征计算方法的疾病预测模型的总体框架的示意图；
图8是根据一示例性实施例示出的基于脑网络统一表征计算方法的疾病预测模型的训练过程的示意图；
图9是根据一示例性实施例示出的一种脑网络统一表征计算装置的框图；
图10是本发明实施例进行仿真实验的实验结果；
图11是根据一示例性实施例示出的一种电子设备的硬件结构图；
图12是根据一示例性实施例示出的一种电子设备的框图。
通过上述附图，已示出本发明明确的实施例，后文中将有更详细的描述，这些附图和文字描述并不是为了通过任何方式限制本发明构思的范围，而是通过参考特定实施例为本领域技术人员说明本发明的概念。
具体实施方式
这里将详细地对示例性实施例执行说明，其示例表示在附图中。下面的描述涉及附图时，除非另有表示，不同附图中的相同数字表示相同或相似的要素。以下示例性实施例中所描述的实施方式并不代表与本发明相一致的所有实施方式。相反，它们仅是与如所附权利要求书中所详述的、本发明的一些方面相一致的装置和方法的例子。
现有的相关技术利用软件构建脑网络，需要手动设置大量的参数，参数选择具有主观性，依赖临床上丰富的经验参考，脑网络构建结果不唯一，最终实验结果难以在相同条件下进行复现和公平比较，导致脑网络的构建不够准确。
同时，现有的相关技术面向大脑部分分割或分类任务，难以精确定位脑区的异常变化，无法有效定位临床意义上的与疾病相关脑区，识别脑网络中的异常连接，不能为疾病诊断提供参考。
由上可知，相关技术中仍存在客观性强，效率低，准确率和自动化程度不高的缺陷。
为此，本申请提供的脑网络统一表征计算方法将原始脑影像数据作为输入，将脑网络构建和图表示学习两个任务相结合，利用深度学习自动特征提取的能力和基于脑网络的图神经网络优化的思想，重构每个样本对应的脑网络，不局限于仅对不同模态图像的配准或对齐，进一步学习脑区边界知识，精确定位到模板划分的脑区空间位置，使用纹理特征相似性和梯度平滑性约束提升区域定位的准确度，提出多分类交叉熵损失函数，将输入多个样本映射到相应疾病类别，利用多层感知器的强大非线性拟合能力，提升了在疾病类别上的高精度预测结果，从多个方面提高脑网络构建的准确性，从而提升 疾病预测的准确性，同时，该脑网络统一表征计算方法适用于统一表征计算装置和基于统一表征计算方法的疾病预测装置、该统一表征计算装置和基于统一表征计算方法的疾病预测装置可部署于电子设备，例如，该电子设备可以是配置冯诺依曼体系结构的计算机设备，该计算机设备包括但不限于台式电脑、笔记本电脑、服务器等等。
请参阅图1，本申请实施例提供了一种脑网络统一表征计算方法，该方法适用于电子设备，例如，该电子设备可以是台式电脑、笔记本电脑、服务器等等。
在下述方法实施例中，为了便于描述，以该方法各步骤的执行主体为电子设备为例进行说明，但是并非对此构成具体限定。
如图1所示，该方法可以包括以下步骤：
步骤110，获取样本的脑影像，并基于参考图像提供的标准脑空间，引导脑影像与标准脑空间进行对齐，得到脑影像的脑区结构形变场。
其中，脑影像可以是结构T1加权影像(T1WI)、弥散加权影像(DTI)和静息状态功能MRI(rs-fMRI)中的任意一种。参考图像则是指人工手动对齐的可提供标准脑空间的MNI脑模板。
在一个可能的实现方式，请参阅图2，在编码阶段，分别对脑影像和参考图像进行下采样，得到脑影像的低维表征和参考图像的低维表征；在解码阶段，由参考图像的低维表征指导脑影像的低维表征进行对齐，得到脑影像的脑区结构形变场。
在一个可能的实现方式，解码阶段的对齐过程可以包括以下步骤：在解码阶段，分别对脑影像和参考图像的低维特征进行上采样，得到所述脑影像的高维特征和所述参考图像的高维特征；通过编解码阶段之间的跳跃连接，将所述脑影像和所述参考图像的各所述低维特征分别在通道方向上与相应的各所述高维特征进行拼接和隐式对齐；将所述脑影像的各所述高维特征与所述参考图像的各所述高维特征进行融合，并对所述脑影像融合后的所述高维特征进行变换调整，得到所述脑区结构形变场。此种方式下，通过缩小语义子空间中的差距，实现了脑影像对参考图像的脑区体素级对齐效果，进而有利于后续精准的脑网络的构建。
其中，图3表示了分别对脑影像和参考图像进行下采样的具体过程，该具体过程包括多次卷积、多次批归一化和多次下采样操作。其中，脑影像和参考图像经过图3中的3次卷积、3次批归一化和2次下采样操作之后，由尺寸大小128×128×128转化为具有32通道的8×8×8的特征图。
步骤130，通过可形变卷积学习脑影像的局部形变，对脑区结构形变场的局部形变差异进行调整，得到优化后的脑区结构形变场，以基于优化后的脑区结构形变场进行脑网络的构建与优化。
具体地，如图4所示，在输入特征图X上使用可形变卷积核进行采样，利用可形变卷积想学习图像纹理的旋转不变性和尺度不变性特征，将输入特征图X和滤波器参数W通过卷积核做矩阵乘法并加上卷积偏置得到输出特征图Y。其中，输入特征图X用于表征优化前的脑区结构形变场，输出特征图Y则用于表征优化后的脑区结构形变场。
具体计算公式如下：
Figure PCTCN2022136710-appb-000001
其中，p＝p 0+p n+Δp n，p n为卷积核中每一个格点的位置，Δp n表示可学习的位置偏移项参数，通常为浮点数，需要重采样映射到整数坐标位置。
在一个可能的实现方式，采用双线性插值来获取特征图上的非整数坐标位置的体素值X(p)，具体计算公式如下：
Figure PCTCN2022136710-appb-000002
其中，q为输入特征图X上的整数坐标，p为需要计算得到的输入特征图X上的浮点坐标，函数F为双线性插值函数。
在一个可能的实现方式，为了保证神经网络的可导性，选择双线性插值函数F为：
F(q，p)＝g(q x，p x)·g(q y，p y)·g( q z，p z)。
其中，函数g由最大值函数构造，函数g的具体计算公式如下：
g(a，b)＝max(0，1-|q x-p x|)。
因此，X(p)可以进一步由以下计算公式得到：
Figure PCTCN2022136710-appb-000003
经过上述过程，通过可形变卷积学习了脑影像的局部形变，对脑区结构形变场的局部形变差异进行了调整，实现了脑区结构形变场的优化，得到更加精确的脑区结构形变场，进而有利于提高后续脑网络构建的准确性。
步骤150，将脑区结构形变场作用于脑影像，得到用于指示脑区分布的移动图像，并通过约束移动图像与参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息。
在一个可能的实现方式，采用交叉注意力机制的加权编码方式联合约束移动图像与参考图像之间的相似性。
在一个可能的实现方式，通过由神经网络构成的多个变换函数，分别与移动图像和参考图像各自进行的多次交叉注意力和自注意力，将移动图像和参考图像分别映射至相应的特征空间，再将移动图像和参考图像分别对应的特征空间进行特征拼接，实现交叉注意力机制的加权编码，最终得到每个脑区中与疾病相关的区域信息。
具体地，如图5所示，
Figure PCTCN2022136710-appb-000004
M为脑影像，M′是指经脑区结构形变场作用于脑影像而形成的移动图像，ψ为脑区结构形变场，F为参考图像，q、k、v分别表示由神经网络构成的多个变换函数，通过与移动图像和参考图像做矩阵相乘，将移动图像和参考图像分别映射到相应维度的特征空间，进而将移动图像映射得到的特征空间M_i和参考图像映射得到的特征空间进行特征拼接，得到M’ i。
在得到M’ i后，便能够按照以下计算公式，计算得到每个脑区中与疾病相关的区域信息Mregion i。
Figure PCTCN2022136710-appb-000005
经过上述过程，采用交叉注意力机制的加权编码方式来联合约束移动图像与参考图像之间的相似性，便能够得到每个脑区的精确的与疾病相关的大 脑区域信息。
步骤170，根据各脑区的区域信息进行初始的脑网络构建，并通过在脑网络的子图及其扩散图之间进行图对比学习，对初始的脑网络进行优化，得到重构脑网络。
其中，重构脑网络用于指示样本的脑网络特征。
在一个可能的实现方式，如图6所示，通过对初始的脑网络进行的采样，构造脑网络的子图，并基于子图的热核特征生成(Diffuse)相应的扩散图，对脑网络的子图及其扩散图中的各节点进行更新节点特征并聚合节点特征，分别得到子图的节点嵌入和扩散图的节点嵌入，并由子图及其扩散图的各节点嵌入提取得到相应的图嵌入，基于子图及其扩散图的节点嵌入和图嵌入，进行子图与所述扩散图之间的对比，最终将初始的脑网络优化为重构脑网络。
在此说明的是，对脑网络的子图及其扩散图中的各节点进行更新节点特征并聚合节点特征是通过图神经网络GCN实现的，具体地，利用GCN学习子图及其扩散图中各节点的节点嵌入，然后将学习到的节点嵌入通过一个共享的全连接层，便能够分别得到子图及其扩散图的节点嵌入的集合，最后使用readout函数得到各自的图嵌入。
首先说明的是，基于子图的热核特征生成相应扩散图的计算公式如下：
S heat＝exp(tAD -1-t)。
其中，A表示子图的邻接矩阵，D表示子图的度矩阵。
其次，脑网络的优化目标表示为如下计算公式：
Figure PCTCN2022136710-appb-000006
其中，
Figure PCTCN2022136710-appb-000007
和
Figure PCTCN2022136710-appb-000008
表示两个视图的节点嵌入，
Figure PCTCN2022136710-appb-000009
和
Figure PCTCN2022136710-appb-000010
表示两个视图的图嵌入，MI表示脑网络互信息函数。该脑网络互信息函数MI的计算公式如下：
Figure PCTCN2022136710-appb-000011
通过上述过程，引入了图对比学习的策略，利用图扩散技术，通过脑网络图与其扩散图的节点嵌入和图嵌入之间的特征相似性指导模型学习同类之间的共性连接和不同类别之间的差异，提升脑网络性能，对初始的脑网络进行优化，得到重构脑网络。
在一种可能的实现方式，得到重构脑网络之后，基于重构脑网络指示的脑网络表征，对样本进行疾病类别预测，得到样本的疾病预测结果。
具体地，疾病类别预测由基于脑网络统一表征计算方法的疾病预测模块实现。其中，该疾病预测模块通过多层感知机聚合蕴含了丰富病理信息的图表征，并与Softmax函数连接，输出样本属于不同疾病类别的概率，进而根据该概率得到样本的疾病预测结果。其中，该疾病预测模块由三层BP神经网络、两层ReLu激活函数、Softmax输出层构成；该Softmax输出层包含四个神经元，分别用于将重构脑网络所指示的样本的脑网络表征映射为四种不同疾病类别的概率。此外，该疾病预测模块还可以包括Dropout层，用于采用Dropout策略防止模型过拟合。
通过上述过程，本发明实施例提供了一种端对端的脑网络统一表征计算方法，直接利用深度学习模型将影像数据映射为脑网络连接，避免了现有技术中繁琐的影像数据处理步骤，自动化程度高，操作简单高效，学习了标准脑空间图像纹理的旋转不变性和尺度不变性特征，提高了脑网络构建的准确性，通过在脑网络的子图及其扩散图之间进行图对比学习，提高了脑网络构建的稳定性，通过多分类交叉熵损失函数，利用多层感知器的强大非线性拟合能力，提升了在分类指标上的高精度预测结果，进一步提高了脑网络统一表征计算的准确性。
图7和图8展示了基于脑网络统一表征计算方法的疾病预测模型的总体框架和训练过程。图7示出了基于脑网络统一表征计算方法的疾病预测模型在一个实施例中的总体框架的示意图，图8示出了基于脑网络统一表征计算方法的疾病预测模型在一个实施例中的训练过程的示意图。在图7中，基于脑网络统一表征计算方法的疾病预测模型包括全局定位模块、局部感知模块、加权编码模块、脑网络重构模块和疾病诊断模块。其中，疾病预测模型中的上述各模块均是经过训练、且具有相应预测能力的机器学习模型。
现结合图7和图8，对基于脑网络统一表征计算方法的疾病预测模型的训练过程进行以下详细地说明：
如图7所示，该方法可以包括以下步骤：
步骤310，通过全局定位模块学习脑影像的区域位置信息，使得脑影像 与标准脑空间对齐，得到脑区结构形变场。
在一个可能的实现方式，将脑影像作为移动图像，根据解剖学脑区知识，使用人工手动对齐的标准脑空间作为参考图像，对参考图像进行卷积和最大池化下采样，得到低维图像表征，根据参考图像的低维图像表征指导移动图像的对齐，缩小语义子空间中的差距，对移动图像的脑区实现体素级对齐效果，构建精准的脑区结构形变场。
步骤330，使用局部感知模块学习脑影像局部微小形变和旋转不变性，优化脑区结构形变场。
步骤350，将优化后的脑区结构形变场作用于脑影像，通过加权编码模块根据变换后的移动图像和参考图像更新全局定位模块和局部感知模块参数。
在一个可能的实现方式，将优化后的脑区结构形变场作用于脑影像，将变换后的移动图像与参考图像输入到基于大脑异常表征学习的加权编码模块，根据大脑异常表征知识进行加权编码，对齐两者的特征空间，对相似性损失进行反向传播，指导全局定位模块和局部感知模块进行参数更新。
具体地，损失函数如下：
Figure PCTCN2022136710-appb-000012
其中，图像纹理特征相似性损失L similarity采用均方体素差异计算，具体计算公式如下：
Figure PCTCN2022136710-appb-000013
在空间形变场φ上，利用梯度正则函数促进光滑位移场平滑性L smooth，计算公式如下：
Figure PCTCN2022136710-appb-000014
通过上述过程，全局定位模块和局部感知模块的参数进行了更新。
步骤370，使用脑网络重构模块通过图对比学习思想优化脑网络。
在一个可能的实现方式，脑网络重构模块通过脑网络互信息对比损失进行训练，其中，脑网络互信息对比损失的计算公式如下：
Figure PCTCN2022136710-appb-000015
对脑网络互信息对比损失进行反向传播，指导脑网络重构模块更新参数。
步骤390，通过疾病诊断模块对样本进行疾病类别预测，根据多分类交叉熵指导参数更新。
在一个可能的实现方式，多分类交叉熵损失的计算公式如下：
Figure PCTCN2022136710-appb-000016
其中，p是一个概率分布，每个元素p i表示样本属于第i个疾病类别的概率，y i表示样本标签，该样本标签指示了样本所属的真实的疾病类别，C表示样本标签数，即疾病类别的个数。
对多分类交叉熵损失进行反向传播，指导疾病预测模型中的上述各个模块进行参数更新。
通过上述过程，将基于脑网络统一表征计算方法的疾病预测模型训练完毕后，该基于脑网络统一表征计算方法的疾病预测模型便具备了对样本进行疾病类别预测的能力，那么，将样本的脑影像输入该疾病预测模型中，便能够得到该样本的疾病预测结果，同时结合从脑网络重构模块中提取得到的该样本的脑网络表征，从而可以分析该样本的脑网络连接差异，与群体脑网络进行比对即可得到该样本的异常脑连接，从而实现智能辅助诊疗。
请参阅图9，在一示例性实施例中，一种脑网络统一表征计算装置900。
所述装置900包括但不限于：全局定位模块910、局部感知模块930、加权编码模块950及脑网络重构模块970。
其中，全局定位模块910，用于获取样本的脑影像，并基于参考图像所提供的标准脑空间，引导脑影像与标准脑空间进行对齐，得到脑影像的脑区结构形变场。
局部感知模块930，用于通过可形变卷积学习脑影像的局部形变，对脑区结构形变场的局部形变差异进行调整，得到优化后的脑区结构形变场。
加权编码模块950，用于将优化后的脑区结构形变场作用于脑影像，得 到用于指示脑区分布的移动图像，并通过约束移动图像与参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息。
脑网络重构模块970，用于根据各脑区的区域信息进行初始的脑网络构建，并通过在脑网络的子图及其扩散图之间进行图对比学习，对初始的脑网络进行优化，得到重构脑网络。需要说明的是，上述实施例所提供的疾病预测装置在进行疾病预测时，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即疾病预测装置的内部结构将划分为不同的功能模块，以完成以上描述的全部或者部分功能。
另外，上述实施例所提供的脑网络统一表征计算装置与脑网络统一表征计算方法的实施例属于同一构思，其中各个模块执行操作的具体方式已经在方法实施例中进行了详细描述，此处不再赘述。
本发明中，采用ADNI数据集进行实验仿真，选取AD、EMCI、SMCI、NC分别包含个50、54、44、44个样本。所有数据预处理统一为128×128×128大小，通过全局定位模块对齐。采用5折交叉验证，使用Python sklearn库随机划分5折。计算每个指标的平均值作为最终结果进行比较。五种评价指标为：准确率(ACC)、灵敏度(SEN)、特异度(SPE)、ROC曲线面积(AUC)。其他两种的方法分别为：经验法(Empirical)和基准方法(Benchmark)，每种方法都采用同一分类器来计算分类性能。其中，Empirical方法使用软件工具箱预处理DTI得到结构连接矩阵，取平均值后得到Empirical脑连接；Benchmark方法是使用简单2层GCN网络，通过融合结构连接矩阵得到Benchmark脑连接。实验结果如图10所示，经过实验证明，本模型通过融合结构和功能影像构建的脑连接比同类其他方法的分类性能上具有很大的优势。
图11根据一示例性实施例示出的一种电子设备的结构示意。
需要说明的是，该电子设备只是一个适配于本申请的示例，不能认为是提供了对本申请的使用范围的任何限制。该电子设备也不能解释为需要依赖于或者必须具有图11示出的示例性的电子设备2000中的一个或者多个组件。
电子设备2000的硬件结构可因配置或者性能的不同而产生较大的差异，如图11所示，电子设备2000包括：电源210、接口230、至少一存储器250、 以及至少一中央处理器(CPU,Central Processing Units)270。
具体地，电源210用于为电子设备2000上的各硬件设备提供工作电压。
接口230包括至少一有线或无线网络接口231，用于与外部设备交互。
当然，在其余本申请适配的示例中，接口230还可以进一步包括至少一串并转换接口233、至少一输入输出接口235以及至少一USB接口237等，如图9所示，在此并非对此构成具体限定。
存储器250作为资源存储的载体，可以是只读存储器、随机存储器、磁盘或者光盘等，其上所存储的资源包括操作系统251、应用程序253及数据255等，存储方式可以是短暂存储或者永久存储。
其中，操作系统251用于管理与控制电子设备2000上的各硬件设备以及应用程序253，以实现中央处理器270对存储器250中海量数据255的运算与处理，其可以是Windows ServerTM、Mac OS XTM、UnixTM、LinuxTM、FreeBSDTM等。
应用程序253是基于操作系统251之上完成至少一项特定工作的计算机程序，其可以包括至少一模块(图11未示出)，每个模块都可以分别包含有对电子设备2000的计算机程序。例如，疾病预测装置可视为部署于电子设备2000的应用程序253。
数据255可以是存储于磁盘中的照片、图片等，还可以是脑影像等，存储于存储器250中。
中央处理器270可以包括一个或多个以上的处理器，并设置为通过至少一通信总线与存储器250通信，以读取存储器250中存储的计算机程序，进而实现对存储器250中海量数据255的运算与处理。例如，通过中央处理器270读取存储器250中存储的一系列计算机程序的形式来完成脑网络统一表征计算方法。
此外，通过硬件电路或者硬件电路结合软件也能同样实现本申请，因此，实现本申请并不限于任何特定硬件电路、软件以及两者的组合。
请参阅图12，本申请实施例中提供了一种电子设备4000，该电子设备400可以包括：台式电脑、笔记本电脑、服务器等。
在图12中，该电子设备4000包括至少一个处理器4001、至少一条通信 总线4002以及至少一个存储器4003。
其中，处理器4001和存储器4003相连，如通过通信总线4002相连。可选地，电子设备4000还可以包括收发器4004，收发器4004可以用于该电子设备与其他电子设备之间的数据交互，如数据的发送和/或数据的接收等。需要说明的是，实际应用中收发器4004不限于一个，该电子设备4000的结构并不构成对本申请实施例的限定。
处理器4001可以是CPU(Central Processing Unit，中央处理器)，通用处理器，DSP(Digital Signal Processor，数据信号处理器)，ASIC(Application Specific Integrated Circuit，专用集成电路)，FPGA(Field Programmable Gate Array，现场可编程门阵列)或者其他可编程逻辑器件、晶体管逻辑器件、硬件部件或者其任意组合。其可以实现或执行结合本申请公开内容所描述的各种示例性的逻辑方框，模块和电路。处理器4001也可以是实现计算功能的组合，例如包含一个或多个微处理器组合，DSP和微处理器的组合等。
通信总线4002可包括一通路，在上述组件之间传送信息。通信总线4002可以是PCI(Peripheral Component Interconnect，外设部件互连标准)总线或EISA(Extended Industry Standard Architecture，扩展工业标准结构)总线等。通信总线4002可以分为地址总线、数据总线、控制总线等。为便于表示，图12中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
存储器4003可以是ROM(Read Only Memory，只读存储器)或可存储静态信息和指令的其他类型的静态存储设备，RAM(Random Access Memory，随机存取存储器)或者可存储信息和指令的其他类型的动态存储设备，也可以是EEPROM(Electrically Erasable Programmable Read Only Memory，电可擦可编程只读存储器)、CD-ROM(Compact Disc Read Only Memory，只读光盘)或其他光盘存储、光碟存储(包括压缩光碟、激光碟、光碟、数字通用光碟、蓝光光碟等)、磁盘存储介质或者其他磁存储设备、或者能够用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算机存取的任何其他介质，但不限于此。
存储器4003上存储有计算机程序，处理器4001通过通信总线4002读取存储器4003中存储的计算机程序。
该计算机程序被处理器4001执行时实现上述各实施例中的脑网络统一 表征计算方法。
此外，本申请实施例中提供了一种存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现上述各实施例中的脑网络统一表征计算方法。
本申请实施例中提供了一种计算机程序产品，该计算机程序产品包括计算机程序，该计算机程序存储在存储介质中。计算机设备的处理器从存储介质读取该计算机程序，处理器执行该计算机程序，使得该计算机设备执行上述各实施例中的脑网络统一表征计算方法。
与相关技术相比，本发明的有益效果是：
1.本发明提出了一个新的脑网络统一表征计算方法，与现有技术相比，本发明是完全端对端的处理流程，直接利用深度学习模型将脑影像数据映射为脑网络，避免传统软件繁琐的影像数据预处理步骤，自动化程度高，操作简单高效。
2.本发明将原始脑影像数据作为模型的输入，将脑网络构建和图表示学习两个任务相结合，利用深度学习自动特征提取的能力和基于脑网络的图神经网络优化的思想，重构每个样本对应的脑网络，有利于对脑疾病患者的异常脑连接进行分析和诊断。
3.本发明提出新的脑结构空间信息引导的全局定位模块，与现有技术相比，本发明不局限于仅对不同模态图像的配准或对齐，进一步学习到脑区边界知识，使得模型精确定位到模板划分的脑区空间位置。
4.本发明的基于大脑异常表征学习的加权编码模块使用了纹理特征相似性和梯度平滑性约束，与现有仅依赖图像一致性的方法相比，提取特征在表征空间更具有高层语义，含有更丰富的空间信息，通过该模块的局部形变微调，提升了区域定位的准确度。
5.本发明提出的基于脑疾病拓扑特征嵌入的脑网络重构模块，利用图的扩散操作使得数据增强，并利用不同类别的脑网络差异，强化同类样本之间的共性连接并凸显不同类别之间的差异，提升了脑网络重构学习的鲁棒性。
6.本发明设计的基于大脑病理知识的疾病诊断模块不仅局限于脑部阿尔茨海默症辅助诊断，也能应用于其他疾病的医学诊疗场景：通过修改模型训练时的对应疾病样本数据和模型输出标签，可以方便应用于其他脑疾病的分 析诊断。
7.本发明提出的脑网络互信息对比损失函数，将图嵌入特征与另一视图的节点特征计算相似性，相比现有技术以节点特征之间和图嵌入特征之间的对比更加高效，提升了脑网络表征学习的稳定性。
8.本发明提出多分类交叉熵损失函数，将输入多个样本映射到相应类别，利用多层感知器的强大非线性拟合能力，提升了在分类指标上的高精度预测结果。
应该理解的是，虽然附图的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，其可以以其他的顺序执行。而且，附图的流程图中的至少一部分步骤可以包括多个子步骤或者多个阶段，这些子步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，其执行顺序也不必然是依次进行，而是可以与其他步骤或者其他步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
上述内容，仅为本发明的较佳示例性实施例，并非用于限制本发明的实施方案，本领域普通技术人员根据本发明的主要构思和精神，可以十分方便地进行相应的变通或修改，故本发明的保护范围应以权利要求书所要求的保护范围为准。

Claims (10)

  1. 一种脑网络统一表征计算方法，其特征在于，所述方法包括：
    获取样本的脑影像，并基于参考图像所提供的标准脑空间，引导所述脑影像与所述标准脑空间进行对齐，得到所述脑影像的脑区结构形变场；
    通过可形变卷积学习所述脑影像的局部形变，对所述脑区结构形变场的局部形变差异进行调整，得到优化后的所述脑区结构形变场；
    将优化后的所述脑区结构形变场作用于所述脑影像，得到用于指示脑区分布的移动图像，并通过约束所述移动图像与所述参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息；
    根据各所述脑区的区域信息进行初始的脑网络构建，并通过在所述脑网络的子图及其扩散图之间进行图对比学习，对初始的脑网络进行优化，得到重构脑网络，所述重构脑网络用于指示所述样本的脑网络统一表征。
  2. 如权利要求1所述的方法，其特征在于，所述得到重构脑网络之后，所述方法还包括：
    基于所述重构脑网络指示的脑网络统一表征，对所述样本进行疾病类别预测，得到所述样本的疾病预测结果。
  3. 如权利要求1所述的方法，其特征在于，所述基于参考图像所提供的标准脑空间，引导所述脑影像与所述标准脑空间进行对齐，得到所述脑影像的脑区结构形变场，包括：
    在编码阶段，分别对所述脑影像和所述参考图像进行卷积和下采样，得到所述脑影像的低维表征和所述参考图像的低维表征；
    在解码阶段，由所述参考图像的低维表征指导所述脑影像的低维表征进行对齐，得到所述脑影像的脑区结构形变场。
  4. 如权利要求3所述的方法，其特征在于，所述在解码阶段，由所述参考图像的低维表征指导所述脑影像的低维表征进行对齐，得到所述脑影像的脑区结构形变场，包括：
    在解码阶段，分别对所述脑影像和所述参考图像的所述低维特征进行上采样，得到所述脑影像的高维特征和所述参考图像的高维特征；
    通过编解码阶段之间的跳跃连接，将所述脑影像和所述参考图像的各所述低维特征分别在通道方向上与相应的各所述高维特征进行拼接和隐式对齐；
    将所述脑影像的各所述高维特征与所述参考图像的各所述高维特征进行融合，并对所述脑影像融合后的所述高维特征进行变换调整，得到所述脑区结构形变场。
  5. 如权利要求1所述的方法，其特征在于，所述通过约束所述移动图像与所述参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息，包括：
    通过由神经网络构成的多个变换函数，与所述移动图像和所述参考图像各自进行的多次交叉注意力和自注意力，将所述移动图像和所述参考图像分别映射至相应的特征空间；
    将所述移动图像和所述参考图像映射得到的特征空间进行特征拼接，得到每个脑区中与疾病相关的区域信息。
  6. 如权利要求1所述的方法，其特征在于，所述根据各所述脑区的区域信息进行初始的脑网络构建，并通过在所述脑网络的子图及其扩散图之间进行的图对比学习，对初始的脑网络进行优化，得到重构脑网络，包括：
    通过对所述脑网络进行的采样，构造所述脑网络的子图，并基于所述子图的热核特征生成相应的扩散图；
    对所述脑网络的子图及其扩散图中的各节点进行更新节点特征并聚合节点特征，分别得到所述子图的节点嵌入和所述扩散图的节点嵌入，并由所述子图及其扩散图的各节点嵌入提取得到相应的图嵌入；
    基于所述子图及其扩散图的节点嵌入和图嵌入，进行所述子图与所述扩散图之间的对比，将初始的所述脑网络优化为所述重构脑网络。
  7. 如权利要求1至6任一项所述的方法，其特征在于，所述方法还包括训练疾病预测模型，所述疾病预测模型包括全局定位模块、局部感知模块、加权编码模块、脑网络重构模块和疾病诊断模块；所述疾病预测模型中的各模块均是经过训练、且具有相应预测能力的机器学习模型。
  8. 一种脑网络统一表征计算装置，其特征在于，所述装置包括：
    全局定位模块，用于获取样本的脑影像，并基于参考图像所提供的标准脑空间，引导所述脑影像与所述标准脑空间进行对齐，得到所述脑影像的脑区结构形变场；
    局部感知模块，用于通过可形变卷积学习所述脑影像的局部形变，对所 述脑区结构形变场的局部形变差异进行调整，得到优化后的所述脑区结构形变场；
    加权编码模块，用于将优化后的所述脑区结构形变场作用于脑影像，得到用于指示脑区分布的移动图像，并通过约束移动图像与参考图像之间的相似性，得到与疾病相关的多个脑区的区域信息；
    脑网络重构模块，用于根据各脑区的区域信息进行初始的脑网络构建，并通过在脑网络的子图及其扩散图之间进行图对比学习，对初始的脑网络进行优化，得到重构脑网络，所述重构脑网络用于指示所述样本的脑网络统一表征。
  9. 一种电子设备，其特征在于，包括：至少一个处理器、至少一个存储器、以及至少一条通信总线，其中，
    所述存储器上存储有计算机程序，所述处理器通过所述通信总线读取所述存储器中的所述计算机程序；
    所述计算机程序被所述处理器执行时实现权利要求1至7中任一项所述的脑网络统一表征计算方法。
  10. 一种存储介质，其上存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现如权利要求1至7中任一项所述的脑网络统一表征计算方法。
PCT/CN2022/136710 2022-12-05 2022-12-05 脑网络统一表征计算方法、装置、电子设备及存储介质 Ceased WO2024119337A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/136710 WO2024119337A1 (zh) 2022-12-05 2022-12-05 脑网络统一表征计算方法、装置、电子设备及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/136710 WO2024119337A1 (zh) 2022-12-05 2022-12-05 脑网络统一表征计算方法、装置、电子设备及存储介质

Publications (1)

Publication Number Publication Date
WO2024119337A1 true WO2024119337A1 (zh) 2024-06-13

Family

ID=91378404

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/136710 Ceased WO2024119337A1 (zh) 2022-12-05 2022-12-05 脑网络统一表征计算方法、装置、电子设备及存储介质

Country Status (1)

Country Link
WO (1) WO2024119337A1 (zh)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118410283A (zh) * 2024-07-01 2024-07-30 小舟科技有限公司 跨域脑电特征分布对齐方法、装置、计算机设备及介质
CN119027680A (zh) * 2024-07-25 2024-11-26 南京航空航天大学 一种基于三元组的高阶个体差异脑网络构建方法
CN119538086A (zh) * 2024-10-21 2025-02-28 北京理工大学 一种基于全局-局部时空语义对齐的多模态情绪识别系统
CN119672291A (zh) * 2024-11-07 2025-03-21 北京理工大学 一种基于多视角自适应特征对齐的光学遥感图像目标检测域适应方法
CN119691571A (zh) * 2024-09-10 2025-03-25 北京理工大学 一种基于脑网络约束和图对比学习的内部言语解码方法
CN120525876A (zh) * 2025-07-23 2025-08-22 南京信息工程大学 基于结构和功能mri影像多模态融合的脑龄预测方法
CN120766980A (zh) * 2025-09-09 2025-10-10 山东建筑大学 一种基于双编码器与扩散模型的脑疾病预测方法
CN121032949A (zh) * 2025-08-14 2025-11-28 烟台大学 影像组学脑网络到形态学脑网络的预测方法、系统和设备
CN120689619B (zh) * 2025-06-14 2026-05-05 中国人民解放军总医院第一医学中心 基于深度学习的三维脑网络动态分割方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100215239A1 (en) * 2009-02-26 2010-08-26 Ramot At Tel Aviv University Ltd. Method and system for characterizing cortical structures
CN112802013A (zh) * 2021-03-12 2021-05-14 清华大学 基于图神经网络与多任务学习的脑疾病检测方法和装置
CN113096127A (zh) * 2021-06-04 2021-07-09 壹药网科技(上海)股份有限公司 一种脑网络演化模型的生成系统及方法
CN113160138A (zh) * 2021-03-24 2021-07-23 山西大学 一种脑部核磁共振图像分割方法及系统
CN113616184A (zh) * 2021-06-30 2021-11-09 北京师范大学 基于多模态磁共振图像的脑网络建模与个体预测方法
EP3940637A1 (en) * 2020-07-15 2022-01-19 Siemens Healthcare GmbH Method and system for characterizing an impact of brain lesions on brain connectivity using mri

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100215239A1 (en) * 2009-02-26 2010-08-26 Ramot At Tel Aviv University Ltd. Method and system for characterizing cortical structures
EP3940637A1 (en) * 2020-07-15 2022-01-19 Siemens Healthcare GmbH Method and system for characterizing an impact of brain lesions on brain connectivity using mri
CN112802013A (zh) * 2021-03-12 2021-05-14 清华大学 基于图神经网络与多任务学习的脑疾病检测方法和装置
CN113160138A (zh) * 2021-03-24 2021-07-23 山西大学 一种脑部核磁共振图像分割方法及系统
CN113096127A (zh) * 2021-06-04 2021-07-09 壹药网科技(上海)股份有限公司 一种脑网络演化模型的生成系统及方法
CN113616184A (zh) * 2021-06-30 2021-11-09 北京师范大学 基于多模态磁共振图像的脑网络建模与个体预测方法

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118410283A (zh) * 2024-07-01 2024-07-30 小舟科技有限公司 跨域脑电特征分布对齐方法、装置、计算机设备及介质
CN119027680A (zh) * 2024-07-25 2024-11-26 南京航空航天大学 一种基于三元组的高阶个体差异脑网络构建方法
CN119691571A (zh) * 2024-09-10 2025-03-25 北京理工大学 一种基于脑网络约束和图对比学习的内部言语解码方法
CN119538086A (zh) * 2024-10-21 2025-02-28 北京理工大学 一种基于全局-局部时空语义对齐的多模态情绪识别系统
CN119672291A (zh) * 2024-11-07 2025-03-21 北京理工大学 一种基于多视角自适应特征对齐的光学遥感图像目标检测域适应方法
CN120689619B (zh) * 2025-06-14 2026-05-05 中国人民解放军总医院第一医学中心 基于深度学习的三维脑网络动态分割方法
CN120525876A (zh) * 2025-07-23 2025-08-22 南京信息工程大学 基于结构和功能mri影像多模态融合的脑龄预测方法
CN121032949A (zh) * 2025-08-14 2025-11-28 烟台大学 影像组学脑网络到形态学脑网络的预测方法、系统和设备
CN120766980A (zh) * 2025-09-09 2025-10-10 山东建筑大学 一种基于双编码器与扩散模型的脑疾病预测方法

Similar Documents

Publication Publication Date Title
WO2024119337A1 (zh) 脑网络统一表征计算方法、装置、电子设备及存储介质
CN109344889B (zh) 脑疾病分类装置、用户终端和计算机可读存储介质
CN110838108B (zh) 基于医疗图像的预测模型构建方法、预测方法及装置
CN111738363B (zh) 基于改进的3d cnn网络的阿尔茨海默病分类方法
JP7611416B2 (ja) リソグラフィマスク生成モデルの訓練方法、装置、機器、及びコンピュータプログラム
Hao et al. Hypergraph convolutional network for longitudinal data analysis in Alzheimer's disease
CN117274662A (zh) 一种改进ResNeXt神经网络的轻量级多模态医学图像分类方法
Gutiérrez-Becker et al. Deep multi-structural shape analysis: application to neuroanatomy
CN114782768A (zh) 预训练网络模型的训练方法、医学图像处理方法和设备
Ma et al. Unsupervised deformable image registration network for 3D medical images
Kim Implementing a hybrid quantum-classical neural network by utilizing a variational quantum circuit for detection of dementia
He et al. Individualized multi-horizon MRI trajectory prediction for Alzheimer’s disease
CN113822323A (zh) 脑部扫描图像的识别处理方法、装置、设备及存储介质
Wang et al. Diegraph: dual-branch information exchange graph convolutional network for deformable medical image registration
Khazane et al. DeepDefacer: automatic removal of facial features via U-Net image segmentation
CN119206334B (zh) 图像分类方法及装置、计算机程序产品、电子设备
CN120374573A (zh) 一种肺炎dr图像的快速语义分割系统及方法
CN120544189A (zh) 一种子宫内膜癌细胞检测方法及系统
CN118411561A (zh) 一种基于视觉自注意力模型的图像分类方法、装置及产品
CN116188367B (zh) 脑网络统一表征计算方法、装置、电子设备及存储介质
CN118887198A (zh) 一种胃部病理切片分割方法及系统
KR20220023841A (ko) 알츠하이머 병 분류를 위한 mr 이미지 분석 시스템 및 방법
CN118351275A (zh) 体外三维器官模型成长分析方法及相关设备
Pálsson et al. Semi-supervised variational autoencoder for survival prediction
CN116205915A (zh) 一种基于掩码的大脑年龄评估方法、系统及电子设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22967499

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22967499

Country of ref document: EP

Kind code of ref document: A1