Red spot squamous skin disease auxiliary diagnosis system

Abstract

The invention belongs to the technical field of auxiliary diagnosis of erythema-squamous skin diseases, and discloses an auxiliary diagnosis system for erythema-squamous skin diseases. The erythema squamous skin disease auxiliary diagnosis system comprises a skin image acquisition module, a skin feature extraction module, an identification module, a skin comparison module, a skin disease judgment module, a treatment scheme editing module and a display module. According to the invention, through the skin feature extraction module, erythema scale skin disease image features can be accurately obtained, and the diagnosis accuracy is improved; meanwhile, the type of the erythema squamous skin disease suffered by the patient can be rapidly recognized through the recognition module, a doctor is assisted in diagnosis and judgment, and the erythema squamous skin disease diagnosis efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of auxiliary diagnosis of erythematosquamous skin diseases, and in particular relates to an auxiliary diagnosis system for erythematosquamous skin diseases. Background Art

[0002] Erythematosquamous dermatosis is a general term for a type of skin disease, among which psoriasis is the most common. The treatment of psoriasis requires the selection of an appropriate treatment plan based on the state of the skin lesions. For those with fewer lesions, glucocorticoids combined with topical calcipotriol cream can be given for sequential treatment. For those with more extensive skin lesions, narrow-spectrum ultraviolet light therapy can be used. Some patients can be given oral acitretin, and for severe cases, immunosuppressants can also be given. Patients with psoriasis should be careful not to use glucocorticoids systemically to avoid erythroderma or pustular psoriasis. Patients with psoriasis also need to avoid upper respiratory tract infections, avoid tension and anxiety, and avoid spicy and irritating foods. However, the existing auxiliary diagnosis system for erythematosquamous dermatosis cannot accurately obtain the image features of erythematosquamous dermatosis; at the same time, for patients, they cannot predict the type of erythematosquamous dermatosis they have before seeing a doctor, and going to the hospital to register for a treatment is not targeted. Doctors also need to exclude one by one during the diagnosis process and finally determine the type of erythematosquamous dermatosis the patient suffers from; on the one hand, it increases the workload of doctors and indirectly affects the quality of doctors' diagnosis; on the other hand, it leads to a sharp drop in the number of patients who can see a doctor within one day, making it impossible for many patients to quickly receive a diagnosis and obtain diagnosis results, wasting patients' time.

[0003] To sum up, the problems existing in the prior art are: the existing auxiliary diagnosis system for erythematosquamous dermatosis cannot accurately obtain the image features of erythematosquamous dermatosis; at the same time, for patients, they cannot predict the type of erythematosquamous dermatosis they have before seeing a doctor, and going to the hospital to register for a doctor is not targeted. Doctors also need to exclude one by one during the diagnosis process and finally determine the type of erythematosquamous dermatosis the patient suffers from; on the one hand, it increases the workload of doctors and indirectly affects the quality of doctors' diagnosis; on the other hand, it leads to a sharp drop in the number of patients who can see a doctor within one day, making it impossible for many patients to quickly undergo a diagnosis and obtain diagnostic results, wasting patients' time. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides an auxiliary diagnosis system for erythematosquamous skin diseases.

[0005] The present invention is implemented in this way: an auxiliary diagnosis system for erythematous and scaly skin diseases comprises:

[0006] Skin image acquisition module, skin feature extraction module, recognition module, skin comparison module, skin disease judgment module, treatment plan editing module, display module;

[0007] A skin image acquisition module, used to acquire patient skin images through a medical camera device;

[0008] A skin feature extraction module, used for extracting color features of skin images through an extraction procedure;

[0009] An identification module, used for identifying the type of erythematosquamous skin disease through an identification procedure;

[0010] A skin comparison module, used for comparing the collected skin image with the erythematous and scaly skin image through a comparison program;

[0011] A skin disease judgment module, used to judge the state of erythematous and scaly skin diseases through a judgment program;

[0012] A treatment plan editing module, used for editing the treatment plan for erythematous and scaly skin diseases through an editing program;

[0013] The display module is used to display skin images, recognition results, comparison results, judgment results, and treatment plans through a display.

[0014] Furthermore, the skin feature extraction module extraction method is as follows:

[0015] (1) constructing an image database, converting the RGB color model of the erythematous and scaly skin disease image into the HSI color model; storing the erythematous and scaly skin disease image in the image database for preservation;

[0016] (2) In the HSI color model, scan each pixel of the erythematosquamous dermatitis image to obtain the HSI color model value (h, s, i) and HSI color feature vector of each pixel;

[0017] (3) According to the skin colors of different types of erythematosquamous skin diseases in the clinical diagnosis of erythematosquamous skin diseases, a naive Bayes classifier is used to classify the HSI color feature vector corresponding to each pixel into different types of skin colors;

[0018] (4) Count the number of pixels in the skin color of each type of disease to obtain the image features of erythematosquamous skin diseases.

[0019] Furthermore, the formula for converting the RGB color model into the HSI color model is as follows:

[0020]

[0021] Where H is hue, θ is phase angle, S is saturation, I is brightness, and min(R,G,B) is a function that takes the minimum value of red R, green G, or blue B.

[0022] Further, the HSI color feature vectors are obtained, which are:

[0023] X HS =[h,s]

[0024] X HI =[h,i]

[0025] X SI =[s,i]

[0026] These three HSI color feature vectors are directly obtained from the HSI color model values ​​(h, s, i).

[0027] Furthermore, the diseased skin color includes 8 types of colors: white, red, light brown, dark brown, light blue-gray, dark blue-gray, purple and black.

[0028] Furthermore, the counting of the number of pixels in the skin color of each type of disease to obtain the image features of the erythematosquamous skin disease also includes constructing a skin color histogram of each type of disease based on the counted number of pixels in the skin color of each type of disease to obtain the histogram features of the erythematosquamous skin disease image.

[0029] Further, the identification module identification method is as follows:

[0030] 1) Collect typical erythematous and scaly skin disease data images, and enhance the erythematous and scaly skin disease data images through image enhancement programs; train based on deep neural networks, output training results, and establish a training model;

[0031] 2) Upload the skin picture you took, call the trained model for analysis and classification, and output the recognition result.

[0032] Furthermore, the collection of typical erythematous and scaly skin disease data images, training based on a deep neural network, outputting training results, and establishing a training model include the following steps:

[0033] Establish model database;

[0034] Collect typical erythematous and scaly skin disease data images and store them in the model database;

[0035] Add corresponding labels according to the different types of pictures in the model database;

[0036] Read the image information in the model database and perform image processing;

[0037] After the image processing is completed, the deep neural network is used for training calculations, the training results are output, and the training model is saved.

[0038] Furthermore, the image processing comprises the following steps:

[0039] Perform image preprocessing;

[0040] Perform data conversion;

[0041] Back up the converted image data.

[0042] Furthermore, after the image processing is completed, training calculation is performed through a deep neural network, the training results are output, and the training model is saved, including the following steps:

[0043] a1. According to the training instructions, retrieve the training data to train the images with the same labels in the processed model database;

[0044] a2. After the training is completed, determine whether the training accuracy meets the requirements. If yes, output the training results, save the training model, and go to step a3; if not, re-enter step a1;

[0045] a3. Based on the training model, use transfer learning to establish corresponding training models for the images of other labels in the model database.

[0046] The advantages and positive effects of the present invention are:

[0047] First, the present invention can accurately obtain the image features of erythematosquamous dermatosis through the skin feature extraction module, thereby improving the accuracy of diagnosis; at the same time, the identification module can quickly identify the type of erythematosquamous dermatosis suffered by the patient, assist the doctor in making a diagnosis, and improve the efficiency and accuracy of erythematosquamous dermatosis diagnosis.

[0048] Second, the present invention proposes a comprehensive auxiliary diagnosis system to address key issues in the diagnosis of erythematosquamous skin diseases, which significantly improves the accuracy and efficiency of diagnosis. Traditional skin disease diagnosis mainly relies on the doctor's experience and naked eye observation, which is easily affected by subjective factors, and the reliability and consistency of the diagnostic results are low. In addition, the existing technology lacks comprehensive energy consumption evaluation of different equipment, making it difficult to achieve comprehensive optimization. These problems lead to deviations in diagnostic results and incomplete energy consumption optimization.

[0049] First, the present invention uses advanced image processing technology to extract accurate color and texture features from the patient's skin image through the skin image acquisition module and the skin feature extraction module. This method overcomes the limitations of traditional manual observation, provides more objective and accurate skin feature data, and lays a solid foundation for subsequent disease identification.

[0050] Secondly, the present invention introduces a recognition module based on machine learning and deep learning, which can automatically identify and classify the types of erythematosquamous dermatosis. This technological advancement significantly improves the accuracy and efficiency of diagnosis, reduces the problems of misdiagnosis and missed diagnosis caused by human judgment, and ensures the high consistency and reliability of the diagnosis results.

[0051] In addition, the present invention realizes accurate comparison of the patient's skin image with the standard image and comprehensive judgment of the disease state through the skin comparison module and the skin disease judgment module. This comprehensive comparison and judgment method not only improves the accuracy of diagnosis, but also can analyze the severity of the disease in detail, providing a scientific basis for the formulation of personalized treatment plans.

[0052] Finally, the treatment plan editing module and display module of the present invention can automatically generate personalized treatment plans based on the disease judgment results, and display diagnosis and treatment information through a friendly user interface. This technological advancement significantly improves the efficiency of the diagnosis and treatment process and the patient's treatment experience, making the diagnosis and treatment process more intelligent, systematic and humane, and promoting the modernization of skin disease diagnosis and treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a structural block diagram of an auxiliary diagnosis system for erythematosquamous skin diseases provided by an embodiment of the present invention.

[0054] Figure 2 It is a flow chart of the skin feature extraction module extraction method provided by an embodiment of the present invention.

[0055] Figure 3 It is a flow chart of the identification module identification method provided by an embodiment of the present invention.

[0056] Figure 1 In: 1. Skin image acquisition module; 2. Skin feature extraction module; 3. Recognition module; 4. Skin comparison module; 5. Skin disease judgment module; 6. Treatment plan editing module; 7. Display module. DETAILED DESCRIPTION

[0057] In order to further understand the content, features and effects of the present invention, the following embodiments are given as examples and described in detail with reference to the accompanying drawings.

[0058] The present invention introduces a recognition module based on machine learning and deep learning, which can automatically identify and classify the types of erythematosquamous dermatosis. The following is a specific implementation scheme.

[0059] First, the system collects a large number of high-quality erythematosus and scaly skin disease images and corresponding diagnostic data. These image datasets come from multiple hospitals and clinics, covering a variety of types and conditions of erythematosus and scaly skin diseases. The images are preprocessed, including noise removal, standardization, and enhancement processing, to ensure the consistency and high quality of the data. Then, these image data are divided into training sets, validation sets, and test sets to provide a basis for subsequent machine learning and deep learning model training.

[0060] After image preprocessing, the system uses image processing algorithms to extract features from the image. Color feature extraction obtains color distribution information of skin images through methods such as color histogram and color moment. Texture feature extraction obtains texture information of the skin surface through methods such as gray-level co-occurrence matrix and local binary pattern (LBP). The extracted feature data will be used to construct feature vectors as input to the machine learning model.

[0061] The system uses convolutional neural networks (CNN) in deep learning to identify and classify erythematosus and scaly skin diseases. Select appropriate network architectures, such as ResNet, DenseNet, etc., to build deep learning models. During the training process, the network parameters are continuously adjusted, the model is trained using image data in the training set, and the model is validated using the validation set. In order to improve the robustness of the model, data enhancement techniques such as rotation, scaling, and flipping are used to expand the training data set.

[0062] After the initial training, the model performance is further optimized through cross-validation and hyperparameter tuning. The model's hyperparameters, such as learning rate, batch size, number of layers, etc., are optimized using grid search, random search, etc. Regularization techniques (such as L2 regularization, Dropout) are used to prevent the model from overfitting and ensure the generalization ability of the model in practical applications.

[0063] The trained and optimized model is evaluated using the test set. The performance of the model is measured by indicators such as accuracy, recall, and F1 score. In order to ensure the practical application effect of the model, the confusion matrix is ​​used to analyze the recognition accuracy of the model in different categories, and to discover and improve the shortcomings of the model in the recognition of specific diseases.

[0064] The optimized model is finally deployed into the system as part of the recognition module. In actual applications, the system automatically identifies and classifies the types of erythematosquamous dermatosis by collecting skin images of patients and using the extracted feature vectors to input into the deep learning model. The recognition results are presented through the display module for doctors’ reference, and combined with the analysis results of other modules, it helps doctors formulate more accurate and personalized treatment plans.

[0065] Through the above-mentioned specific implementation scheme, the system can efficiently and accurately identify the types of erythematosquamous skin diseases, significantly improve the accuracy and efficiency of diagnosis, reduce misdiagnosis and missed diagnosis, and ensure high consistency and reliability of diagnostic results.

[0066] The following are two specific embodiments of the auxiliary diagnosis system for erythematous and scaly skin diseases:

[0067] Example 1: Optimization of hospital dermatology diagnosis process

[0068] 1) Skin image acquisition: When a patient visits the dermatology department of a hospital, the doctor uses a medical camera to capture images of the patient's erythematous and scaly areas. The camera has high resolution and can clearly capture skin details.

[0069] 2) Skin feature extraction: The collected skin images are processed by the skin feature extraction module. This module uses advanced image processing technology to automatically extract key features such as color and texture in the image. These features are the basis for subsequent recognition and analysis.

[0070] 3) Skin disease identification and comparison: The identification module uses the trained machine learning model to identify the type of erythematosus and scaly skin diseases based on the extracted features. At the same time, the skin comparison module compares the collected skin images with the erythematosus and scaly skin images in the database to assist doctors in making more accurate diagnoses.

[0071] 4) Symptom judgment: The skin disease judgment module judges the patient's erythematous and scaly skin disease status based on the results of identification and comparison, combined with the patient's medical history and other relevant information, such as the severity of the disease and whether there are other complications.

[0072] 5) Treatment plan editing and display: Based on the disease judgment results, the treatment plan editing module automatically generates or assists doctors in formulating personalized treatment plans. Finally, through the display module, doctors can view the patient's skin image, recognition results, comparison results, judgment results, and treatment plans, so as to have a more comprehensive understanding of the patient's condition and formulate a reasonable treatment plan.

[0073] Example 2: Remote medical diagnosis service

[0074] 1) Self-service image collection by patients: Patients use smartphones or special cameras at home to take pictures of their own erythematosus and scaly areas and upload the images to the cloud server.

[0075] 2) Cloud feature extraction and recognition: After the server receives the image, the skin feature extraction module automatically extracts the image features. Then, the recognition module recognizes the image and determines the type of erythematosus and scaly skin disease.

[0076] 3) Online skin comparison and disease judgment: The skin comparison module compares and analyzes the images uploaded by the patient with the standard images in the database. The skin disease judgment module makes a preliminary judgment on the patient's condition based on the recognition and comparison results.

[0077] 4) Remote treatment plan recommendations: Based on the diagnosis results, the treatment plan editing module provides patients with preliminary treatment recommendations or guides patients to go to the nearest medical institution for further examination and treatment.

[0078] 5) Result display and feedback: Patients can view the recognition results, comparison results, disease diagnosis results and treatment plan recommendations through smartphones or computers. At the same time, the system also provides online consultation services, so patients can ask questions to doctors at any time and get professional answers.

[0079] like Figure 1 As shown, the auxiliary diagnosis system for erythematosquamous skin diseases provided by the embodiment of the present invention includes: a skin image acquisition module 1, a skin feature extraction module 2, a recognition module 3, a skin comparison module 4, a skin disease judgment module 5, a treatment plan editing module 6, and a display module 7.

[0080] The skin image acquisition module 1 is used to acquire the patient's skin image through a medical camera device;

[0081] Skin feature extraction module 2, used to extract skin image color features through an extraction program;

[0082] Identification module 3, used to identify the type of erythematosquamous skin disease through an identification program;

[0083] The skin comparison module 4 is used to compare the collected skin image with the erythematous and scaly skin image through a comparison program;

[0084] The skin disease judging module 5 is used to judge the state of erythematous and scaly skin disease through a judging procedure;

[0085] The treatment plan editing module 6 is used to edit the treatment plan for erythematous and scaly skin diseases through an editing program;

[0086] The display module 7 is used to display the skin image, recognition result, comparison result, judgment result and treatment plan through the display.

[0087] The auxiliary diagnosis system for erythematous and scaly skin diseases realizes the auxiliary diagnosis and treatment plan formulation of erythematous and scaly skin diseases through the collaborative work of multiple modules. The following is the detailed working principle of the system:

[0088] System structure and module functions

[0089] 1) Skin image acquisition module:

[0090] A high-resolution image of the patient's skin is captured by a medical imaging device (such as a digital camera or a microscope).

[0091] Aim the camera at the patient's skin area to obtain clear skin surface images, which are stored in the system database for subsequent processing.

[0092] 2) Skin feature extraction module:

[0093] Color features and texture features are extracted from the collected skin images.

[0094] Image processing algorithms (such as color histogram and texture analysis algorithms) are used to analyze skin images and extract specific color distribution and texture features, which are of diagnostic significance for erythematosquamous skin diseases.

[0095] 3) Identification module:

[0096] Identify and classify types of erythematosquamous skin diseases.

[0097] Based on the extracted skin features, machine learning or deep learning algorithms (such as support vector machines, convolutional neural networks) are used for classification and recognition to classify skin disease images into specific types of erythematosus and scaly skin diseases.

[0098] 4) Skin comparison module:

[0099] The patient's skin images were compared with images of erythematosquamous skin diseases in the database.

[0100] Image comparison algorithms (such as structural similarity index and feature matching algorithm) are used to compare the similarity between patient images and standard images to help confirm the type of disease and the severity of the disease.

[0101] 5) Skin disease judgment module:

[0102] Determine the symptoms and condition of erythematous and scaly skin diseases.

[0103] Combining the results of the recognition module with the contrast results of the comparison module, medical rules and empirical formulas are applied to determine the specific state of the disease (such as mild, moderate, severe) and provide preliminary diagnostic opinions.

[0104] 6) Treatment plan editing module:

[0105] Edit and develop treatment plans for erythematosquamous skin diseases.

[0106] Based on the condition given by the judgment module, combined with medical treatment standards and expert advice, edit a personalized treatment plan, including drug therapy, physical therapy and life advice, and generate a detailed treatment plan.

[0107] 7) Display module:

[0108] Skin images, recognition results, comparison results, judgment results and treatment plans are displayed on the monitor.

[0109] The processing results of each module are displayed through a user-friendly interface, which is convenient for doctors and patients to view and provides clear and intuitive diagnostic information and treatment suggestions.

[0110] Workflow:

[0111] 1) Image acquisition: The doctor or operator uses the skin image acquisition module to obtain the patient's skin image.

[0112] 2) Feature extraction: The system automatically calls the skin feature extraction module to analyze the collected images and extract key features.

[0113] 3) Disease type identification: The recognition module classifies the extracted features and identifies the specific type of erythematosquamous skin disease.

[0114] 4) Image comparison: The skin comparison module compares the patient's image with the standard image to confirm the type and severity of the disease.

[0115] 5) Symptom judgment: The skin symptom judgment module comprehensively analyzes the recognition and comparison results to judge the symptom status.

[0116] 6) Plan formulation: The treatment plan editing module formulates a personalized treatment plan based on the judgment results.

[0117] 7) Result display: The display module displays all the results on the monitor for doctors’ reference and patients’ understanding.

[0118] Through the collaborative work of the above steps and modules, the system can efficiently and accurately assist doctors in the diagnosis and formulation of treatment plans for erythematosquamous dermatoses, thereby improving the accuracy of diagnosis and treatment effectiveness.

[0119] like Figure 2 As shown, the skin feature extraction module 2 provided by the present invention has the following extraction method:

[0120] S101, constructing an image database, converting the RGB color model of the erythematous and scaly skin disease image into an HSI color model; storing the erythematous and scaly skin disease image in the image database for preservation;

[0121] S102, scanning each pixel of the erythematosquamous dermatitis image in the HSI color model to obtain the HSI color model value (h, s, i) and the HSI color feature vector of each pixel;

[0122] S103, according to the skin colors of different types of diseases in the clinical diagnosis of erythematosquamous skin diseases, using a naive Bayes classifier, classifying the HSI color feature vector corresponding to each pixel point into the skin colors of different types of diseases;

[0123] S104, counting the number of pixels in the skin color of each type of disease to obtain the image features of erythematosquamous skin diseases.

[0124] The formula for converting the RGB color model provided by the present invention into the HSI color model is as follows:

[0125]

[0126] Where H is hue, θ is phase angle, S is saturation, I is brightness, and min(R,G,B) is a function that takes the minimum value of red R, green G, or blue B.

[0127] The HSI color feature vectors provided by the present invention are:

[0128] X HS =[h,s]

[0129] X HI =[h,i]

[0130] X SI =[s,i]

[0131] These three HSI color feature vectors are directly obtained from the HSI color model values ​​(h, s, i).

[0132] The diseased skin colors provided by the present invention include 8 types of colors: white, red, light brown, dark brown, light blue-gray, dark blue-gray, purple and black.

[0133] The present invention provides a method of counting the number of pixels in the skin color of each type of disease to obtain the image features of erythematosquamous skin diseases, which also includes constructing a skin color histogram of each type of disease based on the counted number of pixels in the skin color of each type of disease to obtain the histogram features of the erythematosquamous skin disease image.

[0134] like Figure 3 As shown, the identification method of the identification module 3 provided by the present invention is as follows:

[0135] S201, collect typical erythematous and scaly skin disease data images, enhance the erythematous and scaly skin disease data images through an image enhancement program; perform training based on a deep neural network, output training results, and establish a training model;

[0136] S202, upload the taken skin picture, call the training model for analysis and classification, and output the recognition result.

[0137] The method provided by the present invention collects typical erythematosus and scaly skin disease data images, performs training based on a deep neural network, outputs training results, and establishes a training model, including the following steps:

[0138] Establish model database;

[0139] Collect typical erythematous and scaly skin disease data images and store them in the model database;

[0140] Add corresponding labels according to the different types of pictures in the model database;

[0141] Read the image information in the model database and perform image processing;

[0142] After the image processing is completed, the deep neural network is used for training calculations, the training results are output, and the training model is saved.

[0143] The image processing provided by the present invention comprises the following steps:

[0144] Perform image preprocessing;

[0145] Perform data conversion;

[0146] Back up the converted image data.

[0147] After the image processing provided by the present invention is completed, training calculation is performed through a deep neural network, the training results are output, and the training model is saved, including the following steps:

[0148] a1. According to the training instructions, retrieve the training data to train the images with the same labels in the processed model database;

[0149] a2. After the training is completed, determine whether the training accuracy meets the requirements. If yes, output the training results, save the training model, and go to step a3; if not, re-enter step a1;

[0150] a3. Based on the training model, use transfer learning to establish corresponding training models for the images of other labels in the model database.

[0151] When the present invention is working, first, a skin image acquisition module 1 is used to acquire a patient's skin image through a medical camera; a skin feature extraction module 2 is used to extract skin image color features through an extraction program; an identification module 3 is used to identify the type of erythematous and scaly skin disease through a recognition program; a skin comparison module 4 is used to compare the acquired skin image with the erythematous and scaly skin image through a comparison program; a skin disease judgment module 5 is used to judge the state of the erythematous and scaly skin disease through a judgment program; then, a treatment plan editing module 6 is used to edit the erythematous and scaly skin disease treatment plan through an editing program; and finally, a display module 7 is used to display the skin image, recognition result, comparison result, judgment result, and treatment plan through a display.

[0152] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any simple modification, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention are within the scope of the technical solution of the present invention.

Claims

1. An auxiliary diagnosis system for erythematous and scaly skin diseases, characterized in that: include: A skin image acquisition module, used to acquire patient skin images through a medical camera device; A skin feature extraction module, used for extracting color features of skin images through an extraction procedure; An identification module, used for identifying the type of erythematosquamous skin disease through an identification procedure; A skin comparison module, used for comparing the collected skin image with the erythematous and scaly skin image through a comparison program; A skin disease judgment module, used to judge the state of erythematous and scaly skin diseases through a judgment program; A treatment plan editing module, used for editing the treatment plan for erythematous and scaly skin diseases through an editing program; The display module is used to display skin images, recognition results, comparison results, judgment results, and treatment plans through a display.

2. The erythematosquamous skin disease auxiliary diagnosis system according to claim 1, characterized in that: The skin feature extraction module extraction method is as follows: (1) constructing an image database, converting the RGB color model of the erythematous and scaly skin disease image into the HSI color model; storing the erythematous and scaly skin disease image in the image database for preservation; (2) In the HSI color model, scan each pixel of the erythematosquamous dermatitis image to obtain the HSI color model value (h, s, i) and HSI color feature vector of each pixel; (3) According to the skin colors of different types of erythematosquamous skin diseases in the clinical diagnosis of erythematosquamous skin diseases, a naive Bayes classifier is used to classify the HSI color feature vector corresponding to each pixel into different types of skin colors; (4) Count the number of pixels in the skin color of each type of disease to obtain the image features of erythematosquamous skin diseases.

3. The auxiliary diagnosis system for erythematosquamous skin diseases according to claim 2, characterized in that: The formula for converting the RGB color model into the HSI color model is as follows: Where H is hue, θ is phase angle, S is saturation, I is brightness, and min(R,G,B) is a function that takes the minimum value of red R, green G, or blue B.

4. The auxiliary diagnosis system for erythematosquamous skin diseases according to claim 2, characterized in that: The HSI color feature vectors are obtained as follows: X HS =[h,s] X HI =[h,i] X SI =[s,i] These three HSI color feature vectors are directly obtained from the HSI color model values ​​(h, s, i).

5. The auxiliary diagnosis system for erythematosquamous skin diseases according to claim 2, characterized in that: The diseased skin color includes 8 types of colors: white, red, light brown, dark brown, light blue-gray, dark blue-gray, purple and black.

6. The auxiliary diagnosis system for erythematosquamous skin diseases according to claim 2, characterized in that: The method of counting the number of pixels in the skin color of each type of disease to obtain the image features of erythematosquamous skin diseases also includes constructing a skin color histogram of each type of disease based on the counted number of pixels in the skin color of each type of disease to obtain the histogram features of the erythematosquamous skin disease image.

7. The erythematosquamous skin disease auxiliary diagnosis system according to claim 1, characterized in that: The identification module identification method is as follows: 1) Collect typical erythematous and scaly skin disease data images, and enhance the erythematous and scaly skin disease data images through image enhancement programs; train based on deep neural networks, output training results, and establish a training model; 2) Upload the skin picture you took, call the trained model for analysis and classification, and output the recognition result.

8. The auxiliary diagnosis system for erythematosquamous skin diseases according to claim 7, characterized in that: The collecting of typical erythematous and scaly skin disease data images, training based on a deep neural network, outputting training results, and establishing a training model include the following steps: Establish model database; Collect typical erythematous and scaly skin disease data images and store them in the model database; Add corresponding labels according to the different types of pictures in the model database; Read the image information in the model database and perform image processing; After the image processing is completed, the deep neural network is used for training calculations, the training results are output, and the training model is saved.

9. The auxiliary diagnosis system for erythematosquamous skin diseases according to claim 8, characterized in that: The image processing comprises the following steps: Perform image preprocessing; Perform data conversion; Back up the converted image data.

10. The erythematosquamous skin disease auxiliary diagnosis system according to claim 8, characterized in that: After the image processing is completed, the training calculation is performed through the deep neural network, the training results are output, and the training model is saved, which includes the following steps: a1. According to the training instructions, retrieve the training data to train the images with the same labels in the processed model database; a2. After the training is completed, determine whether the training accuracy meets the requirements. If yes, output the training results, save the training model, and go to step a3; if not, re-enter step a1; a3. Based on the training model, use transfer learning to establish corresponding training models for the images of other labels in the model database.

Images