A multi-group contrast brain functional connectivity batch modeling and mediation effect analysis system

Abstract

The application discloses a multi-group contrast brain function connection batch modeling and mediation effect analysis system, belongs to the field of medical image analysis and neural cognitive technology, and comprises a data preprocessing module, a multi-group contrast batch statistical module, a function connection batch modeling module, a mediation effect analysis module and a result output and visualization module. The multi-group contrast batch statistical module supports batch statistical difference analysis of brain structure and resting state function indexes of multiple groups of subjects, and matches corresponding multiple comparison correction methods according to different modal data. The function connection batch modeling module can take significant brain regions with inter-group differences as seed points, batch construct whole brain function connection models, complete inter-group comparison, and automatically identify the mediation effect type of brain image indexes between grouping factors and cognitive functions through structural equation modeling and bootstrap testing. The application improves the analysis efficiency of brain mechanism research of metabolic diseases, and provides reliable technical support for neural cognitive biomarker identification and mechanism research of diseases such as metabolic dysfunction related fatty liver disease.

Description

Technical Field

[0001] This invention relates to the fields of medical image analysis and neurocognitive technology, and in particular to a system for batch modeling and mediating effect analysis of multi-group comparative brain functional connectivity. Background Technology

[0002] Metabolic dysfunction-associated fatty liver disease (MASLD) is a prevalent chronic liver disease worldwide. Recent studies have found a close association between it and an increased risk of cognitive impairment. Multimodal magnetic resonance imaging (MRI) of the brain is a core tool for exploring its neural mechanisms. Current brain imaging data analysis largely relies on step-by-step manual processing, which has several limitations: First, there is a lack of automated batch analysis processes for multi-stratified patient groups. The step-by-step processing and statistics of structural and functional multimodal data need to be completed manually, which is inefficient and prone to human error. Secondly, functional connectivity analysis relies on manual selection of seed points, making it difficult to achieve batch modeling of a large number of differential brain regions, resulting in poor reproducibility of the results. Third, the analysis of the mediating effect of brain imaging-cognition requires the integration of data across tools, which is cumbersome and makes it difficult to verify multiple neural pathways in batches. Fourth, existing systems cannot simultaneously cover the entire process of multimodal index extraction, multi-group batch statistics, functional connectivity modeling, and mediation effect testing, making it difficult to meet the needs of efficient and accurate analysis in the study of hierarchical brain mechanisms of metabolic diseases such as MASLD. Summary of the Invention

[0003] The purpose of this invention is to propose a multi-group comparative brain functional connectivity batch modeling and mediation effect analysis system, which realizes the fully automated analysis of multimodal brain magnetic resonance imaging, supports multi-group batch statistical comparison, batch modeling of functional connectivity and automatic testing of mediation effects, improves the analysis efficiency and consistency of results in brain imaging research, and provides a reliable tool for the study of the neurocognitive mechanisms of metabolic diseases.

[0004] To achieve the above objectives, this invention provides a multi-group comparative brain functional connectivity batch modeling and mediation effect analysis system, comprising: The data preprocessing module is used to standardize and preprocess multimodal brain magnetic resonance imaging data and cognitive behavior data of multiple groups of subjects, and extract multidimensional quantitative indicators of brain structure and brain function. The multi-group comparison batch statistical module is used to perform batch statistical difference analysis on the extracted brain structure and function indicators among multiple groups, automatically include confounding covariates and perform multiple comparison correction, and screen out brain regions with significant differences between groups. The functional connectivity batch modeling module is used to calculate the whole-brain functional connectivity strength in batches and perform inter-group difference tests, using brain regions with significant differences between groups as seed points. The mediation effect analysis module is used to construct a structural equation model of the mediation effect between grouping factors, brain imaging indicators, and cognitive behavior indicators, and to test the mediating role of brain imaging indicators in the association between grouping and cognition. The results output and visualization module is used to generate statistical analysis reports and brain imaging visualization maps.

[0005] Preferably, the data preprocessing module includes a structural image preprocessing unit, a resting-state functional image preprocessing unit, and a behavioral data standardization unit; the structural image preprocessing unit is used to process T1-weighted structural images and extract structural indicators such as cortical thickness and brain region volume; the resting-state functional image preprocessing unit is used to process resting-state functional magnetic resonance imaging and extract functional indicators such as low-frequency amplitude and local consistency; the behavioral data standardization unit is used to standardize and encode cognitive scale data.

[0006] Preferably, the structural image preprocessing unit completes brain tissue segmentation and brain region atlas division based on the cortical surface analysis process, outputs the thickness and surface area of ​​the whole brain cortex region, as well as the volume of subcortical nuclei and cerebellar structures, and calculates and estimates the total intracranial volume as a correction covariate for volume analysis.

[0007] Preferably, the resting-state functional image preprocessing unit sequentially performs temporal correction, head motion correction, spatial standardization and smoothing denoising, calculates the low-frequency amplitude and local consistency index of the whole brain voxel dimension, and generates the average head motion displacement parameter as a covariate for functional analysis.

[0008] Preferably, the multi-group comparison batch statistical module supports two-group independent sample tests and three-group one-way covariance analysis modes, automatically incorporating age, gender, and education level as general covariates; it uses error detection rate correction for cortical morphology indicators, Gaussian random field correction for voxel-level functional imaging indicators, and Bonferroni correction for pairwise comparisons between groups.

[0009] Preferably, the functional connectivity batch modeling module automatically extracts the average time series of the seed point region, calculates its correlation coefficient with the whole brain voxel time series, and obtains the functional connectivity value through Fisher Z transformation, thus completing the inter-group difference test of whole brain functional connectivity for all seed points in batches.

[0010] Preferably, the mediation effect analysis module uses a structural equation model of mediation effect to construct the mediation path, and uses bootstrap repeated sampling to test the significance of the indirect effect, automatically identifying the types of complete mediation, partial mediation and no mediation effect.

[0011] Preferably, the structural equation model of the mediation effect uses grouping factors as independent variables, brain imaging quantitative indicators as mediating variables, and cognitive behavior scores as dependent variables, automatically controls confounding covariates, and outputs model fit index and path coefficients.

[0012] Preferably, the results output and visualization module generates standardized reports that include inter-group differences, correlation analysis, and mediation effect results, while also outputting brain region statistical parameter maps, correlation scatter plots, functional connectivity network diagrams, and mediation path diagrams.

[0013] Therefore, the present invention employs the above-mentioned multi-group comparative brain functional connectivity batch modeling and mediation effect analysis system, which has the following advantages: (1) It integrates the entire process of multimodal brain image preprocessing, multi-group batch statistics, functional connectivity batch modeling and mediation effect analysis, without the need for manual switching between tools, which greatly improves analysis efficiency; (2) Automated covariate control and multiple comparison correction are adopted to reduce human error and improve the repeatability of results; (3) It supports multi-group comparative analysis stratified by disease subtype, which can accurately identify disease-specific neuroimaging biomarkers, especially suitable for brain mechanism research of metabolic diseases such as MASLD.

[0014] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0015] Figure 1 This is a block diagram of a multi-group comparative brain functional connectivity batch modeling and mediation effect analysis system in an embodiment of the present invention; Figure 2 A bar chart comparing the neurocognitive test scores of the three groups of participants; Figure 3 A bar chart showing the results of the analysis of differences in cortical thickness in the brain regions of the three groups of subjects; Figure 4 Statistical map of brain regions showing differences in low-frequency amplitude between a group with metabolic dysfunction-related fatty liver disease complicated with type 2 diabetes and a normal control group; Figure 5 Statistical map of brain regions showing localized concordance differences between the group with metabolic dysfunction-related fatty liver disease complicated with type 2 diabetes and the normal control group; Figure 6 A network diagram showing the differences in functional connectivity between the two groups of patients and the normal control group; Figure 7 Scatter plot showing the correlation between local consistency of the right superior gyrus and Montreal cognitive assessment scores. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0017] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0018] Example like Figures 1-7 As shown, this embodiment proposes a multi-group comparative brain functional connectivity batch modeling and mediation effect analysis system. This embodiment uses a neurocognitive study of patients with metabolic dysfunction-associated fatty liver disease (MASLD) as its application scenario. The system analyzes multimodal magnetic resonance imaging and cognitive data from 53 MASLD patients and 37 healthy controls. The MASLD patients were divided into a subgroup with type 2 diabetes (22 patients) and a subgroup without type 2 diabetes (31 patients) based on whether they also had type 2 diabetes. Imaging data were acquired using a 3T MRI scanner equipped with a 32-channel head coil. The acquisition sequences included high-resolution T1-weighted structural images and resting-state functional magnetic resonance imaging.

[0019] I. System Overall Architecture: The multi-group comparison brain functional connectivity batch modeling and mediation effect analysis system of the present invention includes a data preprocessing module, a multi-group comparison batch statistics module, a functional connectivity batch modeling module, a mediation effect analysis module, and a result output and visualization module. Data from each module flows sequentially to achieve fully automated analysis.

[0020] II. Specific implementation process of each module: 1. Data Preprocessing Module: The data preprocessing module includes a structural image preprocessing unit, a resting-state functional image preprocessing unit, and a behavioral data standardization unit.

[0021] (1) Structural image preprocessing unit: Based on the cortical surface analysis workflow, T1-weighted structural images were processed sequentially, including skull dissection, brain tissue segmentation, gray-white matter demarcation, cortical surface reconstruction, and brain region segmentation. The Destrieux brain atlas was used to delineate cortical regions. The average cortical thickness of each subject's bilateral hemispheres, the thickness and surface area of ​​74 cortical regions, the volume of 45 subcortical and cerebellar structures, and the refined volume of the hippocampal subregion and amygdala were extracted. At the same time, the total intracranial volume was estimated and used as a covariate for subsequent volume analysis.

[0022] (2) Resting-state functional image preprocessing unit: Time-level correction, head motion correction, spatial standardization to standard brain template, Gaussian smoothing are performed in sequence to remove linear drift and physiological noise, and the low-frequency amplitude and local consistency index of whole brain voxels are calculated; at the same time, the average head motion displacement of each subject is calculated as a covariate for subsequent functional analysis.

[0023] (3) Standardized behavioral data unit: The data of cognitive scales such as the Mini-Mental State Examination, Montreal Cognitive Assessment, and Digit Span Test (including forward, backward and total scores) are coded and standardized and cleaned to generate a behavioral indicator matrix that can be used for statistical analysis.

[0024] 2. Multi-group comparison batch statistics module: The multi-group comparison batch statistics module supports two analysis modes: two-group comparison and three-group comparison. This embodiment adopts the three-group comparison mode, with the groups being the MASLD group with type 2 diabetes, the MASLD group without type 2 diabetes, and the normal control group.

[0025] The batch statistical module for multiple comparisons automatically uses age, gender, and education level as general covariates and matches corresponding statistical methods for different types of indicators: covariance analysis is used for cortical thickness and surface area indicators, Bonferroni correction is used for pairwise comparisons between groups, and error detection rate correction is used for overall multiple comparisons; for brain region volume indicators, the estimated total intracranial volume is added to the covariates; for voxel-level functional imaging indicators such as low-frequency amplitude and local consistency, covariance analysis combined with post-hoc tests is used, and Gaussian random field correction is used for multiple comparisons.

[0026] The batch statistical module for multiple comparisons outputs the statistical values ​​of inter-group differences, mean differences, 95% confidence intervals and corrected p-values ​​of all brain region indicators, and automatically filters out brain regions with significant differences.

[0027] In this embodiment, the comparison results of the three sets of cognitive scores are as follows: Figure 2 As shown, after adjusting for confounding factors using analysis of covariance, the Montreal Cognitive Assessment score of the MASLD group with type 2 diabetes was significantly lower than that of the normal control group and the MASLD group without type 2 diabetes; the results of the analysis of differences in cortical thickness in brain regions are as follows. Figure 3As shown, the mean cortical thickness of both hemispheres and the cortical thickness of multiple brain regions, such as the left middle frontal gyrus and precentral gyrus, in the MASLD group with type 2 diabetes were significantly lower than those in the other two groups.

[0028] The brain regions showing differences in low-frequency amplitude between the MASLD group with type 2 diabetes and the normal control group, such as... Figure 4 As shown, the low-frequency amplitude of brain regions such as the right inferior temporal gyrus and cerebellar vermis was significantly increased in the patient group, while the low-frequency amplitude of brain regions such as the right precuneus and supplementary motor area was significantly decreased; brain regions with localized differences in consistency include... Figure 5 As shown, the patient group showed increased local consistency in brain regions such as the left calcarine cortex and the right thalamus, while decreased local consistency in brain regions such as the right superior parietal gyrus and the left inferior parietal gyrus.

[0029] 3. Functional connection to batch modeling module: The functional connectivity batch modeling module automatically reads the brain regions with significant differences output by the batch statistical module and uses them as seed points. It batch extracts the average time series of all voxels within each seed point, calculates the Pearson correlation coefficient between this time series and the time series of the remaining voxels in the whole brain, and obtains the functional connectivity value after Fisher Z transformation. Then, it performs inter-group difference tests on the whole-brain functional connectivity map of each seed point, uses Gaussian random field for multiple comparison correction, and outputs the brain regions with significantly enhanced or weakened functional connectivity.

[0030] The functional connection differences in this embodiment are as follows: Figure 6 As shown, Figure 6 Groups A, B, and C are MASLD patients with type 2 diabetes. Figure 6 In the middle D group, the MASLD group without type 2 diabetes was represented. In the MASLD group with type 2 diabetes, the functional connectivity between the right inferior temporal gyrus and the left anterior cingulate cortex, between cerebellar vermis area 6 and the left precuneus, and between the right cerebellar Crus area I and cerebellar vermis areas 4-5 was significantly enhanced.

[0031] 4. Mediation Effect Analysis Module: The mediation effect analysis module constructs mediation paths based on structural equation modeling, using grouping factors as independent variables, significantly different brain imaging indicators as mediators, and cognitive behavioral scores as dependent variables. Age, gender, education level, and mean head movement displacement are automatically included as covariates. The module employs a bootstrap method with 5000 repeated samplings to test the significance of indirect effects. Based on the significance of direct and indirect effects, it automatically identifies complete mediation, partial mediation, or no mediation effect, outputting path coefficients, 95% confidence intervals, p-values, and model fit indices.

[0032] In this embodiment, the system automatically verified three core mediation pathways: the marginal mediating role of right superior lateral gyrus local consistency in the difference in Montreal Cognitive Assessment scores between the MASLD group with type 2 diabetes and the normal control group; the mediating role of cerebellar-vermis functional connectivity in the difference between group and digit span scores; and the complete mediating role of right precuneus low-frequency amplitude in the difference in digit span scores between the two MASLD subgroups. The correlation between right superior lateral gyrus local consistency and Montreal Cognitive Assessment scores is as follows: Figure 7 As shown, the two are significantly positively correlated.

[0033] 5. Results Output and Visualization Module: The module generates standardized statistical reports, including results of differences between all groups, partial correlation analysis results, and mediation effect analysis results; it also generates visualization maps, including brain region statistical parameter maps (overlaid on the surface of a standard brain template), cognitive-brain index correlation scatter plots, mediation effect path maps, and functional connectivity network maps, supporting batch export and viewing of all results.

[0034] III. Implementation Results: This system enables the simultaneous preprocessing of multimodal brain images, batch statistical analysis of multiple groups, functional connectivity modeling, and mediation effect analysis without the need for manual step-by-step processing. The analysis cycle is significantly shortened compared to traditional manual processes, and human error introduced by manual operation is avoided, resulting in significantly improved consistency of results. This embodiment successfully identified specific cortical thinning, abnormal brain activity, and altered functional connectivity in MASLD patients with type 2 diabetes, verifying the complete mediating role of low-frequency amplitude in the right precuneus in working memory impairment. This provides precise technical support for the study of the mechanisms and biomarker identification of MASLD-related brain injury.

[0035] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A system for batch modeling and mediating effect analysis of multi-group comparative brain functional connectivity, characterized in that, include: The data preprocessing module is used to standardize and preprocess multimodal brain magnetic resonance imaging data and cognitive behavior data of multiple groups of subjects, and extract multidimensional quantitative indicators of brain structure and brain function. The multi-group comparison batch statistical module is used to perform batch statistical difference analysis on the extracted brain structure and function indicators among multiple groups, automatically include confounding covariates and perform multiple comparison correction, and screen out brain regions with significant differences between groups. The functional connectivity batch modeling module is used to calculate the whole-brain functional connectivity strength in batches and perform inter-group difference tests, using brain regions with significant differences between groups as seed points. The mediation effect analysis module is used to construct a structural equation model of the mediation effect between grouping factors, brain imaging indicators, and cognitive behavior indicators, and to test the mediating role of brain imaging indicators in the association between grouping and cognition. The results output and visualization module is used to generate statistical analysis reports and brain imaging visualization maps.

2. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 1, characterized in that: The data preprocessing module includes a structural image preprocessing unit, a resting-state functional image preprocessing unit, and a behavioral data standardization unit; The structural image preprocessing unit is used to process T1-weighted structural images and extract structural indicators such as cortical thickness and brain region volume; the resting-state functional image preprocessing unit is used to process resting-state functional magnetic resonance imaging and extract functional indicators such as low-frequency amplitude and local consistency; the behavioral data standardization unit is used to standardize and encode cognitive scale data.

3. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 2, characterized in that: The structural image preprocessing unit completes brain tissue segmentation and brain region mapping based on the cortical surface analysis workflow, outputs the thickness and surface area of ​​the whole brain cortex region, as well as the volume of subcortical nuclei and cerebellar structures, and calculates and estimates the total intracranial volume as a correction covariate for volume analysis.

4. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 2, characterized in that: The resting-state functional image preprocessing unit sequentially performs temporal correction, head motion correction, spatial normalization, and smoothing and denoising. It calculates the low-frequency amplitude and local consistency index of the whole brain voxel dimension, and generates the average head motion displacement parameter as a covariate for functional analysis.

5. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 1, characterized in that: The multi-group comparison batch statistics module supports two-group independent sample tests and three-group one-way covariance analysis modes, automatically incorporating age, gender, and education level as general covariates; it uses error detection rate correction for cortical morphology indicators, Gaussian random field correction for voxel-level functional imaging indicators, and Bonferroni correction for pairwise comparisons between groups.

6. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 1, characterized in that: The functional connectivity batch modeling module automatically extracts the average time series of seed point regions, calculates the correlation coefficient between the seed point and the whole brain voxel time series, and obtains the functional connectivity value through Fisher Z transformation. It then performs batch tests on the differences in whole brain functional connectivity among all seed points.

7. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 1, characterized in that: The mediation effect analysis module uses a structural equation model of mediation effect to construct the mediation path, and uses bootstrap repeated sampling to test the significance of indirect effects, automatically identifying the types of complete mediation, partial mediation and no mediation effect.

8. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 7, characterized in that: The structural equation model of the mediation effect uses grouping factors as independent variables, brain imaging quantitative indicators as mediating variables, and cognitive behavior scores as dependent variables. It automatically controls confounding covariates and outputs model fit index and path coefficients.

9. The system for batch modeling and mediation effect analysis of multi-group comparative brain functional connectivity according to claim 1, characterized in that: The results output and visualization module generates standardized reports that include inter-group differences, correlation analysis, and mediation effect results, while also outputting brain region statistical parameter maps, correlation scatter plots, functional connectivity network diagrams, and mediation path diagrams.

Images