Method for detecting functional networks and metabolic coupling of ad based on multivariate analysis

Abstract

This invention relates to a method for detecting AD functional network and metabolic coupling based on multivariate analysis, comprising: acquiring... 18 F-FDG PET and fMRI image data were collected, reconstructed, and preprocessed; the reconstructed and preprocessed images were then analyzed. 18 Feature extraction was performed on F-FDG PET and fMRI image data to obtain whole-brain glucose metabolism rate and functional connectivity strength features, respectively. Dimensionality reduction was applied to the functional connectivity strength features. Multivariate analysis was then performed on the glucose metabolism rate features and the dimensionality-reduced functional connectivity strength features to obtain canonical correlation variables. These canonical correlation variables correspond to spatial distribution patterns of glucose metabolism and functional connectivity, and the coupling relationship between the functional network and metabolism in Alzheimer's disease was obtained based on these distribution patterns. Compared with existing technologies, this invention reveals the relationship between glucose metabolism and functional activity, providing a strong reference for the early detection and diagnosis of Alzheimer's disease.

Description

Technical Field

[0001] This invention relates to the field of medical image analysis, and in particular to a method for detecting AD functional network and metabolic coupling based on multivariate analysis. Background Technology

[0002] Alzheimer's disease (AD) is a common and complex neurodegenerative disease among the elderly, primarily affecting the central nervous system in the brain, causing degenerative changes. By the time patients exhibit obvious clinical symptoms of Alzheimer's disease, the disease has often progressed to an advanced stage and is completely untreatable. Therefore, early diagnosis and intervention for Alzheimer's disease have become the focus of most research.

[0003] In the preclinical stage, Aβ amyloid protein typically deposits in the brain of patients for 10–15 years, impairing the normal function of neurons and synapses. As the disease progresses, neurofibrillary tangles and more severe neuronal damage gradually appear. Since there are currently no effective treatments, early diagnosis and intervention for Alzheimer's disease are crucial. Decreased brain glucose metabolism and weakened functional activity are considered markers of early neuronal and synaptic dysfunction and have predictive value for cognitive decline. Previous studies have primarily focused on metabolic and functional changes in specific brain regions (such as the default mode network hub). However, due to the heterogeneity of brain regions in Alzheimer's patients, there is currently a lack of effective methods to comprehensively detect Alzheimer's disease by integrating changes in the coupling of whole-brain glucose metabolism and functional activity. Summary of the Invention

[0004] The purpose of this invention is to provide a more systematic method for detecting AD functional network and metabolic coupling based on multivariate analysis to reveal the relationship between glucose metabolism and functional activities.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for detecting AD functional network and metabolic coupling based on multivariate analysis includes the following steps:

[0007] Get 18 F-FDG PET and fMRI image data were collected, and then reconstructed and preprocessed.

[0008] For reconstruction and preprocessing 18 F-FDG PET and fMRI image data were used to extract features, and the glucose metabolism rate and functional connectivity strength characteristics of the whole brain were obtained respectively.

[0009] The functional connection strength characteristics are subjected to dimensionality reduction processing;

[0010] Multivariate analysis was performed on glucose metabolic rate characteristics and functional connectivity strength characteristics after dimensionality reduction to obtain canonical correlation variables.

[0011] The typical related variables correspond to the spatial distribution pattern of glucose metabolism and functional connectivity, and the functional network and metabolic coupling relationship of Alzheimer's disease are obtained based on the spatial distribution pattern.

[0012] Furthermore, the specific steps for obtaining glucose metabolic rate characteristics include:

[0013] Based on the reconstruction and preprocessing 18 The average glucose uptake rate of each brain region was calculated using F-FDG PET image data and regional interest analysis.

[0014] The average glucose uptake rate of each brain region is used as a characteristic of the glucose metabolism rate of each brain region.

[0015] Furthermore, the specific steps for obtaining functional connectivity strength characteristics include:

[0016] The average fMRI time-series signal of each brain region was extracted from the reconstructed and preprocessed fMRI image data;

[0017] Based on the correlation coefficient method, the correlation coefficient between the average fMRI time-series signals of every two brain regions is calculated to construct a whole-brain functional connectivity matrix.

[0018] Based on all the connection edges of the whole-brain functional connectivity matrix, the functional connectivity strength features are obtained.

[0019] Furthermore, the brain regions are divided according to the Power brain region template.

[0020] Furthermore, the correlation coefficient method is the Pearson correlation coefficient method.

[0021] Furthermore, the median absolute deviation statistic method is used for dimensionality reduction.

[0022] Furthermore, the expression for the dimensionality reduction process is:

[0023] median(|X i -median(X)|)

[0024] In the formula, X i Let be a vector representing the functional connectivity features of the i-th subject, and let be a vector representing the median of all functional connectivity features.

[0025] Furthermore, the multivariate analysis is a sparse canonical correlation analysis, and the optimization equation for the sparse canonical correlation analysis is:

[0026]

[0027]

[0028] In the formula, u1 and u2 are linear transformation vectors that maximize the correlation coefficients of canonical variables X1u1 and X2u2, X1 and X2 are feature matrices of two modes of glucose metabolic rate characteristics and functional connectivity strength characteristics, and ||·||1 and ||·||2 represent L 1 and L 2 The norms, c1 and c2, are regularization parameters whose values ​​are determined by grid search.

[0029] Furthermore, the specific steps for obtaining the canonical correlation variables include:

[0030] Based on glucose metabolism rate characteristics and functional connectivity strength characteristics after dimensionality reduction, a resampling method is used to construct several resampled feature datasets from the two types of features.

[0031] For each resampled feature dataset, multivariate analysis is used to obtain typical correlation variable pairs and their corresponding weights.

[0032] Based on the characteristics of the aforementioned canonical correlation variables and their corresponding weights, the canonical correlation variables are selected with relatively large and stable weights.

[0033] Furthermore, it also includes using a permutation test to assess the significance of the multivariate analysis treatment, the specific steps of which include:

[0034] The order of fixed functional connectivity strength features remains unchanged. The order of glucose metabolism rate features is randomly permuted several times. Sparse canonical correlation analysis is performed after each permutation to obtain several canonical correlation coefficients.

[0035] The significance of the multivariate analysis was evaluated by comparing several canonical correlation coefficients with the original correlation coefficients.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] (1) Based on the integrated FDG-PET / fMRI technology, this invention measures the synchronously changing glucose metabolism rate characteristics and functional connectivity strength characteristics, and detects the activity relationship of the whole brain function-metabolism coupling. Compared with the previous univariate analysis of a single or a few brain regions, this invention can more systematically reveal the relationship between glucose metabolism and functional activity behind high-dimensional data.

[0038] (2) The activity relationship of whole brain function-metabolism coupling detected by the present invention can be linked to cognitive impairment, find new imaging biomarkers related to cognitive decline, and help provide a strong reference for the early diagnosis and assessment of Alzheimer's disease. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0040] Figure 2 This is a schematic diagram of the method flow according to an embodiment of the present invention. Detailed Implementation

[0041] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0042] This embodiment provides a method for detecting the coupling of AD functional networks and metabolism based on multivariate analysis, such as... Figure 1 As shown, the method includes the following steps:

[0043] S1, Obtain 18 The image data of F-FDG PET and fMRI were reconstructed and preprocessed.

[0044] This study recruited 101 volunteers from a community surrounding a hospital, including 37 Alzheimer's patients, 25 patients with mild cognitive impairment, and 39 healthy participants. All participants underwent CDR cognitive ability assessment and MMSE mental status scale evaluation, and their images were collected using a 3T PET / MR scanner. 18 Imaging data from F-FDG PET and fMRI.

[0045] The method in this embodiment is implemented using a computer. The PC acquires the... 18 The F-FDG PET and fMRI image data were processed using reconstruction and preprocessing methods. 18 The image data from F-FDG PET and fMRI were processed.

[0046] S2, for reconstruction and preprocessing 18 F-FDG PET and fMRI image data were used to extract features, obtaining the glucose metabolism rate characteristics and functional connectivity strength characteristics of the whole brain, respectively.

[0047] After reconstructing and preprocessing the data, this embodiment respectively... 18Feature extraction and sparse canonical correlation analysis were performed on F-FDG PET and fMRI data. Specifically, this embodiment uses Power brain region templates to extract features from fMRI and... 18 The mean fMRI time-series signal and mean glucose uptake rate of each brain region were extracted from the F-FDGPET data. The Pearson correlation coefficient between time series data of every two brain regions was calculated to obtain the whole-brain functional connectivity matrix, where the mean glucose uptake rate was extracted using region interest analysis. Therefore, for each subject, the collected data... 18 Based on F-FDG PET and fMRI data, this embodiment extracted glucose metabolism features and 34,716 functional connectivity strength features (i.e., all edges of the functional connectivity matrix, the number of which is 264 × (264-1) / 2) for 264 brain regions. The whole-brain functional connectivity matrix and glucose uptake rate of each brain region obtained above are as follows: Figure 2 As shown.

[0048] S3. Perform dimensionality reduction processing on the functional connection strength features.

[0049] Before performing canonical correlation analysis, this embodiment uses the median absolute deviation statistic to reduce the dimensionality of functional connectivity features, which is defined as median(|X i -median(X)|), where X i This represents a vector of functional connectivity features for the i-th subject. Functional connectivity features with the median absolute deviation in the top 10% of the total are selected for subsequent sparse canonical correlation analysis.

[0050] S4. Perform multivariate analysis on the functional connectivity strength characteristics and glucose metabolism rate characteristics after dimensionality reduction to obtain typical correlation variables.

[0051] Sparse canonical correlation analysis is a regularized multivariate statistical analysis method that aims to find two sets of linear transformations that maximize the linear relationship between the new variables (canonical variables) after these transformations. Specifically, given the characteristic matrices X1 and X2 of two modes, the sparse canonical correlation method aims to find linear transformation vectors u1 and u2 that maximize the correlation coefficients of the canonical variables X1u1 and X2u2. Its optimization equation is:

[0052]

[0053]

[0054] Where, ||·||1 and ||·‖|2 represent L respectively. 1 and L 2 Norms. c1 and c2 are regularization parameters, whose values ​​are determined through grid search.

[0055] Following sparse canonical correlation analysis, this embodiment employs a permutation test to assess the statistical significance of canonical correlation variables. Specifically, this embodiment keeps the order of functional connectivity features constant, then randomly permutes the order of glucose metabolism features 1000 times, and performs sparse canonical correlation analysis after each permutation. The correlation coefficient r obtained from these 1000 canonical correlation analyses is... i The correlation coefficients were used to obtain the original functional linkage and glucose metabolism data. For comparison, the p-value is calculated as r. i Greater than The number of substitutions was divided by the number of substitutions (i.e., 1000). The canonical correlation coefficients between functional connectivity and glucose metabolism were calculated as linear combinations of functional connectivity features and glucose metabolism features, respectively.

[0056] To identify functional connectivity and glucose metabolism features that significantly contribute to canonical correlation variables, this embodiment uses the Bootstrap resampling method to further select features with larger and more stable weights. Specifically, one-third of the samples were randomly replaced with the remaining samples 1000 times, and canonical correlation analysis was performed after each random sampling. Since resampling affects the order of canonical correlation variable pairs and the sign of feature weights, this embodiment uses the Procrustes rotation method to match the canonical correlation variable pairs obtained after resampling with those obtained from the original data. If a feature's 95% and 99% confidence intervals do not contain 0, that feature is marked as significant. To investigate the relationship between the functional connectivity and glucose metabolism features detected in this embodiment and cognitive decline in Alzheimer's disease patients, this embodiment uses Spearman partial correlation analysis to assess the relationship between the MMSE clinical score and the extracted canonical correlation coefficients (age, gender, and education level as covariates).

[0057] S5. The spatial distribution pattern of glucose metabolism and functional connectivity associated with the typical related variables, and the functional network and metabolic coupling relationship of Alzheimer's disease based on the spatial distribution pattern.

[0058] Through sparse canonical correlation analysis, this embodiment selected three pairs of significant canonical correlation variables, such as Figure 2As shown, each pair of canonical correlation variables corresponds to a set of associated spatial distribution patterns of glucose metabolism and functional connectivity. The first pair of canonical correlation variables captured a significant association between glucose metabolism rate in subcortical regions (including the thalamus and caudate nucleus) and functional connectivity between the default mode network (MMN) and the tegmental network (TMN), and between the subcortical network and the ventral attention network. The second pair of canonical correlation variables captured a significant correlation between glucose metabolism in the TMN, default mode network, attention network, and salience network and functional connectivity between the TMN. The third pair of canonical correlation variables captured a close association between glucose metabolism in the default mode network and the frontoparietal network and functional connectivity between the default mode network and the tegmental network, and between the subcortical network and the ventral attention network. Furthermore, the canonical correlation coefficients of these three pairs of functional connectivity and glucose metabolism were significantly lower in the Alzheimer's disease patient group than in the healthy control group, and the canonical correlation coefficient of the third pair of functional connectivity and glucose metabolism in patients with mild cognitive impairment was also significantly lower than in healthy controls. This embodiment further explored the correlation between canonical correlation coefficients of functional connectivity and glucose metabolism and clinical cognitive scores. It was found that the third pair of canonical correlation coefficients was significantly correlated with clinical scores, indicating that covariate abnormalities in glucose metabolism and functional connectivity in patients with mild cognitive impairment and Alzheimer's disease are related to cognitive decline. This functional metabolic canonical correlation coefficient can serve as a potential multimodal imaging biomarker for the diagnosis and assessment of Alzheimer's disease.

[0059] To explore the topological relationship between whole-brain glucose metabolism and functional activity in Alzheimer's disease patients and to apply it to the detection of Alzheimer's disease, this embodiment uses simultaneous FDG-PET / fMRI technology to measure synchronously changing glucose metabolism and functional connectivity strength. Multivariate analysis (sparse canonical correlation analysis) was employed to detect changes in whole-brain function-metabolism coupling and its relationship with cognitive impairment, identifying novel imaging biomarkers associated with cognitive decline. Compared to previous univariate analyses focusing on single or a few brain regions, whole-brain function-metabolism coupling analysis can more systematically reveal the relationship between glucose metabolism and functional activity behind high-dimensional data. Complex cognitive functions depend on the interaction and synergy of multiple brain regions. Changes in local brain region metabolism may affect functional activity in distant brain regions. Whole-brain multivariate analysis can more comprehensively reveal the relationship between functional activity and metabolic processes. Compared to univariate analysis methods, multivariate analysis helps filter redundant information from high-dimensional data, extracts the most relevant features between modalities, and has stronger generalization and interpretability. Furthermore, compared to previous detection methods based on asynchronous FDG-PET and fMRI, the use of synchronous integrated FDG-PET / fMRI technology can simultaneously measure covariant glucose metabolism processes and functional activities, minimizing the influence of interference factors caused by physiological or psychological changes in subjects at different scanning stages, while also reducing errors caused by image registration.

[0060] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0061] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0062] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0063] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0064] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0065] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0066] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting AD functional network and metabolic coupling based on multivariate analysis, characterized in that, Includes the following steps: Get 18 F-FDG PET and fMRI image data were collected, and then reconstructed and preprocessed. For reconstruction and preprocessing 18 F-FDG PET and fMRI image data were used to extract features, and the glucose metabolism rate and functional connectivity strength characteristics of the whole brain were obtained respectively. The functional connection strength characteristics are subjected to dimensionality reduction processing; Multivariate analysis was performed on glucose metabolic rate characteristics and functional connectivity strength characteristics after dimensionality reduction to obtain canonical correlation variables. The specific steps for obtaining canonical correlation variables include: Based on glucose metabolism rate characteristics and functional connectivity strength characteristics after dimensionality reduction, a resampling method is used to construct several resampled feature datasets from the two types of features. For each resampled feature dataset, multivariate analysis is used to obtain typical correlation variable pairs and their corresponding weights. Based on the aforementioned canonical correlation variable pairs and their corresponding weights, the selected variables are those with relatively large and stable weights. The multivariate analysis is a sparse canonical correlation analysis, and the optimization equation for the sparse canonical correlation analysis is: In the formula, and To make canonical correlation variables and The linear transformation vector that maximizes the correlation coefficient. This is a feature matrix representing the glucose metabolism rate. The feature matrix represents the functional connectivity strength characteristics. and They represent and Norm, and This is a regularization parameter, the value of which is determined through grid search; The typical related variables correspond to the spatial distribution pattern of glucose metabolism and functional connectivity, and the functional network and metabolic coupling relationship of Alzheimer's disease are obtained based on the spatial distribution pattern.

2. The method for detecting AD functional network and metabolic coupling based on multivariate analysis according to claim 1, characterized in that, The specific steps to obtain glucose metabolic rate characteristics include: Based on the reconstruction and preprocessing 18 The average glucose uptake rate of each brain region was calculated using F-FDG PET image data and regional interest analysis. The average glucose uptake rate of each brain region is used as a characteristic of the glucose metabolism rate of each brain region.

3. The method for detecting AD functional network and metabolic coupling based on multivariate analysis according to claim 1, characterized in that, The specific steps for obtaining functional connectivity strength characteristics include: The mean fMRI time-series signal of each brain region was extracted from the reconstructed and preprocessed fMRI image data; Based on the correlation coefficient method, the correlation coefficient between the average fMRI time-series signals of every two brain regions is calculated to construct a whole-brain functional connectivity matrix. Based on all the connection edges of the whole-brain functional connectivity matrix, the functional connectivity strength features are obtained.

4. A method for detecting AD functional network and metabolic coupling based on multivariate analysis according to claim 2 or 3, characterized in that, The brain regions were divided according to the Power brain region template.

5. The method for detecting AD functional network and metabolic coupling based on multivariate analysis according to claim 3, characterized in that, The correlation coefficient method mentioned is the Pearson correlation coefficient method.

6. The method for detecting AD functional network and metabolic coupling based on multivariate analysis according to claim 1, characterized in that, Dimensionality reduction was performed using the median absolute deviation statistic method.

7. The method for detecting AD functional network and metabolic coupling based on multivariate analysis according to claim 6, characterized in that, The expression for the dimensionality reduction process is: In the formula, Indicates the first A vector of functional connectivity strength features for each subject. A vector representing the median of all functional connectivity strength features.

8. The method for detecting AD functional network and metabolic coupling based on multivariate analysis according to claim 1, characterized in that, It also includes using a permutation test to assess the significance of the multivariate analysis treatment, the specific steps of which include: The order of fixed functional connectivity strength features remains unchanged. The order of glucose metabolism rate features is randomly permuted several times. Sparse canonical correlation analysis is performed after each permutation to obtain several canonical correlation coefficients. The significance of the multivariate analysis was evaluated by comparing several canonical correlation coefficients with the original correlation coefficients.

Images