A mild cognitive impairment auxiliary diagnosis method based on multi-modal causal feature coupling

Abstract

The application discloses a mild cognitive impairment auxiliary diagnosis method and device based on multi-modal causal feature coupling, and relates to the technical field of neural image intelligent analysis. The method comprises the following steps: based on an inter-brain region connection analysis method, constructing a training data set according to original DTI data, original resting-state fMRI data and a preset mild cognitive impairment label; based on a multi-modal Granger causal decoupling loss function, optimizing and training a multi-modal Granger causal decoupling module according to the training data set, and obtaining an optimized causal decoupling module and a causal factor; based on a cross-entropy loss function, optimizing and training a multi-modal causal feature coupling module according to the training data set and the causal factor, and obtaining an optimized causal feature coupling module; and performing auxiliary diagnosis based on the optimized causal decoupling module and the optimized causal feature coupling module. The application is a mild cognitive impairment auxiliary diagnosis method based on multi-modal deep fusion information, which has strong explainability and strong robustness.

Description

Technical Field

[0001] This invention relates to the field of intelligent neuroimaging analysis technology, and in particular to an auxiliary diagnostic method and device for mild cognitive impairment based on multimodal causal feature coupling. Background Technology

[0002] In recent years, deep learning technology has made significant progress in medical image analysis and auxiliary diagnosis of neurological diseases. For classification tasks of Mild Cognitive Impairment (MCI) and Alzheimer's Disease (AD), studies based on multimodal brain network analysis have found that MCI patients exhibit specific abnormal patterns in both structural connectivity (SC) and functional connectivity (FC). The coupling relationship between SC and FC reflects the interdependence between structure and function, and is an important basis for distinguishing different disease states.

[0003] While deep learning models demonstrate high classification accuracy in relevant tasks, the decision-making process is often considered a "black box," lacking clear causal explanations and visualization mechanisms. Although existing deep learning models can automatically learn high-dimensional and complex features, their decision-making processes lack traceable causal explanations and transparency. In clinical applications, physicians typically need to understand which specific brain regions, connectivity patterns, or biological mechanisms the model bases its judgments on. However, most existing methods only provide classification results, failing to offer verifiable feature importance analysis or biological explanations, leading to insufficient clinical trust and limiting their widespread application in diagnostic and treatment decisions.

[0004] Multimodal imaging, such as diffusion tensor imaging and functional magnetic resonance imaging (fMRI), has been widely used in the clinical diagnosis of brain diseases, providing complementary structural and functional information. However, existing multimodal fusion methods mostly remain at the feature level, performing simple stitching and weighted summation operations, lacking modeling of the potential nonlinear dependencies and causal relationships between SC and FC. This shallow fusion approach often results in extracted multimodal features lacking biological relevance, making it difficult to effectively characterize high-order coupling patterns between brain structure and function, thus affecting the accuracy and stability of classification.

[0005] Furthermore, while some studies have introduced deep learning models to learn features from multimodal brain networks, most methods still rely on statistical correlation modeling to establish the relationship between features and diseases, lacking the characterization and utilization of potential causal effects. Structural and functional images differ significantly in acquisition methods, temporal resolution, spatial resolution, and physiological basis, and image data inevitably contains noise and individual differences. Most existing technologies rely on relatively simple methods such as stitching, weighted summation, or self-attention to integrate multimodal brain connectivity features for the auxiliary diagnosis of mild cognitive impairment. However, they lack effective noise reduction and heterogeneity decoupling mechanisms, resulting in insufficient robustness of models across centers, devices, or populations.

[0006] In the existing technology, there is a lack of an auxiliary diagnostic method for mild cognitive impairment that is highly interpretable and robust based on multimodal deep fusion information. Summary of the Invention

[0007] To address the limitations of existing multimodal fusion methods, such as limited fusion effectiveness, lack of interpretability in the decision-making process, and insufficient robustness due to the lack of effective denoising and heterogeneity decoupling mechanisms, this invention provides a method and device for auxiliary diagnosis of mild cognitive impairment based on multimodal causal feature coupling. The technical solution is as follows:

[0008] On the one hand, a method for assisting in the diagnosis of mild cognitive impairment based on multimodal causal feature coupling is provided. This method is implemented by a device for assisting in the diagnosis of mild cognitive impairment and includes:

[0009] Acquire raw DTI data and raw resting-state fMRI data;

[0010] Based on the brain region connectivity analysis method, a training dataset was constructed using raw DTI data, raw resting-state fMRI data, and pre-defined mild cognitive impairment labels.

[0011] Based on the multimodal Granger causal decoupling loss function, the multimodal Granger causal decoupling module is optimized and trained according to the training dataset to obtain the optimized causal decoupling module and causal factors;

[0012] Based on the cross-entropy loss function, the multimodal causal feature coupling module is optimized and trained according to the training dataset and causal factors to obtain the optimized causal feature coupling module.

[0013] The system acquires DTI data and resting-state fMRI data of the patients to be evaluated; based on the optimized causal decoupling module and the optimized causal feature coupling module, it performs classification and prediction based on the DTI data and resting-state fMRI data to obtain auxiliary diagnostic results for patients with mild cognitive impairment.

[0014] On the other hand, a diagnostic aid for mild cognitive impairment based on multimodal causal feature coupling is provided. This device is applied to a diagnostic aid for mild cognitive impairment based on multimodal causal feature coupling. The device includes:

[0015] The data acquisition module is used to acquire raw DTI data and raw resting-state fMRI data;

[0016] The data preprocessing module is used to construct a training dataset based on the brain region connectivity analysis method, using raw DTI data, raw resting-state fMRI data, and preset mild cognitive impairment labels.

[0017] The first training module is used to optimize the multimodal Granger causal decoupling module based on the multimodal Granger causal decoupling loss function and the training dataset, so as to obtain the optimized causal decoupling module and causal factors.

[0018] The second training module is used to optimize the multimodal causal feature coupling module based on the cross-entropy loss function, the training dataset, and causal factors, to obtain the optimized causal feature coupling module.

[0019] The auxiliary diagnostic module is used to acquire DTI data and resting-state fMRI data of the patient to be evaluated; based on the optimized causal decoupling module and the optimized causal feature coupling module, classification and prediction are performed according to the DTI data and resting-state fMRI data to obtain auxiliary diagnostic results for patients with mild cognitive impairment.

[0020] On the other hand, a mild cognitive impairment auxiliary diagnostic device is provided, the mild cognitive impairment auxiliary diagnostic device comprising: a processor; a memory, the memory storing computer-readable instructions, which, when executed by the processor, implement any of the methods in the above-described mild cognitive impairment auxiliary diagnostic method based on multimodal causal feature coupling.

[0021] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, the at least one instruction being loaded and executed by a processor to implement any of the above-described auxiliary diagnostic methods for mild cognitive impairment based on multimodal causal feature coupling.

[0022] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0023] This invention proposes an auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling. It decouples latent features based on a multimodal Granger causal decoupling module, uses a Hodge-Laplacian encoder to perform multi-order causal feature coupling, and obtains fusion features rich in node features and causal connection information. A classifier is then used to achieve accurate prediction of mild cognitive impairment.

[0024] This invention enables the simultaneous capture of potential causal relationships between structural and functional connections in the brain, and decouples effective causal features related to disease classification, thereby enhancing the interpretability of the model and the accuracy of MCI classification. The multi-order Hodge-Laplacian encoder fully utilizes high-order topological information of the brain network, improving the efficiency and completeness of causal information propagation within the graph. The overall method achieves effective coupling of structural and functional connections, contributing to a more accurate understanding of brain structure and functional mechanisms. This invention is a highly interpretable and robust auxiliary diagnostic method for mild cognitive impairment based on multimodal deep fusion information. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of an auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling provided by an embodiment of the present invention;

[0027] Figure 2 This is a block diagram of a mild cognitive impairment auxiliary diagnostic device based on multimodal causal feature coupling provided in an embodiment of the present invention;

[0028] Figure 3 This is a schematic diagram of the structure of an auxiliary diagnostic device for mild cognitive impairment provided in an embodiment of the present invention. Detailed Implementation

[0029] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0030] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0031] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0032] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0033] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0034] This invention provides an auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling. This method can be implemented using an auxiliary diagnostic device for mild cognitive impairment, which can be a terminal or a server. Figure 1 The flowchart shown is for an auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling. The processing flow of this method may include the following steps:

[0035] S1. Acquire raw DTI data and raw resting-state fMRI data;

[0036] In one feasible implementation, the original data of the present invention comes from the original diffusion tensor imaging (DTI) data and resting-state functional magnetic resonance imaging (fMRI) data of subjects collected in hospitals or from publicly available datasets.

[0037] S2. Based on the brain region connectivity analysis method, a training dataset is constructed according to the original DTI data, the original resting-state fMRI data and the preset mild cognitive impairment labels;

[0038] Optionally, based on brain region connectivity analysis, a training dataset is constructed from raw DTI data, raw resting-state fMRI data, and pre-defined mild cognitive impairment labels, including:

[0039] The raw DTI data and raw resting-state fMRI data were preprocessed to obtain processed DTI data and processed resting-state fMRI data.

[0040] Based on the brain region fiber tract counting method, a structural connectivity matrix is ​​constructed from the processed DTI data;

[0041] The structural connection matrix is ​​normalized to obtain the structural node feature matrix;

[0042] Based on the brain region correlation coefficient calculation method, a functional connectivity matrix was constructed from the processed resting-state fMRI data.

[0043] The functional connection matrix is ​​normalized to obtain the functional node feature matrix;

[0044] A training dataset is constructed based on the structural connectivity matrix, functional connectivity matrix, structural node feature matrix, functional node feature matrix, and pre-defined mild cognitive impairment labels.

[0045] In one feasible implementation, the present invention performs a conventional preprocessing procedure on the raw DTI data, including head motion correction, construction of fiber tracing atlases, and partitioning using Anatomical Automatic Labeling (AAL) or other standard brain region templates; the number of fiber bundles between each pair of brain regions is calculated to obtain a structural connectivity matrix. .

[0046] For the original resting-state fMRI dataset, time-slice correction, head motion correction, spatial normalization, denoising, and bandpass filtering were performed. The mean blood-oxygen-level dependent (BOLD) signal of each brain region was extracted, and the correlation coefficient between brain region pairs was calculated to obtain the functional connectivity matrix. .

[0047] right and Normalization is performed as the feature matrix of the structural nodes. and functional node feature matrix .

[0048] Based on the results of the above steps, a training dataset is constructed using the pre-defined mild cognitive impairment label Y. The training dataset is divided into a training set and a test set in an 8:2 ratio to ensure that subsequent training and evaluation do not interfere with each other. The training set (… , , , , ) and test set ( , , , , ).

[0049] S3. Based on the multimodal Granger causal decoupling loss function, the multimodal Granger causal decoupling module is optimized and trained according to the training dataset to obtain the optimized causal decoupling module and causal factors.

[0050] The multimodal Granger causal decoupling module includes a feature extraction module and a causal decoupling module.

[0051] The feature extraction module includes a brain structure connectivity graph encoder and a brain function connectivity graph encoder; the brain structure connectivity graph encoder includes two graph convolutional network layers, one normalization layer and one activation function layer; the structure of the brain function connectivity graph encoder is the same as that of the brain structure connectivity graph encoder.

[0052] The causal decoupling module includes a causal factor perceptron and a non-causal factor perceptron.

[0053] In one feasible implementation, the multimodal Granger causality decoupling module aims to decouple potential brain connectivity patterns that are relevant to and unrelated to the target task (i.e., the predefined mild cognitive impairment label Y, representing the MCI category). Therefore, this module decouples the SC and FC causal connectivity patterns associated with MCI diagnosis, enhancing the model's interpretability.

[0054] The multimodal Granger causality decoupling module comprises a feature extraction module and a causality decoupling module. The feature extraction module includes a brain structural connectivity graph encoder and a brain functional connectivity graph encoder. Each graph encoder consists of a Graph Convolutional Network (GCN) layer, a LayerNorm normalization layer, and a LeakyReLU activation function layer. The causality decoupling module consists of two multilayer perceptrons, each responsible for perceiving the relationship between causal and non-causal factors.

[0055] Granger causality analysis decouples task-related structural and functional connectivity features while eliminating noise and non-causal features. This technique is a core means of solving the heterogeneity and noise interference of multimodal data. Without this feature, the task relevance and stability of features cannot be guaranteed.

[0056] Optionally, based on the multimodal Granger causal decoupling loss function, the multimodal Granger causal decoupling module is optimized and trained according to the training dataset to obtain the optimized causal decoupling module and causal factors, including:

[0057] Feature extraction is performed based on training data to obtain latent features; latent features include latent brain structural connectivity features and latent brain functional connectivity features.

[0058] Causal relationships are decoupled based on latent characteristics to obtain causal and non-causal factors; causal factors include structurally connected causal factors and functionally connected causal factors; non-causal factors include structurally connected non-causal factors and functionally connected non-causal factors.

[0059] Based on the multimodal Granger causal decoupling loss function, the causal decoupling loss is calculated according to causal and non-causal factors.

[0060] Based on the causal decoupling loss, the parameters of the multimodal Granger causal decoupling module are optimized to obtain an optimized causal decoupling module.

[0061] In one feasible implementation, the GCN operation extends the traditional convolution operator to graph data by defining filters in the graph domain. The computation process for extracting latent features can be expressed as equations (1) and (2):

[0062] (1);

[0063] (2);

[0064] in, Indicates feature normalization; For activation functions; Represented as Feature normalization; for Feature normalization; This is the first learnable parameter matrix.

[0065] The latent structural features of SC and FC are obtained through a graph encoder. and potential functional characteristics .

[0066] This invention designs a multimodal Granger causality decoupling loss function to separate features that are related to and unrelated to MCI. Latent features and by potential structural causal factors and functional causal factors and structural non-causal factors and functional non-causal factors It contains only causal factors. and It has a causal relationship with label Y.

[0067] The initially separated causal and non-causal factors are decoupled using multimodal Granger causality decoupling loss. Optimization. The method proposed in this invention needs to ensure that on label Y, and and and They are independent of each other, and and It has a direct causal effect on label Y.

[0068] Therefore, the multimodal Granger causal decoupling loss function is shown in equation (3) below:

[0069] (3);

[0070] in, This indicates mutual information computation.

[0071] The parameters of the GCN and the decoupled module are updated using gradient descent until convergence.

[0072] S4. Based on the cross-entropy loss function, the multimodal causal feature coupling module is optimized and trained according to the training dataset and causal factors to obtain the optimized causal feature coupling module.

[0073] The multimodal causal feature coupling module includes a feature coupler and a classifier.

[0074] The feature coupler is a Hodge-Laplace encoder;

[0075] The classifier is a multilayer perceptron.

[0076] In one feasible implementation, in order to better fuse the decoupled multimodal causal features and enhance the information transmission capability of causal structural connections, this invention employs a Hodge-Laplacian encoder. This encoder adjusts the propagation of information in multi-level convolutions to obtain fused features. This technology effectively captures structural information at different levels within the image. This feature ensures a balance between classification performance and causal interpretability, and is a key element in enabling the clinical auxiliary diagnostic application of this invention.

[0077] Optionally, based on the cross-entropy loss function, and according to the training dataset and causal factors, the multimodal causal feature coupling module is optimized and trained to obtain an optimized causal feature coupling module, including:

[0078] Based on the training dataset and causal factors, multi-level graph structure information is captured to obtain fused features;

[0079] A fully connected nonlinear mapping is performed based on the fusion features to obtain the classification result;

[0080] Based on the cross-entropy loss function, the cross-entropy loss is calculated according to the training data and classification results.

[0081] Based on the cross-entropy loss, the parameters of the multimodal causal feature coupling module are optimized to obtain the optimized causal feature coupling module.

[0082] In one feasible implementation, the core of the Hodge-Laplacian encoder operation is to propagate node features through the adjacency matrix, thereby incorporating causal factors. and As an adjacency matrix, the node features are those from the training dataset. and For each The order is calculated as follows: (4), (5), (6):

[0083] (4);

[0084] (5);

[0085] (6);

[0086] in, Represents a Laguerre polynomial. Indicates the first ( ) Node features; This is the second learnable parameter matrix.

[0087] The fused features are input into a multilayer perceptron (MLP), trained using the cross-entropy loss function, and the calculated loss is used to back-optimize the parameters in the multimodal causal fusion module and the classifier module.

[0088] S5. Obtain DTI data and resting-state fMRI data of the patient to be evaluated; based on the optimized causal decoupling module and the optimized causal feature coupling module, perform classification and prediction based on DTI data and resting-state fMRI data to obtain auxiliary diagnostic results for patients with mild cognitive impairment.

[0089] In one feasible implementation, the present invention uses test set data to calculate metrics such as accuracy, recall, and area under the curve (AUC) for performance evaluation. The present invention includes a complete implementation process encompassing multimodal graph feature encoding, causal decoupling, and causal feature coupling to classification prediction; its overall technical features demonstrate the systematic nature and feasibility of the present invention, providing a comprehensive guarantee for achieving its technical effects.

[0090] This invention proposes an auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling. It decouples latent features based on a multimodal Granger causal decoupling module, uses a Hodge-Laplacian encoder to perform multi-order causal feature coupling, and obtains fusion features rich in node features and causal connection information. A classifier is then used to achieve accurate prediction of mild cognitive impairment.

[0091] This invention enables the simultaneous capture of potential causal relationships between structural and functional connections, and decouples effective causal features related to disease classification, thereby enhancing the interpretability of the model and the accuracy of MCI classification. The multi-order Hodge-Laplacian encoder fully utilizes high-order topological information of the brain network, improving the efficiency and completeness of causal information propagation within the graph. The overall approach achieves effective coupling from structural to functional connections, contributing to a more accurate understanding of brain functional mechanisms. This invention is a highly interpretable and robust auxiliary diagnostic method for mild cognitive impairment based on multimodal deep fusion information.

[0092] Figure 2 This is a block diagram of an auxiliary diagnostic device for mild cognitive impairment based on multimodal causal feature coupling, provided by an embodiment of the present invention. This device is used in an auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling. (Refer to...) Figure 2 The device includes a data acquisition module 210, a data preprocessing module 220, a first training module 230, a second training module 240, and an auxiliary diagnostic module 250. Wherein:

[0093] Data acquisition module 210 is used to acquire raw DTI data and raw resting-state fMRI data;

[0094] The data preprocessing module 220 is used to construct a training dataset based on the original DTI data, the original resting-state fMRI data and the preset mild cognitive impairment labels, using the brain region connectivity analysis method.

[0095] The first training module 230 is used to optimize the multimodal Granger causal decoupling module based on the multimodal Granger causal decoupling loss function and the training dataset to obtain the optimized causal decoupling module and causal factors.

[0096] The second training module 240 is used to optimize the training of the multimodal causal feature coupling module based on the cross-entropy loss function, the training dataset, and causal factors, to obtain the optimized causal feature coupling module.

[0097] The auxiliary diagnostic module 250 is used to acquire DTI data and resting-state fMRI data of the patient to be evaluated; based on the optimized causal decoupling module and the optimized causal feature coupling module, it performs classification prediction based on DTI data and resting-state fMRI data to obtain auxiliary diagnostic results for patients with mild cognitive impairment.

[0098] Optionally, the data preprocessing module 220 is further used for:

[0099] The raw DTI data and raw resting-state fMRI data were preprocessed to obtain processed DTI data and processed resting-state fMRI data.

[0100] Based on the brain region fiber tract counting method, a structural connectivity matrix is ​​constructed from the processed DTI data;

[0101] The structural connection matrix is ​​normalized to obtain the structural node feature matrix;

[0102] Based on the brain region correlation coefficient calculation method, a functional connectivity matrix was constructed from the processed resting-state fMRI data.

[0103] The functional connection matrix is ​​normalized to obtain the functional node feature matrix;

[0104] A training dataset is constructed based on the structural connectivity matrix, functional connectivity matrix, structural node feature matrix, functional node feature matrix, and pre-defined mild cognitive impairment labels.

[0105] The multimodal Granger causal decoupling module includes a feature extraction module and a causal decoupling module.

[0106] The feature extraction module includes a brain structure connectivity graph encoder and a brain function connectivity graph encoder; the brain structure connectivity graph encoder includes two graph convolutional network layers, one normalization layer, and one activation function layer; the structure of the brain function connectivity graph encoder is the same as that of the brain structure connectivity graph encoder.

[0107] The causal decoupling module includes a causal factor perceptron and a non-causal factor perceptron.

[0108] Optionally, the first training module 230 is further used for:

[0109] Feature extraction is performed based on training data to obtain latent features; latent features include latent brain structural connectivity features and latent brain functional connectivity features.

[0110] Causal relationships are decoupled based on latent characteristics to obtain causal and non-causal factors; causal factors include structurally connected causal factors and functionally connected causal factors; non-causal factors include structurally connected non-causal factors and functionally connected non-causal factors.

[0111] Based on the multimodal Granger causal decoupling loss function, the causal decoupling loss is calculated according to causal and non-causal factors.

[0112] Based on the causal decoupling loss, the parameters of the multimodal Granger causal decoupling module are optimized to obtain an optimized causal decoupling module.

[0113] The multimodal causal feature coupling module includes a feature coupler and a classifier.

[0114] The feature coupler is a Hodge-Laplace encoder;

[0115] The classifier is a multilayer perceptron.

[0116] Optionally, the second training module 240 is further used for:

[0117] Based on the training dataset and causal factors, multi-level graph structure information is captured to obtain fused features;

[0118] A fully connected nonlinear mapping is performed based on the fusion features to obtain the classification result;

[0119] Based on the cross-entropy loss function, the cross-entropy loss is calculated according to the training data and classification results.

[0120] Based on the cross-entropy loss, the parameters of the multimodal causal feature coupling module are optimized to obtain the optimized causal fusion module.

[0121] This invention proposes an auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling. It decouples latent features based on a multimodal Granger causal decoupling module, uses a Hodge-Laplacian encoder to perform multi-order causal feature coupling, and obtains fusion features rich in node features and causal connection information. A classifier is then used to achieve accurate prediction of mild cognitive impairment.

[0122] This invention enables the simultaneous capture of potential causal relationships between structural and functional connections, and decouples effective causal features related to disease classification, thereby enhancing the interpretability of the model and the accuracy of MCI classification. The multi-order Hodge-Laplacian encoder fully utilizes high-order topological information of the brain network, improving the efficiency and completeness of causal information propagation within the graph. The overall approach achieves effective coupling from structural to functional connections, contributing to a more accurate understanding of brain functional mechanisms. This invention is a highly interpretable and robust auxiliary diagnostic method for mild cognitive impairment based on multimodal deep fusion information.

[0123] Figure 3 This is a schematic diagram of the structure of an auxiliary diagnostic device for mild cognitive impairment provided in an embodiment of the present invention, as shown below. Figure 3As shown, the auxiliary diagnostic device for mild cognitive impairment may include the above-mentioned Figure 2 The illustrated auxiliary diagnostic device for mild cognitive impairment based on multimodal causal feature coupling. Optionally, the auxiliary diagnostic device 310 for mild cognitive impairment may include a first processor 2001.

[0124] Optionally, the mild cognitive impairment assistive diagnostic device 310 may also include a memory 2002 and a transceiver 2003.

[0125] The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus.

[0126] The following is combined Figure 3 A detailed introduction to each component of the mild cognitive impairment auxiliary diagnostic device 310:

[0127] The first processor 2001 is the control center of the mild cognitive impairment auxiliary diagnostic device 310. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0128] Optionally, the first processor 2001 can perform various functions of the mild cognitive impairment auxiliary diagnostic device 310 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0129] In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, for example... Figure 3 CPU0 and CPU1 are shown in the diagram.

[0130] In a specific implementation, as one example, the mild cognitive impairment assistive diagnostic device 310 may also include multiple processors, such as... Figure 3The first processor 2001 and the second processor 2004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, a processor can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).

[0131] The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0132] Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently, and may be connected via the interface circuit of the mild cognitive impairment assistive diagnostic device 310. Figure 3 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0133] The transceiver 2003 is used to communicate with network devices or with terminal devices.

[0134] Alternatively, transceiver 2003 may include a receiver and a transmitter. Figure 3 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0135] Optionally, the transceiver 2003 can be integrated with the first processor 2001, or it can exist independently and be connected to the interface circuit of the mild cognitive impairment auxiliary diagnostic device 310. Figure 3 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0136] It should be noted that, Figure 3 The structure of the mild cognitive impairment assistive diagnostic device 310 shown in the figure does not constitute a limitation on the router. Actual mild cognitive impairment assistive diagnostic devices may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0137] Furthermore, the technical effects of the mild cognitive impairment auxiliary diagnostic device 310 can be referred to the technical effects of the mild cognitive impairment auxiliary diagnostic method based on multimodal causal feature coupling described in the above method embodiments, and will not be repeated here.

[0138] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or it may be any conventional processor, etc.

[0139] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0140] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0141] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0142] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0143] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0144] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0145] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0146] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0147] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0148] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0149] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0150] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for assisting in the diagnosis of mild cognitive impairment based on multimodal causal feature coupling, characterized in that, The method includes: Acquire raw DTI data and raw resting-state fMRI data; Based on the brain region connectivity analysis method, a training dataset was constructed using raw DTI data, raw resting-state fMRI data, and pre-defined mild cognitive impairment labels. Based on the multimodal Granger causal decoupling loss function, the multimodal Granger causal decoupling module is optimized and trained using the training dataset to obtain the optimized causal decoupling module and causal factors, including: Feature extraction is performed based on training data to obtain latent features; the latent features include latent brain structural connectivity features and latent brain functional connectivity features. Causal relationships are decoupled based on latent characteristics to obtain causal factors and non-causal factors; the causal factors include structurally connected causal factors and functionally connected causal factors; the non-causal factors include structurally connected non-causal factors and functionally connected non-causal factors. Based on the multimodal Granger causal decoupling loss function, the causal decoupling loss is calculated according to causal and non-causal factors. Based on the causal decoupling loss, the parameters of the multimodal Granger causal decoupling module are optimized to obtain the optimized causal decoupling module. The multimodal Granger causal decoupling module includes a feature extraction module and a causal decoupling module. The feature extraction module includes a brain structure connectivity graph encoder and a brain function connectivity graph encoder; the brain structure connectivity graph encoder includes two graph convolutional network layers, one normalization layer, and one activation function layer; the structure of the brain function connectivity graph encoder is the same as that of the brain structure connectivity graph encoder. The causal decoupling module includes a causal factor perceptron and a non-causal factor perceptron. Based on the cross-entropy loss function, the multimodal causal feature coupling module is optimized and trained according to the training dataset and causal factors to obtain the optimized causal feature coupling module. The multimodal causal feature coupling module includes a feature coupler and a classifier. The feature coupler is a Hodge-Laplac encoder; The classifier is a multilayer perceptron; The system acquires DTI data and resting-state fMRI data of the patients to be evaluated; based on the optimized causal decoupling module and the optimized causal feature coupling module, it performs classification and prediction based on the DTI data and resting-state fMRI data to obtain auxiliary diagnostic results for patients with mild cognitive impairment.

2. The auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling according to claim 1, characterized in that, The method based on brain region connectivity analysis constructs a training dataset based on raw DTI data, raw resting-state fMRI data, and pre-defined mild cognitive impairment labels, including: The raw DTI data and raw resting-state fMRI data were preprocessed to obtain processed DTI data and processed resting-state fMRI data. Based on the brain region fiber tract counting method, a structural connectivity matrix is ​​constructed from the processed DTI data; The structural connection matrix is ​​normalized to obtain the structural node feature matrix; Based on the brain region correlation coefficient calculation method, a functional connectivity matrix was constructed from the processed resting-state fMRI data. The functional connection matrix is ​​normalized to obtain the functional node feature matrix; A training dataset is constructed based on the structural connectivity matrix, functional connectivity matrix, structural node feature matrix, functional node feature matrix, and pre-defined mild cognitive impairment labels.

3. The auxiliary diagnostic method for mild cognitive impairment based on multimodal causal feature coupling according to claim 1, characterized in that, The method involves optimizing the multimodal causal feature coupling module based on the cross-entropy loss function, the training dataset, and causal factors, to obtain an optimized causal feature coupling module, including: Based on the training dataset and causal factors, multi-level graph structure information is captured to obtain fused features; A fully connected nonlinear mapping is performed based on the fusion features to obtain the classification result; Based on the cross-entropy loss function, the cross-entropy loss is calculated according to the training data and classification results. Based on the cross-entropy loss, the parameters of the multimodal causal feature coupling module are optimized to obtain the optimized causal feature coupling module.

4. A diagnostic aid for mild cognitive impairment based on multimodal causal feature coupling, wherein the diagnostic aid for mild cognitive impairment based on multimodal causal feature coupling is used to implement the diagnostic aid for mild cognitive impairment based on multimodal causal feature coupling as described in any one of claims 1-3, characterized in that, The device includes: The data acquisition module is used to acquire raw DTI data and raw resting-state fMRI data; The data preprocessing module is used to construct a training dataset based on the brain region connectivity analysis method, using raw DTI data, raw resting-state fMRI data, and preset mild cognitive impairment labels. The first training module is used to optimize the multimodal Granger causal decoupling module based on the multimodal Granger causal decoupling loss function and the training dataset, so as to obtain the optimized causal decoupling module and causal factors. The second training module is used to optimize the multimodal causal feature coupling module based on the cross-entropy loss function, the training dataset, and causal factors, to obtain the optimized causal feature coupling module. The auxiliary diagnostic module is used to acquire DTI data and resting-state fMRI data of the patient to be evaluated; based on the optimized causal decoupling module and the optimized causal feature coupling module, classification and prediction are performed according to the DTI data and resting-state fMRI data to obtain auxiliary diagnostic results for patients with mild cognitive impairment.

5. The mild cognitive impairment auxiliary diagnostic device based on multimodal causal feature coupling according to claim 4, characterized in that, The data preprocessing module is further used for: The raw DTI data and raw resting-state fMRI data were preprocessed to obtain processed DTI data and processed resting-state fMRI data. Based on the brain region fiber tract counting method, a structural connectivity matrix is ​​constructed from the processed DTI data; The structural connection matrix is ​​normalized to obtain the structural node feature matrix; Based on the brain region correlation coefficient calculation method, a functional connectivity matrix was constructed from the processed resting-state fMRI data. The functional connection matrix is ​​normalized to obtain the functional node feature matrix; A training dataset is constructed based on the structural connectivity matrix, functional connectivity matrix, structural node feature matrix, functional node feature matrix, and pre-defined mild cognitive impairment labels.

6. A diagnostic aid for mild cognitive impairment, characterized in that, The mild cognitive impairment auxiliary diagnostic device includes: processor; A memory storing computer-readable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 3.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that can be invoked by a processor to execute the method as described in any one of claims 1 to 3.

Images