A monkeypox virus classification method and device based on image comparison and a storage medium

Abstract

The application belongs to the technical field of image recognition, and discloses a monkeypox virus classification method and system based on image comparison and a storage medium, which comprises an individual reference modeling module, collects normal skin and initial lesion images of a healthy side of a patient, generates a healthy texture template and an initial lesion reference library, performs gray scale normalization on the healthy texture template and the initial lesion reference library through adaptive histogram stretching, and outputs a comparison reference image group; a symptom density quantification module, acquires a current monitoring image of the patient, segments a lesion area in combination with the comparison reference image group, extracts layered features of the lesion area to generate a three-dimensional symptom density vector, extracts texture roughness of a surface of the lesion area, and constitutes a lesion quantification data set; the accuracy of disease monitoring is improved, and a clinical nursing process is optimized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image recognition, and more particularly, to a monkeypox virus classification method and device based on image comparison and a storage medium. BACKGROUND

[0002] A patent with publication number CN119780184A provides an integrated biosensing platform based on microneedles and hydrogels and its application in detecting monkeypox virus. The platform includes a three-electrode microneedle, a hydrogel, a burning needle, and a micro electrochemical workstation. The microneedle electrode of the working electrode is combined with a nano-enzyme and an A29 protein antibody. The monkeypox virus in the sample is extracted through the specific binding of the antibody with the monkeypox virus. The burning needle is used to connect the three-electrode microneedle and the micro electrochemical workstation and transmit the electrochemical signal. After the three-electrode microneedle is inserted into the hydrogel, the nano-enzyme catalyzes the TMB contained in the hydrogel and generates a blue oxidation product, thereby generating an electrochemical signal. The signal is collected by the micro electrochemical workstation through the burning needle. The electrochemical signal is collected through Bluetooth and the colorimetric image of the hydrogel is obtained, thereby realizing the detection of the monkeypox virus. The integrated biosensing platform of the invention has a low detection limit and has a broad application prospect in the rapid on-site detection of monkeypox virus.

[0003] The existing monkeypox virus classification method based on image comparison has the following main problems:

[0004] The lesion observation lacks continuity, and the time dimension is fragmented and fuzzy: In the prior art, the lesion evolution stage depends on fixed-point observation or experience judgment, and lacks a continuous time modeling mechanism combined with label dynamic updating, making it difficult to finely track the disease evolution and reflect the gradual change process or mutation signal between stages. Especially when the multi-stage boundary is fuzzy, the traditional method cannot provide quantitative basis with time anchor;

[0005] The feature analysis is isolated and lacks a multi-dimensional collaborative modeling mechanism: Traditional lesion recognition methods mostly rely on a single feature (such as rash density or erythema size), and lack the ability to model the collaborative evolution of multiple key symptom indicators, making it difficult to capture the associated change trend of multiple symptoms within the same stage, thereby limiting the comprehensive understanding and prediction ability of the lesion evolution mechanism;

[0006] The disease condition judgment logic is rough, and lacks dynamic modeling capability for trend direction: the existing method often relies on static threshold or manual experience to judge the disease stage, ignores the difference in evolution speed and feature performance of different individuals, and cannot dynamically capture the rate change of the lesion from one state to another state, and lacks sensitive response capability to the stage switching process; the concept of "trend direction" is not introduced, and it is difficult to judge whether the current feature change is deteriorating or improving. Especially in the early or potential deterioration stage, there is a lack of early warning mechanism for trend change, and there is a risk of delayed intervention.

[0007] In view of this, the present application provides a monkeypox virus classification method, device and storage medium based on image comparison to solve the above problems. SUMMARY

[0008] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a monkeypox virus classification method based on image comparison, comprising:

[0009] S1, collecting normal skin and initial lesion images of the healthy side of the patient to generate a healthy texture template and an initial lesion reference library; performing gray scale normalization on the healthy texture template and the initial lesion reference library through adaptive histogram stretching, and outputting a comparison reference image group;

[0010] S2, obtaining the current monitoring image of the patient, segmenting the lesion area in combination with the comparison reference image group, extracting the hierarchical features of the lesion area to generate a three-dimensional symptom density vector, and extracting the texture roughness of the surface of the lesion area to form a lesion quantization data set;

[0011] S3, presetting a staging feature matrix of each stage of monkeypox virus infection, determining the main staging label based on the matching degree of the lesion quantization data set and the staging feature matrix;

[0012] S4, taking the main staging label as a time sequence anchor point, dynamically fitting the three-dimensional symptom density vector of continuous monitoring to generate a lesion evolution trajectory curve, and calculating the stage transition rate based on the slope change of the lesion evolution trajectory curve to obtain a disease deterioration risk coefficient;

[0013] S5, integrating the main staging label, the lesion evolution trajectory curve and the disease deterioration risk coefficient, generating a visual disease progression atlas to obtain a standardized clinical staging archive.

[0014] Preferably, the method for generating the healthy texture template and the initial lesion reference library comprises:

[0015] Collecting the image of normal skin on the healthy side of the patient and the image of the initial lesion, performing regional alignment and cutting processing on the image of normal skin on the healthy side to obtain a healthy texture block matching the size and shape of the initial lesion area, and performing brightness and color normalization processing on the healthy texture block; performing lesion area extraction on the image of the initial lesion, and performing standardization processing on the extracted initial lesion area image to generate an initial lesion reference library containing color features, texture features and area contours.

[0016] Preferably, the method for obtaining the comparison reference image group comprises:

[0017] Converting the healthy texture block and the initial lesion image from a color image to a color space with separated brightness and color, and extracting the brightness channel information in the healthy texture block and the initial lesion image; performing an adaptive histogram stretching operation on the brightness channel information; determining the stretching boundary value based on the local brightness distribution of the image, mapping the original brightness of the healthy texture block and the initial lesion image to a standard brightness interval, and then obtaining the comparison reference image group.

[0018] Preferably, the method for obtaining the lesion quantification data set comprises:

[0019] Obtaining the current monitoring image of the patient and performing preprocessing, identifying the abnormal area in the current monitoring image of the patient as the lesion area through differential analysis based on the healthy texture template and the initial lesion image in the comparison reference image group;

[0020] Performing hierarchical feature extraction on the lesion area to obtain the epidermal elevation degree, the depth of erythema infiltration and the blister density, and then constructing a three-dimensional symptom density vector; extracting the texture roughness of the surface of the lesion area, and combining the three-dimensional symptom density vector to form the lesion quantification data set.

[0021] Preferably, the method for determining the main staging label comprises:

[0022] Predefining a staging feature matrix for each stage of monkeypox virus infection, performing similarity matching between the lesion quantification data set and the staging feature matrix, and taking the similarity matching result as the main staging label corresponding to the current monitoring image of the patient.

[0023] Preferably, the method for obtaining the lesion evolution trajectory curve comprises:

[0024] Recording the three-dimensional symptom density vector at fixed time intervals, and forming a continuous three-dimensional symptom density vector sequence after data cleaning; associating the main staging label with the three-dimensional symptom density vector sequence to form a time sequence anchor point, for example, when the main staging label is first determined in a certain monitoring process, the corresponding time point is the time sequence anchor point of the label; if the subsequent main staging label changes, the corresponding time sequence anchor point will be added, and all time sequence anchor points are collected to form a multi-anchor time sequence axis.

[0025] The three-dimensional symptom density vector sequence is divided by a sliding window, fitting parameters are dynamically adjusted in combination with a multi-anchor time axis, and a window data set is obtained; the dimensions of the window data set are fitted by a second-order polynomial to generate fitting curves, which are combined into a lesion evolution trajectory curve.

[0026] Preferably, the method for obtaining the disease deterioration risk coefficient comprises:

[0027] The first-order derivatives of the fitting curves contained in the lesion evolution trajectory curve are calculated respectively to obtain the slopes of the fitting curves at different time points; the slope change rates of the fitting curves are obtained based on the slope difference between adjacent two time points; and the slope change rates of all the fitting curves are weighted to obtain a stage transition rate.

[0028] Based on the slopes of the fitting curves at different time points, a direction coefficient of disease characteristic change is defined, and a characteristic deterioration mark is generated for a characteristic showing a deterioration trend; the direction coefficient is determined by the slope of each fitting curve; when the slope is greater than 0, it reflects that the value of the corresponding characteristic increases, and the monkeypox disease shows a deterioration trend; when the slope is less than or equal to 0, it reflects that the value of the corresponding characteristic decreases, and the monkeypox disease tends to improve.

[0029] A preset clinical benchmark threshold is combined with the stage transition rate and the characteristic deterioration mark to generate a deterioration risk coefficient.

[0030] Preferably, the method for obtaining the standardized clinical staging archive comprises:

[0031] The main staging label and the disease deterioration risk coefficient are associated to the lesion evolution trajectory curve by taking the time point as an index, and a disease course progress atlas is generated by visual rendering; and based on the disease course progress atlas, a standardized clinical staging archive is constructed.

[0032] An apparatus comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, a monkeypox virus classification method based on image comparison is implemented.

[0033] A storage medium, wherein the storage medium stores a computer program, and when the computer program is executed, a monkeypox virus classification method based on image comparison is implemented.

[0034] The present application provides a monkeypox virus classification method based on image comparison, an apparatus and a storage medium, which has the following beneficial effects:

[0035] The time anchor points are set according to the change of the main staging label, and a time axis with a multi-anchor structure is formed, effectively solving the fragmentation and ambiguity of the traditional lesion observation method in the time dimension. The structure not only clearly divides the evolution stage of the lesion, but also supports the fitting parameter updating mechanism across the staging, improving the sensitivity and adaptability of the lesion stage switching;

[0036] By fitting the three-dimensional symptom density vector in each sliding window, a multi-dimensional trajectory curve is formed, which can not only analyze the change trend of each feature individually, but also observe the synergistic evolution relationship between them (such as the synchronous rise of the skin elevation degree and the blister density in a certain stage). This synergistic feature fusion mode improves the accuracy and interpretability of staging recognition and disease trend judgment;

[0037] The direction coefficient mechanism is innovatively introduced, the slope of the fitting curve is used to judge whether each feature is in a deterioration trend (such as the continuous rise of the blister density), and the feature deterioration mark is weighted and fused into the risk calculation process. Compared with single threshold judgment, this mechanism is more consistent with the complexity of disease changes in actual clinical practice under the joint action of multiple symptoms;

[0038] The stage transition rate and the feature deterioration direction coefficient are fused to establish a deterioration risk coefficient model, which supports the quantitative classification of the evolution state of the disease, and provides continuous, objective and quantitative risk warning information for doctors to assist clinical decision-making and intervention timing selection. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 A monkeypox virus classification method based on image comparison according to the present application is shown in the flowchart;

[0040] Figure 2 A monkeypox virus classification system structure according to the present application is shown in the schematic diagram;

[0041] Figure 3 A device according to the present application is shown in the schematic diagram;

[0042] Figure 4 A storage medium according to the present application is shown in the schematic diagram. DETAILED DESCRIPTION

[0043] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0044] Embodiment 1, please refer to Figure 1 In this embodiment, a monkeypox virus classification method based on image comparison includes:

[0045] Monkeypox is a zoonosis caused by the monkeypox virus, and the monkeypox epidemic mainly concentrates in Central and West Africa, but in recent years it has shown a significant cross-regional transmission trend worldwide. Clinically, the symptoms of monkeypox are similar to smallpox, but the disease course is usually milder, mainly manifested as fever, headache, myalgia, lymphadenopathy, fatigue, and characteristic skin lesions. Skin lesions usually go through an evolution process from prodromal stage, papule stage, pustule stage, scab stage, and healing stage. Each stage has certain common characteristics, but also shows great heterogeneity due to individual differences, virus strain variation, and different immune status.

[0046] Although the skin lesions of monkeypox have a relatively clear evolution process, there are still many technical shortcomings in the monitoring and evaluation of skin lesion evolution and disease risk in the clinical diagnosis and treatment of monkeypox:

[0047] Lesion observation lacks continuity and time dimension is fragmented and fuzzy: In existing technologies, lesion evolution stages mostly rely on fixed-point observation or experience judgment, lacking a continuous time modeling mechanism combined with tag dynamic updating, making it difficult to track the gradual change process or mutation signal between stages. Especially when the boundaries of multiple stages are fuzzy, traditional methods cannot provide quantitative basis with time anchor;

[0048] Feature analysis is isolated and lacks multi-dimensional collaborative modeling mechanism: Traditional lesion recognition methods mostly rely on a single feature (such as rash density or erythema size), lacking the ability to model the collaborative evolution of multiple key symptom indicators, making it difficult to capture the associated change trend of multiple symptoms within the same stage, thus limiting the comprehensive understanding and prediction ability of lesion evolution mechanism;

[0049] Disease judgment logic is rough and lacks dynamic modeling ability for trend direction: Existing methods often rely on static thresholds or artificial experience for disease stage judgment, ignoring the differences in evolution speed and feature performance of different individuals, and failing to dynamically capture the rate of change of lesions from one state to another, lacking the ability to respond sensitively to the stage switching process. The concept of "trend directionality" is not introduced, making it difficult to determine whether the current feature change is deteriorating or improving. Especially in the early or potential deterioration stage, there is a lack of early warning mechanism for trend changes, which poses a risk of delayed intervention.

[0050] Therefore, it is urgent to build an intelligent diagnosis and treatment method that combines image features, time series evolution information, and multi-feature collaborative judgment, which can realize continuous modeling, accurate staging, and dynamic risk assessment of monkeypox skin lesion evolution, to improve the objectivity, forward-looking nature, and timeliness of clinical judgment.

[0051] In order to effectively solve the above problems, the present application provides a monkeypox virus classification method based on image comparison, which comprises:

[0052] S1, collecting normal skin images of the healthy side of the patient and initial lesion images, generating a healthy texture template and an initial lesion reference library; performing gray scale normalization on the healthy texture template and the initial lesion reference library through adaptive histogram stretching, and outputting a comparison reference image group;

[0053] S2, acquiring the current monitoring image of the patient, segmenting the lesion area in combination with the comparison reference image group, extracting the hierarchical features of the lesion area to generate a three-dimensional symptom density vector, and extracting the texture roughness of the surface of the lesion area to form a lesion quantification data set;

[0054] S3, presetting a staging feature matrix of each stage of monkeypox virus infection, determining the main staging label based on the matching degree of the lesion quantification data set and the staging feature matrix;

[0055] S4, taking the main staging label as a time sequence anchor point, dynamically fitting the three-dimensional symptom density vector of continuous monitoring to generate a lesion evolution trajectory curve, calculating the stage transition rate based on the slope change of the lesion evolution trajectory curve, and obtaining a disease deterioration risk coefficient;

[0056] S5, integrating the main staging label, the lesion evolution trajectory curve and the disease deterioration risk coefficient, generating a visual disease progression atlas to obtain a standardized clinical staging archive.

[0057] The method for generating the healthy texture template and the initial lesion reference library comprises:

[0058] Collecting normal skin images of the healthy side of the patient and initial lesion images, the normal skin images of the healthy side being healthy images of uninfected areas symmetrical to the initial lesion area; performing region alignment and cutting processing on the normal skin images of the healthy side to obtain a healthy texture block matching the size and shape of the initial lesion area, and performing brightness and color normalization processing on the block; extracting the initial lesion area from the initial lesion images, and performing standardization processing on the extracted initial lesion area images to generate an initial lesion reference library containing color features, texture features and area contours.

[0059] The method for obtaining the comparison reference image group comprises:

[0060] Converting the healthy texture block and the initial lesion image from a color image into a color space with separated brightness and color, and extracting the brightness channel information in the healthy texture block and the initial lesion image; performing adaptive histogram stretching operation on the brightness channel information; determining the stretching boundary value based on the local brightness distribution of the image, mapping the original brightness of the healthy texture block and the initial lesion image to a standard brightness interval, and then obtaining the comparison reference image group.

[0061] The method for obtaining the lesion quantification dataset comprises:

[0062] A current monitoring image of the patient is obtained and preprocessed to keep the brightness color distribution of the current monitoring image consistent with the comparison reference image group; based on the healthy texture template in the comparison reference image group and the lesion initial image, an abnormal area in the current monitoring image of the patient is identified as a lesion area through differential analysis;

[0063] The lesion area is subjected to hierarchical feature extraction to obtain epidermal elevation, erythema infiltration depth and blister density, and then a three-dimensional symptom density vector is constructed; a local binary pattern (LBP) texture roughness factor is introduced to extract the texture roughness of the surface of the lesion area, and the three-dimensional symptom density vector is combined to form the lesion quantification dataset.

[0064] The method for determining the main staging label comprises:

[0065] A staging feature matrix of each stage of the monkeypox virus infection is preset, for example, according to the development stages of monkeypox lesions in clinical data, the stages can be divided into prodromal stage, papule stage, pustule stage, scab stage and healing stage; the skin wounds corresponding to each stage have differences in morphology, density and texture, so the lesion area, gray mean value, texture statistical feature and surface texture roughness of the monkeypox lesion at different development stages can be selected as parameters to form the staging feature matrix; the lesion quantification dataset is subjected to similarity matching with the staging feature matrix, and the similarity matching result is taken as the main staging label (prodromal stage, papule stage, pustule stage, scab stage or healing stage) corresponding to the current monitoring image of the patient, so as to reflect the development stage of the monkeypox lesion corresponding to the current condition of the patient.

[0066] The method for obtaining the lesion evolution trajectory curve comprises:

[0067] The three features contained in the three-dimensional symptom density vector, i.e. epidermal elevation, erythema infiltration depth and blister density, are recorded at fixed time intervals, and after data cleaning, a continuous three-dimensional symptom density vector sequence is formed;

[0068] For example: first, the continuous monitoring data collection work is carried out, and the three core features of the three-dimensional symptom density vector, i.e. epidermal elevation, erythema infiltration depth and blister density, are recorded at fixed time intervals of 24 hours; after the above data collection is completed, the data cleaning process is immediately entered: the Z-score standardization method can be used to accurately identify and correct abnormal values; in view of the slight deviation of the monitoring time, the linear interpolation algorithm is used to map the data to the standard time axis, so as to ensure the coherence and consistency of the time series data, and finally the processed three-dimensional symptom density vector sequence is outputted;

[0069] The main stage label is associated with the three-dimensional symptom density vector sequence to form time-series anchor points. For example, when the main stage label is determined for the first time in a monitoring process, the corresponding time point is the time-series anchor point of that label. If the main stage label is changed in the future, a corresponding time-series anchor point will be added. All time-series anchor points are collected to form a multi-anchor time-series axis.

[0070] These time-series anchor points construct the time framework of lesion evolution, not only clearly delineating different stages of lesion evolution, but also providing key time node basis for subsequent dynamic fitting. Within the same major stage, given the significant stage homogeneity of lesion evolution, the trajectory can be reconstructed through standardized curve fitting methods (such as quadratic polynomial fitting). When the monitoring data crosses the boundaries of different stages, the system will automatically perform dynamic update operations of the fitting parameters according to the preset trigger mechanism, so as to accurately match the dynamic law of lesion evolution in the new stage.

[0071] For example, if a patient's primary stage is labeled "eruptive stage," after the target time-series anchor point is determined, continuous monitoring data is selected according to the set window size for parameter fitting. For instance, during the 10th monitoring, the data from the previous 5 monitoring sessions are used to fit curves of epidermal elevation, erythema infiltration depth, and herpes density. The curves show that the epidermal elevation speed is accelerating, the erythema infiltration depth is deepening, and the herpes density is increasing at a faster rate, which is consistent with the characteristics of the eruptive stage. This allows for clear tracking of the lesion's evolution process and provides basic data for subsequent calculations of the transition rate.

[0072] A sliding window is used to divide the three-dimensional symptom density vector sequence, and the fitting parameters are dynamically adjusted in combination with a multi-anchor time axis to obtain a window dataset. For example, the three-dimensional symptom density vector sequence is first divided into windows. For a certain time node in the multi-anchor time axis, the epidermal elevation, erythema infiltration depth, and herpes density data of the previous 5-7 consecutive monitoring are selected as the window dataset. The features of each dimension of the window dataset are fitted with quadratic polynomials to generate fitting curves, which are combined to form a lesion evolution trajectory curve, which can reflect the temporal changes of single features and the collaborative evolution of multiple features.

[0073] For example, considering the continuous evolution characteristics of the three-dimensional symptom density vector, a weighted least squares method is used for quadratic polynomial fitting (taking into account both linear trends and curvature changes, consistent with the clinical pattern of "slow development-acceleration-leveling" in monkeypox lesions). The mathematical expression is: ,in, For time anchor points First Fitted values ​​for each feature; Index of features , respectively representing the three features contained in the three-dimensional symptom density vector: epidermal elevation, erythema infiltration depth, and herpes density; , and are fitting coefficients, reflects the curvature, reflects the linear trend, is a constant term; by fitting the epidermal elevation, erythema infiltration depth and blister density with a quadratic polynomial, fitting curves , and are generated, which are combined to form a three-dimensional lesion evolution trajectory curve, which intuitively reflects the time series change of a single feature (such as the rising and falling trend of blister density over time) and the coordinated evolution of multiple features (such as the synchronous rise of epidermal elevation and blister density in the papule stage).

[0074] The method for obtaining the risk coefficient of disease deterioration includes:

[0075] The stage transition rate is used to quantify the speed of transition of monkeypox disease from the current stage to the adjacent stage (such as papule stage to ulcer stage), and the core is to reflect the “severity of stage switching” through the slope change rate of the lesion evolution trajectory curve, and the steps are as follows:

[0076] The first derivative of the fitting curve contained in the lesion evolution trajectory curve is calculated respectively to obtain the slope of each fitting curve at different time points; based on the slope difference of the adjacent two time points, the slope change rate of the fitting curve is obtained; the slope change rates of all fitting curves are weighted to obtain the stage transition rate; the greater the stage transition rate, the faster the change speed of the lesion feature, and the higher the possibility of stage transition;

[0077] Based on the slope of each fitting curve at different time points, the direction coefficient of the change of the disease feature is defined, and the feature deterioration marker is generated for the feature showing a deterioration trend; the direction coefficient is defined as , which is used to determine whether the epidermal elevation, erythema infiltration depth and blister density show a deterioration trend; wherein, is the direction coefficient of the epidermal elevation at time point ; is the direction coefficient of the erythema infiltration depth at time point ; is the direction coefficient of the blister density at time point ;

[0078] and The values of and are 0 or 1, which are determined by the slope of each fitting curve; when the slope is greater than 0, it reflects that the value of the corresponding feature increases, and the monkeypox disease shows a deterioration trend; when the slope is less than or equal to 0, it reflects that the value of the corresponding feature decreases, and the monkeypox disease tends to improve; for example: = ; wherein, the slope of the fitting curve corresponding to the skin elevation degree at the time point ;

[0079] The preset stage transition critical threshold and the single feature deterioration critical slope are taken as preset clinical benchmark thresholds, combined with the stage transition rate and the feature deterioration marker, to generate a deterioration risk coefficient, and the higher the deterioration risk coefficient, the higher the risk of monkeypox deterioration;

[0080] The deterioration risk coefficient ; wherein, is the weight coefficient of the stage transition rate, is the weight coefficient of the feature deterioration degree, such as , ; The stage transition rate at the time point ; The preset stage transition critical threshold; is 1 when , to avoid risk coefficient overflow caused by excessively large stage transition rate; The slope of the fitting curve corresponding to the th feature at the time point ; The preset single feature deterioration critical slope; indicates that only the slope proportion of the feature showing a deterioration trend is retained;

[0081] For example: preset clinical benchmark thresholds: preset stage transition critical threshold , preset single feature deterioration critical slope ; weight distribution: , ;

[0082] The monitoring data at the time point : the stage transition rate ; the slopes of the fitting curves corresponding to the skin elevation degree, the erythema infiltration depth and the blister density are respectively: (the skin elevation degree increases, a deterioration trend), (the erythema infiltration depth decreases, a recovery trend), (the blister density increases, a deterioration trend); based on this, the direction coefficient ;

[0083] At this time ; ; ; ; from which the risk coefficient at this time can be obtained;

[0084] The quantitative assessment of the disease evolution is achieved by the deterioration risk coefficient, for example: if , it reflects low risk (stable or improved condition); if , it reflects medium risk (need to strengthen monitoring); if , it reflects high risk (need for clinical intervention).

[0085] The method for obtaining the standardized clinical staging archive includes:

[0086] The main staging labels and the disease deterioration risk coefficients are associated with the lesion evolution trajectory curve indexed by time points, and the disease progression atlas is generated by visual rendering; based on the disease progression atlas, the standardized clinical staging archive (such as patient basic information, disease stage statistics, risk record, trajectory evolution and clinical suggestion summary, etc.) is constructed.

[0087] The preset clinical benchmark threshold is set by the staff, and the average value of multiple clinical benchmark indicators is taken as the preset clinical benchmark threshold by collecting different clinical benchmark indicators; similarly, the preset stage transition critical threshold and the preset single feature deterioration critical slope are set.

[0088] This embodiment sets time sequence anchor points according to the changes of the main staging labels, forms a time axis with a multi-anchor point structure, effectively solves the fragmentation and ambiguity of the traditional lesion observation method in the time dimension, and not only clearly divides the evolution stage of the lesion, but also supports the fitting parameter updating mechanism across the stages, improving the sensitivity and adaptability of the lesion stage switching;

[0089] By fitting the three-dimensional symptom density vector in each sliding window, a multi-dimensional trajectory curve is formed, which can not only analyze the change trend of each feature individually, but also observe the synergistic evolution relationship between them (such as the synchronous rise of the skin elevation degree and the blister density in a certain stage), and this synergistic feature fusion mode improves the accuracy and interpretability of the staging recognition and disease trend judgment;

[0090] The direction coefficient mechanism is innovatively introduced, which judges whether each feature is in a deterioration trend (such as continuous rise of blister density) by the slope of the fitting curve, and the feature deterioration mark is weighted and fused into the risk calculation process, compared with single threshold judgment, this mechanism is more consistent with the complexity of disease changes under the joint action of multiple symptoms in actual clinical practice;

[0091] The stage transition rate and the feature deterioration direction coefficient are fused to establish a deterioration risk coefficient model, which supports the quantitative classification of the disease evolution state, and provides continuous, objective and quantitative risk warning information for doctors to assist clinical decision-making and intervention timing selection.

[0092] Embodiment 2, please refer to Figure 2As shown, the embodiment does not describe part of the embodiment 1, provides a monkeypox virus classification system based on image comparison, comprising:

[0093] An individual reference modeling module collects normal skin and initial lesion images of the healthy side of the patient, generates a healthy texture template and an initial lesion reference library, and performs gray scale normalization on the healthy texture template and the initial lesion reference library through adaptive histogram stretching to output a comparison reference image group;

[0094] A symptom density quantification module acquires the current monitoring image of the patient, segments the lesion area in combination with the comparison reference image group, extracts the hierarchical features of the lesion area to generate a three-dimensional symptom density vector, and extracts the texture roughness of the lesion area surface to form a lesion quantification dataset;

[0095] A stage recognition transition module presets a staging feature matrix for each stage of monkeypox virus infection, determines the main staging label based on the matching degree of the lesion quantification dataset and the staging feature matrix;

[0096] A time sequence evolution tracking module takes the main staging label as a time sequence anchor point, dynamically fits the three-dimensional symptom density vector of continuous monitoring to generate a lesion evolution trajectory curve, and calculates the stage transition rate based on the slope change of the lesion evolution trajectory curve to obtain a disease deterioration risk coefficient;

[0097] A clinical care suggestion module integrates the main staging label, the lesion evolution trajectory curve, and the disease deterioration risk coefficient, generates a visual disease progression atlas to obtain a standardized clinical staging profile.

[0098] Embodiment 3, please refer to Figure 3 As shown, the present application also provides a device 500. The device 500 can include one or more processors and one or more memories. Among them, the memory stores computer readable code, which when run by one or more processors, can execute a monkeypox virus classification method based on image comparison as described above.

[0099] As Figure 3 shown, the device 500 can include a bus 501, one or more CPUs 502, a ROM 503, a RAM 504, a communication port 505 connected to a network, an input / output 506, a hard disk 507, etc. The storage device in the device 500, such as the ROM 503 or the hard disk 507, can store a monkeypox virus classification method based on image comparison provided by the present application. Further, the device 500 can also include a user interface 508. Of course, Figure 2 The architecture shown is only exemplary, and when implementing different devices, one or more components in the electronic device shown can be omitted Figure 2 according to actual needs.

[0100] Embodiment 4, please refer to Figure 4 Fig. 25 shows a storage medium 250 according to an embodiment of the present application. The storage medium 250 stores computer readable instructions. When the computer readable instructions are run by a processor, an image comparison based monkeypox virus classification method according to an embodiment of the present application described above with reference to the accompanying drawings can be executed. The storage medium 250 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM), cache memory, and the like. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, and the like.

[0101] Those skilled in the art can appreciate that the units and algorithm steps of the examples described in combination with the embodiments disclosed in the present application can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0102] In several embodiments provided by the present application, it should be understood that the disclosed system, device and method can be realized in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only one, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0103] The above description is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0104] Finally: the above description is only the preferred embodiment of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A monkeypox virus classification method based on image comparison, characterized in that, include: S1. Collect images of the patient's healthy skin and initial lesions on the healthy side, and generate healthy texture templates and initial baseline libraries for lesions; By adaptive histogram stretching, grayscale normalization is performed on the healthy texture template and the initial benchmark library of lesions, and a set of comparison benchmark images is output. S2. Obtain the patient's current monitoring image, segment the lesion area by comparing it with the baseline image group, extract the layered features of the lesion area to generate a three-dimensional symptom density vector, extract the texture roughness of the lesion area surface, and form a lesion quantification dataset. S3. Pre-set the staging feature matrix for each stage of monkeypox virus infection, and determine the main stage label based on the matching degree between the lesion quantification dataset and the staging feature matrix. S4. Using the main stage label as the time-series anchor point, dynamically fit the continuously monitored three-dimensional symptom density vector to generate the lesion evolution trajectory curve, and calculate the stage transition rate based on the slope change of the lesion evolution trajectory curve to obtain the disease deterioration risk coefficient. The methods for obtaining the risk factor for disease worsening include: The first derivative of each fitted curve contained in the lesion evolution trajectory curve is calculated to obtain the slope of each fitted curve at different time points; the slope change rate of the fitted curve is obtained based on the slope difference between two adjacent time points; the slope change rates of all fitted curves are weighted to obtain the stage transition rate. Based on the slope of each fitted curve at different time points, a directional coefficient for the change of disease characteristics is defined, and a feature deterioration label is generated for features that show a worsening trend. The directional coefficient is determined by the slope of each fitted curve. When the slope is greater than 0, it reflects that the value of the corresponding feature is increasing and the monkeypox disease is worsening. When the slope is less than or equal to 0, it reflects that the value of the corresponding feature is decreasing and the monkeypox disease is improving. Preset clinical baseline thresholds, and combine them with stage transition rates and characteristic deterioration markers to generate a deterioration risk coefficient; S5 integrates the main stage label, lesion evolution trajectory curve, and disease progression risk coefficient to generate a visualized disease progression map and obtain standardized clinical staging records.

2. The monkeypox virus classification method based on image comparison according to claim 1, characterized in that, The method for generating healthy texture templates and initial baseline libraries of lesions includes: Images of normal skin on the healthy side of the patient and images of the initial lesion were acquired. The normal skin image on the healthy side was processed by region alignment and cropping to obtain a healthy texture patch that matches the size and shape of the initial lesion region. The brightness and color of the patch were then normalized. The lesion region was extracted from the initial lesion image. The extracted initial lesion region image was then standardized to generate an initial lesion baseline library containing color features, texture features, and region contours.

3. The monkeypox virus classification method based on image comparison according to claim 2, characterized in that, The method for obtaining the comparison reference image set includes: The healthy texture patch and the initial lesion image are converted from color images to a color space that separates luminance and color, and the luminance channel information in the healthy texture patch and the initial lesion image is extracted. An adaptive histogram stretching operation is performed on the luminance channel information. The stretching boundary value is determined based on the local luminance distribution of the image, and the original luminance of the healthy texture patch and the initial lesion image is mapped to the standard luminance range, thereby obtaining the comparison benchmark image group.

4. The monkeypox virus classification method based on image comparison according to claim 3, characterized in that, The methods for obtaining the lesion quantification dataset include: The current monitoring image of the patient is acquired and preprocessed. Based on the healthy texture template and the initial image of the lesion in the comparison reference image group, the abnormal area in the current monitoring image of the patient is identified as the lesion area through differential analysis. Layered feature extraction was performed on the lesion area to obtain epidermal elevation, erythema infiltration depth and herpes density, and then a three-dimensional symptom density vector was constructed. The texture roughness of the lesion area surface was extracted and combined with the three-dimensional symptom density vector to form a lesion quantification dataset.

5. The monkeypox virus classification method based on image comparison according to claim 4, characterized in that, The method for determining the primary installment label includes: A staging feature matrix for each stage of monkeypox virus infection is pre-defined. The lesion quantification dataset is matched with the staging feature matrix for similarity, and the similarity matching result is used as the main stage label corresponding to the patient's current monitoring image.

6. The monkeypox virus classification method based on image comparison according to claim 5, characterized in that, The method for obtaining the lesion evolution trajectory curve includes: Three-dimensional symptom density vectors are recorded at fixed time intervals. After data cleaning, a continuous three-dimensional symptom density vector sequence is formed. The main stage label is associated with the three-dimensional symptom density vector sequence to form a time-series anchor point. If the main stage label changes in the future, a corresponding time-series anchor point will be added. All time-series anchor points are collected to form a multi-anchor time-series axis. The three-dimensional symptom density vector sequence is divided into sliding windows, and the fitting parameters are dynamically adjusted by combining multiple anchor point time axes to obtain the window dataset. The features of each dimension of the window dataset are fitted with quadratic polynomials to generate fitting curves, which are then combined to form the lesion evolution trajectory curve.

7. The monkeypox virus classification method based on image comparison according to claim 6, characterized in that, The method for obtaining standardized clinical staging records includes: Using time points as an index, the main stage label and the risk coefficient of disease progression are linked to the lesion evolution trajectory curve, and a disease progression map is generated through visualization rendering; based on the disease progression map, a standardized clinical staging file is constructed.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the electronic device executes the computer program, it implements the monkeypox virus classification method based on image comparison as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed, implements the image comparison-based monkeypox virus classification method according to any one of claims 1-7.

Images