Alzheimer's disease early screening method based on magnetic resonance imaging

Abstract

The invention discloses an Alzheimer's disease early screening method based on magnetic resonance imaging, and relates to the technical field of magnetic resonance imaging, and the method comprises the steps: obtaining information identity data and magnetic resonance scanning image data of a to-be-detected person; preprocessing the magnetic resonance scanning image data, and calculating a brain area volume and a DTI-ALPS index according to the preprocessed magnetic resonance scanning image data; constructing a brain network and calculating global network attributes and regional node attributes according to the brain network; substituting the information identity data, the brain area volume, the DTI-ALPS index, the global network attribute and the area node attribute of the to-be-tested person into an Alzheimer's disease screening model to calculate diagnosis parameters; and comparing the diagnosis parameter with the optimal diagnosis threshold value to obtain a screening result. According to the method, the brain colloidal lymphatic network function is evaluated by means of multiple high-sensitivity nerve image indexes, and the early screening capability of the Alzheimer's disease is improved.

Description

Technical Field

[0001] The present invention relates to a method for early screening of Alzheimer's disease based on magnetic resonance imaging, belonging to the technical field of magnetic resonance imaging. Background Art

[0002] Alzheimer's disease (AD) is a neurodegenerative disease, mainly manifested as progressive cognitive impairment, affecting memory, thinking and behavior. The deposition of Aβ and tau proteins is considered to be its pathological feature. As the most common type of dementia, the diagnosis and early identification of AD are crucial for improving the quality of life of patients and delaying the progression of the disease.

[0003] Currently, clinically, symptom self-reporting (interview between patients and their families) and cognitive assessment tools (various neuropsychological tests to quantify cognitive ability) are relied on to screen potential AD patients. Subsequently, biomarker detection (cerebrospinal fluid amyloid and tau protein detection) or genetic testing (susceptibility genes such as APOEε4) is used for further diagnosis.

[0004] However, clinical assessment may be limited by the patient's state of consciousness and expression ability. Especially, AD patients often overestimate their cognitive ability, and the early and non-obvious symptoms of AD may be confused with normal aging or other types of dementia. Neuropsychological tests usually cause fatigue in patients and assessors due to consuming a large amount of time, and the assessment results may be affected by the patient's education level. Cerebrospinal fluid collection and genetic testing are invasive methods, which may cause discomfort and risks to patients, and patients and their families may feel anxious due to the results. In addition, imaging examinations can effectively assist in the diagnosis of AD. Positron emission tomography (PET) can detect amyloid proteins and other pathological features in the brain. However, its high cost and radiation risk limit its wide application. Magnetic resonance imaging (MRI) can use a variety of imaging techniques (such as structural MRI, functional MRI (fMRI), diffusion-weighted imaging (DWI), etc.) to provide information in different dimensions, which is helpful for comprehensively evaluating the functional and structural changes of the brain.

[0005] Recent studies have found that the deposition of Aβ and tau is due to the reduction of their clearance efficiency rather than the increase of production. This involves the brain's clearance system - the glymphatic network, which plays a key role in clearing brain waste and amyloid proteins. Currently, dynamic MRI using tracers can effectively reflect the function of the glymphatic network. However, this examination is also limited in promotion due to its invasiveness. Some studies have pointed out that the choroid plexus volume and DTI-ALPS index can indirectly reflect the function of the brain's glymphatic network. An increase in the choroid plexus volume and a decrease in the ALPS index can be observed in early AD patients, and these changes may occur before the deposition of Aβ and tau.

[0006] Therefore, there is an urgent need to provide an objective, convenient and non-invasive early screening method for AD. Summary of the Invention

[0007] The purpose of the present invention is to overcome the deficiencies in the prior art and provide an early screening method for Alzheimer's disease based on magnetic resonance imaging. By means of multiple highly sensitive neuroimaging indicators, the function of the brain glymphatic network is evaluated, and the early screening ability for Alzheimer's disease is improved.

[0008] To achieve the above object, the present invention is implemented by the following technical solutions: The present invention provides an early screening method for Alzheimer's disease based on magnetic resonance imaging, including: Obtaining the identity information data and magnetic resonance scan image data of the subject to be tested; Preprocessing the magnetic resonance scan image data to obtain preprocessed magnetic resonance scan image data; Calculating the brain region volume and DTI-ALPS index according to the preprocessed magnetic resonance scan image data; Constructing a brain network using the preprocessed magnetic resonance scan image data and calculating the global network attributes and regional node attributes according to the brain network; Constructing an Alzheimer's disease screening model and determining the optimal diagnostic threshold; Substituting the identity information data, brain region volume, DTI-ALPS index, global network attributes and regional node attributes of the subject to be tested into the Alzheimer's disease screening model to calculate the diagnostic parameters; Comparing the diagnostic parameters with the optimal diagnostic threshold to obtain the screening result.

[0009] Further, the magnetic resonance scan image data includes T2-weighted imaging, T2 fluid-attenuated inversion recovery imaging, 3D-MPRAGE T1-weighted imaging, diffusion tensor imaging and blood oxygenation level-dependent imaging.

[0010] Further, the constructing an Alzheimer's disease screening model and determining the optimal diagnostic threshold includes: Obtaining the identity information data of multiple subjects, and the identity information data includes gender, age and years of education; Performing magnetic resonance scans on all subjects to obtain the magnetic resonance scan image data of all subjects; Dividing the subjects into an Alzheimer's disease group and a healthy control group according to the identity information data and magnetic resonance scan image data of the subjects; Preprocessing the magnetic resonance scan image data to obtain preprocessed magnetic resonance scan image data; Calculating the brain region volume and DTI-ALPS index according to the preprocessed magnetic resonance scan image data; Construct a brain network using preprocessed magnetic resonance imaging (MRI) data and calculate global network properties and regional node properties based on the brain network; Take identity information data, brain region volume, DTI-ALPS index, global network properties, and regional node properties as modalities, and perform comparative analysis on each modality of the Alzheimer's disease group and the healthy control group to obtain an Alzheimer's disease screening model and the optimal diagnostic threshold.

[0011] Furthermore, the division of the subjects into the Alzheimer's disease group and the healthy control group according to the identity information data and MRI data of the subjects includes: Use the Montreal Cognitive Assessment Scale to evaluate the general cognitive ability of the subjects to obtain the Montreal Cognitive Assessment Scale score; Judge whether there are brain structural abnormalities in the subjects according to T2-weighted imaging; Evaluate the white matter hyperintensity score of the subjects according to T2 fluid-attenuated inversion recovery imaging; Subjects who meet the NIA / AA Alzheimer's disease diagnostic criteria, Montreal Cognitive Assessment Scale score ≤ 18, white matter hyperintensity score < 2, and no brain structural abnormalities are classified into the Alzheimer's disease group, and subjects who have no complaint of cognitive impairment, Montreal Cognitive Assessment Scale score ≥ 26, white matter hyperintensity score < 2, and no brain structural abnormalities are classified into the healthy control group. Furthermore, the preprocessing of the MRI data to obtain preprocessed MRI data includes: Take 3D-MPRAGE T1-weighted imaging as the input and use FreeSurfer software for processing and output to obtain a cortical partition statistical table of the brain structure; Take diffusion tensor imaging as the input and use FSL software to output x 、 y and z the FA maps and diffusion coefficients in the three directions of the Take blood oxygenation level-dependent imaging as the input and use the DPABI toolbox on Matlab for processing to obtain a brain time series map; The cortical partition statistical table of the brain structure, x 、 y and z the FA maps and diffusion coefficients in the three directions of the

[0012] Furthermore, the calculation of the brain region volume and DTI-ALPS index according to the preprocessed MRI data includes: Extract the brain region volumes from the cortical partition statistical table of the brain structure of the subject to be tested, where the brain region volumes include hippocampal volume, choroid plexus volume, lateral ventricle volume, cortical gray volume, and total brain volume; Take x , y and z the FA maps and diffusion coefficients in the three directions of the x, y, and z axes as inputs, and output the diffusion coefficients of the projection fibers of the bilateral cerebral hemispheres on the x and y axes and the diffusion coefficients of the association fibers of the bilateral cerebral hemispheres on the x and z axes based on the FSLeyes software; Calculate the DTI-ALPS index based on the diffusion coefficients of the projection fibers of the bilateral cerebral hemispheres on the x and y axes and the diffusion coefficients of the association fibers of the bilateral cerebral hemispheres on the x and z axes.

[0013] Furthermore, the construction of the brain network using the preprocessed magnetic resonance scan image data and the calculation of the global network attributes and regional node attributes based on the brain network are implemented based on the GRETNA toolbox, including: Take the brain time series map of the subject to be tested as an input, extract the average time series of each node in the brain time series of the subject to be tested through AAL atlas mapping, and calculate the Pearson correlation coefficient between every two nodes; Take the Pearson correlation coefficient between every two nodes as an edge, and construct a brain network according to the nodes; Calculate the global network attributes and regional node attributes of all nodes according to the brain network. The global network attributes include clustering coefficient, characteristic path length, global efficiency, and local efficiency. The regional node attributes include the betweenness centrality and degree centrality of the brain region structure corresponding to each regional node.

[0014] Furthermore, the nodes are obtained by dividing the regions of interest of the brain. Each region of interest corresponds to a node.

[0015] Furthermore, taking the identity information data, brain region volumes, DTI-ALPS index, global network attributes, and regional node attributes as modalities, comparing and analyzing each modality of the Alzheimer's disease group and the healthy control group to obtain the Alzheimer's disease screening model and the optimal diagnostic threshold, including: Take the identity information data, brain region volumes, and DTI-ALPS index as modalities, and perform between-group comparisons on each modality of the Alzheimer's disease group and the healthy control group respectively, and select the modalities with significant between-group differences as alternative modalities; Take the global network attributes and regional node attributes of all nodes as modalities, use the identity information data as a covariate, and compare each modality of the Alzheimer's disease group and the healthy control group using a two-sample t-test, and select the modalities with significant between-group differences as alternative modalities; Integrate all alternative modalities. Taking the Alzheimer's disease group and the healthy control group as dependent variables, perform univariate logistic regression analysis on all alternative modalities to obtain significant modalities; Use the significant modalities as dependent variables to perform multivariate logistic regression to obtain a regression equation; Use the regression equation as the Alzheimer's disease screening model, and the cut-off value of the regression equation as the optimal diagnostic threshold.

[0016] Furthermore, the expression of the Alzheimer's disease screening model is: P = 0.143×AGE + 0.529×EDU + (-0.595×LING.R) + (-55.55×E loc ) + (-2.9×HV) + 7.828×CPV + (-1.032×ALPS) + 9.353 where P represents the diagnostic parameter; AGE represents age; EDU represents years of education; LING.R represents the centrality of the right lingual gyrus; E loc represents the local efficiency; HV represents the hippocampal volume; CPV represents the choroid plexus volume; ALPS represents the DTI-ALPS index; The optimal diagnostic threshold is 1.0531036.

[0017] Compared with the prior art, the beneficial effects achieved by the present invention are: The early screening method for Alzheimer's disease based on magnetic resonance imaging technology provided by the present invention utilizes multiple highly sensitive neuroimaging indicators. By quantifying the structural volumes of regions such as the choroid plexus and lateral ventricles, it can effectively capture the subtle brain structural changes in early Alzheimer's disease, making up for the deficiency of relying solely on visual identification of brain atrophy in the early diagnosis of Alzheimer's disease. At the same time, the combination of the DTI-ALPS index and the topological properties of the brain network provides various parameters such as the glymphatic function and brain function of the brain to comprehensively evaluate the pathological changes of Alzheimer's disease, improving the accuracy of early screening for Alzheimer's disease; The present invention has the advantages of non-invasiveness, no radiation exposure, painlessness, etc., effectively improving the acceptance of the tested patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the construction process of the Alzheimer's disease screening model in the early screening method for Alzheimer's disease based on magnetic resonance imaging in an embodiment of the present invention; Figure 2 It is a schematic diagram of the volume segmentation of brain regions in the early screening method for Alzheimer's disease based on magnetic resonance imaging in an embodiment of the present invention; Figure 3Schematic diagram of calculating the DTI-ALPS index in an embodiment of the present invention for the early screening method of Alzheimer's disease based on magnetic resonance imaging; Figure 4 Schematic diagram of the brain network structure in an embodiment of the present invention for the early screening method of Alzheimer's disease based on magnetic resonance imaging, where L represents the left brain and R represents the right brain. Detailed implementation manner

[0019] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention. Embodiment 1

[0020] As Figure 1 shown, the embodiment of the present invention provides a method for constructing an Alzheimer's disease screening model and the optimal diagnostic threshold, specifically including the following steps: In this embodiment, an Alzheimer's disease screening model is constructed based on 96 subjects and the optimal diagnostic threshold is determined. Specifically: The identity information data of a total of 229 subjects (by default, all subjects are right-handed) are collected. The identity information data includes gender, age, and education years. Then, the Montreal Cognitive Assessment Scale (MoCA) is used to evaluate the general cognitive ability of all subjects to obtain the MoCA score.

[0021] In this embodiment, the MoCA score is also corrected according to the education years of the subjects. The specific operation is as follows: If the education years of the subject are less than 12 years, the MoCA score is increased by 1 point. If the education years of the participant are less than 6 years, the MoCA score is increased by 2 points to obtain the final MoCA score.

[0022] All subjects are scanned by whole-brain magnetic resonance (MR) to obtain magnetic resonance scan image data. The magnetic resonance scan image data includes T2 weighted imaging (T2WI), T2-fluid attenuated inversion recovery (T2 FLAIR), 3D-MPRAGE T1 weighted imaging (T1WI), Diffusion tensor imaging (DTI), and blood-oxygen-level-dependent (BOLD).

[0023] In this embodiment, the magnetic resonance scan image data are all obtained by scanning with a GE Signa HDxt 3.0T magnetic resonance machine, and an 8-channel head coil is used. The specific scan parameters are as follows: 3D-MPRAGE T1-weighted imaging: repetition time 6.50 ms, echo time 2.80 ms, flip angle 8°, inversion time 900 ms, field of view (FOV) 256 mm × 256 mm, number of slices 176, slice thickness 1 mm, slice gap 1 mm, scan time 4 min 36 s.

[0024] T2 fluid-attenuated inversion recovery (FLAIR) imaging: repetition time 8000 ms, echo time 126 ms, FOV 240×240 mm, slice thickness 3 mm.

[0025] Diffusion tensor imaging: repetition time 17000 ms, echo time 85.4 ms, inversion angle 90°, matrix 128×128, FOV 256 mm×256 mm, slice thickness 2 mm, slice gap 0, b-value 1000 s / mm2, number of diffusion directions 30, scan time 9 min 47 s.

[0026] Blood oxygenation level-dependent (BOLD) imaging: repetition time 2000 ms, echo time 30 ms, flip angle 90°, FOV 256 mm×256 mm, slice thickness 4 mm, slice gap 0, number of slices 33, scan time 8 min.

[0027] Based on the T2-weighted imaging, it is determined whether there are brain structure abnormalities in the subjects, and the white matter hyperintensity score of the subjects is evaluated according to the T2 FLAIR imaging.

[0028] 96 subjects were divided into an Alzheimer's disease group and a healthy control group. Subjects who met the diagnostic criteria of the National Institute on Aging / Alzheimer's Association (NIA / AA), had a Montreal Cognitive Assessment (MoCA) score ≤ 18, a white matter hyperintensity score < 2, and no brain structure abnormalities were included in the Alzheimer's disease group. Subjects who had no complaints of cognitive impairment, had a MoCA score ≥ 26, a white matter hyperintensity score < 2, and no brain structure abnormalities were included in the healthy control group. In this example, there were 69 people in the Alzheimer's disease group and 27 people in the healthy control group.

[0029] The magnetic resonance imaging (MRI) data were preprocessed to obtain preprocessed MRI data, specifically including: Taking the 3D-MPRAGE T1-weighted imaging as the input, using FreeSurfer 7.2.0 software for image processing and outputting the cortical parcellation statistics of the brain structure, and the cortical parcellation statistics of the brain structure were saved in the stats / aseg.stats result file.

[0030] Taking diffusion tensor imaging as the input, preprocessing is performed using FSL software, including motion correction, brain extraction, and calculation of diffusion metrics, and FA maps and diffusion coefficients in the three directions of the x, y, and z axes are output.

[0031] Taking blood oxygenation level-dependent imaging as the input, preprocessing is performed using the DPABI toolbox on Matlab (http: / / rfmri.org / DPABI), including removing the first 10 time points, head motion and time slice correction [According to the standard of head translation < 3 mm and rotation < 3°, all participants were included in the group], registration, DARTEL segmentation, removing linear drift and regression of covariates, normalization and resampling, smoothing, filtering, and a brain time series map is output.

[0032] Cortical parcellation statistical tables of brain structures, x 、 y and z The FA maps and diffusion coefficients in the three directions of the x, y, and z axes and the brain time series map constitute the preprocessed magnetic resonance scan image data.

[0033] Combined with Figure 3 , the brain regional volumes and DTI-ALPS indices are calculated based on the preprocessed magnetic resonance scan image data, including: Bilateral choroid plexus, bilateral hippocampus, bilateral lateral ventricles, cortical gray matter volume, and total brain volume are respectively extracted from the cortical parcellation statistical tables of brain structures, with the unit of cm 3 , and the segmentation results are as Figure 2 shown. The volumes of the choroid plexus, hippocampus, and lateral ventricles are calculated as the sum of bilateral volumes, and the volumes of all regions are finally expressed as the ratio to the total brain volume to eliminate individual differences.

[0034] Taking x 、 y and the FA maps and diffusion coefficients in the three directions of the x, y, and z axes as inputs into FSLeyes, ROI placement is performed in FSLeyes (on the body of the lateral ventricle level of the FA map, regions of interest with a radius of 4 mm are respectively placed on the same x axis of the projection fibers and association fibers in the bilateral cerebral hemispheres) and diffusion coefficients of the projection fibers and association fibers in the bilateral cerebral hemispheres are extracted in x 、 y and the three directions of the x, y, and z axes.

[0035] Then, based on the diffusion coefficients of the projection fibers in the bilateral cerebral hemispheres in x 、 y axes and the diffusion coefficients of the association fibers in the bilateral cerebral hemispheres in x 、 zThe diffusion coefficient on the x axis is used to calculate the unilateral ALPS index through mean(Dxproj, Dxasso) / mean(Dyproj, Dzasso). The specific calculation method is to project the bilateral cerebral hemisphere fibers onto the x axis with the diffusion coefficient Dxproj and the association fibers of the bilateral cerebral hemispheres onto the y axis with the diffusion coefficient Dxasso. The sum of them is divided by 2 as the numerator. The bilateral cerebral hemisphere projection fibers on the z axis with the diffusion coefficient Dyproj and the association fibers of the bilateral cerebral hemispheres on the

[0036] axis with the diffusion coefficient Dzasso are added and divided by 2 as the denominator to obtain the DTI-ALPS index. The level of the DTI-ALPS index reflects the ability of the perivascular drainage pathway to excrete brain waste. Figure 4 Combined with the preprocessed magnetic resonance scanning image data, a brain network is constructed, and global network properties and regional node properties are calculated based on the brain network, including:

[0037] Taking the brain time series map as the input, the brain is divided into 90 regions of interest (45 in each hemisphere) using the AAL atlas based on the GRETNA toolbox. These regions are defined as 90 nodes in the brain network. Then, the average time series of each node is extracted from the brain time series map, and the Pearson correlation coefficient between two nodes is calculated. The matrix of these correlation coefficients is defined as the edges of the network to evaluate the functional connection between nodes, resulting in a 90×90 matrix, which is the brain network.

[0038] The clustering coefficient refers to the degree of clustering among nodes in the entire network, reflecting the local connectivity of the network. The global efficiency refers to the global efficiency of parallel information transmission in the network, reflecting the information transmission efficiency of the entire functional network. Both are used as measures of network functional integration. The characteristic path length refers to the degree of proximity between network nodes. The local efficiency refers to the communication efficiency between the first neighboring nodes when a given node is deleted, reflecting the fault tolerance rate of the functional network. Both are used as measures of network functional separation. The betweenness centrality characterizes the influence of a certain node on the information flow between other nodes, reflecting the information transmission rate of the node. The degree centrality represents the number of direct connections of a given node to other nodes, reflecting its information communication ability in the functional network.

[0039] Taking identity information data, brain region volume, DTI-ALPS index, global network attributes, and regional node attributes as modalities, the Alzheimer's disease screening model and the optimal diagnostic threshold are obtained by analyzing each modality of the Alzheimer's disease group and the healthy control group, specifically including: First, conduct the preliminary screening of identity information data, brain region volume, and DTI-ALPS index: Use IBM SPSS 20 software to compare the identity information data, brain region volumes, and DTI-ALPS indices between the Alzheimer's group and the healthy control group. Use the K-S analysis to test the normality of the data, and use the chi-square test for categorical variables (gender) and the two-sample t-test or U-test for continuous variables. The results show that except for gender (p = 0.49), there are significant inter-group differences in other modalities (p < 0.05), which are used as alternative modalities.

[0040] Second, conduct the preliminary screening of network topological attributes: Use the GRETNA toolbox, with age, gender, and years of education as covariates, and use the two-sample t-test to compare the differences in global network attributes and local node attributes between the Alzheimer's group and the healthy control group. The Bonferroni method is used to correct the analysis of node attributes (a global attribute p < 0.05 and a node attribute p < 0.0005 are considered to have statistical differences). The results show that there are significant inter-group differences in the global efficiency (p = 0.018), local efficiency (p = 0.013) of the global network attributes, and the DC of the left lingual gyrus, right lingual gyrus, and left fusiform gyrus of the regional node attributes (p < 0.0002), which are used as alternative modalities.

[0041] Finally, construct an Alzheimer's disease screening model and the optimal diagnostic threshold: Using IBM SPSS 20 software, after adjusting for age, gender, and years of education, with the AD and HC groups as the dependent variables, perform univariate logistic regression analysis on all alternative modalities (volumes of each region, DTI-ALPS index, and properties of each network) to select significant modalities (p < 0.05). Subsequently, incorporate the significant modalities into a multivariate logistic regression analysis, and use the forward or backward method to select the final modalities to be included in the model (age, education level, hippocampal volume, choroid plexus volume, local efficiency, right lingual gyrus degree centrality, and DTI-ALPS index). The results are shown in Table 1:

[0042] Therefore, the expression of the Alzheimer's disease screening model finally obtained in this embodiment is: P = 0.143×AGE + 0.529×EDU + (-0.595×LING.R) + (-55.55×E loc ) + (-2.9×HV) + 7.828×CPV + (-1.032×ALPS) + 9.353 Among them, P represents the diagnostic parameter; AGE represents age; EDU represents years of education; LING.R represents the right lingual gyrus degree centrality; E loc represents local efficiency; HV represents hippocampal volume; CPV represents choroid plexus volume; ALPS represents the DTI-ALPS index.

[0043] The cut-off value of the regression equation is used as the optimal diagnostic threshold. In this embodiment, the optimal diagnostic threshold is 1.0531036. Example 2

[0044] This embodiment provides a method for early screening of Alzheimer's disease based on magnetic resonance imaging. Based on the Alzheimer's disease screening model and the optimal diagnostic threshold described in Example 1, according to the Alzheimer's disease screening model in Example 1, the finally determined modalities are age, education level, hippocampal volume, choroid plexus volume, local efficiency, right lingual gyrus degree centrality, and DTI-ALPS index.

[0045] Specifically, it includes the following steps: Obtain the age, education level, and magnetic resonance scan image data of the person to be tested. The magnetic resonance scan image data includes 3D-MPRAGE T1-weighted imaging, diffusion tensor imaging, and blood oxygenation level-dependent images.

[0046] Preprocess the magnetic resonance scan image data to obtain preprocessed magnetic resonance scan image data. Specifically: Taking 3D-MPRAGE T1-weighted imaging as the input, the FreeSurfer 7.2.0 software was used for image processing output to obtain the cortical parcellation statistical table of the brain structure, and the cortical parcellation statistical table of the brain structure was saved in the stats / aseg.stats result file.

[0047] Taking diffusion tensor imaging as the input, preprocessing was performed using FSL software, including motion correction, brain extraction, and diffusion index calculation, and FA maps and diffusion coefficients in the three directions of the x, y, and z axes were output.

[0048] Taking blood oxygenation level-dependent imaging as the input, preprocessing was performed using the DPABI toolbox on Matlab (http: / / rfmri.org / DPABI), including removing the first 10 time points, head motion and time slice correction [According to the standard of head translation < 3 mm and rotation < 3°, all participants were included in the group], registration, DARTEL segmentation, removing linear drift and regression on covariates, normalization and resampling, smoothing, and filtering, and a brain time series map was output.

[0049] The cortical parcellation statistical table of the brain structure, x 、 y and z the FA maps and diffusion coefficients in the three directions of the x, y, and z axes and the brain time series map constitute the preprocessed magnetic resonance scan image data.

[0050] Based on the preprocessed magnetic resonance scan image data, the brain regional volume and DTI-ALPS index were calculated, including: Bilateral choroid plexus, bilateral hippocampus, and total brain volume were separately extracted from the cortical parcellation statistical table of the brain structure, with the unit of cm 3 , and the segmentation results are as Figure 2 shown. The choroid plexus and hippocampus were calculated as the sum of bilateral volumes, and the final volume of all regions was expressed as the ratio to the total brain volume to eliminate individual differences.

[0051] Taking x 、 y and the FA maps and diffusion coefficients in the three directions of the x, y, and z axes were input into FSLeyes, and ROI placement was performed in FSLeyes (at the level of the body of the lateral ventricle in the FA map, an ROI with a radius of 4 mm was placed on the same x axis of the projection fibers and association fibers in the bilateral cerebral hemispheres) and diffusion coefficients were extracted to obtain the diffusion coefficients of the projection fibers and association fibers in the bilateral cerebral hemispheres in x 、 y and the three directions of the z axis.

[0052] Then, according to the projection fibers in the bilateral cerebral hemispheres in x 、y Diffusion coefficient on the axis and commissural fibers of bilateral cerebral hemispheres in x 、 z The diffusion coefficient on the axis is used to calculate the unilateral ALPS index by mean(Dxproj, Dxasso) / mean(Dyproj, Dzasso). The specific calculation method is to add the diffusion coefficient Dxproj of the projection fibers of bilateral cerebral hemispheres on x the axis and the diffusion coefficient Dxasso of the commissural fibers of bilateral cerebral hemispheres on x the axis, and then divide the sum by 2 as the numerator. Add the diffusion coefficient Dyproj of the projection fibers of bilateral cerebral hemispheres on y the axis and the diffusion coefficient Dzasso of the commissural fibers of bilateral cerebral hemispheres on z the axis, and divide the sum by 2 as the denominator to obtain the DTI-ALPS index. The level of the DTI-ALPS index reflects the ability of the perivascular drainage pathway to excrete brain waste.

[0053] Using the preprocessed magnetic resonance scanning image data to construct a brain network and calculate the global network properties and regional node properties according to the brain network, including: Taking the brain time series map as the input, based on the GRETNA toolbox, the brain is divided into 90 regions of interest (45 in each hemisphere) using the AAL atlas. These regions are defined as 90 nodes in the brain network. Then, the average time series of each node is extracted from the brain time series map, and the Pearson correlation coefficient between two nodes is calculated. The connection matrix of these correlation coefficients is defined as the edge of the network to evaluate the functional connection between nodes, resulting in a 90×90 matrix, which is the brain network.

[0054] At sparsity levels from 0.05 to 0.4 with an interval of 0.01, calculate the topological properties of the above matrix, including global network properties and regional node properties. Specifically, the local efficiency is calculated for the global network properties, and the right lingual gyrus degree centrality is calculated for the regional node properties.

[0055] Substitute the age, education level, hippocampal volume, choroid plexus volume, local efficiency, right lingual gyrus degree centrality, and DTI-ALPS index of the subject to be tested into the Alzheimer's disease screening model to calculate the diagnostic parameter.

[0056] Compare the diagnostic parameter with the optimal diagnostic threshold to obtain the screening result. The screening result can be used as a reference for medical staff to diagnose whether the subject to be tested has Alzheimer's disease based on the screening result in the later stage.

[0057] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. An early screening method for Alzheimer's disease based on magnetic resonance imaging, characterized in that, Comprising: Obtaining the identity information data and magnetic resonance scan image data of the person to be tested; Preprocessing the magnetic resonance scan image data to obtain preprocessed magnetic resonance scan image data; Calculating the brain region volume and DTI-ALPS index based on the preprocessed magnetic resonance scan image data; Constructing a brain network using the preprocessed magnetic resonance scan image data and calculating the global network attributes and regional node attributes based on the brain network; Constructing an Alzheimer's disease screening model and determining the optimal diagnostic threshold; Substituting the identity information data, brain region volume, DTI-ALPS index, global network attributes and regional node attributes of the person to be tested into the Alzheimer's disease screening model to calculate the diagnostic parameters; Comparing the diagnostic parameters with the optimal diagnostic threshold to obtain the screening result.

2. The early screening method for Alzheimer's disease based on magnetic resonance imaging according to claim 1, wherein The magnetic resonance scan image data includes T2-weighted imaging, T2 fluid-attenuated inversion recovery imaging, 3D-MPRAGE T1-weighted imaging, diffusion tensor imaging and blood oxygenation level-dependent imaging.

3. The early screening method for Alzheimer's disease based on magnetic resonance imaging according to claim 1, wherein The constructing an Alzheimer's disease screening model and determining the optimal diagnostic threshold includes: Obtaining the identity information data of multiple subjects, where the identity information data includes gender, age and years of education; Performing magnetic resonance scans on all subjects to obtain the magnetic resonance scan image data of all subjects; Dividing the subjects into an Alzheimer's disease group and a healthy control group according to the identity information data and magnetic resonance scan image data of the subjects; Preprocessing the magnetic resonance scan image data to obtain preprocessed magnetic resonance scan image data; Calculating the brain region volume and DTI-ALPS index based on the preprocessed magnetic resonance scan image data; Constructing a brain network using the preprocessed magnetic resonance scan image data and calculating the global network attributes and regional node attributes based on the brain network; Using the identity information data, brain region volume, DTI-ALPS index, global network attributes and regional node attributes as modalities, and performing comparative analysis on each modality of the Alzheimer's disease group and the healthy control group to obtain the Alzheimer's disease screening model and the optimal diagnostic threshold.

4. The method for early screening of Alzheimer's disease based on magnetic resonance imaging according to claim 3, wherein The dividing the subjects into an Alzheimer's disease group and a healthy control group according to the identity information data and magnetic resonance scan image data of the subjects includes: Using the Montreal Cognitive Assessment Scale to evaluate the general cognitive ability of the subjects to obtain the Montreal Cognitive Assessment Scale score; Judging whether the subject has brain structure abnormalities based on the T2-weighted imaging; Evaluating the white matter hyperintensity score of the subject based on the T2 fluid-attenuated inversion recovery imaging; Assigning the subjects who meet the NIA / AA Alzheimer's disease diagnostic criteria, Montreal Cognitive Assessment Scale score ≤ 18, white matter hyperintensity score < 2, and no brain structure abnormalities to the Alzheimer's disease group, and assigning the subjects with no complaint of cognitive impairment, Montreal Cognitive Assessment Scale score ≥ 26, white matter hyperintensity score < 2, and no brain structure abnormalities to the healthy control group.

5. The early screening method for Alzheimer's disease based on magnetic resonance imaging according to claim 1 or 3, characterized in that The preprocessing the magnetic resonance scan image data to obtain preprocessed magnetic resonance scan image data includes: Taking the 3D-MPRAGE T1-weighted imaging as the input, and using the FreeSurfer software for processing and outputting to obtain the cortical partition statistical table of the brain structure; Taking diffusion tensor imaging as the input and using FSL software to output x 、 y and z FA maps and diffusion coefficients in the three directions of the axial axis; Taking blood oxygenation level-dependent imaging as the input, it is processed using the DPABI toolbox on Matlab to obtain the brain time series graph; The cortical division statistical table of the brain structure, x , y and z the FA maps, diffusion coefficients, and brain time series maps in three directions of the axis constitute the preprocessed magnetic resonance scan image data.

6. The early screening method for Alzheimer's disease based on magnetic resonance imaging according to claim 1 or 3, characterized in that, Calculating the brain region volume and DTI-ALPS index according to the preprocessed magnetic resonance scanning image data, including: Extracting the brain region volume from the cortical division statistical table of the brain structure of the subject to be tested, and the brain region volume includes hippocampal volume, choroid plexus volume, lateral ventricle volume, cortical gray volume, and total brain volume; Taking x , y and z the FA maps and diffusion coefficients in the three directions of the axis as inputs, the diffusion coefficients of the bilateral cerebral hemisphere projection fibers on the x and y axes and the diffusion coefficients of the bilateral cerebral hemisphere association fibers on the x and z axes are obtained based on the output of the FSLeyes software; Calculating the DTI-ALPS index according to the diffusion coefficients of the bilateral cerebral hemisphere projection fibers on the x and y axes and the diffusion coefficients of the bilateral cerebral hemisphere association fibers on the x and z axes.

7. The method for early screening of Alzheimer's disease based on magnetic resonance imaging according to claim 1 or 3, characterized in that, Constructing a brain network using the preprocessed magnetic resonance scanning image data and calculating the global network attributes and regional node attributes based on the GRETNA toolbox, including: Taking the brain time series graph of the subject to be tested as the input, extracting the average time series of each node in the brain time series of the subject to be tested through AAL atlas mapping, and calculating the Pearson correlation coefficient between every two nodes; Taking the Pearson correlation coefficient between every two nodes as the edge, and constructing a brain network according to the nodes; Calculating the global network attributes and regional node attributes of all nodes according to the brain network, and the global network attributes include clustering coefficient, characteristic path length, global efficiency, and local efficiency, and the regional node attributes include betweenness centrality and degree centrality of the brain region structure corresponding to each regional node.

8. The early screening method for Alzheimer's disease based on magnetic resonance imaging according to claim 7, characterized in that The nodes are obtained by dividing the regions of interest of the brain, and each region of interest corresponds to a node.

9. The early screening method for Alzheimer's disease based on magnetic resonance imaging according to claim 3, wherein Taking the identity information data, brain region volume, DTI-ALPS index, global network attributes, and regional node attributes as modalities, comparing and analyzing each modality of the Alzheimer's disease group and the healthy control group to obtain the Alzheimer's disease screening model and the optimal diagnostic threshold, including: Taking the identity information data, brain region volume, and DTI-ALPS index as modalities, respectively performing between-group comparisons on each modality of the Alzheimer's disease group and the healthy control group, and selecting the modalities with significant between-group differences as alternative modalities; Taking the global network attributes and regional node attributes of all nodes as modalities, using the identity information data as a covariate, comparing each modality of the Alzheimer's disease group and the healthy control group using a two-sample t-test, and selecting the modalities with significant between-group differences as alternative modalities; Combining all alternative modalities, taking the Alzheimer's disease group and the healthy control group as dependent variables, and performing univariate logistic regression analysis on all alternative modalities to obtain significant modalities; Taking the significant modalities as dependent variables to perform multivariate logistic regression to obtain the regression equation; Taking the regression equation as the Alzheimer's disease screening model and the cut-off value of the regression equation as the optimal diagnostic threshold.

10. The method for early screening of Alzheimer's disease based on magnetic resonance imaging according to claim 9, characterized in that, The expression of the Alzheimer's disease screening model is: P = 0.143×AGE + 0.529×EDU + (-0.595×LING.R) + (-55.55×E loc ) + (-2.9×HV) + 7.828×CPV + (-1.032×ALPS) + 9.353 Among them, P represents the diagnostic parameter; AGE represents age; EDU represents the years of education; LING.R represents the centrality of the right lingual gyrus; E loc represents the local efficiency; HV represents the hippocampal volume; CPV represents the choroid plexus volume; ALPS represents the DTI-ALPS index; The optimal diagnostic threshold is 1.0531036.

Images