Alzheimer's disease early screening method based on three-dimensional space modeling

Abstract

The application provides an Alzheimer's disease early screening method based on three-dimensional space modeling, and the method comprises the following steps: constructing a 3D virtual scene containing fixed background elements, wherein the fixed background elements at least include a floor, a blanket, a wall corner and a group of predefined fixed interference objects; setting at least N sofas as target stimuli in the 3D virtual scene; generating a test scene, introducing a visual interference variable by independently changing the sofa topology structure, adjusting the scene lighting condition and the shadow projection position, wherein the sofa topology structure and the lighting condition are parameterized and adjusted according to a predefined experimental design matrix, and the parameter change ensures that the scene satisfies an orthogonal or partial orthogonal statistical relationship between the stimulus attribute and the interference variable. The application can systematically generate standardized and multi-angle visual stimulation materials under the premise that the background and interference conditions are highly controllable, and is suitable for the field of Alzheimer's disease early screening.

Description

Technical Field

[0001] This invention relates to the field of visualization and digital platform development technology, and in particular to an Alzheimer's disease auxiliary assessment method based on three-dimensional spatial modeling and visualization technology, and a 3D scene modeling and image generation method for cognitive psychology or computer vision testing. Background Technology

[0002] Alzheimer's disease (AD) is a neurodegenerative disease with insidious onset and progressive development, characterized by memory impairment, executive function decline, visuospatial impairment, and orientation difficulties. With the increasing aging of the global population, early screening and intervention for Alzheimer's disease has become an important issue in the field of public health. Studies have shown that early identification and intervention can effectively slow disease progression and improve patients' quality of life.

[0003] Currently, clinical screening for Alzheimer's disease primarily relies on neuropsychological scales, such as the Montreal Cognitive Assessment (MoCA) and the Mini-Mental State Examination (MMSE). These scales comprehensively evaluate the cognitive state of test subjects through multi-dimensional testing, including language, memory, visuospatial, and executive function. However, traditional scale assessment methods have the following limitations: 1. Reliance on professional assessors; the test needs to be conducted one-on-one by trained medical personnel, which is time-consuming and the assessment results are easily affected by the subjective factors of the assessor, resulting in insufficient standardization.

[0004] 2. The testing process is fixed and rigid; traditional tests are mostly based on paper questionnaires or fixed question and answer patterns, making it difficult to dynamically adjust the test difficulty or content and lacking personalized assessment capabilities.

[0005] 3. The assessment methods for visual space and executive function are limited; existing tests that assess abilities such as visual space, graphics drawing, and multi-task execution rely heavily on manual scoring, which is inefficient and inconsistent.

[0006] 4. Data is difficult to manage and analyze digitally; paper-based test results are not conducive to long-term tracking, big data statistical analysis, and cross-institutional sharing, which limits screening efficiency and scientific research value.

[0007] In recent years, with the development of artificial intelligence and computer vision technologies, existing technologies have applied machine learning methods to cognitive assessment, such as semantic understanding detection based on speech analysis and visuospatial ability assessment based on hand-drawn image recognition. However, existing technical solutions still have the following problems: 1. Most systems only analyze a single modality (such as speech or handwriting) and lack the fusion and evaluation of multimodal information; 2. Automatic scoring of hand-drawn graphics is mostly based on simple rules or traditional image processing methods, and has limited ability to recognize complex features such as drawing completeness and spatial relationships; 3. The system is not usually deeply integrated with standardized cognitive assessment scales (such as MoCA), and its clinical effectiveness and practicality need to be verified.

[0008] Therefore, there is an urgent need for an Alzheimer's disease assisted screening system that can integrate multimodal information collection, intelligent analysis and standardized assessment to improve screening efficiency, reduce reliance on professionals, and provide objective and quantifiable data support for early identification. Summary of the Invention

[0009] This invention relates to a method for early screening of Alzheimer's disease based on three-dimensional spatial modeling, in order to solve at least one of the above-mentioned technical problems.

[0010] To solve the above-mentioned technical problems, the present invention adopts the following technical means: A test scenario generation method, the method comprising: Construct a 3D virtual scene containing fixed background elements, wherein the fixed background elements include at least a floor, a blanket, a corner, and a set of predefined fixed interference objects, wherein the fixed interference objects include items placed on a table, on the floor, and in a corner; In the 3D virtual scene, at least N sofas are set as target stimuli, and the three-dimensional models of the sofas are selected from a predefined sofa model library containing at least N+1 different topological structures; where N is a positive integer greater than or equal to 4. A test scenario is generated with the sofa positions fixed. Visual interference variables are introduced by independently changing the topology of each sofa, adjusting the scene lighting conditions and shadow projection positions. The sofa topology and lighting conditions are parameterized according to a predefined experimental design matrix, and the parameter changes ensure that the scene satisfies an orthogonal or partially orthogonal statistical relationship between stimulus attributes and interference variables. The lighting condition adjustment includes changing at least one of the light source brightness or direction. Two-dimensional images of the test scene instance are generated from M fixed spatial observation points, and the spatial coordinate relationship of the M observation points is fixed; where M is a positive integer greater than or equal to 7.

[0011] In some embodiments, the test scenario uses the center of the coffee table as the center and a predetermined radius to determine a horizontal circular reference line; The fixed spatial observation points are seven in number. The arrangement of the seven fixed spatial observation points is centered on a preset starting angle on a horizontal circular reference line to achieve a uniform distribution of the seven observation points, and adjacent observation points are located on different circular lines in the vertical direction.

[0012] The present invention also provides a test scenario generation system for implementing the method described above, the system comprising: Scene configuration module; the scene configuration module is used to load the basic 3D virtual scene template and receive configuration instructions for sofa parameters and lighting parameters; An instantiation generation module is used to generate and store multiple different test scenario instances according to configuration instructions; A batch rendering engine is used to read each test scene instance, automatically call the renderer to render from M fixed observation points, and output M images corresponding to each instance.

[0013] In some embodiments, each test scenario instance corresponds to a set of seven images, which are rendered from the same instance from seven fixed observation points.

[0014] This invention also provides an early screening method for Alzheimer's disease based on three-dimensional spatial modeling, which applies the test scenario generation system described above. The method includes: Show the subjects a rendered image of the first observation point corresponding to the target test scenario instance and the preset control duration; After the first image is displayed, a selection interface containing four candidate images is presented, one of which is the correct target image and three are incorrect interference images. The correct target image is the second observation point rendered image of the same instance as the first image, and the incorrect interference images are instances from other test scene instances. Record the subjects' selection results and generate an evaluation score based on whether the correct target image was selected.

[0015] In some embodiments, generating an evaluation score based on whether the correct target image is selected specifically includes: If the subject selects the correct target image, they will receive a positive score for that question. If the subject selects an incorrect distracting image, the corresponding negative score will be deducted from that question. The scores for each question are added together to obtain the initial total score.

[0016] In some embodiments, the first observation point and the second observation point are both arbitrary points among M fixed observation points, and the preset control duration is 8 seconds.

[0017] In some embodiments, the test scene instance corresponding to the erroneous interference image, compared with the scene instance corresponding to the correct image, satisfies the requirement of minimizing a preset difference in sofa topology and positional distribution to control the interference intensity. set up: Sc={s1,s2,s3,s4} is the set of structural features of the four sofas in the correct option; Se={s1′,s2′,s3′,s4′} is the set of structural features of the four sofas in the incorrect options; d(si,si′) is the difference function describing the topology, size parameters or shape of the corresponding sofa; The scene-level difference function is then defined as follows:

[0018] in, Let be the weight coefficient of the i-th sofa; When the selected error option satisfies: When D(Sc, Se)≤δ, where δ is a pre-set difference threshold, the incorrect option is considered to meet the condition of minimizing the preset difference.

[0019] This invention also provides an early screening system for Alzheimer's disease based on three-dimensional spatial modeling, comprising: The display control unit is used to precisely control the display timing and duration of the first image and the selection interface; The option logic unit is used to automatically select the correct target image and the incorrect interference image from the image dataset product and construct a selection interface; An input capture unit is used to receive the subject's selection operation; The scoring output unit is used to automatically calculate and present the evaluation result based on the selected operation and the correct answer.

[0020] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described above.

[0021] In one or more of the above specific embodiments, the Alzheimer's disease early screening method based on three-dimensional spatial modeling provided by the present invention relates to the fields of computer vision and three-dimensional scene modeling technology. It constructs a three-dimensional virtual scene containing fixed background elements and a set of predefined interference objects; sets at least four sofas as target stimuli in this scene; generates test scene instances with visual interference variables by independently changing the topological structure of each sofa position and adjusting the scene lighting conditions; and renders corresponding two-dimensional image sets from seven preset fixed spatial observation points. The present invention also discloses a system implemented based on this method, the generated image dataset product, and a method and interactive testing system for assessing psychological cognition or machine vision capabilities. The present invention can systematically generate standardized, multi-view visual stimulus materials under highly controllable background and interference conditions, and is applicable to fields such as early screening for Alzheimer's disease, screening for mild cognitive impairment, visual cognition experiments, and computer vision model testing. It has the advantages of strong repeatability, high parameter controllability, and good scalability. Attached Figure Description

[0022] Embodiments of the invention will now be described by way of example only, with reference to the accompanying schematic diagrams, wherein: Figure 1 A flowchart of the test scenario generation method provided by the present invention; Figure 2 This is a schematic diagram of the layout of the test scenario in an embodiment of the present invention; Figure 3 A structural block diagram of the test scenario generation system provided by this invention; Figure 4 A flowchart of the Alzheimer's disease early screening method based on three-dimensional spatial modeling provided by the present invention; Figures 5-10 This is a simulation diagram of an embodiment of the present invention; Figure 11 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0023] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0024] In one specific implementation, such as Figure 1 As shown, the test scenario generation method provided by this invention includes the following steps: S110: Construct a 3D virtual scene containing fixed background elements, wherein the fixed background elements include at least a floor, a blanket, a corner, and a set of predefined fixed interference objects, wherein the fixed interference objects include items placed on a table, on the ground, and in a corner; S120: In the 3D virtual scene, at least N sofas are set as target stimuli, and the three-dimensional models of the sofas are selected from a predefined sofa model library containing at least N+1 different topological structures; where N is a positive integer greater than or equal to 4; S130: Generate a test scene with fixed sofa positions. Introduce visual interference variables by independently changing the topology of each sofa, adjusting the scene lighting conditions and shadow projection positions. The sofa topology and lighting conditions are parameterized according to a predefined experimental design matrix, and the parameter changes ensure that the scene satisfies an orthogonal or partially orthogonal statistical relationship between stimulus attributes and interference variables. The lighting condition adjustment includes changing at least one of the light source brightness or direction. S140: Render two-dimensional images of the test scene instance from M preset fixed spatial observation points, respectively, with the spatial coordinate relationship of the M observation points fixed; where M is a positive integer greater than or equal to 7.

[0025] The above test scenario uses the center of the coffee table as the center and a predetermined radius to determine a horizontal circular reference line; there are seven fixed spatial observation points. The arrangement of the seven fixed spatial observation points is centered on a preset starting angle on the horizontal circular reference line to achieve a uniform distribution of the seven observation points, and adjacent observation points are located on different circular lines in the vertical direction.

[0026] The following is combined with Figure 2 This paper will use a specific scenario as an example to introduce the overall implementation process of the test scenario generation method provided by the present invention.

[0027] In this specific scenario, the test scenario generation method includes the following steps: S1: Construct a 3D virtual scene containing fixed background elements, wherein the fixed background elements include at least a floor, a blanket, a corner, and a set of predefined fixed distractions; wherein the set of predefined fixed distractions includes items placed on a table, on the floor, and in a corner; S2: In the 3D virtual scene, at least four sofas are set as target stimuli; wherein, the three-dimensional model shape of the sofas as target stimuli is selected from a predefined sofa model library, which contains at least five sofa models with different topological structures. S3: Generate a specific test scene instance where the sofa's position is fixed. Visual interference variables are introduced by independently changing the topology of each sofa in its fixed position (e.g., adjusting its size, shape, texture, etc.) and adjusting the lighting conditions and shadow projection positions in the scene. The sofa topology and lighting conditions are parametrically adjusted according to a predefined experimental design matrix. The adjustment of lighting conditions includes changing at least one of the brightness or direction of the light source. The sofa topology and lighting condition parameters are varied according to the predefined experimental design matrix to ensure that the generated scene satisfies an orthogonal or partially orthogonal statistical relationship between stimulus attributes and interference variables. S4: For the generated test scene instance, render corresponding two-dimensional images from seven preset fixed spatial observation point positions. The spatial coordinate relationship of the seven observation points is preset and remains consistent in all instances; the arrangement of the seven fixed spatial observation points includes: Horizontal arrangement: Using the center of the coffee table in the 3D virtual scene as the center, determine a horizontal circular reference line with a predetermined radius; Vertical arrangement: Using a 45-degree downward angle as the reference pitch angle, and adding +5 degrees and -5 degrees of pitch offset to this reference angle to generate three parallel horizontal loops; Camera position selection: On the horizontal circular reference line, select a starting angle according to the actual scene layout, and with this angle as the center, distribute the seven observation points evenly along the horizontal circular reference line; in the vertical direction, the seven observation points are respectively located on the three horizontal circular lines, and adjacent observation points are located on different circular lines in the vertical direction to achieve sequential and spaced distribution in the vertical direction.

[0028] It should be noted that a parameterized test scenario was designed to study the impact of sofa type and lighting changes on visual perception. The scenario includes four sofas in fixed positions, each with six independently selectable topological structures. Global lighting conditions are adjustable to introduce visual interference variables. The specific parameterization scheme and experimental design matrix are as follows.

[0029] 1. Parameter definition and adjustment method 1.1 Sofa Type Parameters Each sofa has 6 different topologies, represented by integer variables: Sofa type S1∈{0,1,2,3,4,5} Sofa types S2∈{0,1,2,3,4,5} Sofas come in 3 types, S3∈{0,1,2,3,4,5} Sofas come in 4 types, S4∈{0,1,2,3,4,5} Adjustment method: Six different sofa models were pre-created in 3D modeling software, ensuring they had the same dimensional references and positional anchor points. Based on the experimental matrix, the corresponding model file was loaded for each fixed position using a script, keeping the sofa position unchanged.

[0030] 1.2 Illumination Parameters Illumination conditions are controlled by two continuous variables. To simplify the experimental design, each variable is used at two levels: Azimuth angle θ of light source: Low horizontal 30 High level 60 (Rotate horizontally around the center of the scene) Light source intensity I: Low level 100 lux, High level 200 lux The two variables are combined into a lighting variable L∈{0,1,2,3}, and the mapping relationship is as follows: L=0L=0:θ=30 I=100 L=1L=1:θ=30 I=200 L=2L=2:θ=60 I=100 L=3L=3:θ=60 I=200 Adjustment method: Set a single main light source in the rendering engine, adjust the azimuth and intensity according to the value of L, and ensure that the shadow casting position changes accordingly. Keep other lighting parameters (such as ambient light and shadow quality) constant.

[0031] 2. Experimental Design Matrix A full factorial design was employed, with all possible parameter combinations forming the experimental matrix. Each experimental condition was represented by a vector (S1, S2, S3, S4, L), where Si ∈ {0, 1, 2, 3, 4, 5} and L ∈ {0, 1, 2, 3}. The total number of experiments was 64 × 4 = 1296 × 4 = 5184. The mathematical expression of the matrix is: M = {(s1, s2, s3, s4, l) | si ∈ {0, 1, 2, 3, 4, 5}, i = 1, ..., 4; l ∈ {0, 1, 2, 3}}.

[0032] Thus, the method provided by this invention, by constructing a 3D virtual scene containing fixed background elements and fixed interfering objects, can eliminate the interference of background environment differences on test results, ensuring the consistency and comparability of test scenes; by selecting at least N+1 different sofa models with different topologies as target stimuli, and N≥4, it can provide rich dimensions of stimulus variation, meet the needs of multivariate testing, and avoid the testing limitations caused by a single stimulus; by parametrically adjusting the sofa topology, lighting conditions, and shadow positions, and ensuring that the stimulus attributes and interfering variables satisfy orthogonal or partially orthogonal statistical relationships, the interfering variables can be independently controllable, facilitating accurate analysis of the impact of single or multivariate interactions on test results, and improving the scientificity and accuracy of the test; by setting M≥7 fixed spatial observation points and rendering two-dimensional images, it can capture the characteristics of the test scene from multiple angles and all directions, simulate the differences in different observation perspectives in real scenes, avoid test bias caused by a single perspective, and provide comprehensive and diverse image data support for subsequent scene-based evaluations (such as AD-assisted evaluation), ensuring the reliability and comprehensiveness of the evaluation results.

[0033] Furthermore, by using the center of the coffee table as the center and determining a horizontal circular reference line with a predetermined radius, the observation points can be distributed around the core area of ​​the scene (the area where the sofa is located). This ensures that each observation point can effectively capture the key features of the target stimulus (the sofa) and avoid invalid images caused by the observation points deviating from the core of the test. The seven observation points are evenly distributed on the horizontal circular reference line, and adjacent observation points are located on different circular lines in the vertical direction. On the one hand, this achieves 360° all-round coverage in the horizontal direction. On the other hand, the height difference in the vertical direction forms a multi-angle and multi-layer observation perspective, which can more comprehensively present the three-dimensional structural features of the sofa and the scene lighting and shadow effects. This further enhances the diversity and comprehensiveness of the rendered images and provides richer perspective data for subsequent evaluation. At the same time, the fixed arrangement of observation points can ensure that the observation conditions of different test scene instances are consistent, ensuring the repeatability and comparability of the test.

[0034] On the other hand, the present invention also provides a test scenario generation system for implementing the method described above, such as... Figure 3 As shown, the system includes: Scene configuration module 310; the scene configuration module is used to load a basic 3D virtual scene template and receive configuration instructions for sofa parameters and lighting parameters; The instantiation generation module 320 is used to generate and store multiple different test scenario instances according to configuration instructions; Batch rendering engine 330 is used to read each test scene instance, automatically call the renderer to render from M fixed observation points, and output M images corresponding to each instance.

[0035] In this way, the scene configuration module can quickly build the basic framework of the test scene by loading 3D virtual scene templates, reducing repetitive modeling work and improving scene construction efficiency. At the same time, it can receive configuration instructions for sofa parameters and lighting parameters, realize flexible adjustment of parameters, meet different testing needs, and enhance the versatility of the system. The instantiation generation module can generate and store multiple different test scene instances in batches according to configuration instructions, which can obtain a large number of test samples at one time, avoiding the tedious operation of manual generation one by one, improving the efficiency of test scene generation, and facilitating subsequent comparative testing of different scene instances. The batch rendering engine can automatically call the renderer to batch render each scene instance from M fixed observation points and output the corresponding images, realizing the automation of the rendering process, greatly reducing the cost of manual operation, improving the efficiency of image generation, and ensuring that the rendering conditions (including observation points, rendering parameters, etc.) of each scene instance are consistent, ensuring the uniformity and comparability of the rendered images, and providing efficient and standardized data support for subsequent image-based evaluation work.

[0036] Specifically, each test scene instance in the batch rendering engine 330 corresponds to a set of seven images, which are rendered from the same instance from seven fixed observation points. Each test scene instance corresponds to a set of seven images from different observation points, which can fully present the visual characteristics of the same test scene from different perspectives, especially the three-dimensional structure, lighting effects, and shadow changes of the target stimulus (the sofa in this example), providing comprehensive image evidence for subsequent evaluations (such as the assessment of the subject's spatial cognition and perspective switching ability). The seven images correspond to seven fixed observation points, ensuring that the image perspectives of the same instance have a fixed pattern of difference, and that the image perspectives of different instances remain consistent, facilitating subsequent comparative analysis of images of different instances. At the same time, a set of seven images can form a complete set of perspectives, avoiding evaluation bias caused by incomplete information from a single perspective image, and improving the accuracy and reliability of the evaluation results.

[0037] Furthermore, this invention also provides an early screening method for Alzheimer's disease based on three-dimensional spatial modeling, applying the test scenario generation system described above, such as... Figure 4 As shown, the method includes the following steps: S410: Show the subject the rendered image of the first observation point corresponding to the target test scenario instance and the preset control duration; S420: After the first image is displayed, a selection interface containing four candidate images is presented, one of which is the correct target image and three are incorrect interference images; the correct target image is the second observation point rendered image of the same instance as the first image, and the incorrect interference images are instances from other test scene instances; S430: Record the subject's selection results and generate an evaluation score based on whether the correct target image was selected. Specifically, in the test scenario, the evaluation score quantifies the subject's (or model's) selection results for a series of questions, each requiring the selection of the correct target image from four candidate images. The score mapping rules are as follows: Scoring: The test consists of 15 questions, with a total score of 100 points.

[0038] Scoring rules: If the subject selects the correct target image, they will receive a positive score for that question. If the subject selects an incorrect distracting image, the corresponding negative score will be deducted from that question. Total score calculation: Add up the scores (positive or negative) of each question to get the initial total score.

[0039] Meaning of full marks: A full score of 100 points means that you have answered all questions correctly and have not answered any questions incorrectly. In this case, the total score is the sum of all positive scores.

[0040] Thus, the method provided by this invention simulates a brief observation scenario in a real-world setting by showing the subject a first observation point image of a fixed duration, testing the subject's ability to quickly capture and remember scene information. It displays a selection interface containing one correct image and three incorrect interference images. By comparing the correct image from different perspectives within the same instance with the incorrect interference images from different instances, it can accurately test the subject's ability to recognize the spatial structure of the scene, the characteristics of the target stimulus, and their perspective switching ability. This is particularly suitable for cognitive function assessment in AD patients, enabling early detection of abnormalities in spatial cognition, memory, and visual recognition. By recording the selection results and generating an assessment score, the assessment process can be standardized and quantified, avoiding the subjectivity of manual assessment and improving the objectivity and accuracy of the assessment results. This provides a scientific and operable assessment method for early screening, auxiliary diagnosis, and disease monitoring of AD.

[0041] Furthermore, adopting a scoring method that adds positive points and deducts points for errors can enhance the discriminatory power of the assessment. It can not only distinguish between correct and incorrect choices, but also reflect the degree of error in the subject's selection through the deduction mechanism (e.g., repeated selection of incorrect interference images will lead to a decrease in the total score), thus more accurately quantifying the subject's cognitive ability level. The initial total score obtained by summing the scores of each question can realize the quantitative presentation of the assessment results, which is convenient for horizontal comparison of assessment results of different subjects and for longitudinal tracking of assessment results of the same subject at different times, providing clear data support for monitoring the progression of Alzheimer's disease. At the same time, standardized scoring rules can ensure the consistency and comparability of assessment results of different assessors and different assessment batches, improving the standardization and practicality of the assessment method.

[0042] Specifically, both the first and second observation points are arbitrary points from among M fixed observation points, with a preset control duration of 8 seconds. The arbitrary selection of both observation points from the M fixed points increases the randomness and diversity of the assessment, avoids assessment bias caused by fixed viewpoint combinations, and more comprehensively tests the subject's adaptability to different viewpoint transitions and scene recognition ability. The preset control duration of 8 seconds, verified through extensive experiments, ensures that the subject has sufficient time to capture key information in the image (such as sofa topology and scene lighting features) while avoiding excessively long durations that could lead to insufficient memory load and ineffective testing of memory ability. Furthermore, the fixed duration of 8 seconds ensures consistent observation conditions for all subjects, guaranteeing the fairness and repeatability of the assessment and further improving the accuracy and reliability of the assessment results.

[0043] The test scene instances corresponding to the above erroneous interference images, compared with the scene instances corresponding to the correct images, satisfy the requirement of minimizing the preset difference in sofa topology and positional distribution to control the interference intensity. set up: Sc={s1,s2,s3,s4} is the set of structural features of the four sofas in the correct option; Se={s1′,s2′,s3′,s4′} is the set of structural features of the four sofas in the incorrect options; d(si,si′) is the difference function describing the topology, size parameters or shape of the corresponding sofa; The scene-level difference function is then defined as follows:

[0044] in, Let be the weight coefficient of the i-th sofa; When the selected error option satisfies: When D(Sc, Se)≤δ, where δ is a pre-set difference threshold, the incorrect option is considered to meet the condition of minimizing the preset difference.

[0045] Thus, the method provided by this invention ensures that the erroneous interference images have appropriate interference by controlling the difference in sofa topology and positional distribution between the scene instances corresponding to the erroneous interference images and the scene instances of the correct images. This prevents the interference from being too weak (where subjects can easily distinguish between correct and incorrect options, making it impossible to effectively test their cognitive abilities) due to excessive difference, or too strong (where subjects cannot distinguish between correct and incorrect options, affecting the effectiveness of the assessment). By defining a scene-level difference function and combining it with a weighting coefficient wi, the impact of different sofa structural features on scene differences can be precisely quantified, achieving refined control of interference intensity. A preset difference threshold δ ensures that the interference intensity of all erroneous interference images remains consistent, avoiding assessment bias caused by differences in interference intensity. This further improves the scientific rigor and accuracy of the assessment method, making it particularly suitable for assessing cognitively sensitive populations such as AD patients, enabling more precise capture of subtle changes in their cognitive abilities.

[0046] This invention also provides an early screening system for Alzheimer's disease based on three-dimensional spatial modeling, comprising: The display control unit is used to precisely control the display timing and duration of the first image and the selection interface; The option logic unit is used to automatically select the correct target image and the incorrect interference image from the image dataset product and construct a selection interface; An input capture unit is used to receive the subject's selection operation; The scoring output unit is used to automatically calculate and present the evaluation result based on the selected operation and the correct answer.

[0047] Thus, in the system provided by this invention, the display control unit precisely controls the display sequence and duration of the images and selection interface, ensuring the standardization of the assessment process, avoiding assessment errors caused by chaotic display sequence or duration deviations, ensuring consistent assessment conditions for all subjects, and improving the fairness and reliability of the assessment. The option logic unit automatically selects correct and incorrect interference images and constructs the selection interface, automating the assessment process, reducing manual intervention and lowering manual operation costs. Simultaneously, it selects interference images through preset logic rules, ensuring consistency in interference intensity and improving the standardization of the assessment. The input capture unit quickly and accurately receives the subjects' selection operations, avoiding assessment deviations caused by input delays or recording errors, and ensuring the authenticity and accuracy of the selection results. The scoring output unit automatically calculates and presents the assessment results, enabling rapid output of assessment results and improving assessment efficiency. Furthermore, the standardized scoring output format facilitates quick reading and analysis of assessment results by medical personnel, providing efficient and convenient system support for early screening, auxiliary diagnosis, and disease management of AD, and enhancing the clinical applicability of the assessment work.

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are for illustrative purposes only and are not intended to limit the invention.

[0049] Example 1: Digital Platform Development Technology for Scene Memory Cognitive Assessment System This scene memory and cognition assessment system operates within the "Future Blue Dolphin" medical and health service system. The system's user terminal side supports mobile smart terminals (smartphones, tablets, etc.) equipped with 5G mobile communication networks. The terminals access the server through mainstream web browsers (such as Safari and Chrome). The server adopts a B / S architecture, with a business logic module developed in Java and a MySQL database (storing the assessment question bank, user answer data, and assessment results) in the backend. The frontend uses HTML5+CSS3+JavaScript to implement page rendering and interaction logic. The timing function and image switching function are triggered and executed by JavaScript.

[0050] Example 2: 3D Indoor Space Modeling and Visualization Technology This embodiment details how to use the method and system of the present invention to generate a standardized multi-view sofa image dataset for visual cognition research in batches.

[0051] 1. System preparation and environment configuration: The system modules of this invention are developed using open-source or commercial 3D modeling and rendering software (such as Blender, Unity, or Unreal Engine) as the underlying rendering engine and based on their APIs or script interfaces.

[0052] Build a basic 3D virtual scene template: Create a virtual room, including a wooden-textured floor, a corner of the room, and a rug with a specific pattern. Predefine the placement of distracting objects within the room, such as: placing a doll in the corner, several cushions on the floor, and a water glass, an orange bowl, and two fox dolls on the table. The 3D models and textures of these distracting objects are built once and remain unchanged in all subsequent scene instances.

[0053] Establish a sofa model library: Design or collect at least five 3D sofa models with significantly different topologies, such as: a two-seater armless sofa, a single-seater high-backed armchair, and a modern minimalist curved sofa. All models need to undergo geometric simplification and UV unwrapping for efficient rendering and subsequent parametric processing.

[0054] 2. Parameter configuration and scene instantiation: Researchers can set the number of scene instances to be generated (e.g., 100) and parameter rules through the graphical interface or configuration file of the scene configuration module.

[0055] Define the experimental design matrix: each row of the matrix represents a combination of parameters for a scenario instance. For example, it may contain the following factors: Factor A: Sofa models placed at the four sofa positions (P1, P2, P3, P4) (selected from the model library).

[0056] Factor B: The brightness or direction of the light source.

[0057] The system automatically executes the instantiation generation module based on the design matrix (which can use full factorial or partial factorial design): For the i-th instance, the sofa models with the corresponding numbers are placed sequentially at positions P1-P4, the light source is adjusted to the specified direction, the ambient light brightness is set, and the shadow calculation of the rendering engine is enabled. After generation, the complete 3D scene description file of the instance is saved.

[0058] 3. Multi-view batch rendering: The batch rendering engine reads all saved scene instance files.

[0059] For each instance file, the engine automatically places the virtual cameras sequentially at seven preset observation points. The observation point layout is based on a virtual hemisphere centered on the scene center (coffee table), including a horizontal reference ring and three vertical ring lines (corresponding to a 45° overhead view and its ±5° deviation). The cameras are evenly distributed horizontally and spaced out vertically, with all camera lenses pointed at the center area of ​​the scene.

[0060] The rendering engine renders an RGB image with a resolution of 1920×1080 pixels from each viewpoint, keeping the rendering settings consistent.

[0061] Finally, each scene instance outputs an image group containing 7 images, and is sorted by "instance". The image is stored using the rule "ID_observation point number.jpg".

[0062] Example 3: Specific Procedures of Scene Memory Cognitive Assessment Method Combined with appendix Figure 5-10 The specific implementation steps of this scenario-based memory and cognitive assessment method are as follows: 1. Tutorial Demonstration and Skip Section After a user accesses the system via a terminal, the system automatically loads the scene memory cognitive assessment tutorial module, which contains three consecutively displayed tutorial pages: The first tutorial page displays a reference image of the room scene, along with text instructions: "You need to take 8 seconds to carefully observe the objects in the scene, and pay special attention to the placement of the sofa; subsequently, you will select the image from four pictures that has the exact same position as all the objects in the initial scene. You will have 20 seconds to answer this question. Just focus on whether the sofa has been changed, and do not worry about changes in lighting or perspective."

[0063] The second tutorial page displays the original scene image and four alternative images (marked with correct / incorrect indicators), along with explanations of the options: "The criteria for determining the correct answer is that the sofa position is consistent with the original image, and changes in lighting / viewpoint do not affect the judgment; the incorrect options are incorrect because the sofa has been replaced and does not match the original scene."

[0064] The third tutorial page displays an example of the test results interface (including visual displays of two results: 13% high risk and 73% low risk), along with a result feedback explanation: "This interface provides a test summary to help understand health risks. If you have any questions, please consult a doctor or health advisor."

[0065] Each tutorial page has a "Skip Tutorial" button at the bottom. When a user clicks this button, the system immediately stops displaying the tutorial and jumps to the formal assessment stage.

[0066] 2. Formal evaluation launch phase After the tutorial, the system loads the assessment launch page, which displays cartoon medical staff images and text prompts: "Question 1: You have 8 seconds to observe the image, and then 20 seconds to select the correct answer from 4 options." A "Start" interactive button is also provided. After the user clicks the "Start" button, the system triggers the formal assessment process, entering the scene image observation stage.

[0067] 3. Scene Image Observation Phase Upon receiving the "Start" command, the system immediately displays an image of the original room scene on the terminal page (including key visual elements such as a white irregularly shaped sofa, a round coffee table, and wooden floors), and starts an 8-second countdown timer in the background (the countdown process is not visually displayed). Users must observe and memorize the objects in the scene (especially the sofa's placement) before the 8-second countdown ends.

[0068] 4. Answering the multiple-choice questions After the 8-second countdown ends, the system automatically switches to the alternative options display page. This page shows four different room scene images as alternatives and starts a 20-second countdown timer in the background, with the countdown progress displayed visually on the left side of the page as a blue progress bar. Of the four alternative images, only one image has a sofa placement that perfectly matches the original scene image; the sofas in the other three images have been changed or rearranged. The user must select the alternative image they believe to be correct before the 20-second countdown ends; if the user does not complete the selection before the 20-second countdown ends, the system will automatically determine that the answer to the question is incorrect.

[0069] 5. Evaluation Result Generation Stage After a single question is answered, the system can repeat steps 3-4 based on a pre-set assessment question bank to complete multiple-question assessments. After all questions are answered, the system calls the results analysis module to generate cognitive assessment results based on the percentage of questions answered correctly out of the total number of questions. A score of 74% or higher indicates a normal population; 60%-74% indicates a low-risk population; 46%-60% indicates a medium-risk population; and below 46% indicates a high-risk population. The system will display the assessment results in percentage values ​​and risk level text visualizations, and simultaneously store user responses and assessment results in the backend database.

[0070] 6. Core Judgment Logic Image matching judgment logic: The system backend pre-stores the sofa position feature vector data of the original scene image. The candidate image must completely match the feature vector data to be judged as the correct answer; slight changes in light intensity and shooting angle are not included in the reference dimensions of feature matching.

[0071] Timing logic: The 8-second observation timer and the 20-second response timer are both implemented by JavaScript. After the timer ends, the page will automatically redirect or an error will be detected.

[0072] The result classification logic is as follows: the percentage threshold for cognitive risk level is: above 74% is considered normal, 60%-74% is considered low-risk, 46%-60% is considered medium-risk, and below 46% is considered high-risk.

[0073] Result classification criteria: This study included 95 participants, who were divided into three groups based on their mmse values: >27, 24 ≤ mmse ≤ 27, and 20 ≤ mmse < 24 (group 1: n=31; group 2: n=35; group 3: n=29). Baseline characteristics of each group are shown in Table 1.

[0074] Table 1

[0075] (1) Group design and sample characteristics To construct a clinically discriminative grading standard, we divided 95 participants into three groups with different clinical cognitive levels based on the internationally recognized Mini-Mental State Examination (MMSE) score, to simulate a continuous spectrum from normal to impaired cognition: Group 1 (Cognitive Normal Group, n=31): MMSE > 27 points; Group 2 (mild cognitive decline group, n=35): 24 points ≤ MMSE ≤ 27 points; Group 3 (significant cognitive decline, n=29): 20 points ≤ MMSE < 24 points, The baseline demographic and neuropsychological characteristics of the three groups of participants are shown in Table 1. The trend of increasing age and decreasing scores on various cognitive tests (MMSE, MoCA) verifies that this grouping can effectively reflect the cognitive function level at different stages.

[0076] (2) Determination of core classification threshold The core indicator used for grading is the accuracy rate of scene memory assessment. As shown in Table 1, the average accuracy rates of the three groups of participants show significant and regular differences: The average accuracy rate of Group 1 (the cognitively normal group) was 88.67%; The average accuracy rate for Group 2 (mild cognitive decline group) was 62.00%; The average accuracy rate for Group 3 (the group with a significant decline in cognitive ability) was 44.75%. Based on this data distribution and referring to the standard grading points for clinical cognitive assessment, we established the following percentage thresholds for four cognitive risk levels: Normal population: accuracy >74%. This threshold is higher than the upper bound of Group2, aiming to identify individuals with intact cognitive function with high specificity.

[0077] Low-risk group: accuracy rate 60%–74%. This range covers the average level of Group 2 (62.00%), representing a mild risk of cognitive decline.

[0078] Medium-risk group: accuracy rate 46%–60%. This range falls between Group 2 and Group 3, indicating a further increase in the risk of cognitive decline.

[0079] High-risk group: accuracy <46%. This threshold is consistent with the average level of Group 3 (44.75%) and is used to identify high-risk individuals with obvious cognitive impairment.

[0080] Preliminary evidence of the effectiveness of the classification To verify the statistical significance of the differences in task accuracy among the above groups, we performed the Kruskal-Wallis test. The results showed that the differences in accuracy among the three groups were highly statistically significant (H = 11.719, P = 0.003). Further post-hoc pairwise comparisons (Bonferroni test) indicated that: The accuracy rate of Group 1 was significantly higher than that of Group 2 and Group 3 (p<0.05). This result statistically confirms that there are fundamental differences in task accuracy among the three clinically significant groups categorized according to the MMSE. The mean accuracy rates of the three groups precisely fall within our defined risk intervals (Group 1 → normal interval, Group 2 → low-risk interval, Group 3 → high-risk interval). This highly consistent correspondence provides strong data support for the criterion validity of the current grading threshold, indicating that this grading standard can effectively distinguish groups with different cognitive function states.

[0081] Example 4: Implementation of Clinical Cognitive Assessment Based on This System The image dataset generated by this invention was applied to screening for mild cognitive impairment (MCI). Subjects completed a multi-view recognition task in a standardized environment, and the system recorded their accuracy and reaction time. A risk assessment model was established by comparing the data with that of a healthy control group. Clinical validation showed that the system has high sensitivity and specificity in distinguishing between MCI and healthy individuals, and can be used as an auxiliary diagnostic tool.

[0082] In one or more of the above specific embodiments, the Alzheimer's disease early screening method based on three-dimensional spatial modeling provided by the present invention relates to the fields of computer vision and three-dimensional scene modeling technology. It constructs a three-dimensional virtual scene containing fixed background elements and a set of predefined interference objects; sets at least four sofas as target stimuli in this scene; generates test scene instances with visual interference variables by independently changing the topological structure of each sofa position and adjusting the scene lighting conditions; and renders corresponding two-dimensional image sets from seven preset fixed spatial observation points. The present invention also discloses a system implemented based on this method, the generated image dataset product, and a method and interactive testing system for assessing psychological cognition or machine vision capabilities. The present invention can systematically generate standardized, multi-view visual stimulus materials under highly controllable background and interference conditions, and is applicable to fields such as early screening for Alzheimer's disease, screening for mild cognitive impairment, visual cognition experiments, and computer vision model testing. It has the advantages of strong repeatability, high parameter controllability, and good scalability.

[0083] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 10As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and model predictions. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The model predictions of the computer device store static and dynamic information data. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps in the above method embodiments.

[0084] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device to which the present invention is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0085] Corresponding to the above embodiments, this invention also provides a computer storage medium containing one or more program instructions. These one or more program instructions are used to execute the method described above.

[0086] The present invention also provides a computer program product, the computer program product including a computer program, the computer program being stored on a non-transitory computer-readable storage medium, and the computer being able to perform the above-described method when the computer program is executed by a processor.

[0087] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0088] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0089] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0090] Among them, non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0091] Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus RAM (DRRAM).

[0092] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0093] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0094] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A test scenario generation method characterized by, The method includes: Construct a 3D virtual scene containing fixed background elements, wherein the fixed background elements include at least a floor, a blanket, a corner, and a set of predefined fixed interference objects, wherein the fixed interference objects include items placed on a table, on the floor, and in a corner; In the 3D virtual scene, at least N sofas are set as target stimuli, and the three-dimensional models of the sofas are selected from a predefined sofa model library containing at least N+1 different topological structures; where N is a positive integer greater than or equal to 4. A test scenario is generated with the sofa positions fixed. Visual interference variables are introduced by independently changing the topology of each sofa, adjusting the scene lighting conditions and shadow projection positions. The sofa topology and lighting conditions are parameterized according to a predefined experimental design matrix, and the parameter changes ensure that the scene satisfies an orthogonal or partially orthogonal statistical relationship between stimulus attributes and interference variables. The lighting condition adjustment includes changing at least one of the light source brightness or direction. Two-dimensional images of the test scene instance are generated from M fixed spatial observation points, and the spatial coordinate relationship of the M observation points is fixed; where M is a positive integer greater than or equal to 7.

2. The method of claim 1, wherein, The test scenario uses the center of the coffee table as the center and a predetermined radius to determine a horizontal circular reference line; The fixed spatial observation points are seven in number. The arrangement of the seven fixed spatial observation points is centered on a preset starting angle on a horizontal circular reference line to achieve a uniform distribution of the seven observation points, and adjacent observation points are located on different circular lines in the vertical direction.

3. A test scenario generation system for implementing the method according to claim 1 or 2, characterized in that The system includes: Scene configuration module; the scene configuration module is used to load the basic 3D virtual scene template and receive configuration instructions for sofa parameters and lighting parameters; An instantiation generation module is used to generate and store multiple different test scenario instances according to configuration instructions; A batch rendering engine is used to read each test scene instance, automatically call the renderer to render from M fixed observation points, and output M images corresponding to each instance.

4. The test scenario generation system of claim 3, wherein, Each test scenario instance corresponds to a set of seven images, which are rendered from the same instance from seven fixed observation points.

5. A method for early screening of Alzheimer's disease based on three-dimensional spatial modeling, using the test scenario generation system as described in claim 3 or 4, characterized in that, The method includes: Show the subjects a rendered image of the first observation point corresponding to the target test scenario instance and the preset control duration; After the first image is displayed, a selection interface containing four candidate images is presented, one of which is the correct target image and three are incorrect interference images. The correct target image is the second observation point rendered image of the same instance as the first image, and the incorrect interference images are instances from other test scene instances. Record the subjects' selection results and generate an evaluation score based on whether the correct target image was selected.

6. The method according to claim 5, characterized in that, An evaluation score is generated based on whether the correct target image is selected, specifically including: If the subject selects the correct target image, they will receive a positive score for that question. If the subject selects an incorrect distracting image, the corresponding negative score will be deducted from that question. The scores for each question are added together to obtain the initial total score.

7. The method according to claim 5, characterized in that, Both the first observation point and the second observation point are arbitrary points among the M fixed observation points, and the preset control duration is 8 seconds.

8. The method according to claim 5, characterized in that, The test scene instance corresponding to the erroneous interference image, compared with the scene instance corresponding to the correct image, satisfies the requirement of minimizing the preset difference in sofa topology and position distribution to control the interference intensity. set up: Sc={s1,s2,s3,s4} is the set of structural features of the four sofas in the correct option; Se={s1′,s2′,s3′,s4′} is the set of structural features of the four sofas in the incorrect options; d(si,si′) is the difference function describing the topology, size parameters or shape of the corresponding sofa; The scene-level difference function is then defined as follows: in, Let be the weight coefficient of the i-th sofa; When the selected error option satisfies: When D(Sc, Se)≤δ, where δ is a pre-set difference threshold, the incorrect option is considered to meet the condition of minimizing the preset difference.

9. An early screening system for Alzheimer's disease based on three-dimensional spatial modeling, characterized in that, The system includes: The display control unit is used to precisely control the display timing and duration of the first image and the selection interface; The option logic unit is used to automatically select the correct target image and the incorrect interference image from the image dataset product and construct a selection interface; An input capture unit is used to receive the subject's selection operation; The scoring output unit is used to automatically calculate and present the evaluation result based on the selected operation and the correct answer.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 5-8.

Images