An alzheimer's disease risk prediction system based on brain images

Abstract

The application provides a kind of Alzheimer's disease risk prediction system based on brain image, comprising: brain image receiving unit, for receiving the brain image of subject;Temporal muscle thickness acquisition unit, for obtaining the temporal muscle thickness of subject according to brain image;Alzheimer's disease risk analysis unit, for comparing gender-specific cutoff value according to the temporal muscle thickness of subject, predict the result of subject suffering from Alzheimer's disease;Specific cutoff value is: male 8.40mm, female 7.51mm;When the temporal muscle thickness is less than or equal to the cutoff value, the subject is at high risk of suffering from Alzheimer's disease;When the temporal muscle thickness is greater than the cutoff value, the subject is at low risk of suffering from Alzheimer's disease.The application can solve the technical problem of lack of non-invasive, simple and easy-to-promote early screening tool for Alzheimer's disease in the prior art.

Description

Technical Field

[0001] This invention relates to the field of neurodegenerative disease screening technology, specifically to an Alzheimer's disease risk prediction system based on cranial imaging. Background Technology

[0002] Alzheimer's disease (AD) is the most common neurodegenerative disease and type of dementia, imposing a huge social and economic burden globally. The current gold standard for AD diagnosis relies on positron emission tomography (PET / CT) to detect Aβ or Tau protein deposition in the brain, cerebrospinal fluid biomarker testing, and neurological function scale scoring. However, these existing technologies all have limitations: PET / CT is expensive and requires sophisticated equipment, making it difficult to implement at the grassroots level; cerebrospinal fluid biomarker testing has significant invasiveness, resulting in many patients not receiving early diagnosis and intervention; and neurological function scale assessments are susceptible to subjective influence from the test takers, leading to inaccurate results and affecting clinical diagnosis.

[0003] Therefore, there is an urgent clinical need for a non-invasive, simple, and easily promoted early screening tool for Alzheimer's disease (AD). Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes an Alzheimer's disease risk prediction system based on cranial imaging, in order to solve the technical problem of the lack of non-invasive, simple, and easily promoted early screening tools for AD.

[0005] The technical solution adopted in this invention is as follows: Firstly, an Alzheimer's disease risk prediction system based on cranial imaging is provided, including: A cranial imaging receiving unit is used to receive cranial images of the subject. Temporalis muscle thickness acquisition unit, used to acquire the temporalis muscle thickness of the subject based on cranial imaging; The Alzheimer's disease risk analysis unit is used to predict the outcome of Alzheimer's disease by comparing the subject's temporalis muscle thickness with a gender-specific cutoff value.

[0006] Furthermore, the cranial imaging includes magnetic resonance imaging (MRI) images.

[0007] Furthermore, the step of obtaining the temporalis muscle thickness of the subject based on cranial images includes: extracting the temporalis muscle tissue using medical image analysis software on an axial view of the cranial images; and obtaining the maximum thickness of the temporalis muscle from a direction perpendicular to the long axis of the temporalis muscle on an image showing the thickest layer of the temporalis muscle as the subject's temporalis muscle thickness.

[0008] Furthermore, the maximum thickness of the left and right temporalis muscles was measured separately, and the average of the two was taken as the subject's temporalis muscle thickness.

[0009] Furthermore, the sex-specific cutoff values ​​are: 8.40 mm for males and 7.51 mm for females; when the temporalis muscle thickness is less than or equal to the cutoff value, the subject is at high risk of developing Alzheimer's disease; when the temporalis muscle thickness is greater than the cutoff value, the subject is at low risk of developing Alzheimer's disease.

[0010] Furthermore, it also includes a results output unit for generating an assessment report containing a conclusion of "screening positive / high risk" or "screening negative / low risk".

[0011] Secondly, a biomarker for predicting the risk of Alzheimer's disease and a method for obtaining cutoff values ​​are provided, including the following steps: We collected cranial imaging and clinical data from patients diagnosed with Alzheimer's disease by specialists and from a sex- and age-matched healthy control group. The temporalis muscle thickness of all subjects was obtained based on cranial imaging data; Statistical analysis showed that the thickness of the temporalis muscle was negatively correlated with the presence of Alzheimer's disease. Using receiver operating characteristic (ROC) curve analysis, with Alzheimer's disease diagnosis as the state variable and temporalis muscle thickness as the test variable, ROC curves were plotted and the area under the curve was calculated. The optimal cutoff value for temporalis muscle thickness to distinguish between Alzheimer's disease and healthy cognition was determined by maximizing the Youden index.

[0012] Furthermore, statistical analysis revealed the correlation between temporalis muscle thickness and Alzheimer's disease, including: first, comparing the differences in temporalis muscle thickness among Alzheimer's disease patients with different CDR grades using the Kolmogorov-Smirnov test; and then performing a correlation analysis using Spearman to determine the correlation between temporalis muscle thickness and CDR and MMSE scores.

[0013] Furthermore, partial correlation analysis was used to validate the correlation analysis results, correcting for the effects of age, gender, education level, family history, and / or BMI on the correlation analysis results.

[0014] Furthermore, the optimal cutoff values ​​include: 8.40 mm for males and 7.51 mm for females.

[0015] As can be seen from the above technical solution, the beneficial technical effects of the present invention are as follows: It can assist doctors in using cranial medical imaging examinations that are already widely used in clinical practice, so as to achieve rapid initial screening of AD without additional expensive equipment or invasive procedures. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0017] Figure 1 This is a schematic diagram of the Alzheimer's disease risk prediction system architecture based on cranial imaging in an embodiment of the present invention; Figure 2 This is an AD-ROC curve of the thickness of the temporalis muscle in a female embodiment of the present invention; Figure 3 This is an AD-ROC curve of the thickness of the male temporalis muscle in an embodiment of the present invention. Detailed Implementation

[0018] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0019] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0020] The inventors of this application discovered through research that sarcopenia and Alzheimer's disease (AD) share a comorbidity mechanism and may be an independent risk factor for AD. Because of its anatomical proximity to the hippocampus and temporal lobe cortex, key pathological areas in AD, the temporal muscle may release pro-inflammatory factors or reduce actin secretion through muscle contraction, exacerbating neuroinflammation and synaptic damage in the adjacent temporal lobe, thereby accelerating the pathological progression of AD. Temporal muscle thickness (TMT) is a non-invasive physiological indicator that can be measured in patients. On a brain magnetic resonance imaging (MRI) image, temporal muscle thickness is displayed perpendicular to the long axis of the temporal muscle, with the lateral fissure and orbital roof as references. The measurable muscle thickness measurement shows high inter-measurement consistency, excellent reliability, and is easily incorporated into routine clinical practice.

[0021] The inventors of this application, through further large-sample retrospective studies, confirmed a negative correlation between temporalis muscle thickness and AD disease in the AD population, and that the degree of thinning was correlated with the clinical severity of AD according to the Mini-Mental State Examination (MMSE) and Clinical Dementia Rating (CDR) stages. Based on this, statistical methods were used to determine the optimal TMT cutoff value for distinguishing AD patients from healthy individuals. The specific experimental verification process for the negative correlation between temporalis muscle thickness and AD disease, and the determination of the optimal TMT cutoff value, is as follows: Step S1: Collect cranial imaging data and clinical data from AD patients diagnosed by specialists and healthy controls matched for sex and age. This study retrospectively collected imaging data of Han Chinese in Southwest China who underwent cranial MRI examinations at the hospital between December 2024 and September 2025. Ultimately, 242 patients with PET-confirmed Alzheimer's disease (AD) (according to CDR classification: 72 mild, 77 moderate, and 93 severe), 28 patients with mild cognitive impairment (MCI), and 100 age- and sex-matched healthy controls were included.

[0022] Step S2: Obtain the temporalis muscle thickness of all subjects based on cranial imaging data. A head MRI was performed using a Siemens MAGNETOM ESSENZA 1.5T medical magnetic resonance imaging system. The original T1-weighted images (T1WI) of the brain in DICOM format (slice thickness 0.8-1.5 mm to ensure measurement accuracy) were imported into an offline workstation, and the temporalis muscle was extracted from the original images using Syngo.via VB20A analysis software.

[0023] The thickness of the temporalis muscle of the subject to be screened is obtained from the original image. On the axial view of the original image, the temporalis muscle tissue is automatically extracted using medical image analysis software (such as Syngovia). The software has a built-in judgment program that can extract the temporalis muscle tissue using the Sylvian fissure and the top of the orbit as anatomical reference landmarks without the need for manual annotation.

[0024] Based on the extracted temporalis muscle tissue, the maximum thickness of the temporalis muscle is measured in a direction perpendicular to the long axis of the temporalis muscle on an image showing the thickest layer of the temporalis muscle. In some embodiments, in order to obtain the optimal TMT cutoff value, the maximum thickness of the left and right temporalis muscles is measured separately, and the average of the two is used as the temporalis muscle thickness of the subject to be screened.

[0025] Step S3: Statistical analysis confirmed a negative correlation between temporalis muscle thickness and Alzheimer's disease (AD). First, intergroup comparisons were performed using the Kolmogorov-Smirnov test to compare the differences in TMT among AD patients with different CDR grades. Then, correlation analysis was conducted using Spearman's correlation analysis to explore the correlation between TMT and CDR scores and MMSE scores. Details are as follows: The detailed procedure for intergroup comparisons using the Kolmogorov-Smirnov test is as follows: Step 1: Construct the empirical cumulative distribution function; sort the data from smallest to largest. For any value x, the empirical CDFS(x) is equal to the number of data points less than or equal to x divided by the total number of data points n, resulting in S1(x).

[0026] Step 2: Determine the distribution to be compared; Two-sample test: Compare with the empirical distribution S2(x) of another sample.

[0027] Step 3: Calculate the core statistic D; for all data points, calculate the absolute value of the difference between the two CDF values ​​and find the maximum value. The formula is: D = max | S1(x) - S2(x) | Step 4: Statistical Inference. Based on the sample size n (or m, n) and the significance level (e.g., p < 0.05), look up the KS test critical value table and compare the calculated D with the critical value. If D is greater than the critical value, there is sufficient evidence to suggest a significant difference between the two distributions; otherwise, there is no significant difference. Specifically: A total of 342 participants were included, including 100 healthy controls, 72 with mild dementia, 77 with moderate dementia, and 93 with severe dementia. Participants in the AD group (p<0.001) were older than those in the healthy controls. Patients in the AD group (p<0.01) had a thinner TMT than those in the healthy controls. There was no significant difference in TMT between moderate and severe AD, but the TMT in moderate AD was thinner than that in mild AD (p<0.05).

[0028] The detailed process of correlation analysis using Spearman is as follows: Step 1: Data Ranking: Rank the original data of the two variables from smallest to largest.

[0029] Step 2: Calculate the difference: Calculate the rank difference (dᵢ) for each pair of data points.

[0030] Step 3: Substitute into the formula: In the above formula, n is the number of data pairs (i.e., the sample size), and dᵢ represents the rank difference of the i-th data pair (i.e., the rank difference between the two variables), which is determined as follows: >0 indicates a positive correlation (a higher ranking in one component tends to lead to a higher ranking in the other). <0 indicates a negative correlation; ≈0 indicates no correlation.

[0031] Data from 242 AD patients were analyzed, and TMT and MMSE scores showed a correlation. =0.265, p<0.01), TMT and CDR scores were negatively correlated ( =-0.263, p<0.01).

[0032] In some embodiments, to correct for potential confounding factors such as age, gender, education level, family history, and / or BMI (body mass index), partial correlation analysis is further used to verify the results of the above correlation analysis. Taking controlling for one variable Z as an example, the simple correlation coefficients r between X and Y, X and Z, and Y and Z are first calculated. XY r XZ and r YZ Calculate using the following formula: In the above formula, n is the sample size, and X... ik It is variable X i The k-th observation, X jk It is variable X j The kth observation, It is variable X i The average value, It is variable X j The average value.

[0033] Then, calculate the partial correlation coefficient between X and Y after controlling Z using the following formula: In the above formula, X represents the thickness of the temporalis muscle, Y represents a clinical scoring indicator, such as the MMSE score or CDR score, and Z represents confounding variables that need to be controlled, including factors that may affect both X and Y, such as age, sex, education level, family history, and / or BMI.

[0034] Partial correlation analysis, after adjusting for potential confounding factors such as age, gender, education level, family history, and BMI, confirmed that TMT and MMSE scores were still correlated (r = 0.255, p < 0.01), while TMT and CDR scores were negatively correlated (r = -0.286, p < 0.01).

[0035] Step S4: Using receiver operating characteristic (ROC) curve analysis, with AD diagnosis as the state variable and temporalis muscle thickness as the test variable, plot the ROC curve and calculate the area under the curve (AUC). Determine the optimal cutoff value for distinguishing AD from health cognition using the Youden index maximization principle. When plotting the ROC curve, the candidate threshold sequence is first determined. In this embodiment, the predicted values ​​of all samples are used as candidate thresholds. Then, the TPR and FPR corresponding to each candidate threshold are calculated.

[0036] The true positive rate (TPR, sensitivity) represents the ability to correctly identify positive results; the higher the better. It is calculated as follows: The false positive rate (FPR, 1 - specificity) represents the probability of incorrectly classifying a negative result as a positive result; the lower the better. It is calculated as follows: For each candidate threshold, count the number of TP (true positives), FP (false positives), TN (true negatives), and FN (false negatives), and substitute them into the above formula to calculate the corresponding TPR and FPR.

[0037] Next, plot the ROC curve: using FPR (horizontal axis, X-axis) as the independent variable and TPR (vertical axis, Y-axis) as the dependent variable, connect the calculated (FPR, TPR) data points sequentially according to the threshold from smallest to largest (or from largest to smallest) to form the ROC curve. Finally, calculate the area under the curve (AUC) to evaluate the diagnostic efficacy of TMT, and determine the optimal cutoff value for distinguishing AD from health cognition using the Youden index maximization principle. First, based on the plotted ROC curve (horizontal axis FPR = 1 - TNR, vertical axis TPR), calculate the J value for each candidate threshold using the following formula: Then find the threshold with the largest J value, which is the optimal diagnostic threshold (this threshold corresponds to the "point closest to the top left corner" on the ROC curve, since the top left corner of the ROC curve represents the ideal state of TPR=1 and TNR=1).

[0038] In the specific trial, 160 female AD patients, 14 MCI patients, and 49 female healthy controls were included. ROC curves were plotted with AD as the dependent variable. Diagnostic efficacy: AUC (Area Under Curve) 0.818 (95% CI: 0.762-0.875), cutoff value ≤7.51 mm, sensitivity 61.3%, specificity 88.9%, and Youden index 0.501. 82 male AD patients, 14 MCI patients, and 51 male healthy controls were also included. ROC curves were plotted with AD as the dependent variable. Diagnostic efficacy: AUC (Area Under Curve) 0.863 (95% CI: 0.804-0.922), cutoff value ≤8.40 mm, sensitivity 86.6%, specificity 73.8%, and Youden index 0.604.

[0039] Using the Youden index maximization principle, the optimal cutoff value for distinguishing AD from healthy cognition was determined: ≤8.40mm for men and ≤7.51mm for women; that is, when the temporalis muscle thickness is less than or equal to this cutoff value, the subject is at high risk of developing Alzheimer's disease; when the temporalis muscle thickness is greater than this cutoff value, the subject is at low risk of developing Alzheimer's disease.

[0040] Based on the above research findings, this embodiment provides an Alzheimer's disease risk prediction system based on cranial imaging. This system achieves effective initial screening for Alzheimer's disease by determining a novel use of temporalis muscle thickness as a biomarker and its specific cutoff value. The system includes: A cranial imaging receiving unit is used to receive cranial images of a subject, including magnetic resonance images. In specific implementations, the method of receiving cranial images is not limited and can be implemented in any feasible manner in the prior art, such as directly inputting the subject's cranial images into the medical image analysis software Syngo.via.

[0041] The temporalis muscle thickness acquisition unit is used to acquire the temporalis muscle thickness of the subject based on cranial images. Specifically, the temporalis muscle tissue is extracted using medical image analysis software (such as Syngovia) on the axial view of the cranial images, and the maximum thickness of the temporalis muscle is acquired from the direction perpendicular to the long axis of the temporalis muscle on the image showing the thickest layer of the temporalis muscle as the subject's temporalis muscle thickness. In some embodiments, the maximum thickness of the left and right temporalis muscles is measured separately, and the average of the two is taken as the subject's temporalis muscle thickness.

[0042] The Alzheimer's disease risk analysis unit is used to predict the outcome of Alzheimer's disease based on the subject's temporalis muscle thickness by comparing it with gender-specific cutoff values ​​(male ≤8.40mm, female ≤7.51mm). The results can be divided into two types: "screening positive / high risk" or "screening negative / low risk".

[0043] The results output unit is used to generate an assessment report containing a conclusion of "screening positive / high risk" or "screening negative / low risk". If the average temporalis muscle thickness is lower than or equal to the cutoff value for the corresponding sex, it is determined to be Alzheimer's disease screening positive / high risk; if it is higher than the cutoff value, it is determined to be screening negative / low risk. The implementation of the results output unit is not limited and can be implemented in any feasible manner in the prior art.

[0044] By using the aforementioned brain imaging-based Alzheimer's disease risk prediction system, doctors can be assisted in conducting rapid initial screening for AD using the widely available brain MRI examinations, without the need for additional expensive equipment or invasive procedures.

[0045] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. An Alzheimer's disease risk prediction system based on cranial imaging, characterized in that, include: A cranial imaging receiving unit is used to receive cranial images of the subject. Temporalis muscle thickness acquisition unit, used to acquire the temporalis muscle thickness of the subject based on cranial imaging; The Alzheimer's disease risk analysis unit is used to predict the outcome of Alzheimer's disease by comparing the subject's temporalis muscle thickness with a gender-specific cutoff value.

2. The Alzheimer's disease risk prediction system based on cranial imaging according to claim 1, characterized in that, The cranial images include magnetic resonance images.

3. The Alzheimer's disease risk prediction system based on cranial imaging according to claim 1, characterized in that, The method of obtaining the temporalis muscle thickness of the subject based on cranial images includes: extracting temporalis muscle tissue using medical image analysis software on an axial view of cranial images; and obtaining the maximum thickness of the temporalis muscle from a direction perpendicular to the long axis of the temporalis muscle on an image showing the thickest layer of the temporalis muscle as the subject's temporalis muscle thickness.

4. The Alzheimer's disease risk prediction system based on cranial imaging according to claim 3, characterized in that, The maximum thickness of the left and right temporalis muscles was measured separately, and the average of the two measurements was taken as the subject's temporalis muscle thickness.

5. The Alzheimer's disease risk prediction system based on cranial imaging according to claim 1, characterized in that, The sex-specific cutoff values ​​are: 8.40 mm for males and 7.51 mm for females. When the temporalis muscle thickness is less than or equal to the cutoff value, the subject is at high risk of developing Alzheimer's disease; when the temporalis muscle thickness is greater than the cutoff value, the subject is at low risk of developing Alzheimer's disease.

6. The Alzheimer's disease risk prediction system based on cranial imaging according to claim 1, characterized in that, It also includes a results output unit for generating an assessment report containing a "screening positive / high risk" or "screening negative / low risk" conclusion.

7. A biomarker for predicting the risk of Alzheimer's disease and a method for obtaining cutoff values, characterized in that, Includes the following steps: We collected cranial imaging and clinical data from patients diagnosed with Alzheimer's disease by specialists and from a sex- and age-matched healthy control group. The temporalis muscle thickness of all subjects was obtained based on cranial imaging data; Statistical analysis showed that the thickness of the temporalis muscle was negatively correlated with the presence of Alzheimer's disease. Using receiver operating characteristic (ROC) curve analysis, with Alzheimer's disease diagnosis as the state variable and temporalis muscle thickness as the test variable, ROC curves were plotted and the area under the curve was calculated. The optimal cutoff value for temporalis muscle thickness to distinguish between Alzheimer's disease and healthy cognition was determined by maximizing the Youden index.

8. The biomarker and cutoff value acquisition method for predicting the risk of Alzheimer's disease according to claim 7, characterized in that, Statistical analysis revealed the correlation between temporalis muscle thickness and Alzheimer's disease, including: first, comparing the differences in temporalis muscle thickness among Alzheimer's disease patients with different CDR grades using the Kolmogorov-Smirnov test; and then performing a Spearman correlation analysis to determine the correlation between temporalis muscle thickness and CDR and MMSE scores.

9. The biomarker and cutoff value acquisition method for predicting the risk of Alzheimer's disease according to claim 8, characterized in that, Partial correlation analysis was used to validate the correlation analysis results, and the influence of age, gender, education level, family history and / or BMI on the correlation analysis results was adjusted.

10. The biomarker and cutoff value acquisition method for predicting the risk of Alzheimer's disease according to claim 7, characterized in that, The optimal cutoff values ​​include: 8.40 mm for males and 7.51 mm for females.

Images