Classification prediction method of denoised MRI image for ADHD

Abstract

The invention relates to the technical field of bioinformatics and deep learning, and discloses a denoised MRI image classification prediction method for ADHD, and the method comprises the steps: collecting an MRI image of a to-be-detected brain, carrying out the online denoising of the collected MRI image through an online training unsupervised denoising network, and carrying out the classification prediction of the denoised MRI image. And then preprocessing an online denoising processing result, and inputting the preprocessed result into an offline trained DSC-Former model to obtain a classification prediction result. The unsupervised denoising network comprises two parallel processing lines, and each line comprises a downsampling module and a noise estimation model which are connected in series; the DSC-Former model comprises four layers of three-dimensional snakelike convolution modules and a Vision Transform module which are connected in sequence, and a SoftMax layer. According to the method, denoising processing is performed firstly, and then classification prediction is performed, so that the classification prediction effect of the MRI image of the ADHD patient is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the field of bioinformatics and deep learning technology, and specifically relates to a classification prediction method for denoised MRI images of ADHD. Background Art

[0002] Attention Deficit Hyperactivity Disorder (ADHD) is a common neurodevelopmental disorder that affects many children and adults worldwide. Its main characteristics include inattention, hyperactivity, and impulsive behavior. The exact diagnosis of ADHD mainly relies on clinical symptom observation, behavioral scale assessment, and detailed medical history collection, and lacks direct biological markers as a basis for diagnosis. This diagnostic method that relies on subjective judgment often leads to a high misdiagnosis rate, affecting treatment outcomes and patient management.

[0003] At the same time, with the development of medical imaging technology, especially the widespread application of magnetic resonance imaging (MRI) in the field of neuroscience, a new perspective has been provided for exploring the neurobiological basis of ADHD. MRI can reveal changes in brain structure and function non-invasively. Recent studies have found that ADHD patients have differences in brain volume, white matter integrity, and functional connectivity of specific brain regions. However, since MRI images are susceptible to noise caused by various factors, such as motion artifacts and hardware limitations, these factors often affect the quality of imaging data and the accuracy of subsequent analysis.

[0004] Deep learning, as a revolutionary technology in the field of artificial intelligence, has been successfully applied to medical image analysis with its powerful data processing and feature learning capabilities, significantly improving the accuracy of disease diagnosis. By constructing a multi-level neural network model, deep learning automatically extracts weak features related to the disease from complex image data to achieve accurate classification of disease status. For example, the invention application "Assisted Screening of Childhood ADHD Based on Multimodal Appearance Features" with publication number CN118588288A has a data acquisition module for collecting video stream data through a camera; a data preprocessing module for dividing the collected video files into frames and obtaining facial areas of interest; a feature extraction module that uses computer vision technology to analyze head posture, expression and gaze distribution, and performs homogeneous vector fusion; a multimodal information classification module that uses a temporal convolutional network (TCN) to pre-train the fusion vector and detection results within a certain time period, and finally determines the classification result of whether the patient has abnormal behavior in the entire time period. However, existing deep learning methods mostly focus on achieving the accuracy of ADHD classification prediction by adopting improved feature extraction networks and target detection methods. Few methods use denoising to improve the image quality of the input image to further improve the classification effect. Therefore, there is a need to improve the existing technology. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a classification prediction method for denoised MRI images of ADHD, which is used to combine MRI imaging technology with a deep learning algorithm for classification prediction of MRI images of ADHD.

[0006] In order to solve the above technical problems, the present invention provides a classification prediction method for denoised MRI images for ADHD, comprising: collecting MRI images of the brain to be detected, using an online trained unsupervised denoising network to perform online denoising on the collected MRI images, and then pre-processing the results of the online denoising and inputting them into an offline trained DSC-Former model to obtain classification prediction results;

[0007] The unsupervised denoising network includes two parallel processing circuits, each of which includes a downsampling module and a noise estimation model connected in series;

[0008] The DSC-Former model includes four layers of three-dimensional snake convolution modules and VisionTransformer modules connected in sequence, and a SoftMax layer.

[0009] As an improvement of the classification prediction method of denoised MRI images for ADHD of the present invention:

[0010] The downsampling module is a three-dimensional convolution kernel structure;

[0011] The noise estimation model in the unsupervised denoising network is composed of a stack of Vision Transformer and ConvFormer.

[0012] As a further improvement of the classification prediction method of denoised MRI images for ADHD of the present invention:

[0013] The online denoising process of the unsupervised denoising network is:

[0014] Each MRI image is trained independently for 50 epochs using the Adam optimizer. During the training process, the loss function is used, and the two downsampled images are back-propagated to approximate the denoised image obtained by the noise estimation model to another downsampled image, thereby obtaining noise estimates for the two downsampled images. After the training is completed, the two output noise estimates are homogenized, and then the difference between the original input MRI image and the homogenized result is used as the result of the online denoising process:

[0015]

[0016] Among them, θ is the noise estimation model, x is the input image, S1(x) and S2(x) are two downsampled images, and f θ (S(x)) is the two-way noise estimate output by the noise estimation model.

[0017] As a further improvement of the classification prediction method of denoised MRI images for ADHD of the present invention:

[0018] The unsupervised denoising network loss function is:

[0019]

[0020] As a further improvement of the classification prediction method of denoised MRI images for ADHD of the present invention:

[0021] The pre-processing is:

[0022] (1) Mapping the denoised MRI image data to a voxel size of (1,1,1);

[0023] (2) Normalize the intensity value of each voxel and map it to the range of 0-511;

[0024] (3) Crop the non-brain background.

[0025] As a further improvement of the classification prediction method of denoised MRI images for ADHD of the present invention:

[0026] The three-dimensional dynamic snake convolution module in the DSC-Former model is characterized by:

[0027] The module contains four branch paths, which use conventional three-dimensional convolution, dynamic snake convolution in the x-direction, dynamic snake convolution in the y-direction, and dynamic snake convolution in the z-direction respectively. The three dynamic snake convolution modules form receptive fields along the X, Y, and Z axes respectively. The feature vectors extracted by the four different convolution layers are concat-operated to output the corresponding feature representation.

[0028] As a further improvement of the classification prediction method of denoised MRI images for ADHD of the present invention:

[0029] The offline training process of the DSC-Former model is as follows:

[0030] Obtain MRI image data of desensitized ADHD patients and a control group from a public dataset ADHD-200 dataset and a cooperative hospital, then obtain denoised MRI images by subjecting each MRI image to an online training process of the unsupervised denoising network, and then perform preprocessing and data augmentation operations, and randomly divide the augmented dataset into a training set, a validation set, and a test set in a ratio of 8:1:1;

[0031] The MRI images in the training set were input into the DSC-Former model in sequence. DSC-Former was trained using the RMSprop optimizer, and a loss function was constructed based on cross entropy to calculate the loss between the recognition result and the label. After the backward propagation process was completed, the DSC-Former model weights were updated. A 10-fold cross-validation model classification performance was used.

[0032] The beneficial effects of the present invention are mainly reflected in:

[0033] 1. The present invention adopts a method of first denoising and then classification prediction, which effectively improves the classification prediction effect of MRI images of ADHD patients.

[0034] 2. Aiming at the problem that inherent noise sources such as hardware equipment noise affect the imaging quality during MRI image acquisition, the present invention designs an unsupervised MRI image denoising network module, adopts a three-dimensional downsampling module, and performs unsupervised denoising on independent MRI image data. It is applicable to MRI images of any data volume and scale, and can reduce the interference of image noise on feature extraction in subsequent classification network recognition, thereby improving classification accuracy. At the same time, the edges of the image after denoising are clearer, which is convenient for doctors to further observe the brain structured information and make judgments.

[0035] 3. The present invention designs a Transformer-based DSC-Former classification network, and combines convolution with self-attention to construct a feature extraction network. At the same time, according to the characteristics of MRI images, a three-dimensional dynamic snake convolution is used as a network downsampling module, so that the model can dynamically adjust the convolution kernel sampling position according to the input features, which is more suitable for brain MRI image data and further reduces the interference of background on feature extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The specific implementation modes of the present invention are further described in detail below with reference to the accompanying drawings.

[0037] Figure 1 This is the network structure diagram of the DSC-Former proposed by the present invention;

[0038] Figure 2 This is a diagram of the structure of the downsampling three-dimensional convolution kernel of the MRI image denoising network proposed by the present invention;

[0039] Figure 3 This is the structural diagram of the MRI image denoising network proposed by the present invention;

[0040] Figure 4 Schematic diagram of Transformer and Convformer structure;

[0041] Figure 5 Schematic diagram of the three-dimensional dynamic snake convolution module. DETAILED DESCRIPTION

[0042] The present invention is further described below in conjunction with specific embodiments, but the protection scope of the present invention is not limited thereto:

[0043] Example 1: A classification prediction method for denoised MRI images of ADHD, such as Figure 1-5 As shown in the figure, the MRI images of ADHD are subjected to unsupervised denoising using the unsupervised MRI image denoising network module to reduce the interference of image noise on feature extraction and improve classification accuracy. Then, the denoised MRI images are classified and predicted based on the Transformer-based DSC-Former classification network. Specifically:

[0044] Step 1: Data collection and dataset construction for ADHD classification task

[0045] Data collection is divided into two major channels: one is part of the data in the public data set ADHD-200 data set; the other is to obtain MRI imaging data of ADHD patients and control groups after desensitization through cooperative hospitals in Zhejiang Province. The specific data set information is shown in Table 1 below:

[0046] Table 1. Datasets for ADHD classification tasks

[0047] Dataset Number of patient groups Number of control groups ADHD-200PKU 113 146 ADHD-200NYU 123 99 ADHD-200OHSU 37 42 ADHD-200Brown 24 27 Data provided by partner hospitals 63 232

[0048] Step 2: Build an unsupervised denoising network

[0049] The unsupervised MRI denoising network structure proposed in this invention is as follows Figure 2 As shown, it includes two parallel processing lines, each of which includes a downsampling module and two noise estimation models composed of a corresponding Vision Transformer and ConvFormer stack. After the MRI image passes through the two downsampling modules, two downsampled images are obtained, that is, two sampling results on an image with the same noise distribution are obtained. The two downsampled images are respectively used as the input of a noise estimation model. During the training process, noise estimation is performed on two paired downsampled images of a single MRI image, so that the denoising result of the downsampled image is as close as possible to the other downsampled image, thereby obtaining a noise estimation of the original image.

[0050] Downsampling module such as Figure 3 As shown in the figure, a three-dimensional convolution kernel structure is used with a convolution kernel size of (2,2,2). Downsampling is performed by averaging the three-dimensional diagonal voxels. Each MRI image is independently used as the input data of the denoising network. After passing through two downsampling convolution kernels, two independent and identically distributed downsampled images of the same image are obtained.

[0051] The noise estimation model is composed of Vision Transformer and ConvFormer stacked together. The specific model structures of Vision Transformer and ConvFormer are as follows: Figure 4 As shown in the figure. In Vision Transformer, the input data passes through the Multi-Head Attention layer and the feed-forward network in sequence. Both the Multi-Head Attention layer and the feed-forward network use residual connections and are normalized using the LayerNorm layer. Compared with Vision Transformer, the ConvFormer module only replaces the Multi-Head Attention layer with the SeparableConvolution layer, and the rest remains the same.

[0052] After the downsampling module, two downsampled images are obtained, which are used as the input of two noise estimation modules respectively. The noise estimation results of the downsampled images are homogenized, and then the difference between the original input MRI image and the homogenized result is used as the final denoised image, thereby realizing the noise estimation of the original image and finally obtaining the denoised image.

[0053]

[0054] Among them, θ is the noise estimation model, x is the input image, S1(x) and S2(x) are two downsampled images, and f θ1 (S(x)) and f θ2 (S(x)) are the two-way noise estimates output by the noise estimation model.

[0055] The denoising network loss function for model training is designed as follows:

[0056]

[0057] Through this loss function, the denoised image of S1(y) is as close to S2(y) as possible, and the denoised image of S2(y) is as close to S1(y) as possible. After training, the noise estimation of the two downsampled images is obtained. Step 3: DSC-Former model construction

[0058] The DSC-Former model proposed in the present invention is as follows Figure 1 As shown. The backbone network of the DSC-Former model is a three-dimensional dynamic snake convolution module and a Vision Transformer module. The three-dimensional dynamic snake convolution module integrates directional snake convolutions in the three directions of X, Y, and Z axes, focusing on extracting slender structures in different directions to adapt to the characteristics of brain MRI images used in ADHD diagnosis. This module uses Leaky RELU as the activation function; the Vision Transformer module uses StarReLU as the model activation function, which can achieve better performance than GELU with less computation. The model reasoning path passes through the snake convolution module and the Vision Transformer module in turn, and the final output layer uses the SoftMax layer.

[0059] The module structure of Dynamic Snake Convolution (DSC) is as follows Figure 5 As shown in the figure, as a downsampling model, each layer extracts feature information from MRI images. DSC is inspired by deformable convolution and introduces a dynamic offset to adjust the position of the convolution kernel. Unlike deformable convolution, DSC introduces a prediction network in three dimensions to predict the dynamic offset of each pixel, and introduces a smoothness term to constrain the change of the offset, so that the convolution kernel can move smoothly along the direction of the tubular structure, like a snake, so that it can better adapt to slender structures and has good feature extraction capabilities for complex brain tissue scenes. Taking the x direction as an example, the calculation process of the receptive field position is shown in formula (3), (xi+c ,y i+c ,z i+c ) is the center of the convolution kernel, thus forming a snake-shaped receptive field along the x direction:

[0060]

[0061] The three-dimensional dynamic snake convolution module used in the present invention integrates conventional three-dimensional (3D) convolution with three directional dynamic snake convolutions - X-type, Y-type and Z-type. Conventional 3D convolution is responsible for refining the key features of the target object in the three-dimensional image. The X-type, Y-type and Z-type dynamic snake convolutions focus on extracting the characteristics of the slender tubular structure along the X, Y and Z axis directions respectively. Finally, the feature vectors extracted from the four different convolution layers are concat together to form a comprehensive feature representation. Figure 5 As shown in the figure, the DSC module is divided into four branches, among which Conv3D is a conventional three-dimensional convolution kernel, and the DSConv3D modules in the other three branches are used to learn feature information in three directions. DSConv3D is formed by the residual connection of two Conv3D modules. The Instance Norm layer is set after the residual part, and the Tanh activation function is used. The Instance Norm layer is set after the remaining convolution kernels, and the Leaky ReLU activation function is used.

[0062] Between each layer of DSC modules, Vision Transformer is used to further learn feature information in a self-attention manner. The output of the 3D dynamic snake convolution module is added to the Instant Norm layer, and Leaky RELU is used as the activation function.

[0063] Vision Transformer extracts the correlation information between the regions of the image data through the self-attention mechanism, thereby extracting the image signs related to ADHD contained therein. At the same time, the Vision Transformer module used in the present invention uses StarReLU as the activation function. The StarReLU calculation process is as follows:

[0064] StarReLU(x)=s·(ReLU(x)) 2 +β (4)

[0065] By introducing two trainable parameters, s and β, better performance than GELU can be achieved with less computation.

[0066] The model reasoning path passes through the four-layer snake convolution module and the Vision Transformer module in sequence, and finally passes through the SoftMax layer to obtain the classification prediction result given by the model.

[0067] Step 4: De-noise, pre-process and normalize the MRI images

[0068] The unsupervised denoising network constructed in step 2 is used to denoise all MRI images in the data set constructed in step 1. The denoising process of each MRI image is a separate training process. During the training process, the loss function of formula (2) is used to make the two downsampled images pass through backpropagation, and the denoised image obtained by the noise estimation model is as close as possible to the other downsampled image, thereby obtaining the noise estimation of the two downsampled images; after the training is completed, the two output noise estimates are homogenized, and then the difference between the original input MRI image and the homogenized result is used as the denoising result of the unsupervised denoising network.

[0069] Then all denoised MRI images are preprocessed and normalized, which mainly includes the following processes:

[0070] (1) Resample all image data to a voxel size of (1,1,1).

[0071] (2) Normalize the intensity value of each voxel and map it to the range of 0-511.

[0072] (3) Integrate the training data, count the foreground and background information of the image, and crop the non-brain background to unify the image dimension. The final input data dimension is (128, 128, 128).

[0073] Step 5: DSC-Former model training

[0074] The training process is as follows:

[0075] (1) Each sample in the dataset that has undergone denoising and preprocessing in step 4 is randomly horizontally flipped and rotated to expand the dataset size. The expanded dataset is randomly divided into a training set, a validation set, and a test set in a ratio of 8:1:1.

[0076] (2) Build the DSC-Former model, randomly initialize the model parameters and load the divided training set and validation set. Input the data in the training set into the DSC-Former model in batches to obtain the inference results, calculate the loss function of the inference classification results and the input labels, complete the backward propagation process, and update the DSC-Former model weights. In the training, cross entropy is used to construct the binary classification task loss function, and the optimizer is set as follows: use the RMSprop optimizer, set the learning rate lr to 1e-4, the decay rate ρ to 0.9, the weight decay weight decay to 1e-4, and ∈ to 1e-8. The training batch size is set to 4, and a total of 200 epochs are trained. The model performance is evaluated on the validation set in each round, and the model weight with the highest AUC is retained. Use the test set to verify the classification AUC, accuracy and recall rate of the optimal weight model.

[0077] (3) In the experiment, 10-fold cross validation was used. The data set was randomly divided into 10 parts. Each part of the data was used as the test set in turn. The remaining data was divided into the validation set and the test set at a ratio of 1:8. The training process of step (2) was repeated. Finally, the average classification index under 10-fold cross validation was obtained.

[0078] Step 6: Use online

[0079] After the model is trained, the online reasoning process is as follows:

[0080] (1) Collect MRI images of the brain of the patient to be tested;

[0081] (2) Use the unsupervised denoising network proposed in the present invention to perform online training and denoise the collected MRI images. The denoising process of each MRI image is a separate training process. Each MRI image is trained independently for 50 epochs to complete the noise estimation process. The loss function used is shown in formula (1), and the Adam optimizer is used for training. After the training is completed, the noise estimates of the two downsampled images are obtained respectively. The noise estimation results of the downsampled images are normalized, and then the difference between the original input MRI image and the normalized result is used as the noise estimate of the original image, thereby completing the denoising based on the original image, as shown in formula (1);

[0082] (3) Preprocess the denoised images, resample all image data to a voxel size of (1,1,1), normalize the intensity value of each voxel, map it to a range of 0-511, and crop the non-brain background;

[0083] (4) The preprocessed image is input into the DSC-Former model trained in step 5, and the classification result is output after passing through the multi-layer dynamic snake convolution module and Vision Transformer.

[0084] experiment:

[0085] 1. Verification of the effectiveness of unsupervised denoising network

[0086] In order to verify the role of the unsupervised denoising network proposed in the present invention in the ADHD classification and identification task, an ablation experiment for the denoising network was designed. In the first case (denoised data), the MRI images in all data sets were trained with the DSC-Former model after the unsupervised denoising network and preprocessing process of the present invention. In the second case (original data), the original MRI images were directly used, and the DSC-Former model was trained after the preprocessing operation of the present invention. Under 10-fold cross-validation, the ADHD classification indicators of the DSC-Former model in both denoised and non-denoised data are shown in Table 2 below:

[0087] Table 2

[0088] method AUC Accuracy (%) Recall rate (%) (Raw data)DSC-Former 0.948 93.4 89.7 (Denoised Data)DSC-Former 0.921 91.3 84.4

[0089] From the above results, it can be seen that by removing the inherent noise in the MRI image through the unsupervised denoising network of the present invention, the DSC-Former model of the present invention can accurately extract the feature information in the image to a greater extent, and obtain a better recognition effect.

[0090] Experiment 2: Comparative experiment

[0091] In order to further verify the classification and identification ability of the proposed method for ADHD patients, this section compares the unsupervised denoising network and DSC-Former model of the present invention with the ADHD classification models proposed in two other related documents on a data set consisting of ADHD-200 and data provided by cooperative hospitals. Document 1 constructed four different convolutional neural network structures, among which the ResNet with the best classification performance is listed here. Document 2 uses RBSVM to perform ADHD classification detection on structural MRI images after feature extraction.

[0092] For the training data set, there are also two cases. In the first case (denoised data), all MRI images in the data set are subjected to the unsupervised denoising network and preprocessing process of the present invention before model training. In the second case (original data), the original MRI images are directly used and the model training is performed after the preprocessing operation of the present invention.

[0093] The classification performance comparison experimental results based on the data set used in this experiment under 10-fold cross validation are shown in Table 3 below:

[0094] Table 3

[0095]

[0096]

[0097] Document 1 See Cicek, Gulay, and Aydin Akan. "Deep learning approach versus traditional machine learning for ADHD classification." 2021MedicalTechnologies Congress (TIPTEKNO). IEEE, 2021.

[0098] Reference 2: Lohani, Dhruv Chandra, and Bharti Rana. "ADHD diagnosis using structural brain MRI and personal characteristic data with machine learning framework." Psychiatry Research: Neuroimaging 334 (2023): 111689.

[0099] The results show that the DSC-Former model of the present invention has an accuracy improvement of 4.8% and 10.2% respectively on the same data set compared to the recognition methods used in Documents 1 and 2, and has achieved the highest AUC. It can be seen that DSC-Former has shown stronger representation ability and recognition accuracy in the ADHD classification task using structural MRI data by integrating three-dimensional dynamic snake convolution and Vision Transformer multi-head self-attention mechanism. When using denoised data, both DSC-Former and ResNet used in Document 1 show better classification performance. It can be seen that the unsupervised denoising network used in the present invention can better extract specific information from the image and achieve higher classification accuracy after removing the inherent noise in the MRI image. Since the model used in Document 2 is aimed at the brain gray matter and cortical thickness features extracted from MRI images, it is not sensitive to denoised image data, and the classification accuracy does not show a significant improvement.

[0100] Finally, it should be noted that the above examples are only some specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments, and there are many variations. All variations that can be directly derived or associated with the content disclosed by a person skilled in the art should be considered as the protection scope of the present invention.

Claims

1. A classification prediction method for denoised MRI images of ADHD, characterized by: The method includes collecting MRI images of the brain to be tested, performing online denoising on the collected MRI images using an online trained unsupervised denoising network, and then preprocessing the results of the online denoising and inputting them into an offline trained DSC-Former model to obtain classification prediction results; The unsupervised denoising network includes two parallel processing circuits, each of which includes a downsampling module and a noise estimation model connected in series; The DSC-Former model includes four layers of three-dimensional snake convolution modules and Vision Transformer modules connected in sequence, and a SoftMax layer.

2. The classification prediction method for denoised MRI images for ADHD according to claim 1, characterized in that: The downsampling module is a three-dimensional convolution kernel structure; The noise estimation model in the unsupervised denoising network is composed of a stack of Vision Transformer and ConvFormer.

3. The classification prediction method for denoised MRI images for ADHD according to claim 2, characterized in that: The online denoising process of the unsupervised denoising network is: Each MRI image is trained independently for 50 epochs using the Adam optimizer. During the training process, the loss function is used, and the two downsampled images are back-propagated to approximate the denoised image obtained by the noise estimation model to another downsampled image, thereby obtaining noise estimates for the two downsampled images. After the training is completed, the two output noise estimates are homogenized, and then the difference between the original input MRI image and the homogenized result is used as the result of the online denoising process: Among them, θ is the noise estimation model, x is the input image, S1(x) and S2(x) are two downsampled images, and f θ (S(x)) is the two-way noise estimate output by the noise estimation model.

4. The classification prediction method for denoised MRI images for ADHD according to claim 3, characterized in that: The loss function of the unsupervised denoising network is:

5. The classification prediction method for denoised MRI images for ADHD according to claim 4, characterized in that: The pre-processing is: (1) Mapping the denoised MRI image data to a voxel size of (1,1,1); (2) Normalize the intensity value of each voxel and map it to the range of 0-511; (3) Crop the non-brain background.

6. The classification prediction method for denoised MRI images for ADHD according to claim 5, characterized in that: The three-dimensional dynamic snake convolution module in the DSC-Former model is characterized by: The module contains four branch paths, which use conventional three-dimensional convolution, dynamic snake convolution in the x-direction, dynamic snake convolution in the y-direction, and dynamic snake convolution in the z-direction respectively. The three dynamic snake convolution modules form receptive fields along the X, Y, and Z axes respectively. The feature vectors extracted by the four different convolution layers are concat-operated to output the corresponding feature representation.

7. The classification prediction method for denoised MRI images for ADHD according to claim 6, characterized in that: The offline training process of the DSC-Former model is as follows: Obtain MRI image data of desensitized ADHD patients and a control group from a public dataset ADHD-200 dataset and a cooperative hospital, then obtain denoised MRI images by subjecting each MRI image to an online training process of the unsupervised denoising network, and then perform preprocessing and data augmentation operations, and randomly divide the augmented dataset into a training set, a validation set, and a test set in a ratio of 8:1:1; The MRI images in the training set were input into the DSC-Former model in sequence. DSC-Former was trained using the RMSprop optimizer, and a loss function was constructed based on cross entropy to calculate the loss between the recognition result and the label. After the backward propagation process was completed, the DSC-Former model weights were updated. A 10-fold cross-validation model classification performance was used.

Images