Construction method and application of metabolic disorder degree evaluation model

Abstract

The invention relates to the technical field of medical care informatics, and discloses a construction method and application of a metabolic disorder degree evaluation model, and the method comprises the steps: learning the facial non-uniformity features, the facial sequence features and the facial symmetry features of a healthy group and a metabolic disorder group through a model based on a facial front infrared image of a target group; a high-precision metabolic disorder degree evaluation model is obtained through training, the evaluation result can be quantitatively output and visually displayed, simple, sensitive and economical metabolic disorder degree evaluation is achieved, medical workers are helped to find patients suffering from chronic diseases in time and formulate proper treatment schemes for the patients, and the treatment efficiency is improved. The assessment process is noninvasive, green, objective and accurate, the psychological burden and economic burden of the patient are effectively relieved, and meanwhile waste of medical resources is prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of medical care informatics, and in particular to a method for constructing a metabolic disorder degree assessment model and its application. Background Art

[0002] Chronic diseases develop slowly and have a long course. From healthy people to mild metabolic abnormalities and then to the occurrence of related chronic diseases, it can be understood as the process of occurrence, development and aggravation of metabolic abnormalities in the body. Metabolic disorders are an important predisposing factor for a variety of chronic diseases, and can significantly increase the risk of cardiovascular and cerebrovascular diseases and diabetes in the long term. Therefore, early screening and intervention of metabolic abnormalities can significantly reduce medical costs and have considerable socioeconomic value. However, patients with metabolic abnormalities generally have low detection rates and insufficient intervention. On the one hand, this is because the disease is hidden at this stage, patients lack obvious symptoms, and the rate of medical consultation and compliance are low; on the other hand, the assessment of metabolic conditions relies on professional medical institutions and personnel, and the process is time-consuming and cumbersome.

[0003] Therefore, there is an urgent need for a method to construct a metabolic disorder assessment model to achieve a simple, sensitive and economical metabolic disorder assessment, to assist medical staff in timely identifying patients with chronic diseases and to develop appropriate treatment plans for patients. Summary of the invention

[0004] The present invention provides a method for constructing a metabolic disorder degree assessment model and application thereof, so as to solve the defect of lacking a simple, sensitive and economical metabolic disorder degree assessment method.

[0005] The present invention provides a method for constructing a metabolic disorder degree assessment model, comprising:

[0006] Acquire frontal infrared images of faces of a target group, wherein the target group includes a healthy group and a metabolic disorder group;

[0007] According to a preset facial segmentation rule, the frontal infrared image of the face of each individual in the target group is divided into multiple regions, and the facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group are extracted from the multiple regions of the frontal infrared image of the face of each individual in the target group;

[0008] According to the facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group, the model is used to learn the facial non-uniformity features, facial sequential features, and facial symmetry features of the healthy group and the metabolic disorder group, and a metabolic disorder degree assessment model is trained.

[0009] In one embodiment, the multiple regions include any one of the following items or any combination thereof: left frontal region, right frontal region, left eye region, right eye region, nose region (including nose bridge region and nose tip region), lip region, mandibular region, left cheek region, right cheek region, left ear region, and right ear region, wherein, in the frontal infrared image of the face, the position of the left frontal region is symmetrical to that of the right frontal region, the position of the left eye region is symmetrical to that of the right eye region, the position of the left cheek region is symmetrical to that of the right cheek region, and the position of the left ear region is symmetrical to that of the right ear region.

[0010] In one embodiment, the preset face segmentation rules include:

[0011] The forehead area enclosed by the line from the hairline to the eyebrows of the individual in the frontal infrared image of the face is divided into a left forehead area and a right forehead area according to the midline of the frontal infrared image of the face;

[0012] The area bounded by the eye sockets of the individual in the frontal infrared image of the face is divided into the eye area, and the midline of the frontal infrared image of the face is used to divide the left eye area and the right eye area;

[0013] The triangular area from the root of the nose to the lower end of the nose tip and the sides of the nose wings in the frontal infrared image of the face is divided into the nose area, including the nose bridge area and the nose tip area;

[0014] The left front cheek and the right front cheek enclosed by the outer side of the nasolabial groove to the outer edge of the cheek and the lower orbit to the two mandibular angles in the face front infrared image are divided into the left front cheek area and the right front cheek area; the area enclosed by the lower orbital edge, the nasolabial line, the mandibular angle line and the front ear line on the left and right sides of the head of the individual in the face front infrared image is divided into the left cheek area and the right cheek area;

[0015] The circular area of ​​the individual including the left and right lip corners and the oral cavity as the boundary in the frontal infrared image of the face is divided into the lip area;

[0016] The area from the lower edge of the lips to the edge of the jaw and the downward line of the corner of the mouth in the frontal infrared image of the face is divided into the jaw area.

[0017] In one embodiment, extracting facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group from multiple regions of the frontal infrared image of the face of each individual in the target group includes:

[0018] For each of the multiple regions of the frontal infrared image of the face of each individual in the target group, the number of pixels in each region and the temperature data of each pixel are obtained;

[0019] According to the number of pixels in each area and the temperature data of each pixel, the average temperature and standard deviation of each area are obtained;

[0020] The facial non-uniformity of each region is obtained according to the temperature mean value and the temperature standard deviation of each region. The collection of facial non-uniformity of multiple regions of an individual forms the facial non-uniformity feature of the individual.

[0021] In one embodiment, the expression for the temperature average is:

[0022]

[0023] In the expression of the average temperature, μ T represents the average temperature of the Tth region, T i represents the temperature of the i-th pixel in the region, and N represents the number of pixels in the region;

[0024] The expression for the temperature standard deviation is:

[0025]

[0026] In the expression of temperature standard deviation, σ T represents the temperature standard deviation of the Tth region, T i Represents the temperature of the i-th pixel in the area, μ T represents the average temperature of the Tth region, and N represents the number of pixels in the region;

[0027] The expression of facial non-uniformity is:

[0028]

[0029] In the expression of facial non-uniformity, NU T represents the facial non-uniformity of the Tth region, μ T represents the average temperature of the Tth region, σ T Represents the temperature standard deviation of the Tth region.

[0030] In one embodiment, extracting facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group from multiple regions of the frontal infrared image of the face of each individual in the target group also includes:

[0031] For the frontal infrared image of each individual in the target group, according to the average temperature of each of the multiple regions, by testing whether there is a significant difference in the average temperature between the multiple regions, multiple significant regions are obtained from the multiple regions;

[0032] Sort and rank multiple significant regions according to their average temperature from low to high to obtain the facial sequence of each individual in the target group;

[0033] According to the average temperature of multiple significant areas of each individual in the healthy group in the target group, the standard facial sequence of the healthy group is obtained;

[0034] According to the facial sequentiality of each individual in the target group and the standard facial sequentiality of the healthy group, the facial sequentiality deviation index of each individual in the target group that deviates from the standard facial sequentiality of the healthy group is obtained, and the facial sequentiality deviation index of each individual in the target group is used as the facial sequentiality feature of each individual in the target group.

[0035] In one embodiment, the step of obtaining the standard facial sequence of the healthy group based on the average temperature values ​​of multiple significant regions of each individual in the healthy group in the target group comprises:

[0036] Taking the multiple significant regions of each individual in the healthy group in the target group as the multiple significant regions of the healthy group, and obtaining the total temperature average of each significant region in the multiple significant regions of the healthy group according to the temperature average of the multiple significant regions of each individual in the healthy group in the target group;

[0037] The multiple significant areas of the healthy group are sorted and ranked from low to high according to their total temperature averages to obtain the standard facial sequentiality of the healthy group.

[0038] In one embodiment, the expression of the facial sequential deviation index is:

[0039]

[0040] In the expression of facial sequential deviation index, SDI i represents the facial sequential deviation index of the i-th individual in the target group, A j represents the rank of the jth significant region of interest in the temperature series representation of the healthy group, B j represents the rank of the jth significant region of interest in the temperature sequence representation of the i-th individual (subject) in the target population, n represents the number of significant regions of interest, and k represents the iteration variable, which ranges from 1 to A positive integer between .

[0041] In one embodiment, extracting facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group from multiple regions of the frontal infrared image of the face of each individual in the target group also includes:

[0042] According to the midline of the frontal infrared image of the face, the nose area, lip area, and jaw area of ​​multiple areas of the frontal infrared image of the face of each individual in the target group are divided left and right to obtain multiple pairs of sub-areas, wherein the multiple pairs of sub-areas include nose symmetrical sub-areas, lip symmetrical sub-areas, jaw symmetrical sub-areas, forehead symmetrical sub-areas formed by left and right forehead areas, eye symmetrical sub-areas formed by left and right eye areas, cheek symmetrical sub-areas formed by left and right cheek areas, and ear symmetrical sub-areas formed by left and right ear areas, and each pair of sub-areas includes a left partition and a right partition;

[0043] For each pair of sub-regions in the frontal infrared image of the face of each individual in the target group, the temperature feature and texture feature of the left partition and the temperature feature and texture feature of the right partition are extracted respectively;

[0044] According to the temperature features and texture features of the left partition and the temperature features and texture features of the right partition in each pair of sub-regions, the Euclidean distance between the left partition and the right partition in each pair of sub-regions is obtained;

[0045] According to the Euclidean distance between the left partition and the right partition in each pair of sub-regions, the mean Euclidean distances of multiple pairs of sub-regions about the left partition and the right partition in the frontal infrared image of the face of each individual in the target group are obtained, and the mean Euclidean distance of each individual in the target group is used as the facial symmetry feature of each individual in the target group.

[0046] In one embodiment, the temperature feature includes any one of the following or any combination thereof: temperature mean, temperature standard deviation, temperature maximum, temperature minimum, wherein the expression of temperature mean is:

[0047]

[0048] In the expression of the mean temperature, mean represents the mean temperature, X(i) represents the temperature of the i-th pixel in the X sub-region, and N p Indicates the number of pixels with temperature in the X sub-area;

[0049] The expression for the temperature standard deviation is:

[0050]

[0051] In the expression of temperature standard deviation, std represents the temperature standard deviation, X(i) represents the temperature of the i-th pixel in the X sub-region, and N p Indicates the number of pixels in the X sub-region, Represents the average temperature of the pixels in the X sub-area;

[0052] The expression for the maximum temperature is:

[0053] max=max(X),

[0054] In the expression of the maximum temperature, max(X) represents the maximum temperature of the pixel point in the X sub-region;

[0055] The expression for the minimum temperature is:

[0056] min=min(X),

[0057] In the expression of minimum temperature, min(X) represents the minimum temperature of the pixel points in the X sub-region;

[0058] The expression of the temperature characteristic of the left partition in each pair of sub-regions is:

[0059]

[0060] The expression of the temperature characteristic of the right partition in each pair of sub-regions is:

[0061]

[0062] In the formula, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, mean l，i represents the mean temperature of the left partition in the i-th sub-region, mean r，i represents the mean temperature of the right partition in the i-th sub-region, std l，i represents the standard deviation of the temperature in the left partition of the i-th sub-region, std r，i represents the standard deviation of the temperature of the right partition in the i-th sub-region, max l，i Indicates the maximum temperature of the left partition in the i-th sub-region, max r，i represents the maximum temperature of the right partition in the i-th sub-region, min l，i Indicates the minimum temperature of the left partition in the i-th sub-region, min r，i Indicates the minimum temperature of the right partition in the i-th sub-region.

[0063] In one embodiment, the texture features include any one of the following or any combination thereof: contrast, correlation, homogeneity, entropy, wherein the expression of contrast is:

[0064]

[0065] In the expression of contrast, contrast represents contrast, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels;

[0066] The expression of correlation is:

[0067]

[0068] In the expression of correlation, correlation means correlation, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g represents the number of gray levels, p x (i) represents the marginal probability distribution of the gray-level co-occurrence matrix in the row direction, p y (j) represents the marginal probability distribution of the gray-level co-occurrence matrix in the column direction, μ x Indicates p x The mean of (i), that is, the mean of the row marginal distribution, μ y Indicates p y The mean of (j), that is, the mean of the column marginal distribution, σ x (i) indicates p x (i) Standard deviation, i.e. the standard deviation of the marginal distribution of the row, σ y (j) indicates p y The standard deviation of (j) is the standard deviation of the column marginal distribution.

[0069] The expression of homogeneity is:

[0070]

[0071] In the expression of homogeneity, homogeneity means homogeneity, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels;

[0072] The expression of entropy is:

[0073]

[0074] In the expression of entropy, energy represents entropy, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels;

[0075] The expression of the texture feature of the left partition in each pair of sub-regions is:

[0076]

[0077] The expression of the texture feature of the right partition in each pair of sub-regions is:

[0078]

[0079] In the formula, represents the texture feature of the left partition in the i-th pair of sub-regions, θ represents the scanning direction, represents the texture feature of the right partition in the i-th pair of sub-regions, contrast l，θ，i Represents the contrast of the left partition in the i-th pair of sub-regions, contrast r，θ，i represents the contrast of the right partition in the i-th pair of sub-regions, correlation l，θ，i Represents the correlation of the left partition in the i-th pair of sub-regions, correlation r，θ，i Represents the correlation of the right partition in the i-th pair of sub-regions, homogenity l，θ，i Indicates the homogeneity of the left partition in the i-th pair of subregions, homogenity r，θ，i represents the homogeneity of the right partition in the i-th pair of subregions, energy l，θ，i represents the entropy of the left partition in the i-th pair of subregions, energy r，θ，i represents the entropy of the right partition in the i-th pair of subregions.

[0080] In one embodiment, the expression for the sum of the features extracted from the left partition of each pair of sub-regions of the frontal infrared image of the face of each individual in the target group is:

[0081]

[0082] The expression for the sum of the features extracted from the right partition of each pair of sub-regions of the frontal infrared image of the face of each individual in the target group is:

[0083]

[0084] In the formula, represents the sum of the features extracted from the left partition of each pair of sub-regions of the frontal infrared image of the face of an individual in the target group, represents the sum of the features extracted from the right partition of each pair of sub-regions of the frontal infrared image of the face of an individual in the target group, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, and θ represents the scanning direction.

[0085] In one embodiment, the expression for the Euclidean distance is:

[0086]

[0087] In the expression of Euclidean distance, SED x，y It represents the Euclidean distance between the sum x of features extracted from the left partition and the sum y of features extracted from the right partition in a pair of sub-regions of the frontal infrared image of a certain individual in the target group. x and y represent two n-dimensional vectors. i The sum of the features extracted from the left partition of a pair of sub-regions in the frontal infrared image of an individual in the target group, the feature value of the i-th dimension in x, y i The eigenvalue of the i-th dimension in the sum y of the features extracted from the right partition of a pair of sub-regions in the frontal infrared image of the face of an individual in the target group, σ i It represents the overall standard deviation of the eigenvalues ​​about the i-th dimension in the left and right partitions of all sub-regions of the healthy population in the target population, and n represents the number of dimensions.

[0088] The present invention also provides a metabolic disorder degree auxiliary assessment system, comprising:

[0089] A data receiving module is used to: receive a frontal infrared image of the face of a person to be tested from at least one terminal;

[0090] An image processing module is used to obtain facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject to be tested based on the frontal infrared image of the face of the subject to be tested and according to any of the above methods for constructing a metabolic disorder degree assessment model;

[0091] An evaluation module, for obtaining the probability that the subject belongs to a metabolic disorder group based on the metabolic disorder degree evaluation model obtained by the metabolic disorder degree evaluation model construction method provided by any of the above items according to the facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the subject's face;

[0092] The data output module is used to output the probability that the subject belongs to the metabolic disorder group to at least one client.

[0093] The present invention also provides an electronic device, comprising a processor and a memory storing a computer program, wherein the processor implements any of the above-mentioned methods for constructing a metabolic disorder degree assessment model when executing the computer program.

[0094] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for constructing any of the above-mentioned metabolic disorder degree assessment models is implemented.

[0095] The present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute any of the above-mentioned methods for constructing a metabolic disorder degree assessment model.

[0096] The present invention provides a method for constructing a metabolic disorder degree assessment model and its application. Based on the frontal infrared image of the face of a target group, the model is used to learn the facial non-uniformity features, facial sequential features, and facial symmetry features of a healthy group and a metabolic disorder group, and a high-precision metabolic disorder degree assessment model is trained. The method can realize simple, sensitive, and economical metabolic disorder degree assessment, which helps to assist medical staff to promptly detect patients with chronic diseases and formulate appropriate treatment plans for patients. The assessment process is non-invasive, green, objective, and accurate, which effectively reduces the psychological and economic burdens of patients and prevents the waste of medical resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0098] Figure 1 A schematic diagram of a process for constructing a metabolic disorder degree assessment model provided by the present invention.

[0099] Figure 2 An example of facial segmentation for a frontal infrared image of the face.

[0100] Figure 3 This is a schematic diagram of the structure of a metabolic disorder degree auxiliary assessment system provided by the present invention.

[0101] Figure 4 This is a schematic structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0102] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0103] Combine the following Figure 1-Figure 4 The method for constructing a metabolic disorder degree assessment model provided by the present invention and its application are described. It should be noted that the execution subject of the method for constructing a metabolic disorder degree assessment model provided by the present invention can be any network side device / terminal side device that meets the technical requirements, such as a construction device for a metabolic disorder degree assessment model.

[0104] Figure 1 Schematic diagram of the method for constructing a metabolic disorder degree assessment model provided by the present invention. Figure 1 The present invention provides a method for constructing a metabolic disorder degree assessment model, which may include:

[0105] Step S110: Acquire a frontal infrared image of the face of a target group, wherein the target group includes a healthy group and a metabolic disorder group.

[0106] In one embodiment, the frontal infrared image of the face of the target group may be taken by medical staff using infrared imaging equipment for the target group.

[0107] Step S120: divide the frontal infrared facial image of each individual in the target group into multiple regions according to a preset facial segmentation rule, and extract facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group from the multiple regions of the frontal infrared facial image of each individual in the target group.

[0108] In one embodiment, the preset face segmentation rule includes:

[0109] The forehead area enclosed by the line from the hairline to the eyebrows of the individual in the frontal infrared image of the face is divided into a left forehead area and a right forehead area according to the midline of the frontal infrared image of the face;

[0110] The area bounded by the eye sockets of the individual in the frontal infrared image of the face is divided into the eye area, and the midline of the frontal infrared image of the face is used to divide the left eye area and the right eye area;

[0111] The triangular area from the root of the nose to the lower end of the nose tip and the sides of the nose wings in the frontal infrared image of the face is divided into the nose area, including the nose bridge area and the nose tip area;

[0112] The left front cheek and the right front cheek enclosed by the outer side of the nasolabial groove to the outer edge of the cheek and the lower orbit to the two mandibular angles in the face front infrared image are divided into the left front cheek area and the right front cheek area; the area enclosed by the lower orbital edge, the nasolabial line, the mandibular angle line and the front ear line on the left and right sides of the head of the individual in the face front infrared image is divided into the left cheek area and the right cheek area;

[0113] The circular area of ​​the individual including the left and right lip corners and the oral cavity as the boundary in the frontal infrared image of the face is divided into the lip area;

[0114] The area from the lower edge of the lips to the edge of the jaw and the downward line of the corner of the mouth in the frontal infrared image of the face is divided into the jaw area.

[0115] See also Figure 2 , the multiple regions include any one of the following or any combination thereof: left forehead region, right forehead region, left eye region, right eye region, nose region, lip region, mandibular region, left cheek region, right cheek region, left ear region, and right ear region, wherein, in the frontal infrared image of the face, the position of the left forehead region is symmetrical to the position of the right forehead region, the position of the left eye region is symmetrical to the position of the right eye region, the position of the left cheek region is symmetrical to the position of the right cheek region, and the position of the left ear region is symmetrical to the position of the right ear region.

[0116] In one embodiment, step S120 can be divided into three parts, corresponding to extracting facial non-uniformity features, facial sequential features, and facial symmetry features.

[0117] First, step S120 may include:

[0118] For each of the multiple regions of the frontal infrared image of the face of each individual in the target group, the number of pixels in each region and the temperature data of each pixel are obtained;

[0119] According to the number of pixels in each area and the temperature data of each pixel, the average temperature and standard deviation of each area are obtained;

[0120] The facial non-uniformity of each region is obtained according to the temperature mean value and the temperature standard deviation of each region. The collection of facial non-uniformity of multiple regions of an individual forms the facial non-uniformity feature of the individual.

[0121] In one embodiment, step S120 may obtain the temperature data of each of the multiple regions according to the following steps:

[0122] For each pair of left and right partitions of the sub-region, the temperature data inside is extracted to form a temperature set (T region ).

[0123] Set window width: The fixed window width is 10°C.

[0124] Dynamically adjust the window level: Initialize the window level to the minimum value of the partition temperature. With a step size of -0.1℃, gradually reduce the window level to generate a layer sequence. The expression is: (L i =min(T region )-i×0.1), where (L i ) is the window level of the i-th step.

[0125] Binarization: High temperature detection binarization: Apply the Otsu adaptive binarization algorithm to set pixels with a temperature higher than a certain threshold to 1 (white) and those lower to 0 (black). Low temperature detection binarization: Apply the Otsu algorithm for binarization.

[0126] Connected region detection: Perform morphological closing operations on the binary image to eliminate noise. Use connected region detection algorithms (such as four-connected or eight-connected) to determine connected regions.

[0127] The expression of the average temperature is:

[0128]

[0129] In the expression of the average temperature, μ T represents the average temperature of the Tth region, T i represents the temperature of the i-th pixel in the region, and N represents the number of pixels in the region;

[0130] The expression for the temperature standard deviation is:

[0131]

[0132] In the expression of temperature standard deviation, σ T represents the temperature standard deviation of the Tth region, T i Represents the temperature of the i-th pixel in the area, μ T represents the average temperature of the Tth region, and N represents the number of pixels in the region;

[0133] The expression of facial non-uniformity is:

[0134]

[0135] In the expression of facial non-uniformity, NU T represents the facial non-uniformity of the Tth region, μ T represents the average temperature of the Tth region, σT Represents the temperature standard deviation of the Tth region.

[0136] Second, step S120 may include:

[0137] For the frontal infrared image of the face of each individual in the target group, according to the average temperature of each area in the multiple areas, by testing whether there is a significant difference in the average temperature between the multiple areas, multiple significant areas are obtained from the multiple areas. In this embodiment, analysis of variance (ANOVA) or Friedman test can be used to determine whether there is a significant difference in the average temperature between the multiple areas, and a significance level (α) (for example, 0.05) is set. According to the test results, areas with significant differences between each other in the post hoc test are screened out and recorded as significant areas (significant ROIs);

[0138] Sort and rank multiple salient regions according to their average temperature from low to high, and obtain the facial sequence of each individual in the target group, wherein the ranking is to assign ranks to the sorted salient regions, and the ranking starts from 1. The salient region with the lowest average temperature has a rank of 1, the second lowest has a rank of 2, and so on;

[0139] According to the temperature average of multiple significant areas of each individual in the healthy group in the target group, the standard facial sequentiality of the healthy group is obtained, wherein in this embodiment, the multiple significant areas of each individual in the healthy group in the target group are used as the multiple significant areas of the healthy group, and according to the temperature average of multiple significant areas of each individual in the healthy group in the target group, the total temperature average of each significant area in the multiple significant areas of the healthy group is obtained, and then the multiple significant areas of the healthy group are sorted and ranked from low to high according to their total temperature averages, so as to obtain the standard facial sequentiality of the healthy group;

[0140] According to the facial sequentiality of each individual in the target group and the standard facial sequentiality of the healthy group, the facial sequentiality deviation index of each individual in the target group that deviates from the standard facial sequentiality of the healthy group is obtained, and the facial sequentiality deviation index of each individual in the target group is used as the facial sequentiality feature of each individual in the target group.

[0141] Among them, the expression of facial sequential deviation index is:

[0142]

[0143] In the expression of facial sequential deviation index, SDI i It represents the facial sequentiality deviation index of the i-th individual in the target group. The SDI value is in the interval [0, 1]. The closer the SDI value is to 1, the greater the deviation of the temperature sequentiality from the standard abdominal sequentiality of the healthy group.j represents the rank of the jth significant region of interest in the temperature series representation of the healthy group, B j represents the rank of the jth significant region of interest in the temperature series representation of the i-th individual in the target population. n represents the number of significant regions of interest, and k is an iteration variable, which ranges from 1 to A positive integer between .

[0144] Thirdly, step S120 may include:

[0145] According to the midline of the frontal infrared image of the face, the nose area, lip area, and jaw area of ​​multiple areas of the frontal infrared image of the face of each individual in the target group are divided left and right to obtain multiple pairs of sub-areas, wherein the multiple pairs of sub-areas include nose symmetrical sub-areas, lip symmetrical sub-areas, jaw symmetrical sub-areas, forehead symmetrical sub-areas formed by left and right forehead areas, eye symmetrical sub-areas formed by left and right eye areas, cheek symmetrical sub-areas formed by left and right cheek areas, and ear symmetrical sub-areas formed by left and right ear areas, and each pair of sub-areas includes a left partition and a right partition;

[0146] For each pair of sub-regions in the frontal infrared image of the face of each individual in the target group, the temperature feature and texture feature of the left partition and the temperature feature and texture feature of the right partition are extracted respectively, wherein the temperature feature includes any one of the following items or any combination thereof: temperature mean, temperature standard deviation, temperature maximum, temperature minimum, and the texture feature includes any one of the following items or any combination thereof: contrast, correlation, homogeneity, entropy;

[0147] According to the temperature features and texture features of the left partition and the temperature features and texture features of the right partition in each pair of sub-regions, the Euclidean distance between the left partition and the right partition in each pair of sub-regions is obtained;

[0148] According to the Euclidean distance between the left partition and the right partition in each pair of sub-regions, the mean Euclidean distances of multiple pairs of sub-regions about the left partition and the right partition in the frontal infrared image of the face of each individual in the target group are obtained, and the mean Euclidean distance of each individual in the target group is used as the facial symmetry feature of each individual in the target group.

[0149] The expression of the mean temperature is:

[0150]

[0151] In the expression of the mean temperature, mean represents the mean temperature, X(i) represents the temperature of the i-th pixel in the X sub-region, and N p Indicates the number of pixels with temperature in the X sub-area;

[0152] The expression for the temperature standard deviation is:

[0153]

[0154] In the expression of temperature standard deviation, std represents the temperature standard deviation, X(i) represents the temperature of the i-th pixel in the X sub-region, and N p Indicates the number of pixels in the X sub-region, Represents the average temperature of the pixels in the X sub-area;

[0155] The expression for the maximum temperature is:

[0156] max=max(X),

[0157] In the expression of the maximum temperature, max(X) represents the maximum temperature of the pixel point in the X sub-region;

[0158] The expression for the minimum temperature is:

[0159] min=min(X),

[0160] In the expression of minimum temperature, min(X) represents the minimum temperature of the pixel points in the X sub-region;

[0161] The expression of the temperature characteristic of the left partition in each pair of sub-regions is:

[0162]

[0163] The expression of the temperature characteristic of the right partition in each pair of sub-regions is:

[0164]

[0165] In the formula, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, mean l，i represents the mean temperature of the left partition in the i-th sub-region, mean r，i represents the mean temperature of the right partition in the i-th sub-region, std l，i represents the standard deviation of the temperature in the left partition of the i-th sub-region, std r，i represents the standard deviation of the temperature of the right partition in the i-th sub-region, max l，o Indicates the maximum temperature of the left partition in the i-th sub-region, max r，i represents the maximum temperature of the right partition in the i-th sub-region, min l，i Indicates the minimum temperature of the left partition in the i-th sub-region, min r，i Indicates the minimum temperature of the right partition in the i-th sub-region.

[0166] The expression of contrast is:

[0167]

[0168] In the expression of contrast, contrast represents contrast, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels;

[0169] The expression of correlation is:

[0170]

[0171] In the expression of correlation, correlation means correlation, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g represents the number of gray levels, p x (i) is the marginal probability distribution of the gray-level co-occurrence matrix in the row direction, p y (j) is the marginal probability distribution of the gray-level co-occurrence matrix in the column direction. x Indicates p x The mean of (i), that is, the mean of the row marginal distribution, μ y Indicates p y The mean of (j), that is, the mean of the column marginal distribution. x (i) indicates p x (i) Standard deviation, i.e. the standard deviation of the marginal distribution of the row. σ y (j) indicates p y The standard deviation of (j), that is, the standard deviation of the column marginal distribution. x μ y σ x (i)σ y (j);

[0172] The expression of homogeneity is:

[0173]

[0174] In the expression of homogeneity, homogeneity means homogeneity, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels;

[0175] The expression of entropy is:

[0176]

[0177] In the expression of entropy, energy represents entropy, P (i，j) represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels.

[0178] The gray-level co-occurrence matrix can describe the probability of occurrence of a gray-level value combination with a certain distribution law in space, reflecting the texture characteristics of the image. Its mathematical description is as follows:

[0179]

[0180] P represents the gray-level co-occurrence matrix, Δ x , Δ y is the offset, and the default value 1 is selected in this embodiment; θ represents the scanning direction, and 0 is usually selected. Reflects the combination of pixel pairs in horizontal, vertical and diagonal directions, i, j = 0, ... N g -1 indicates the pixel gray level, N g Represents the number of gray levels. (i，j) It represents the joint probability density of gray level (i, j). Therefore, the co-occurrence matrix can reflect the comprehensive information of the image about direction, adjacent interval and change amplitude.

[0181] Since some texture features may be mirror images on the left and right sides, horizontally flipping the right sub-region can more accurately align the directionality difference and improve the accuracy of symmetry detection. Therefore, before calculating the symmetry of each pair of sub-regions, this embodiment keeps the left sub-region unchanged, flips the right sub-region horizontally, and then extracts texture features from the left and right sub-regions.

[0182] Among them, the expression of the texture feature of the left partition in each pair of sub-regions is:

[0183]

[0184] The expression of the texture feature of the right partition in each pair of sub-regions is:

[0185]

[0186] In the formula, represents the texture feature of the left partition in the i-th pair of sub-regions, θ represents the scanning direction, represents the texture feature of the right partition in the i-th pair of sub-regions, contrast l，θ，i Represents the contrast of the left partition in the i-th pair of subregions, contrast r，θ，irepresents the contrast of the right partition in the i-th pair of sub-regions, correlation l，θ，i Represents the correlation of the left partition in the i-th pair of sub-regions, correlation r，θ，i Represents the correlation of the right partition in the i-th pair of sub-regions, homogenity l，θ，i Indicates the homogeneity of the left partition in the i-th pair of subregions, homogenity r，θ，i represents the homogeneity of the right partition in the i-th pair of subregions, energy l，θ，i represents the entropy of the left partition in the i-th pair of subregions, energy r，θ，i represents the entropy of the right partition in the i-th pair of subregions.

[0187] Among them, the expression of the sum of the features extracted from the left partition of each pair of sub-regions of the frontal infrared image of the face of each individual in the target group is:

[0188]

[0189] The expression for the sum of the features extracted from the right partition of each pair of sub-regions of the frontal infrared image of the face of each individual in the target group is:

[0190]

[0191] In the formula, represents the sum of the features extracted from the left partition of each pair of sub-regions of the frontal infrared image of the face of an individual in the target group, represents the sum of the features extracted from the right partition of each pair of sub-regions of the frontal infrared image of the face of an individual in the target group, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, and θ represents the scanning direction.

[0192] Among them, the expression of Euclidean distance is:

[0193]

[0194] In the expression of Euclidean distance, SED x，y It represents the Euclidean distance between the sum x of features extracted from the left partition and the sum y of features extracted from the right partition in a pair of sub-regions of the frontal infrared image of a certain individual in the target group. x and y represent two n-dimensional vectors. i The sum of the features extracted from the left partition of a pair of sub-regions in the frontal infrared image of an individual in the target group, the feature value of the i-th dimension in x, y iThe eigenvalue of the i-th dimension in the sum y of the features extracted from the right partition of a pair of sub-regions in the frontal infrared image of the face of an individual in the target group, σ i represents the overall standard deviation of the eigenvalues ​​of the left and right partitions of all sub-regions of the healthy population in the target population about the i-th dimension, and n represents the number of dimensions. Since each dimension is divided by its standard deviation, the standardized Euclidean distance is insensitive to scale. Since the magnitudes of the above different features in GLCM vary greatly, the standardized Euclidean distance can ensure that the contributions of each feature are equal.

[0195] Step S120 can also set a suitable standardized Euclidean distance threshold SED according to actual application requirements and data analysis results. threshold If SED(F L , F R )>SED threshol , then the symmetry of the pair of sub-regions is considered to be poor. The standardized Euclidean distance calculation results of each pair of sub-regions can be listed to determine whether their symmetry meets the requirements. Then the frontal infrared thermal image of the face is displayed, and the areas with poor symmetry are marked with colors or markers.

[0196] Step S130, based on the facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group, the facial non-uniformity features, facial sequential features, and facial symmetry features of the healthy group and the metabolic disorder group are learned by using the model to train a metabolic disorder degree assessment model, wherein the model framework and classification algorithm of the metabolic disorder degree assessment model can be selected according to actual conditions.

[0197] This embodiment uses a target group formed by 108 healthy subjects and 104 metabolic syndrome patients from Dongzhimen Hospital of Beijing University of Chinese Medicine, and uses a stratified random sampling method to randomly divide them into a training set and a validation set at a ratio of 7:3, with 148 cases in the training set and 64 cases in the validation set. The metabolic disorder degree assessment model is trained by the above-mentioned method for constructing a metabolic disorder degree assessment model, and the AUC (area under the ROC curve) of the model on the training set is 0.74, and the AUC on the validation set is 0.65, with good evaluation performance.

[0198] The present invention believes that the human body surface temperature is the result of the combined effects of heat production and heat dissipation, and the local body temperature is affected by multiple factors such as the functional activities of adjacent organs, blood perfusion, neural regulation, and fat distribution, and is a comprehensive reflection of the body's intrinsic metabolic state. From the perspective of the "holistic view" of traditional Chinese medicine, the human head and face are closely connected with the meridians, qi and blood, and internal organs of the whole body, and the local appearance of the human body is a microcosm of the health status of the whole body. Therefore, the temperature distribution characteristics of the head and face captured by infrared thermal imaging technology can preliminarily reflect the comprehensive influence of various metabolic factors, provide intuitive information on the metabolic status of the human body, and provide a new means for the assessment of metabolic disorders.

[0199] The present invention provides a method for constructing a metabolic disorder degree assessment model and its application. Based on the frontal infrared image of the face of a target group, the model is used to learn the facial non-uniformity features, facial sequential features, and facial symmetry features of a healthy group and a metabolic disorder group, and a high-precision metabolic disorder degree assessment model is trained. The assessment results can be quantified and output and visualized, thereby achieving a simple, sensitive, and economical metabolic disorder degree assessment, which helps to assist medical staff in timely detecting patients with chronic diseases and formulating appropriate treatment plans for patients, thereby improving the accuracy and efficiency of disease assessment.

[0200] The auxiliary evaluation system for the degree of metabolic disorder provided by the present invention is described below. The auxiliary evaluation system for the degree of metabolic disorder described below and the method for constructing the evaluation model for the degree of metabolic disorder described above can be referred to each other.

[0201] Reference Figure 4 The present invention provides a metabolic disorder degree auxiliary assessment system, which may include:

[0202] A data receiving module is used to: receive a frontal infrared image of the face of a person to be tested from at least one terminal;

[0203] An image processing module is used to obtain facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject to be tested based on the frontal infrared image of the face of the subject to be tested and according to any of the above methods for constructing a metabolic disorder degree assessment model;

[0204] An evaluation module, for obtaining the probability that the subject belongs to a metabolic disorder group based on the metabolic disorder degree evaluation model obtained by the metabolic disorder degree evaluation model construction method provided by any of the above items according to the facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the subject's face;

[0205] The data output module is used to output the probability that the subject belongs to the metabolic disorder group to at least one client.

[0206] Figure 4An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to perform the following steps:

[0207] receiving a frontal infrared image of the face of a person to be tested from at least one terminal;

[0208] Based on the frontal infrared image of the face of the subject to be tested, according to the method for constructing the metabolic disorder degree assessment model provided in any of the above items, the facial non-uniformity feature, the facial sequential feature, and the facial symmetry feature of the frontal infrared image of the face of the subject to be tested are obtained;

[0209] According to the facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject, a metabolic disorder degree assessment model is obtained by the metabolic disorder degree assessment model construction method provided by any of the above items, to obtain the probability that the subject belongs to the metabolic disorder group;

[0210] The probability that the subject belongs to a metabolic disorder group is output to at least one client.

[0211] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0212] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and when the computer program is executed by a processor, the computer can perform the following steps:

[0213] receiving a frontal infrared image of the face of a person to be tested from at least one terminal;

[0214] Based on the frontal infrared image of the face of the subject to be tested, according to the method for constructing the metabolic disorder degree assessment model provided in any of the above items, the facial non-uniformity feature, the facial sequential feature, and the facial symmetry feature of the frontal infrared image of the face of the subject to be tested are obtained;

[0215] According to the facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject, a metabolic disorder degree assessment model is obtained by the metabolic disorder degree assessment model construction method provided by any of the above items, to obtain the probability that the subject belongs to the metabolic disorder group;

[0216] The probability that the subject belongs to a metabolic disorder group is output to at least one client.

[0217] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented to perform the following steps when executed by a processor:

[0218] receiving a frontal infrared image of the face of a person to be tested from at least one terminal;

[0219] Based on the frontal infrared image of the face of the subject to be tested, according to the method for constructing the metabolic disorder degree assessment model provided in any of the above items, the facial non-uniformity feature, the facial sequential feature, and the facial symmetry feature of the frontal infrared image of the face of the subject to be tested are obtained;

[0220] According to the facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject, a metabolic disorder degree assessment model is obtained by the metabolic disorder degree assessment model construction method provided by any of the above items, to obtain the probability that the subject belongs to the metabolic disorder group;

[0221] The probability that the subject belongs to a metabolic disorder group is output to at least one client.

[0222] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0223] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0224] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a metabolic disorder degree assessment model, characterized in that: include: Acquire frontal infrared images of faces of a target group, wherein the target group includes a healthy group and a metabolic disorder group; According to a preset facial segmentation rule, the frontal infrared image of the face of each individual in the target group is divided into multiple regions, and the facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group are extracted from the multiple regions of the frontal infrared image of the face of each individual in the target group; According to the facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group, the model is used to learn the facial non-uniformity features, facial sequential features, and facial symmetry features of the healthy group and the metabolic disorder group, and a metabolic disorder degree assessment model is trained.

2. The method for constructing a metabolic disorder degree assessment model according to claim 1, characterized in that: The multiple regions include any one of the following or any combination thereof: left forehead region, right forehead region, left eye region, right eye region, nose region, lip region, mandibular region, left cheek region, right cheek region, left ear region, and right ear region, wherein, in the frontal infrared image of the face, the position of the left forehead region is symmetrical to the position of the right forehead region, the position of the left eye region is symmetrical to the position of the right eye region, the position of the left cheek region is symmetrical to the position of the right cheek region, and the position of the left ear region is symmetrical to the position of the right ear region; The preset face segmentation rules include: The forehead area enclosed by the line from the hairline to the eyebrows of the individual in the frontal infrared image of the face is divided into a left forehead area and a right forehead area according to the midline of the frontal infrared image of the face; The area bounded by the eye sockets of the individual in the frontal infrared image of the face is divided into the eye area, and the midline of the frontal infrared image of the face is used to divide the left eye area and the right eye area; The triangular area from the root of the nose to the lower end of the nose tip and the sides of the nose wings in the frontal infrared image of the face is divided into the nose area, including the nose bridge area and the nose tip area; The left front cheek and the right front cheek enclosed by the outer side of the nasolabial groove to the outer edge of the cheek and the lower orbit to the two mandibular angles in the face front infrared image are divided into the left front cheek area and the right front cheek area; the area enclosed by the lower orbital edge, the nasolabial line, the mandibular angle line and the front ear line on the left and right sides of the head of the individual in the face front infrared image is divided into the left cheek area and the right cheek area; The circular area of ​​the individual including the left and right lip corners and the oral cavity as the boundary in the frontal infrared image of the face is divided into the lip area; The area from the lower edge of the lips to the edge of the jaw and the downward line of the corner of the mouth in the frontal infrared image of the face is divided into the jaw area.

3. The method for constructing a metabolic disorder degree assessment model according to claim 2, characterized in that: The facial non-uniformity features, facial sequential features, and facial symmetry features of each individual in the target group are extracted from multiple regions of the frontal infrared image of the face of each individual in the target group, including: For each of the multiple regions of the frontal infrared image of the face of each individual in the target group, the number of pixels in each region and the temperature data of each pixel are obtained; According to the number of pixels in each area and the temperature data of each pixel, the average temperature and standard deviation of each area are obtained; According to the temperature average and temperature standard deviation of each area, the facial non-uniformity of each area is obtained, and the collection of facial non-uniformity of multiple areas of an individual forms the facial non-uniformity feature of the individual; The expression of the average temperature is: In the expression of the average temperature, μ T represents the average temperature of the Tth region, T i represents the temperature of the i-th pixel in the region, and N represents the number of pixels in the region; The expression for the temperature standard deviation is: In the expression of temperature standard deviation, σ T represents the temperature standard deviation of the Tth region, T i Represents the temperature of the i-th pixel in the area, μ T represents the average temperature of the Tth region, and N represents the number of pixels in the region; The expression of facial non-uniformity is: In the expression of facial non-uniformity, NU T represents the facial non-uniformity of the Tth region, μ T represents the average temperature of the Tth region, σ T Represents the temperature standard deviation of the Tth region.

4. The method for constructing a metabolic disorder degree assessment model according to claim 3, characterized in that: The facial non-uniformity feature, facial sequential feature, and facial symmetry feature of each individual in the target group are extracted from multiple regions of the frontal infrared image of the face of each individual in the target group, and further include: For the frontal infrared image of each individual in the target group, according to the average temperature of each of the multiple regions, by testing whether there is a significant difference in the average temperature between the multiple regions, multiple significant regions are obtained from the multiple regions; Sort and rank multiple significant regions according to their average temperature from low to high to obtain the facial sequence of each individual in the target group; According to the average temperature of multiple significant areas of each individual in the healthy group in the target group, the standard facial sequence of the healthy group is obtained; According to the facial sequentiality of each individual in the target group and the standard facial sequentiality of the healthy group, the facial sequentiality deviation index of each individual in the target group that deviates from the standard facial sequentiality of the healthy group is obtained, and the facial sequentiality deviation index of each individual in the target group is used as the facial sequentiality feature of each individual in the target group.

5. The method for constructing a metabolic disorder degree assessment model according to claim 4, characterized in that: The method of obtaining the standard facial sequence of the healthy group according to the average temperature of multiple significant areas of each individual in the healthy group in the target group includes: Taking the multiple significant regions of each individual in the healthy group in the target group as the multiple significant regions of the healthy group, and obtaining the total temperature average of each significant region in the multiple significant regions of the healthy group according to the temperature average of the multiple significant regions of each individual in the healthy group in the target group; Sort and rank multiple significant areas of the healthy group according to their total temperature averages from low to high to obtain the standard facial sequence of the healthy group; And, the expression of facial sequential deviation index is: In the expression of facial sequential deviation index, SDI i represents the facial sequential deviation index of the i-th individual in the target group, A j represents the rank of the jth significant region of interest in the temperature series representation of the healthy group, B j represents the rank of the jth significant region of interest in the temperature series representation of the i-th individual in the target population, n represents the number of significant regions of interest, and k represents the iteration variable.

6. The method for constructing a metabolic disorder degree assessment model according to claim 5, characterized in that: The facial non-uniformity feature, facial sequential feature, and facial symmetry feature of each individual in the target group are extracted from multiple regions of the frontal infrared image of the face of each individual in the target group, and further include: According to the midline of the frontal infrared image of the face, the nose area, lip area, and jaw area of ​​multiple areas of the frontal infrared image of the face of each individual in the target group are divided left and right to obtain multiple pairs of sub-areas, wherein the multiple pairs of sub-areas include nose symmetrical sub-areas, lip symmetrical sub-areas, jaw symmetrical sub-areas, forehead symmetrical sub-areas formed by left and right forehead areas, eye symmetrical sub-areas formed by left and right eye areas, cheek symmetrical sub-areas formed by left and right cheek areas, and ear symmetrical sub-areas formed by left and right ear areas, and each pair of sub-areas includes a left partition and a right partition; For each pair of sub-regions in the frontal infrared image of the face of each individual in the target group, the temperature feature and texture feature of the left partition and the temperature feature and texture feature of the right partition are extracted respectively; According to the temperature features and texture features of the left partition and the temperature features and texture features of the right partition in each pair of sub-regions, the Euclidean distance between the left partition and the right partition in each pair of sub-regions is obtained; According to the Euclidean distance between the left partition and the right partition in each pair of sub-regions, the mean Euclidean distances of multiple pairs of sub-regions about the left partition and the right partition in the frontal infrared image of the face of each individual in the target group are obtained, and the mean Euclidean distance of each individual in the target group is used as the facial symmetry feature of each individual in the target group.

7. The method for constructing a metabolic disorder degree assessment model according to claim 6, characterized in that: The temperature characteristics include any one of the following or any combination thereof: temperature mean, temperature standard deviation, temperature maximum, and temperature minimum. The expression of temperature mean is: In the expression of the mean temperature, mean represents the mean temperature, X(i) represents the temperature of the i-th pixel in the X sub-region, and N p Indicates the number of pixels with temperature in the X sub-area; The expression for the temperature standard deviation is: In the expression of temperature standard deviation, std represents the temperature standard deviation, X(i) represents the temperature of the i-th pixel in the X sub-region, and N p Indicates the number of pixels in the X sub-region, Represents the average temperature of the pixels in the X sub-area; The expression for the maximum temperature is: max=max(X), In the expression of the maximum temperature, max(X) represents the maximum temperature of the pixel point in the X sub-region; The expression for the minimum temperature is: min=min(X), In the expression of minimum temperature, min(X) represents the minimum temperature of the pixel points in the X sub-region; The expression of the temperature characteristic of the left partition in each pair of sub-regions is: The expression of the temperature characteristic of the right partition in each pair of sub-regions is: In the formula, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, mean l,i represents the mean temperature of the left partition in the i-th sub-region, mean r,i represents the mean temperature of the right partition in the i-th sub-region, std l,i represents the standard deviation of the temperature in the left partition of the i-th sub-region, std r,i represents the standard deviation of the temperature of the right partition in the i-th sub-region, max l,i Indicates the maximum temperature of the left partition in the i-th sub-region, max r,i represents the maximum temperature of the right partition in the i-th sub-region, min l,o Indicates the minimum temperature of the left partition in the i-th sub-region, min r,i represents the minimum temperature of the right partition in the i-th sub-region; Texture features include any one of the following or any combination thereof: contrast, correlation, homogeneity, and entropy, where the expression of contrast is: In the expression of contrast, contrast represents contrast, P (i,j) Represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels; The expression of correlation is: In the expression of correlation, correlation represents correlation, p (i,j) Represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels, p x (i) represents the marginal probability distribution of the gray-level co-occurrence matrix in the row direction, p y (j) represents the marginal probability distribution of the gray-level co-occurrence matrix in the column direction, μ x Indicates p x The mean μ of (i) y Indicates p y The mean of (j), σ x (i) indicates p x (i) Standard deviation, σ y (j) indicates p y (j) standard deviation; The expression of homogeneity is: In the expression of homogeneity, homogenity represents homogeneity, P (i,j) Represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels; The expression of entropy is: In the expression of entropy, energy represents entropy, P (i,j) Represents the joint probability density of gray level (i, j), i, j = 0, ... N g -1 indicates the pixel gray level, N g Indicates the number of gray levels; The expression of the texture feature of the left partition in each pair of sub-regions is: The expression of the texture feature of the right partition in each pair of sub-regions is: In the formula, represents the texture feature of the left partition in the i-th pair of sub-regions, θ represents the scanning direction, represents the texture feature of the right partition in the i-th pair of sub-regions, contrast l,θ,i Represents the contrast of the left partition in the i-th pair of subregions, contrast r,θ,i represents the contrast of the right partition in the i-th pair of sub-regions, correlation l,θ,i Represents the correlation of the left partition in the i-th pair of sub-regions, correlation r,θ,i Represents the correlation of the right partition in the i-th pair of sub-regions, homogenity l,θ,i Indicates the homogeneity of the left partition in the i-th pair of subregions, homogenity r,θ,i represents the homogeneity of the right partition in the i-th pair of subregions, energy l,θ,i represents the entropy of the left partition in the i-th pair of subregions, energy r,θ,i represents the entropy of the right partition in the i-th pair of subregions; The expression for the sum of the features extracted from the left partition of each pair of sub-regions of the frontal infrared image of the face of each individual in the target group is: The expression for the sum of the features extracted from the right partition of each pair of sub-regions of the frontal infrared image of the face of each individual in the target group is: In the formula, represents the sum of the features extracted from the left partition of each pair of sub-regions of the frontal infrared image of the face of an individual in the target group, represents the sum of the features extracted from the right partition of each pair of sub-regions of the frontal infrared image of the face of an individual in the target group, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, represents the temperature feature set of the left partition, represents the temperature feature set of the right partition, and θ represents the scanning direction; The expression of Euclidean distance is: In the expression of Euclidean distance, SED x,y It represents the Euclidean distance between the sum x of features extracted from the left partition and the sum y of features extracted from the right partition in a pair of sub-regions of the frontal infrared image of a certain individual in the target group. x and y represent two n-dimensional vectors. i The sum of the features extracted from the left partition of a pair of sub-regions in the frontal infrared image of an individual in the target group, the feature value of the i-th dimension in x, y i The eigenvalue of the i-th dimension in the sum y of the features extracted from the right partition of a pair of sub-regions in the frontal infrared image of the face of an individual in the target group, σ i It represents the overall standard deviation of the eigenvalues ​​about the i-th dimension in the left and right partitions of all sub-regions of the healthy population in the target population, and n represents the number of dimensions.

8. A metabolic disorder degree auxiliary assessment system, characterized in that: include: A data receiving module is used to: receive a frontal infrared image of the face of a person to be tested from at least one terminal; An image processing module, for obtaining facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject to be tested according to the method for constructing a metabolic disorder degree assessment model provided in any one of claims 1 to 7 based on the frontal infrared image of the face of the subject to be tested; An evaluation module, for obtaining the probability that the subject belongs to a metabolic disorder group based on the metabolic disorder degree evaluation model obtained by the method for constructing a metabolic disorder degree evaluation model provided by any one of claims 1 to 7, based on facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the subject's face; The data output module is used to output the probability that the subject belongs to the metabolic disorder group to at least one client.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the following steps are implemented: receiving a frontal infrared image of the face of a person to be tested from at least one terminal; Based on the frontal infrared image of the face of the subject to be tested, according to the method for constructing a metabolic disorder degree assessment model provided by any one of claims 1 to 7, obtaining facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject to be tested; According to facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject, a metabolic disorder degree assessment model is obtained by the method for constructing a metabolic disorder degree assessment model provided by any one of claims 1 to 7, to obtain the probability that the subject belongs to a metabolic disorder group; The probability that the subject belongs to a metabolic disorder group is output to at least one client.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the following steps are implemented: receiving a frontal infrared image of the face of a person to be tested from at least one terminal; Based on the frontal infrared image of the face of the subject to be tested, according to the method for constructing a metabolic disorder degree assessment model provided by any one of claims 1 to 7, obtaining facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject to be tested; According to facial non-uniformity features, facial sequential features, and facial symmetry features of the frontal infrared image of the face of the subject, a metabolic disorder degree assessment model is obtained by the method for constructing a metabolic disorder degree assessment model provided by any one of claims 1 to 7, to obtain the probability that the subject belongs to a metabolic disorder group; The probability that the subject belongs to a metabolic disorder group is output to at least one client.

Images