Diabetes wound image analysis and intelligent rehabilitation method, system and device
Identifying the characteristics of diabetic wounds through image analysis technology, solving the problem of subjectivity and low monitoring frequency of wound assessment, real-time feedback on wound health status and personalized rehabilitation suggestions, and improving the rehabilitation effect of diabetic patients.
Patent Information
- Application Number
- CN202411817252.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-11
- Publication Date
- 2025-03-14
AI Technical Summary
In the rehabilitation of diabetes, wound assessment depends on the subjective judgment of medical staff, low monitoring frequency and insufficient data integration, resulting in inconsistent evaluation results and delayed treatment timing.
Through image acquisition, preprocessing, feature extraction and classification analysis, convolutional neural networks and deep learning models are used to identify the color, area, depth and edge features of the wound, and feedback the health status of the wound in real time, and provide data support for medical staff.
Real-time and accurate assessment of diabetic wounds is achieved, the frequency of monitoring of wound changes and data integration level is improved, thereby improving the recovery process of diabetic patients.
Smart Images
Figure CN119626471A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent medical technology, and more specifically, to a method, system and device for diabetic wound image analysis and intelligent rehabilitation. Background Art
[0002] Wound healing is a complex biological process, in which multiple factors such as inflammatory cells, repair cells, extracellular matrix and cytokines participate in, coordinate and regulate each other. The skin of diabetic patients is easily damaged, and the damage often recurs, persists and does not heal, forming stubborn and difficult-to-heal ulcers. This is mainly because the tissue cells of diabetic patients are in a high-sugar environment for a long time, which makes local skin tissue cells and proteins glycosylated, causing vascular endothelial dysfunction and impaired response; macrophage dysfunction, prolonged and unbalanced inflammatory response; lack or damage of growth factors, reduction of gelatin tissue, disorder of epidermal barrier function, and reduction of granulation tissue, which make the wound difficult to repair.
[0003] At present, there are the following problems in diabetic wound rehabilitation:
[0004] 1. Wound assessment often relies on the subjective judgment of medical staff, which may lead to inconsistent assessment results;
[0005] 2. Low monitoring frequency: Traditional methods require regular follow-up visits, which makes it difficult to monitor wound changes in real time, delaying treatment opportunities;
[0006] 3. Insufficient data integration: There is a lack of effective tools to integrate wound images with other health data of patients, which affects the formulation of personalized treatment plans. Summary of the invention
[0007] The purpose of the present invention is to provide a method, system and device for diabetic wound image analysis and intelligent rehabilitation, which can identify the color, area, depth and edge characteristics of the wound, provide timely feedback on the health status of the wound, and provide data support for medical staff, thereby improving the rehabilitation process of diabetic patients.
[0008] The above technical objectives of the present invention are achieved through the following technical solutions:
[0009] In a first aspect, the present invention provides a diabetic wound image analysis and intelligent rehabilitation method, comprising the following steps:
[0010] S1. Image acquisition: Use a professional camera to take high-resolution images of diabetic wounds and upload them to the cloud for processing via a mobile application;
[0011] S2. Image preprocessing: using filtering algorithms to remove noise from wound images, and applying histogram equalization or adaptive histogram equalization to enhance the contrast of wound images;
[0012] S3. Feature extraction, using a convolutional neural network to extract features from the wound image, wherein the features include wound color features, wound shape features, and wound depth features, wherein the wound shape features include wound area features and edge features, and the wound depth features are the depth of the wound estimated by optical or ultrasonic imaging technology;
[0013] S4. Classification analysis: Use the trained deep learning model to classify the images, input the wound images into the deep learning model, extract high-level features, output the scores of each category, and judge the degree of wound healing and infection risk.
[0014] Use the Softmax function to convert the output into a probability distribution:
[0015]
[0016] Where P(y|x) is the probability of each category y for a given input image x; f(x,y) is the score of the model predicting that the input x is category y; K is the total number of categories;
[0017] Select the category with the highest probability as the final judgment result:
[0018]
[0019] in, is the predicted category; P(y|x) is the probability of each category y given the input image x;
[0020] Provide corresponding evaluation results according to the predicted category;
[0021] S5. Feedback and suggestions: Convert the evaluation results into a visual report, including the wound status and characteristics, and generate personalized rehabilitation suggestions based on the evaluation results and patient historical data.
[0022] In combination with the first aspect, the present invention is further configured as follows: in step S2, the filtering algorithm includes Gaussian filtering and median filtering.
[0023] In combination with the first aspect, the present invention is further configured as follows: in step S4, the deep learning model is ResNet and U-Net.
[0024] In combination with the first aspect, the present invention is further configured as follows: in step S4, the categories include good healing, moderate healing, high infection risk and low infection risk.
[0025] In a second aspect, the present invention further provides a diabetic wound image analysis and intelligent rehabilitation system, comprising:
[0026] The image acquisition module uses a professional camera to take high-resolution images of diabetic wounds and uploads them to the cloud for processing via a mobile app;
[0027] An image preprocessing module uses a filtering algorithm to remove noise from the wound surface image, and applies histogram equalization or adaptive histogram equalization to enhance the contrast of the wound surface image;
[0028] A feature extraction module, which uses a deep learning model to extract features from the wound image, wherein the features include wound color features, wound shape features, and wound depth features, wherein the wound shape features include wound area features and edge features, and the wound depth features are used to estimate the depth of the wound using optical or ultrasonic imaging technology;
[0029] The classification analysis module uses the trained deep learning model to classify images, inputs the wound images into the deep learning model, extracts high-level features, outputs the score of each category, and determines the degree of wound healing and infection risk;
[0030] Use the Softmax function to convert the output into a probability distribution:
[0031]
[0032] Where P(y|x) is the probability of each category y for a given input image x; f(x,y) is the score of the model predicting that the input x is category y; K is the total number of categories;
[0033] Select the category with the highest probability as the final judgment result:
[0034]
[0035] in, is the predicted category; P(y|x) is the probability of each category y given the input image x;
[0036] Provide corresponding evaluation results according to the predicted category;
[0037] The result feedback and suggestion module converts the evaluation results into a visual report, including the status and characteristics of the wound, and generates personalized rehabilitation suggestions based on the evaluation results and patient historical data.
[0038] In combination with the second aspect, the present invention is further configured as follows: in the image preprocessing module, the filtering algorithm includes Gaussian filtering and median filtering.
[0039] In combination with the second aspect, the present invention is further configured as follows: in the feature extraction module, the deep learning model is a convolutional neural network.
[0040] In combination with the second aspect, the present invention is further configured as follows: in the classification analysis module, the deep learning model is ResNet and U-Net.
[0041] In combination with the second aspect, the present invention is further configured as follows: in the classification analysis module, the categories include good healing, moderate healing, high infection risk and low infection risk.
[0042] In a third aspect, the present invention further provides a diabetic wound image analysis and intelligent rehabilitation device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the first aspect.
[0043] In summary, the present invention has the following beneficial effects: the present invention can identify the color, area, depth and edge features of the wound through feature extraction, timely feedback the health status of the wound, and provide data support for medical staff, thereby improving the rehabilitation process of diabetic patients. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 It is a flow chart of the diabetic wound image analysis and intelligent rehabilitation method based on artificial intelligence in an embodiment of the present invention;
[0045] Figure 2 It is a module diagram of the artificial intelligence-based diabetic wound image analysis and intelligent rehabilitation system in an embodiment of the present invention. DETAILED DESCRIPTION
[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. 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.
[0047] Example:
[0048] A diabetic wound image analysis and intelligent rehabilitation method, such as Figure 1 The following steps are included
[0049] S1. Image acquisition: Use a professional camera to take high-resolution images of diabetic wounds and upload them to the cloud for processing via a mobile application;
[0050] S2. Image preprocessing: using filtering algorithms (Gaussian filtering and median filtering) to remove noise in the wound surface image, and applying histogram equalization or adaptive histogram equalization to enhance the contrast of the wound surface image;
[0051] S3. Feature extraction: extracting features from wound images using a deep learning model (such as a convolutional neural network, CNN), wherein the features include wound color features, wound shape features, and wound depth features, wherein the wound shape features include wound area features and edge features, and the wound depth features are the depth of the wound estimated by optical or ultrasonic imaging technology;
[0052] S4. Classification analysis: Use a trained algorithm model (such as ResNet, U-Net, etc.) to input the wound image. After preprocessing, the image is input into the model in the form of a tensor, usually a three-dimensional array (width, length, number of channels). The model outputs categories. For diabetic wounds, the categories may include: well-healed, moderately healed, high risk of infection, and low risk of infection.
[0053] To classify an image, the input image x is passed through a deep learning model to extract high-level features, and the output of the model is the score for each category. Usually, the last layer of the model is a fully connected layer, and the dimension of the output is equal to the number of categories.
[0054] Use the Softmax function to convert the output into a probability distribution:
[0055]
[0056] Where P(y|x) is the probability of each category y for a given input image x; f(x,y) is the score of the model predicting that the input x is category y; K is the total number of categories;
[0057] Select the category with the highest probability as the final judgment result:
[0058]
[0059] in, is the predicted category; P(y|x) is the probability of each category y given the input image x;
[0060] Based on the predicted category, the corresponding evaluation results are provided; for example, if the model output is "high risk of infection", the system will provide early warning information to medical staff and recommend further examination or adjustment of treatment plans.
[0061] S5. Feedback and suggestions: Convert the evaluation results into a visual report, including the wound status and characteristics, and generate personalized rehabilitation suggestions based on the evaluation results and patient historical data.
[0062] The embodiment of the present invention also provides a diabetic wound image analysis and intelligent rehabilitation system, such as Figure 2 As shown, it includes the following modules:
[0063] The image acquisition module uses a professional camera to take high-resolution images of diabetic wounds and uploads them to the cloud for processing via a mobile app;
[0064] An image preprocessing module uses filtering algorithms (Gaussian filtering and median filtering) to remove noise from wound surface images, and applies histogram equalization or adaptive histogram equalization to enhance the contrast of wound surface images;
[0065] A feature extraction module, which uses a deep learning model (such as a convolutional neural network, CNN) to extract features from the wound image, wherein the features include wound color features, wound shape features, and wound depth features, wherein the wound shape features include wound area features and edge features, and the wound depth features are the depth of the wound estimated by optical or ultrasonic imaging technology;
[0066] The classification analysis module uses a trained algorithm model (such as ResNet, U-Net, etc.) to input the wound image. After preprocessing, the image is input into the model in the form of a tensor, usually a three-dimensional array (width, length, number of channels). The model outputs categories. For diabetic wounds, the categories may include: well-healed, moderately healed, high risk of infection, and low risk of infection.
[0067] To classify the image, input the wound image, extract the high-level features through the deep learning model, and the output of the model is the score of each category. Usually, the last layer of the model is a fully connected layer, and the output dimension is equal to the number of categories.
[0068] Use the Softmax function to convert the output into a probability distribution:
[0069]
[0070] Where P(y|x) is the probability of each category y for a given input image x; f(x,y) is the score of the model predicting that the input x is category y; K is the total number of categories;
[0071] Select the category with the highest probability as the final judgment result:
[0072]
[0073] in, is the predicted category; P(y|x) is the probability of each category y given the input image x;
[0074] Based on the predicted category, the corresponding evaluation results are provided; for example, if the model output is "high risk of infection", the system will provide early warning information to medical staff and recommend further examination or adjustment of treatment plans.
[0075] The result feedback and suggestion module converts the evaluation results into a visual report, including the status and characteristics of the wound, and generates personalized rehabilitation suggestions based on the evaluation results and patient historical data.
[0076] An embodiment of the present invention further provides a diabetic wound image analysis and intelligent rehabilitation device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.
[0077] The device provided by the present invention can display temperature, inflammatory factors, glucose, pH value, and oxygen on the display screen after being attached to the patient's wound, and upload the data to the medical staff's mobile phone APP, and automatically analyze the wound and give suggestions.
[0078] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A diabetic wound image analysis and intelligent rehabilitation method, characterized by: The steps include: S1. Image acquisition: Use a professional camera to take high-resolution images of diabetic wounds and upload them to the cloud for processing via a mobile application; S2. Image preprocessing: using filtering algorithms to remove noise from wound images, and applying histogram equalization or adaptive histogram equalization to enhance the contrast of wound images; S3. Feature extraction, using a convolutional neural network to extract features from the wound image, wherein the features include wound color features, wound shape features, and wound depth features, wherein the wound shape features include wound area features and edge features, and the wound depth features are the depth of the wound estimated by optical or ultrasonic imaging technology; S4. Classification analysis: Use the trained deep learning model to classify the images, input the wound images into the deep learning model, extract high-level features, output the scores of each category, and judge the degree of wound healing and infection risk. Use the Softmax function to convert the output into a probability distribution: Where P(y|x) is the probability of each category y for a given input image x; f(x,y) is the score of the model predicting that the input x is category y; K is the total number of categories; Select the category with the highest probability as the final judgment result: in, is the predicted category; P(y|x) is the probability of each category y given the input image x; Provide corresponding evaluation results according to the predicted category; S5. Feedback and suggestions: Convert the evaluation results into a visual report, including the wound status and characteristics, and generate personalized rehabilitation suggestions based on the evaluation results and patient historical data.
2. The method for diabetic wound image analysis and intelligent rehabilitation according to claim 1, characterized in that: In step S2, the filtering algorithm includes Gaussian filtering and median filtering.
3. The method for diabetic wound image analysis and intelligent rehabilitation according to claim 1, characterized in that: In step S4, the deep learning model is ResNet and U-Net.
4. The method for diabetic wound image analysis and intelligent rehabilitation according to claim 1, characterized in that: In step S4, the categories include good healing, moderate healing, high risk of infection, and low risk of infection.
5. A diabetic wound image analysis and intelligent rehabilitation system, characterized by: include: The image acquisition module uses a professional camera to take high-resolution images of diabetic wounds and uploads them to the cloud for processing via a mobile app; An image preprocessing module uses a filtering algorithm to remove noise from the wound surface image, and applies histogram equalization or adaptive histogram equalization to enhance the contrast of the wound surface image; A feature extraction module, which uses a convolutional neural network to extract features from the wound image, wherein the features include wound color features, wound shape features, and wound depth features, wherein the wound shape features include wound area features and edge features, and the wound depth features are used to estimate the depth of the wound using optical or ultrasonic imaging technology; The classification analysis module uses the trained deep learning model to classify images, inputs the wound images into the deep learning model, extracts high-level features, outputs the score of each category, and determines the degree of wound healing and infection risk; Use the Softmax function to convert the output into a probability distribution: Where P(y|x) is the probability of each category y for a given input image x; f(x,y) is the score of the model predicting that the input x is category y; K is the total number of categories; Select the category with the highest probability as the final judgment result: in, is the predicted category; P(y|x) is the probability of each category y given the input image x; Provide corresponding evaluation results according to the predicted category; The result feedback and suggestion module converts the evaluation results into a visual report, including the status and characteristics of the wound, and generates personalized rehabilitation suggestions based on the evaluation results and patient historical data.
6. The diabetic wound image analysis and intelligent rehabilitation system according to claim 5, characterized in that: In the image preprocessing module, the filtering algorithm includes Gaussian filtering and median filtering.
7. The diabetic wound image analysis and intelligent rehabilitation system according to claim 5 is characterized by: In the classification analysis module, the deep learning models are ResNet and U-Net.
8. The diabetic wound image analysis and intelligent rehabilitation system according to claim 5, characterized in that: In the classification analysis module, the categories include good healing, moderate healing, high risk of infection, and low risk of infection.
9. A diabetic wound image analysis and intelligent rehabilitation device, comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method of claim 1.