A univariate neurodegenerative biomarker for early detection of AD

Abstract

The present invention discloses a univariate neurodegenerative biomarker for early detection of AD, belonging to the technical field of Alzheimer's disease recognition. By assigning different weights to different regions of interest, a reliable and robust univariate morphometric index is generated to reflect the morphological changes of the brain caused by Alzheimer's disease. Compared with traditional univariate biomarkers, the present invention has higher statistical power and can better characterize the morphological changes of the cerebral cortex caused by Alzheimer's disease.

Description

Technical Field

[0001] The present invention relates to the field of Alzheimer's disease identification, and particularly to a univariate neurodegenerative biomarker for early detection of AD. Background Art

[0002] Alzheimer's disease (AD) is a common neurodegenerative disease, which is highly prevalent in the elderly population over 65 years old. With the rapid development of social economy and medicine, the global population is showing an aging trend, and the threat of Alzheimer's disease to human health is becoming increasingly obvious.

[0003] Structural magnetic resonance imaging (sMRI) is an important basis for clinicians to diagnose Alzheimer's disease patients. Through sMRI, clinicians can diagnose Alzheimer's disease faster and more accurately, so as to intervene and treat Alzheimer's disease patients as early as possible.

[0004] However, in daily practice, the inventors found that the existing technical solutions have the following problems:

[0005] Due to the complex geometry of the brain and individual differences, it is difficult to establish a corresponding relationship of the overall cerebral cortical geometry between individuals while maintaining the local morphological structure. In addition, in previous studies, regions of interest (ROIs) were formed by performing statistical between-group difference analysis on the morphological measurements of the entire cortical surface. These regions of interest (ROIs) are only affected by cortical morphological measurements and do not reflect abnormalities in cortical anatomical regions. Therefore, the lack of anatomical information will lead to unreliable regions of interest (ROIs), which is not conducive to revealing the relationship between morphological abnormalities in anatomical regions and AD-related symptoms.

[0006] In view of this, it is necessary to provide a new technical solution to solve the above problems. Summary of the Invention

[0007] The present invention provides a univariate neurodegenerative biomarker for early detection of AD, which has higher statistical power and can better characterize the morphological changes of the cerebral cortex caused by Alzheimer's disease.

[0008] A univariate neurodegenerative biomarker for early detection of AD, comprising:

[0009] Extract the gray matter part in the structural magnetic resonance image to be analyzed and perform surface structure reconstruction, and calculate the surface thickness information of the gray matter part;

[0010] Divide the surface of the gray matter part into 34 anatomical regions;

[0011] Register the surface thickness information of the gray matter part and the anatomical regions for all individuals;

[0012] Extract the essential morphological structure features among groups;

[0013] Based on the essential morphological structure features among groups, obtain the group statistical differences of all points in each anatomical region, and select the n regions with the most significant morphological changes as the regions of interest;

[0014] According to the significance of the statistical differences, assign different weights to different regions of interest for weighted calculation;

[0015] Calculate the univariate morphometric index.

[0016] Preferably, the registration of the surface thickness information of the gray matter part and the anatomical region for all individuals includes:

[0017] Based on the constructed gray matter cortical triangular mesh and generate the corresponding spherical mesh based on the FreeSurfer software;

[0018] Estimate the spherical harmonic coefficients according to the spherical harmonic basis functions and by the least squares method;

[0019] Use a unified unit sphere as a template to register the left hemisphere cortical surface of different individuals according to the spherical harmonic basis functions on the template surface.

[0020] Preferably, in the estimation of the spherical harmonic coefficients according to the spherical harmonic basis functions and by the least squares method, the spherical harmonic coefficients are:

[0021]

[0022] where f lm is the spherical harmonic coefficient, is the spherical coordinate, is the spherical harmonic function.

[0023] Preferably, the spherical harmonic function is:

[0024]

[0025] where, Y lm is the spherical harmonic function, is the Legendre polynomial of order m.

[0026] Preferably, the obtaining of the group statistical differences of all points in each anatomical region based on the essential morphological structure features among groups and selecting the n regions with the most significant morphological changes as the regions of interest includes:

[0027] Compare the statistical results with a single sample to test the significance of the overall atrophy of each anatomical region;

[0028] Statistically analyze the differences obtained for each anatomical region;

[0029] Sort the statistical differences in anatomical regions and select the top n anatomical regions with the most significant differences as the regions of interest.

[0030] Preferably, according to the significance of statistical differences, the weights assigned to different regions of interest for weighted calculation are as follows:

[0031]

[0032] where p k is the p-value of the statistical difference when the kth region of interest is compared with a single sample.

[0033] Preferably, the calculation of the univariate morphometric index includes:

[0034] Calculate the average group statistical difference between the Alzheimer's disease group and the cognitively unimpaired group in the regions of interest;

[0035] Calculate the statistical difference between an individual subject and the cognitively unimpaired group;

[0036] Obtain the univariate morphometric index by comparing the similarity between the individual atrophy degree and the atrophy degree of the Alzheimer's disease group on predefined regions of interest.

[0037] Preferably, the calculation method of the univariate morphometric index is:

[0038]

[0039] where UMI is the univariate morphometric index, w k is the weight, is the atrophy degree of the Alzheimer's disease group, is the individual atrophy degree.

[0040] Compared with the prior art, the present application has at least the following beneficial effects:

[0041] By using the generated univariate morphometric index as a biomarker for early detection of Alzheimer's disease, compared with traditional univariate biomarkers, it has higher statistical power and can better characterize the morphological changes in the cerebral cortex caused by Alzheimer's disease. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Some specific embodiments of the present invention will be described in detail hereinafter with reference to the accompanying drawings in an exemplary and non-limiting manner. The same reference numerals in the drawings denote the same or similar components or parts. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings:

[0043] Figure 1Schematic diagram of the overall process of the present invention;

[0044] Figure 2 Schematic diagram of the anatomical region registration of the present invention;

[0045] Figure 3 Schematic diagram of the construction of the area weighted model of the present invention;

[0046] Figure 4 Comparison chart of the univariate morphometric index based on the generalized linear model and the univariate morphometric index based on the original thickness information;

[0047] Figure 5 Chart of the top six most significantly different regions of interest selected;

[0048] Figure 6 Comparison chart of the weighted univariate morphometric index and the unweighted univariate morphometric index;

[0049] Figure 7 Comparison chart of the univariate morphometric index and the existing commonly used Alzheimer's disease recognition methods. Specific implementation manners

[0050] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be clearly and completely described below in conjunction with specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the scope of protection of the present application.

[0051] The present invention selects the anatomical regions with significant statistical differences on the cortical surface as the regions of interest by analyzing the between-group differences. Since the atrophy degrees of each region of interest are different, that is, the atrophy degrees affected by Alzheimer's disease (AD) are different, if equal treatment is performed on each region of interest, the generated univariate morphometric index will not be able to accurately reflect the main impact of Alzheimer's disease on the cortical structure. The present invention performs different treatments on different regions of interest to generate a reliable and robust univariate morphometric index, so as to better reflect the morphological changes of the brain caused by Alzheimer's disease.

[0052] As Figure 1 shown, a univariate neurodegenerative biomarker for early detection of AD includes the following steps:

[0053] S1. Extract the gray matter part in the structural magnetic resonance imaging to be analyzed and perform surface structure reconstruction, and calculate the surface thickness information of the gray matter part.

[0054] Specifically, the structural magnetic resonance images to be analyzed are preprocessed using FreeSurfer software, segmented into white matter, gray matter, and cerebrospinal fluid, and the gray matter part is selected using FreeSurfer software for surface structure reconstruction to form the triangular mesh of the cerebral gray matter cortex and the spherical mesh it constitutes, and the thickness of the gray matter cortex is calculated.

[0055] S2. Divide the surface of the gray matter part into 34 anatomical regions.

[0056] Specifically, the Desikan / Killiany parcellation method is used to label the regions of the gray matter part during the processing of structural magnetic resonance images, and the surface of the gray matter part is divided into 34 anatomical regions.

[0057] S3. Register the surface thickness information and anatomical regions of the gray matter part of all individuals.

[0058] Specifically, as Figure 2 shown, it includes the following steps:

[0059] S31. Based on the constructed triangular mesh of the gray matter cortex and the corresponding spherical mesh generated based on FreeSurfer software.

[0060] S32. Estimate the spherical harmonic coefficients according to the spherical harmonic basis functions and by the least squares method.

[0061] Specifically, change the coordinates of the gray matter cortex mesh M to the spherical coordinate system of:

[0062]

[0063] According to the triangular mesh of the gray matter cortex and its spherical representation, combined with the spherical harmonic function Y lm , we can obtain the spherical harmonic coefficients:

[0064]

[0065] where the spherical harmonic function of order l and degree m is defined as

[0066]

[0067] where P l m is the Legendre polynomial of order m.

[0068] S33. Use a unified unit sphere as a template to register the cortical surfaces of the left hemispheres of different individuals according to the spherical harmonic basis functions on the template surface.

[0069] We use weighted spherical harmonic representation for registration:

[0070]

[0071] Among them, Y' lm is the spherical harmonic basis function of the unit sphere (template sphere).

[0072] In this way, the registration of the left hemisphere cortical surface of all individuals is completed. By using the same method when calculating the thickness feature, the thickness information of each registered vertex can be obtained.

[0073] In addition, after the above steps are completed, the correspondence of anatomical information among all subjects is established.

[0074] Specifically, after registration, since the positions of the newly registered vertices are different from those before, relying on the k-nearest neighbor (KNN) algorithm on the spherical parameter domain and the area-weighted model of the triangular mesh, the anatomical information of the sample points to be registered is redefined.

[0075] As Figure 3 shown, (a) is the mapping result on the original left cerebral cortex unit sphere, (b) shows the spherical coordinates of the local mapping result, the anatomical characteristics of three vertices A, B, and C are known, and then we use the k-nearest neighbor (KNN) algorithm to determine the position of the registration point in the original unit sphere, (c) indicates that the newly generated registration point P is located in the triangular element S ABC and (d) is the area-weighted model.

[0076] To determine the anatomical property of point P, we connect point P with the three vertices of the triangular element to form three small triangles S 1 , S 2 , S 3 . S 1 , S 2 , S 3 are the regions corresponding to the three vertices A, B, and C respectively. Among them, S 3 has the largest area. Therefore, we redefine the anatomical property of point P as the anatomical property of vertex C. In this way, the anatomical property of each vertex can be redefined. Finally, according to the anatomical partition information of FreeSurfer, there are a total of 34 anatomical regions.

[0077] S4. Extract the essential morphological structure features among groups.

[0078] Specifically, considering the influence of age and gender factors among individuals, the generalized linear model in SurfStat software can be used to eliminate the individual differences caused by age and gender and obtain the essential morphological structure features of the group.

[0079] In this paper, we used a generalized linear model to exclude the influence of individual differences in age and gender to obtain the intrinsic group morphological structure among the same group. To explore the influence of these factors on the univariate morphological measurement index, we directly used the original thickness characteristics for the control group without using the generalized linear model. The comparison results are as Figure 4 shown.

[0080] Experiments show that using the generalized linear model can eliminate the influence of individual differences in age and gender. It enables us to extract the intrinsic morphological changes of the cortical surface caused by Alzheimer's disease. Using the generalized linear model can improve the stability of the univariate morphological measurement index and enhance the statistical discrimination ability.

[0081] S5. Based on the morphological structure characteristics of the essence between groups, obtain the group statistical differences of all points in each anatomical region, and select the n regions with the most significant morphological changes as the regions of interest.

[0082] Specifically, it includes the following steps:

[0083] S51. Compare the statistical results with a single sample to test the significance of overall atrophy in each anatomical region.

[0084] Specifically, compare the group statistical results of all points in each anatomical region with a single-sample t to test the significance of overall atrophy in each anatomical region. Among them, the single-sample t is based on 0 as the standard.

[0085] S52. Statistically analyze the differences obtained in each anatomical region.

[0086] S53. Rank the statistical differences in the anatomical regions and select the top n anatomical regions with the most significant differences as the regions of interest.

[0087] Specifically, rank the statistical differences and select the top n anatomical regions with the most significant statistical differences as the regions of interest. This means that the regions of interest reflect the regions with the most significant morphological changes induced by AD.

[0088] As Figure 5 shown, in this embodiment, the top six anatomical regions with the most significant statistical differences are selected as the regions of interest.

[0089] S6. According to the significance of the statistical differences, assign different weights to different regions of interest for weighted calculation.

[0090] Specifically, since the degree of atrophy in each region of interest is different, in order to generate a more stable univariate morphological measurement index, we assigned different weights to different regions of interest according to the significance of the statistical differences.

[0091] The expression of the weight is:

[0092]

[0093] Among them, p k refers to the p-value of the statistical difference when the k-th region of interest is compared with a single sample.

[0094] In this paper, to highlight the different degrees of influence of Alzheimer's disease on six regions of interest, we assigned different weights to the six regions of interest according to the statistical significance of the overall atrophy in each region of interest. To verify the effect of weighting, we calculated the minimum sample size in weighted and unweighted forms respectively, where the unweighted form has a weight value w k = 1 to calculate the univariate morphometric index, and the results are as Figure 6 shown.

[0095] This indicates that weighted univariate morphometric index measurement may increase the statistical analysis sensitivity for detecting morphological changes. On the other hand, the weighting method for different regions of interest helps to enhance the influence of Alzheimer's disease on the morphological structure, thereby improving the generalization performance of the univariate morphometric index generation algorithm.

[0096] S7. Calculate the univariate morphometric index.

[0097] Specifically, it includes the following steps:

[0098] S71. Calculate the average group statistical difference between the Alzheimer's disease group and the cognitively unimpaired group in the region of interest;

[0099] S72. Calculate the statistical difference between the individual subject and the cognitively unimpaired group;

[0100] S73. Obtain the univariate morphometric index by comparing the similarity between the individual atrophy degree and the atrophy degree of the Alzheimer's disease group on the predefined region of interest.

[0101] Among them, the specific calculation method of the univariate morphometric index is:

[0102]

[0103] Among them, m k is the total number of vertices in the k-th region of interest, is the z-score of the i-th vertex in the k-th region of interest of the test individual, which is called the individual atrophy degree, is the z-score of the i-th vertex in the k-th region of interest of the Alzheimer's disease group, that is, the atrophy degree of the Alzheimer's disease group.

[0104] We will demonstrate the reduction of the minimum sample size by using the univariate morphometric index as a biomarker in clinical research.

[0105]

[0106] Among them, σ represents the standard deviation of biomarker changes, m and b respectively refer to the mean values of the biomarker at 24 months and at baseline for the longitudinal data, and C is a constant.

[0107] We estimated the minimum sample sizes for univariate morphometric indices, left hemisphere cortical volume measurements, and clinical scores over 24 months, such as the Clinical Dementia Rating Scale, the Mini-Mental State Examination, and the Alzheimer's Disease Assessment Scale - Cognitive, and the results are as Figure 7 shown.

[0108] Regardless of whether it is based on univariate morphometric indices, volume measurements, or clinical scores, the longitudinal Aβ+AD group has the smallest minimum sample size, followed by the longitudinal Aβ+MCI group, and the longitudinal Aβ+CU group has the largest minimum sample size. This indicates that the morphological changes in the Aβ+AD group are relatively the largest, followed by the Aβ+MCI group, and the morphological changes in the Aβ+CU group are relatively the smallest at baseline and at the 24-month follow-up. At the same time, the results show that the minimum sample sizes for volume measurements and clinical scores are both larger than our univariate morphometric indices, indicating that the univariate morphometric indices may be better able to detect the underlying morphological changes induced by Alzheimer's disease and can sensitively identify the degree of cortical morphometric abnormalities caused by neurodegenerative diseases.

[0109] For ease of description, spatial relative terms, such as "above", "over", "on the upper surface", "upper", etc., can be used here to describe the spatial positional relationship of a device or feature shown in the figure with other devices or features. It should be understood that the spatial relative terms are intended to encompass different orientations in use or operation in addition to the orientation depicted in the figure for the device. For example, if the device in the drawing is inverted, a device described as "above" or "over" other devices or structures will then be positioned "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both the orientation of "above" and "below". The device can also be positioned in other different ways (rotated 90 degrees or in other orientations), and the corresponding interpretations of the spatial relative descriptions used here will be made.

[0110] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they specify the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0111] It should be noted that the terms "first", "second", etc. in the description and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order different from those illustrated or described herein.

[0112] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for determining a univariate neurodegenerative biomarker for early detection of AD, characterized in that, comprising: extracting the gray matter part in the structural magnetic resonance image to be analyzed and performing surface structure reconstruction, and calculating the surface thickness information of the gray matter part; dividing the surface of the gray matter part into 34 anatomical regions; registering the surface thickness information of the gray matter part and the anatomical regions of all individuals; extracting the essential morphological structure features between groups; based on the essential morphological structure features between groups, obtaining the group statistical differences of all points in each anatomical region, and selecting the n regions with the most significant morphological changes as the regions of interest; assigning different weights to different regions of interest according to the significance of the statistical differences for weighted calculation; calculating a univariate morphological measurement index; The registering the surface thickness information of the gray matter part and the anatomical regions of all individuals includes: generating a corresponding spherical mesh based on the constructed gray matter cortex triangular mesh and based on the FreeSurfer software; estimating the spherical harmonic coefficients according to the spherical harmonic basis functions and by the least squares method; using a unified unit sphere as a template to register the left hemisphere cortex surfaces of different individuals according to the spherical harmonic basis functions on the template surface; In the estimating the spherical harmonic coefficients according to the spherical harmonic basis functions and by the least squares method, the spherical harmonic coefficients are: ， wherein, is the spherical harmonic coefficient, is the spherical coordinate, is the spherical harmonic function; The spherical harmonic functions are: ， Among them, , is a spherical harmonic function, is the Legendre polynomial of order m; In the assigning different weights to different regions of interest according to the significance of the statistical differences for weighted calculation, the weights are: ， Among them, is the p-value of the statistical difference when the k-th region of interest is compared with a single sample; The calculating a univariate morphological measurement index includes: calculating the average group statistical difference between the Alzheimer's disease group and the cognitively unimpaired group in the regions of interest; calculating the statistical difference between the individual subject and the cognitively unimpaired group; obtaining the univariate morphological measurement index by comparing the similarity between the individual atrophy degree and the atrophy degree of the Alzheimer's disease group on the predefined regions of interest.

2. The method for determining a univariate neurodegenerative biomarker for early detection of AD according to claim 1, characterized in that, The obtaining the group statistical differences of all points in each anatomical region based on the essential morphological structure features between groups and selecting the n regions with the most significant morphological changes as the regions of interest includes: comparing the statistical results with a single sample to test the significance of the overall atrophy of each anatomical region; statistically analyzing the differences of each obtained anatomical region; ranking the statistical differences of the anatomical regions and selecting the top n anatomical regions with the most significant differences as the regions of interest.

3. The method for determining a univariate neurodegenerative biomarker for early detection of AD according to claim 1, characterized in that, The calculation method of the univariate morphological measurement index is: ， wherein, is a univariate morphometric index, is the weight, is the atrophy degree of the Alzheimer's disease group, is the individual atrophy degree.

Images