A depression subtype classification system based on cross-species multi-modal brain images

Abstract

This invention discloses a depression subtype classification system based on cross-species multimodal brain imaging. The method first acquires and preprocesses cross-species multimodal brain imaging data and depression brain imaging data; it then constructs a multimodal mapping process between the human and chimpanzee brains to obtain homologous brain maps of the two brains; it uses whole-brain probabilistic fiber tracing to construct connectivity features between the human and chimpanzee brains, extracting human-specific connectivity features through cross-species comparison; it constructs a depression subtype model based on these human-specific connectivity features; and it validates the subtype results based on functional imaging data indicators and brain functional connectivity. This invention, from the perspective of species evolution, constructs human-specific connectivity features through cross-species comparison, achieving depression subtype classification and providing a new perspective and method for understanding depression subtypes and their biological basis.

Description

Technical Field

[0001] This invention relates to the field of neuroimaging technology, and specifically to a method for classifying depression using cross-species brain imaging features, namely a depression subtype classification system based on cross-species multimodal brain imaging. Background Technology

[0002] Major Depressive Disorder (MDD) is a common debilitating mental illness caused by a combination of factors, and its etiology and pathophysiology remain unclear. Typically, the diagnosis of MDD is guided by subjective behavioral symptoms. Currently, physicians' selection of scales and diagnostic criteria is influenced by their professional experience, and there is no unified standard within the industry. Furthermore, some questions on the scales involve personal privacy, and some patients may conceal their actual situation when answering questions, leading to diagnostic biases. Due to these factors, classifying depression is difficult, and there are no objective indicators that show significant heterogeneity in assessing phenotypic manifestations, etiologies, and longitudinal trajectories.

[0003] Studies have shown that patients with depression exhibit abnormalities in the occipital lobe, accompanied by visual cortical dysfunction. The occipital lobe is involved in the perception and processing of visual information and complex visual perception processes, and also has strong connections with the limbic system (especially the hippocampus), parietal lobe, and temporal lobe. Simultaneously, numerous studies have shown that MDD (Madness Depression) is also related to histological changes in the higher cortical areas of the brain, such as the frontal lobe, parietal lobe, temporal lobe, and default mode network. The higher association cortex is a structural and functional system in the cerebral cortex that plays a connecting and integrative role. It is the part of the cerebral cortex that performs higher cognitive functions, controlling and organizing speech and thought, planning purposeful behavior, and regulating conscious activity. These higher areas have undergone particularly significant evolutionary changes; therefore, analyzing the pathogenesis and classification of depression from an evolutionary perspective is crucial.

[0004] Research into the underlying mechanisms of MDD faces numerous challenges in cross-species classification of depression. First, there is a lack of reliable homologous brain maps for cross-species comparison; second, the accuracy of cross-species brain mapping models is a significant factor influencing subsequent classification; and finally, while researchers both domestically and internationally have identified multiple subtypes and subgroups that MDD may include, there is no effective solution for further subdividing these subtypes and subgroups from a cross-species perspective. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention aims to propose a subtype classification system for depression based on cross-species multimodal brain imaging, comprising the following steps:

[0006] The data preprocessing module preprocesses cross-species multimodal brain imaging data and depression brain imaging data respectively, to obtain preprocessed structural image data, diffusion-weighted image data and functional image data;

[0007] The cross-species mapping module constructs a multimodal mapping process for the human brain and the chimpanzee brain, obtaining homologous brain maps of the chimpanzee brain and the human brain.

[0008] The feature extraction module obtains whole-brain fiber connections in the human and chimpanzee brains based on probabilistic fiber tracing, filters the groups to obtain human-specific connection features, and obtains the brain regions associated with these human-specific structural connections.

[0009] The depression classification module extracts the resting-state fMRI time series corresponding to human-specific connections and calculates Pearson correlations to obtain the functional connectivity feature matrix of each sample. It then uses a clustering method to perform unsupervised clustering of the functional connections of patients with depression to obtain the depression classification.

[0010] The validation module compares the classification results with the control group based on functional imaging data indicators and brain functional connectivity, and uses the significant differences in the comparison results as the validation criteria for each classification.

[0011] The data preprocessing module acquires standard brain magnetic resonance imaging (MRI) data from both humans and chimpanzees, including T1w images and diffusion-weighted imaging. Brain imaging data for depression includes T1w images and human resting-state fMRI data. Specifically, the preprocessing process involves denoising, uniform grayscale correction, descaling, gray-white matter segmentation, and surface reconstruction of the T1w structural images to obtain groove depth, curvature, thickness, surface area, and volume features, and registering the individual voxel spatial structural images to the MNI standard space. The diffusion MRI data undergoes denoising, head motion correction, eddy current correction, descaling, and diffusion tensor fitting. The resting-state MRI data undergoes time correction, head motion correction, spatial standardization, temporal detrending, spatial smoothing, and regression analysis to exclude head motion parameters, white matter, and cerebrospinal fluid signals.

[0012] The cross-species mapping module, which constructs a multimodal mapping between the human brain atlas and the chimpanzee brain atlas, mainly includes the following steps:

[0013] A) Projecting standard human brain atlases from voxels onto the surface of the human brain to obtain surface-based human brain atlases and extracting corresponding brain regions;

[0014] B) Using the distribution of myelin, groove depth, curvature, thickness, surface area, and volume of the brains of humans and chimpanzees as multimodal features to drive surface-based multimodal brain mapping, the transformation relationship between the two species is calculated.

[0015] C) Apply the above transformation relationship to surface-based human brain maps to achieve brain map mapping between two species and obtain homologous whole-brain maps of chimpanzees.

[0016] In the feature extraction module, whole-brain probabilistic fiber tracing is used in both humans and chimpanzees, and the fibers are standardized, denoised, and false positives are removed to obtain whole-brain fiber connections in both the human and chimpanzee brains. Based on the homologous brain maps between humans and chimpanzees obtained by the cross-species mapping module, whole-brain structural connectivity networks based on brain regions are constructed for cross-species comparative analysis. The constructed whole-brain structural connections for both species are statistically analyzed in groups to calculate the proportion of subjects with structural connections between brain regions. In the human brain, connections with a group threshold greater than 60% are selected, and in the chimpanzee brain, connections with a group threshold of 0% are selected, thereby filtering out human-specific connectivity features and obtaining the brain regions associated with these human-specific structural connections.

[0017] In the aforementioned depression classification module, constructing a feature-based depression classification model mainly includes the following steps:

[0018] A) Based on the brain regions with unique human brain structural connections obtained by the feature extraction module, extract the resting-state fMRI time series corresponding to the unique human connections;

[0019] B) Calculate the Pearson correlation between the time series corresponding to human-specific connections and use Fisher transform to convert them into the corresponding Z values, thereby obtaining the correlation coefficient of each connection for each subject and obtaining the functional connectivity feature matrix of each sample.

[0020] C) The dimensionality reduction method is used on the above functional connectivity feature matrix to select all features with feature values ​​greater than the preset value to form a feature extraction matrix. Then, the clustering method is used to perform unsupervised clustering of the functional connectivity of patients with depression to obtain the depression classification.

[0021] D) The optimal classification model is obtained by using intra-cluster error variance, profile value, and graphical tool elbow method as the judgment criteria.

[0022] In the aforementioned verification module, the verification criteria are resting-state functional magnetic resonance imaging (fMRI) indicators, including local consistency (Reho), amplitude of low-frequency fluctuations (ALFF), fractional amplitude of low-frequency fluctuations (fALFF), and functional connectivity density (FCD). The module is characterized by using the brain imaging research analysis tool SPM to extract the resting-state fMRI indicators for each subject, performing a two-sample t-test on the classification of each indicator against the control group, and obtaining comparison results for different classifications after voxel-based multiple comparison correction. The significant differences in the comparison results are used as the verification criteria for each classification. Attached Figure Description

[0023] To more clearly illustrate the implementation of the present invention or the existing technical solutions, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0024] Figure 1 This is a flowchart illustrating a subtype classification system for depression based on cross-species multimodal brain imaging, provided in an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of the process of constructing a multimodal mapping between a human brain map and a chimpanzee brain map using a cross-species mapping module provided in an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram of a feature-based depression classification process provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] In one embodiment, such as Figure 1 As shown, a depression subtype classification system based on cross-species multimodal brain imaging is provided, including a preprocessing module 10, a cross-species mapping module 20, a feature extraction module 30, a depression subtyping module 40, and a validation module 50. The preprocessing module 10 uses standard brain magnetic resonance imaging (MRI) data from both humans and chimpanzees, including T1w images and diffusion-weighted imaging. The brain imaging data for depression includes T1w images and human resting-state fMRI data. Preprocessing is performed on the T1w images, diffusion-weighted images, and resting-state fMRI data respectively. The cross-species mapping module 20 uses brain features from both humans and chimpanzees to construct a multimodal mapping between the human and chimpanzee brains, obtaining corresponding atlases of the chimpanzee and human brains. The feature extraction module 30 uses whole-brain probabilistic fiber tracing, and after standardization, denoising, and false-positive removal, compares the structural connectivity differences between the human and chimpanzee brains in a group setting to obtain features specific to the human brain. The depression subtyping module 40 uses extracted human-specific connections as features to perform dimensionality reduction and clustering on depression data to obtain a subtyping model. The validation module 50 uses resting-state functional magnetic resonance imaging (fMRI) indicators and brain functional connectivity to validate the subtyping model.

[0029] The aforementioned depression subtype classification system based on cross-species multimodal brain imaging preprocesses structural and functional image data, as well as depression brain image data, through a preprocessing module. A cross-species mapping module obtains a map corresponding to the chimpanzee and human brains. A feature extraction module then extracts human-specific features from cross-species probabilistic fiber tracing. A depression subtyping module then classifies depression based on these extracted features. Finally, a verification module further confirms the accuracy of the subtyping. This system achieves depression subtype classification based on cross-species multimodal brain imaging, improving the accuracy and reliability of such classification.

[0030] In embodiments of the present invention, such as Figure 2 As shown, the cross-species mapping module 20 further includes a human brain map transformation module 201, an inter-species mapping module 202, and a chimpanzee brain map output module 203.

[0031] The human brain mapping projection module 201 projects a standard human brain template from a voxel onto the surface of the human brain, thereby obtaining the sub-region of interest corresponding to the human brain template and extracting the corresponding brain region;

[0032] The inter-species mapping module 202 uses the distribution of myelin, groove depth, curvature, thickness, surface area and volume of the brains of humans and chimpanzees as features to drive multimodal brain atlas mapping, calculate the transformation relationship between the two species, and obtain the mapping model between the human brain and the chimpanzee brain.

[0033] The chimpanzee brain map output module 203 uses the mapping model in 202 to obtain the mapping of a specified region of the human brain in the chimpanzee brain, realizes the mapping of the corresponding region, and obtains the corresponding whole brain map of the chimpanzee.

[0034] In embodiments of the present invention, such as Figure 3 As shown, the depression subtyping module 40 further includes depression data feature sequence extraction 401, feature matrix construction 402, clustering subtyping 403, and optimal subtyping model determination 404:

[0035] Depression data feature sequence extraction 401: Based on the brain regions with human-specific structural connections obtained by feature extraction module 30, the resting-state fMRI time series corresponding to human-specific connections are extracted.

[0036] Feature matrix construction 402: Pearson correlation is calculated between time series corresponding to human-specific connections and Fisher transform is used to convert them into corresponding Z values, thereby obtaining the correlation coefficient of each connection for each subject and obtaining the functional connectivity feature matrix of each sample.

[0037] Clustering classification 403: The above functional connectivity feature matrix is ​​subjected to dimensionality reduction method, all features with feature values ​​greater than preset values ​​are selected to form a feature extraction matrix, and the functional connectivity of patients with depression is unsupervised clustering using clustering method to obtain the depression classification.

[0038] The optimal classification model was determined by using intra-cluster error variance, profile value, and graphical tool elbow method as the judgment criteria to obtain the optimal classification model.

[0039] Please see Figures 1-3 Based on the above scheme, to facilitate a better understanding of the depression subtype classification system based on cross-species multimodal brain imaging provided by the embodiments of the present invention, the steps of the scheme are described in detail below:

[0040] (1) Data Acquisition

[0041] We acquired MRI data of the brains of 40 healthy humans, 46 healthy chimpanzees, and brain imaging data of 1,279 individuals with depression from multiple centers.

[0042] (2) Data preprocessing

[0043] The preprocessing steps for structural image data mainly include: denoising, grayscale uniform field correction, descaling, gray-white matter segmentation, and surface reconstruction to obtain groove depth, curvature, thickness, surface area, and volume features, and registering individual voxel spatial structural images to the MNI standard space; the preprocessing steps for diffusion magnetic resonance imaging data mainly include: denoising, head motion correction, eddy current correction, descaling, and diffusion tensor fitting; the preprocessing steps for resting-state magnetic resonance imaging data mainly include: time correction, head motion correction, spatial normalization, time series detrending, spatial smoothing, and elimination of head motion parameters, white matter, and cerebrospinal fluid signals through regression.

[0044] (3) Multimodal mapping between human brain atlas and chimpanzee brain atlas

[0045] To obtain brain maps of the human brain surface, a voxel-to-surface projection method was used. The Freesurfer and Connectome Workbench toolkit were used to convert various brain file formats such as .annot, .func.gii, and .label.gii. The standard AAL template was projected onto the human brain surface to obtain the ROI corresponding to the AAL template. A total of 80 brain regions were extracted. For some regions that overlapped under the cortex, the regions closer to the cortex were selected as the standard.

[0046] Multimodal brain mapping is driven by features such as myelin distribution, sulcus depth, curvature, thickness, surface area, and volume in the brains of humans and chimpanzees. These features are used as multimodal inputs to map the human brain to a chimpanzee brain atlas, calculating the correspondence and transformation relationships between the two species, which are then used for mapping corresponding regions across species. Using these transformation relationships, the mapping of a specified region in the human brain to the chimpanzee brain can be obtained, achieving the mapping of corresponding regions and obtaining a whole-brain AAL atlas of the chimpanzee.

[0047] (4) Extraction of features unique to the human brain

[0048] Whole-brain probabilistic fiber tracing was used in both humans and chimpanzees. Based on the obtained homologous brain maps of the two species, whole-brain structural connectivity networks based on brain regions were constructed to obtain the structural connections between ROIs in the AAL brain maps of the two species: 80*80*40 for human brains and 80*80*46 for chimpanzee brains. For the connection matrix of the samples, the connections of each ROI for each sample were first standardized by dividing by the total number of connections through that ROI. Then, connections with a probability greater than 0.05% were taken as valid connections for noise reduction. Next, the connections between each pair of ROIs were statistically analyzed to obtain an upper triangular matrix of unique connection probabilities between each ROI, thus obtaining the average connection probability between ROIs of the two species.

[0049] For the structural connectivity results described above, the constructed whole-brain structural connectivity data for both species were statistically analyzed at the group level to calculate the proportion of subjects with structural connections between brain regions. Humans and chimpanzees were compared at the group level. In the human brain, connections with a group threshold greater than 60% were selected, while in the chimpanzee brain, connections with a group threshold of 0% were selected. Human-specific connectivity features were identified, and the brain regions associated with these human-specific structural connections were obtained. A total of 223 human-specific inter-ROI structural connections and their corresponding coordinates were identified, and these human-specific connections were used as features for further analysis.

[0050] (5) Construct a feature-based depression classification model

[0051] Based on the brain regions with unique human brain structural connections obtained above, time series of AAL templates corresponding to resting-state fMRI of the depression brain imaging dataset were extracted. Pearson correlations were calculated between the time series corresponding to human-specific connections, and Fisher transform was used to convert them into corresponding Z values, thereby obtaining the correlation coefficient of each connection for each subject, and finally obtaining the functional connectivity feature matrix of each sample.

[0052] First, PCA is used to reduce the dimensionality of the functional connectivity feature matrix obtained above, and all features with eigenvalues ​​greater than 1 are selected. Then, unsupervised clustering of the functional connectivity of MDD patients is performed using the k-means clustering method to obtain 2-10 subtypes of MDD.

[0053] Then, the optimal number of clusters is selected using two indicators: intra-cluster error variance (SSE) and silhouette index. The intra-cluster error variance is visualized based on the number of clusters using graphical tools.

[0054] Finally, since the inflection point where the decreasing curve tends to flatten when the intra-cluster error variance k is 3 and the profile value is the highest when k is 3, k is determined to be 3 as the number of clusters, resulting in three subtypes, corresponding to 486, 520, and 273 subjects respectively.

[0055] (6) Validation of typing results

[0056] Resting-state functional magnetic resonance imaging (fMRI) regional consistency (Reho), amplitude of low-frequency fluctuations (ALFF), fractional amplitude of low-frequency fluctuations (fALFF), and functional connectivity density (FCD) were used as validation criteria for each subtype. First, resting-state fMRI indicators were extracted for each subject using SPM. A two-sample t-test was performed between the subtype of each indicator and the control group, followed by voxel-based multiple comparison correction (FWE corrected p < 0.05) to obtain comparison results for different subtypes. The feasibility of the subtypes was further validated by comparing the significant differences between different subtypes and the control group.

[0057] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0058] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A classification system for depression subtypes based on cross-species multimodal brain imaging, characterized in that, Includes the following steps: S1, the data preprocessing module, acquires cross-species multimodal brain imaging data and depression brain imaging data and preprocesses them to obtain preprocessed structural image data, diffusion-weighted image data and functional image data; S2, the cross-species mapping module, constructs a multimodal mapping process for the human brain and the chimpanzee brain to obtain homologous brain maps of the chimpanzee brain and the human brain. S3, Feature Extraction Module, obtains whole-brain fiber connections of human and chimpanzee brains based on probabilistic fiber tracing, filters the groups to obtain human-specific connection features, and obtains the brain regions associated with these human-specific structural connections. S4, the depression subtyping module, extracts the resting-state fMRI time series corresponding to human-specific connections and calculates Pearson correlation to obtain the functional connectivity feature matrix of each sample. It then uses a clustering method to perform unsupervised clustering of the functional connectivity of patients with depression to obtain the depression subtyping. S5, the verification module, compares the subtyping results with the control group based on functional imaging data indicators and brain functional connectivity, and uses the significant difference in the comparison results as the verification standard for each subtyping.

2. The depression subtype classification system based on cross-species multimodal brain imaging according to claim 1, characterized in that, The acquired cross-species data consisted of standard brain magnetic resonance imaging (MRI) data from both humans and chimpanzees, including T1w images and diffusion-weighted imaging. Brain imaging data for depression included T1w images and human resting-state fMRI data. The data preprocessing module specifically performed the following steps: denoising, uniform grayscale correction, descaling, gray-white matter segmentation, and surface reconstruction on the T1w structural images to obtain groove depth, curvature, thickness, surface area, and volume features; and registering the individual voxel spatial structural images to the MNI standard space. The diffusion MRI data underwent denoising, head motion correction, eddy current correction, descaling, and diffusion tensor fitting. The resting-state MRI data underwent time correction, head motion correction, spatial standardization, temporal detrending, spatial smoothing, and regression analysis to exclude head motion parameters, white matter, and cerebrospinal fluid signals.

3. The depression subtype classification system based on cross-species multimodal brain imaging according to claim 2, characterized in that, In S2, the cross-species mapping module constructs a multimodal mapping process between the human brain atlas and the chimpanzee brain atlas, mainly including the following steps: S2-1, Projecting a voxel-based human brain atlas onto a human brain surface template to obtain a surface-based human brain atlas; S2-2 uses the distribution of myelin, groove depth, curvature, thickness, surface area and volume of the brains of humans and chimpanzees as multimodal features to drive surface-based multimodal brain mapping and calculate the transformation relationship between the two species. S2-3 utilizes the transformation relationship in S2-2 to apply to surface-based human brain maps, enabling the mapping of brain maps between two species and obtaining a homologous whole-brain map of the chimpanzee.

4. The depression subtype classification system based on cross-species multimodal brain imaging according to claim 3, characterized in that, In S3, whole-brain probabilistic fiber tracing was used in both humans and chimpanzees. The tracing was standardized, denoised, and false positives were removed to obtain whole-brain fiber connections in the human and chimpanzee brains. Based on the homologous brain maps between the two species obtained in S2-3, whole-brain structural connectivity networks based on brain regions were constructed for cross-species comparative analysis. The constructed whole-brain structural connectivity networks for the two species were statistically analyzed in groups to calculate the proportion of subjects with structural connectivity between brain regions. In the human brain, connections with a group threshold greater than 60% were selected, and in chimpanzees, connections with a group threshold of 0% were selected. Through comparative analysis, human-specific connectivity features were screened out, and the brain regions associated with these human-specific structural connections were obtained.

5. A classification system for depression subtypes based on cross-species multimodal brain imaging according to claim 4, characterized in that, In S4, the depression subtyping model module mainly includes the following steps: S4-1: Based on the brain regions with unique human brain structural connections obtained in S3, extract the resting-state fMRI time series corresponding to the unique human connections. S4-2 calculates the Pearson correlation between time series of brain regions connected by human-specific structures and uses Fisher transform to convert them into corresponding Z values, thereby obtaining the correlation coefficient of each connection for each subject and obtaining the functional connectivity feature matrix of each sample. S4-3, the dimensionality reduction method is used on the above functional connectivity feature matrix to select all features with eigenvalues ​​greater than the preset value to form a feature extraction matrix, and the clustering method is used to perform unsupervised clustering of the functional connectivity of patients with depression to obtain the depression classification. S4-4 uses intra-cluster error variance, profile value, and graphical tool elbow method as judgment criteria to obtain the optimal fractal model.

6. A classification system for depression subtypes based on cross-species multimodal brain imaging according to claim 5, characterized in that, In S5, resting-state functional magnetic resonance imaging (fMRI) parameters were extracted for each subject, including: local consistency (Reho), amplitude of low-frequency fluctuations (ALFF), fractional amplitude of low-frequency fluctuations (fALFF), and functional connectivity density (FCD). The classification of each parameter was compared with the control group using a two-sample t-test. After voxel-based multiple comparison correction, the comparison results of different classifications were obtained, and the significant differences in the comparison results were used as the validation criteria for each classification.

Images