Chronic wound depth analysis system based on image recognition

Abstract

The invention belongs to the technical field of chronic wound depth analysis, and relates to a chronic wound depth analysis system based on image recognition. According to the method, more comprehensive and richer wound information is provided by acquiring the chronic wound image of the target patient, misdiagnosis and missed diagnosis can be reduced, the depth of the chronic wound can be judged more accurately, and by analyzing the initial depth level of the chronic wound image of the target patient, the accuracy of the chronic wound is improved. The chronic wound in-depth detection method of the target patient is determined, the in-depth detection requirement of the chronic wound of the target patient is judged, a more personalized detection scheme can be formulated, unnecessary examination and treatment are reduced, the economic burden and psychological stress of the patient are reduced, and the detection accuracy is improved by determining the chronic wound in-depth detection method of the target patient and performing detection and correction. The final depth grade of the chronic wound image of the target patient is analyzed, interference factors in the image can be recognized and corrected, the diagnosis accuracy is improved, and an important basis is provided for formulating a personalized treatment scheme.

Description

Technical Field

[0001] The invention belongs to the technical field of chronic wound depth analysis and relates to a chronic wound depth analysis system based on image recognition. Background Art

[0002] With the aging of the population, the rising incidence of chronic diseases such as diabetes, and the frequent occurrence of accidental injuries such as traffic accidents and work-related injuries, the number of patients with chronic wounds is increasing. Chronic wounds usually refer to skin tissue damage caused by various reasons, which still cannot heal within the expected time after conventional treatment, such as diabetic foot ulcers, pressure sores, and lower limb venous ulcers. Such wounds not only bring physical pain and a decline in the quality of life to patients, but also increase medical costs and social burdens. Therefore, the deep analysis system of chronic wounds based on image recognition is of great significance and role.

[0003] In the prior art, there are also some related solutions involving chronic wound analysis, for example, the invention patent application of a chronic wound recognition system, method, device and medium based on multimodal fusion with Chinese patent publication number CN118570841A, which specifically relates to the field of image processing technology, wherein the recognition module uses a laser camera, an RGB camera and an infrared camera to scan the chronic wound to be identified, and obtains 3D point cloud data, RGB image and infrared image of the chronic wound to be identified respectively; the processing module aligns the 3D point cloud data, RGB image and infrared image, and converts the 3D point cloud data into a depth image, performs scale normalization processing on the depth image, RGB image and infrared image, and obtains the multimodal data to be detected; the recognition result acquisition module inputs the multimodal data to be detected into the pre-trained multimodal fusion model, and outputs the recognition result of the chronic wound to be identified. The above system adopts a composite data acquisition scheme, which can comprehensively utilize various types of image acquisition data, improve the accuracy of chronic wound recognition, and timely and accurate chronic wound recognition.

[0004] Another Chinese patent application with publication number CN118213092B is for a remote medical monitoring system for chronic wound diseases. The system includes a patient user terminal, a medical user terminal and a cloud server. The patient user terminal and the medical user terminal are respectively connected to the cloud server in communication; the patient user terminal includes a wound image acquisition module and a basic information acquisition module. The wound image acquisition module is used to obtain the patient's preliminary chronic wound image information and pre-process the patient's preliminary chronic wound image information; the basic information acquisition module is used to obtain the patient's basic information; the cloud server includes a sample acquisition module, a wound analysis model construction module, a model training module, a wound analysis module, a storage module and an electronic medical record module; the medical user terminal includes a medical diagnosis module, a patient monitoring module, and a doctor-patient interaction module; the present invention can remotely and effectively monitor the wounds of patients with chronic wound diseases, reduce the number of times patients go back and forth to the hospital, and improve the quality of life of patients.

[0005] Although the above schemes have proposed some solutions for chronic wound analysis, they still have certain limitations: on the one hand, the existing schemes adopt a composite data acquisition scheme, which can comprehensively utilize various types of image acquisition data to improve the accuracy of chronic wound recognition and timely and accurate chronic wound recognition. However, the existing schemes ignore the analysis from multiple angles such as the geometric features, texture features and color features of chronic wounds, and thus cannot identify chronic wounds more comprehensively and accurately based on the shape and size, subtle structure and surface changes, and color information of chronic wounds, thereby failing to reduce unnecessary examinations and treatments and reduce the economic burden and psychological pressure on patients.

[0006] On the other hand, existing solutions can effectively monitor the wounds of patients with chronic wound diseases remotely, reduce the number of times patients go to and from the hospital, and improve their quality of life. However, existing solutions lack accurate judgment of the depth of chronic wounds, which is not conducive to selecting the most appropriate treatment method, and cannot avoid unnecessary surgical trauma and increase patient satisfaction and trust. Summary of the invention

[0007] In view of this, in order to solve the problems raised in the above background technology, a chronic wound depth analysis system based on image recognition is proposed.

[0008] The purpose of the present invention can be achieved through the following technical scheme: The present invention provides a chronic wound in-depth analysis system based on image recognition, including: a chronic wound preliminary detection module, a chronic wound preliminary analysis module, a chronic wound in-depth detection demand judgment module, a chronic wound in-depth detection module, a chronic wound in-depth analysis module and a cloud database.

[0009] The chronic wound preliminary detection module is used to record the patient who currently needs to detect the depth of the chronic wound as the target patient, use professional camera equipment to shoot the target patient's wound site, obtain the target patient's chronic wound images from various angles, and fuse them to obtain the target patient's chronic wound image.

[0010] The chronic wound preliminary analysis module is used to obtain the geometric feature data, texture feature data and color feature data of the chronic wound image of the target patient, analyze the geometric feature vector, texture feature vector and color feature vector of the chronic wound image of the target patient, and obtain the preliminary depth level of the chronic wound image of the target patient based on this, where the depth level includes shallow ulcer, half-thickness ulcer, full-thickness ulcer and deep ulcer.

[0011] The chronic wound in-depth detection demand judgment module is used to judge the target patient's chronic wound depth in-depth detection demand. If in-depth detection is required, the chronic wound in-depth detection module is executed.

[0012] The chronic wound depth detection module is used to determine the target patient's chronic wound depth detection method based on the target patient's preliminary depth level of the chronic wound image and basic medical record information, and perform real-time detection during the detection process to obtain the artifact existence index of the chronic wound site during the target patient's chronic wound depth detection process.

[0013] The chronic wound depth analysis module is used to analyze the final depth level of the chronic wound images of the target patient.

[0014] The cloud database is used to store the permitted area of ​​chronic wound images without affecting the normal life conditions of patients, store the hue mean, saturation mean and brightness mean of the normal part images of the target patients, store the minimum preliminary depth level corresponding to the in-depth detection requirements of the depth of chronic wounds, and store the in-depth detection methods of each chronic wound corresponding to each preliminary depth level.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The present invention obtains chronic wound images of target patients from various angles and fuses them to obtain chronic wound images of target patients. Image fusion can integrate image information from different angles into a single image, thereby providing more comprehensive and richer wound information, helping to reduce misdiagnosis and missed diagnosis, and helping to more accurately judge the depth of chronic wounds, which is conducive to selecting the most appropriate treatment method, avoiding unnecessary surgical trauma, and alleviating patients' pain.

[0016] 2. The present invention analyzes the geometric feature vectors, texture feature vectors and color feature vectors of the chronic wound image of the target patient, thereby obtaining the preliminary depth level of the chronic wound image of the target patient and judging the need for in-depth detection of the depth of the chronic wound of the target patient, which helps to have a more comprehensive understanding of the morphology, structure and color changes of the chronic wound, thereby obtaining the depth level of the chronic wound, and further helps to formulate a more personalized detection plan, which is conducive to reducing unnecessary examinations and treatments, and reducing the economic burden and psychological pressure on patients.

[0017] 3. The present invention determines the depth detection method of the chronic wound of the target patient, performs detection and correction, and analyzes the final depth level of the chronic wound image of the target patient. The artifact existence index obtained during the real-time detection process helps to identify and correct interference factors in the image, thereby improving the accuracy of diagnosis. Accurate chronic wound depth assessment provides an important basis for formulating personalized treatment plans, and precise wound assessment and personalized treatment plans can improve the treatment effect, thereby increasing patient satisfaction and trust. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 It is a schematic diagram of system module connection of the present invention.

[0020] Figure 2 It is a schematic diagram of the system implementation flow of the present invention.

[0021] Figure 3 It is a schematic diagram of the texture extraction process of the gray level co-occurrence matrix of the present invention. DETAILED DESCRIPTION

[0022] 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.

[0023] See also Figure 1As shown, the present invention provides a chronic wound in-depth analysis system based on image recognition, and the specific modules are distributed as follows: a chronic wound preliminary detection module, a chronic wound preliminary analysis module, a chronic wound in-depth detection demand judgment module, a chronic wound in-depth detection module, a chronic wound in-depth analysis module and a cloud database. Among them, the connection between the modules is as follows: the chronic wound preliminary detection module is connected to the chronic wound preliminary analysis module, the chronic wound preliminary analysis module is connected to the chronic wound in-depth detection demand judgment module, the chronic wound in-depth detection demand judgment module is connected to the chronic wound in-depth detection module, the chronic wound in-depth detection module is connected to the chronic wound in-depth analysis module, and the cloud database is respectively connected to the chronic wound preliminary analysis module, the chronic wound in-depth detection demand judgment module and the chronic wound in-depth analysis module.

[0024] The chronic wound preliminary detection module is used to record the patient who currently needs to detect the depth of the chronic wound as the target patient, use professional camera equipment to shoot the target patient's wound site, obtain the target patient's chronic wound images from various angles, and fuse them to obtain the target patient's chronic wound image.

[0025] It should be noted that the system implementation flow diagram is as follows Figure 2 shown.

[0026] As a preferred feasibility example, the chronic wound image of the target patient is specifically acquired in the following manner: segmenting the chronic wound images of the target patient at various angles to obtain sub-regions belonging to the chronic wound images of the target patient at various angles, and then selecting different fusion methods according to the different texture features and color features of the sub-regions belonging to the chronic wound images of the target patient at various angles, further obtaining the fused sub-regions belonging to the chronic wound image of the target patient, and reconstructing them to obtain the chronic wound image of the target patient.

[0027] It should be noted that the specific method of segmenting the chronic wound images at various angles of the target patient to obtain the sub-regions belonging to the chronic wound images at various angles of the target patient is: segmenting the chronic wound images at various angles of the target patient according to a plane grid form, and recording each grid in the obtained chronic wound images at various angles of the target patient as the sub-regions belonging to the chronic wound images at various angles of the target patient.

[0028] It should be further explained that the fusion method includes but is not limited to weighted average method, maximum method, minimum method and median method.

[0029] The specific content of the weighted average method is: weighted averaging is performed on each pixel of the sub-region to which the chronic wound image of each angle of the target patient belongs, and the obtained average is used as each pixel value of the fusion sub-region to which the chronic wound image of the target patient belongs.

[0030] The specific content of the maximum value method is: taking the maximum value of each pixel of the sub-region belonging to the chronic wound image of each angle of the target patient, and the obtained maximum value is used as the pixel value of the fusion sub-region belonging to the chronic wound image of the target patient.

[0031] The specific content of the minimum method is: taking the minimum value of each pixel in the sub-region belonging to the chronic wound image of each angle of the target patient, and the obtained minimum value is used as the pixel value of the fusion sub-region belonging to the chronic wound image of the target patient.

[0032] The specific content of the median method is: taking the median of each pixel in the sub-region belonging to the chronic wound image of each angle of the target patient, and the obtained median is used as the pixel value of the fusion sub-region belonging to the chronic wound image of the target patient.

[0033] The present invention obtains chronic wound images of target patients from various angles and fuses them to obtain chronic wound images of target patients. Image fusion can integrate image information from different angles into a single image, thereby providing more comprehensive and richer wound information, helping to reduce misdiagnosis and missed diagnosis, and helping to more accurately judge the depth of chronic wounds, which in turn is conducive to selecting the most appropriate treatment method, avoiding unnecessary surgical trauma, and alleviating patients' pain.

[0034] The chronic wound preliminary analysis module is used to obtain the geometric feature data, texture feature data and color feature data of the chronic wound image of the target patient, analyze the geometric feature vector, texture feature vector and color feature vector of the chronic wound image of the target patient, and obtain the preliminary depth level of the chronic wound image of the target patient based on this, where the depth level includes shallow ulcer, half-thickness ulcer, full-thickness ulcer and deep ulcer.

[0035] As a preferred feasibility example, the geometric feature data of the chronic wound image of the target patient includes area and fractal dimension.

[0036] It should be further explained that the reason for using area and fractal dimension as geometric feature data of chronic wound images of target patients is: (1) The area size can reflect the scope of the injury to a certain extent. A larger wound area often means deeper or more extensive tissue damage.

[0037] (2) Fractal dimension can measure the degree of irregularity of a pattern. The more complex the shape of the wound and the more jagged the edges, the higher the fractal dimension. It can reflect the geometric complexity of the wound and its relationship with depth changes from a more detailed perspective. For example, a wound with a high fractal dimension may have more significant depth differences in different local areas.

[0038] It should be further explained that the specific method for obtaining the area of ​​the chronic wound image of the target patient is: importing the chronic wound image of the target patient into the OpenCV library, and further using it to process to obtain the area of ​​the chronic wound image of the target patient.

[0039] It should be explained that OpenCV is an open source computer vision and image processing library, which is widely used in various fields, such as video surveillance, medical image processing, robot vision, autonomous driving, etc. OpenCV provides a wealth of functions and tools for image and video processing, feature detection, object recognition, machine learning and other tasks.

[0040] The specific method for obtaining the fractal dimension of the chronic wound surface image of the target patient is: using image analysis software to directly analyze the chronic wound surface image of the target patient to obtain the fractal dimension of the chronic wound surface image of the target patient.

[0041] In a specific example, the image analysis software is ImageJ, and the specific steps are: first open the chronic wound image, then select the FracLac plug-in in the plug-in menu and start the corresponding analysis interface. Some parameters can be set in the interface, such as the starting value, ending value and step size of the box size. After clicking the "Start Analysis" button, the software will cover the wound area in turn according to the set different box sizes, and count the corresponding number of boxes. Finally, based on the collected data, the fractal dimension is automatically calculated through the built-in mathematical model and the results are displayed.

[0042] The texture feature data of the chronic wound image of the target patient includes contrast, energy, correlation and entropy.

[0043] It should be further explained that the reasons for using contrast, energy, correlation and entropy as texture feature data of chronic wound images of target patients are: (1) Contrast reflects the degree of contrast of local brightness changes in the image. For chronic wound images, wounds with rough texture and deep depth often have higher contrast because the grayscale difference of tissues at different depths is large, resulting in obvious local brightness contrast.

[0044] (2) Energy indicates the uniformity of the image grayscale distribution and the texture coarseness and fineness. A high energy value indicates a more uniform image texture, which may correspond to a shallower wound. This is because shallower wounds mainly involve the epidermis and part of the superficial dermis, with relatively regular tissue texture and more consistent grayscale distribution.

[0045] (3) Correlation measures the linear relationship of pixel grayscale in a specific direction. Regular textures have strong correlation, while deep trauma destroys the regular structure of the tissue, making the texture irregular and the correlation weakened.

[0046] (4) Entropy can measure the randomness of image texture. For complex and irregular wound textures, the entropy value is usually higher. It can describe the relationship between texture complexity and depth from the perspective of information theory.

[0047] It needs to be further explained that the specific method of obtaining the contrast, energy, correlation and entropy of the chronic wound image of the target patient is: grayscale processing is performed on the chronic wound image of the target patient to obtain the grayscale image of the chronic wound of the target patient, and then the grayscale co-occurrence matrix corresponding to the grayscale image of the chronic wound of the target patient is obtained.

[0048] Among them, the gray level co-occurrence matrix is ​​a matrix that describes the spatial relationship between gray levels in an image and can be used to extract a variety of texture features.

[0049] It should be noted that the texture extraction process diagram of the gray level co-occurrence matrix is ​​as follows: Figure 3 shown.

[0050] According to the calculation formula , get the contrast of chronic wound images of target patients ,in are the grayscale levels of the target patient’s chronic wound grayscale image, , , is the number of gray levels of the target patient’s chronic wound grayscale image, Grayscale The frequency of co-occurrence.

[0051] According to the calculation formula Get the energy of chronic wound images of target patients .

[0052] According to the calculation formula Get the correlation of chronic wound images of target patients ,in Grayscale The average value of Grayscale The standard deviation of .

[0053] According to the calculation formula Get the entropy of the chronic wound image of the target patient ,in A small constant is set to prevent zero values ​​in logarithmic operations.

[0054] The color feature data of the chronic wound image of the target patient includes hue mean, hue variance, saturation mean, saturation variance, brightness mean and brightness variance.

[0055] It should be further explained that the specific method for obtaining the hue mean, hue variance, saturation mean, saturation variance, brightness mean and brightness variance of the chronic wound image of the target patient is as follows: the chronic wound image of the target patient is imported into the OpenCV library, and the R, G, and B three-channel values ​​corresponding to each pixel point of the chronic wound image of the target patient are obtained by processing it, which are recorded as ,in , is the number of each pixel, is the number of pixels, and the red, green, and blue three-channel values ​​corresponding to each pixel of the chronic wound image of the target patient are normalized to obtain the normalized corresponding values ​​of the red, green, and blue three-channels corresponding to each pixel of the chronic wound image of the target patient. .

[0056] It should be explained that the specific content of the normalization process is: according to the calculation formula Get the normalized corresponding values ​​of the red, green and blue channels of each pixel of the chronic wound image of the target patient .

[0057] According to the standard calculation formula, the hue of each pixel in the chronic wound image of the target patient can be obtained. , Saturation and brightness , and the mean value processing can be performed on them to obtain the hue mean, saturation mean and brightness mean of the chronic wound image of the target patient. According to the calculation formula Get the hue variance of the chronic wound image of the target patient , Similarly, the saturation variance of the chronic wound image of the target patient can be obtained and brightness variance .

[0058] It should be noted that the standard formula for obtaining the brightness of each pixel of the chronic wound image of the target patient is: .

[0059] The standard formula for obtaining the saturation of each pixel of the chronic wound image of the target patient is: .

[0060] The standard formula for obtaining the hue of each pixel of the chronic wound image of the target patient is: .

[0061] As a preferred feasibility example, the geometric feature vector of the chronic wound image of the target patient is specifically analyzed by extracting the area and fractal dimension of the chronic wound image of the target patient, which are respectively recorded as , analyzing the geometric feature vectors of chronic wound images of target patients ,in The permitted area of ​​chronic wound images extracted from the cloud database without affecting the patient's normal life. are the weight factors of the geometric eigenvectors corresponding to the set area and fractal dimension respectively.

[0062] A specific feasibility example, .

[0063] As a basic geometric feature, area can intuitively reflect the size and scope of trauma and is an important reference for clinical diagnosis and treatment. However, relying solely on area may not be able to fully describe the complexity and irregularity of trauma. Therefore, the weight factor of the geometric feature vector corresponding to the area is set to 0.33.

[0064] The fractal dimension can reflect the complexity and irregularity of the wound edge, and help reveal the microstructure and pathological characteristics of the wound. The change of fractal dimension may be more sensitive than the area or have a higher diagnostic value. Therefore, the weight factor of the geometric eigenvector corresponding to the set fractal dimension is assigned to 0.67.

[0065] As a preferred feasibility example, the texture feature vector of the chronic wound image of the target patient is specifically analyzed by extracting the contrast, energy, correlation and entropy of the chronic wound image of the target patient, which are respectively recorded as , , , , analyzing the texture feature vector of the chronic wound image of the target patient ,in The weight coefficients of the texture feature vector corresponding to the set contrast, energy, correlation and entropy respectively.

[0066] A specific feasibility example, and .

[0067] Contrast reflects the clarity of the image and the depth of the texture grooves. High contrast means a clear image and deep texture grooves, while low contrast means a blurred image and shallow texture grooves. In chronic wound images, contrast can help identify the clarity of the wound edge and subtle changes in texture. Therefore, the weight coefficient of the texture feature vector corresponding to the set contrast is assigned to 0.33.

[0068] Energy reflects the uniformity of the grayscale distribution and the coarseness of the texture. High energy indicates a stable and uniform grayscale distribution and a fine texture, while low energy indicates an uneven grayscale distribution and a coarse texture. In chronic wound images, energy can be used to evaluate the texture uniformity and coarseness of the wound area. Therefore, the weight coefficient of the texture feature vector corresponding to the set energy is assigned to 0.27.

[0069] Correlation describes the linear relationship between pixel values ​​in an image. A high correlation indicates a strong linear relationship between pixel values ​​and a more regular image texture, while a low correlation indicates a weaker linear relationship between pixel values ​​and a more complex image texture. In chronic wound images, correlation can be used to analyze the texture regularity and complexity of the wound area. Therefore, the weight coefficient of the texture feature vector corresponding to the set correlation is assigned to 0.15.

[0070] Entropy reflects the amount of information contained in an image. High entropy means that the image is rich in information and has complex and varied textures, while low entropy means that the image has less information and has simple textures. In chronic wound images, entropy can be used to evaluate the texture complexity and amount of information in the wound area. Therefore, the weight coefficient of the texture feature vector corresponding to the set entropy is assigned to 0.25.

[0071] As a preferred feasibility example, the color feature vector of the chronic wound image of the target patient is specifically analyzed by extracting the hue mean, hue variance, saturation mean, saturation variance, brightness mean and brightness variance of the chronic wound image of the target patient, which are respectively recorded as , analyzing the color feature vector of the chronic wound image of the target patient ,in are the mean hue, saturation, and brightness of the normal part images of the target patient extracted from the cloud database, respectively.

[0072] As a preferred feasibility example, the preliminary depth level of the chronic wound image of the target patient is specifically obtained by recording the geometric feature vector, texture feature vector and color feature vector of the chronic wound image of the target patient as input vectors, inputting them into the chronic wound learning model, and directly obtaining the preliminary depth level of the chronic wound image of the target patient based on them.

[0073] It should be further explained that the specific method of obtaining the chronic wound learning model is as follows: (1) Dataset acquisition preparation: Collect a large number of chronic wound images with accurate chronic wound depth level annotations, and record them as chronic wound image datasets. According to the above steps, extract and construct the geometric feature vectors, texture feature vectors and color feature vectors of each chronic wound image in the chronic wound image dataset, and use them as input features. The depth level of each chronic wound image in the chronic wound image dataset is used as the output label to form a training dataset.

[0074] (2) Model selection and training: Select a suitable machine learning model or deep learning model, use the training data set to train the selected model, adjust the model parameters, and enable the model to learn the mapping relationship between the geometric feature vector, texture feature vector, and color feature vector of the chronic wound image and the depth level of the chronic wound. The trained model is then recorded as the chronic wound learning model.

[0075] It should be explained that the machine learning model includes but is not limited to support vector machines, decision trees, and multi-layer perceptrons, and the deep learning model includes but is not limited to convolutional neural networks, etc.

[0076] The chronic wound in-depth detection demand judgment module is used to judge the target patient's chronic wound depth in-depth detection demand. If in-depth detection is required, the chronic wound in-depth detection module is executed.

[0077] As a preferred feasibility example, the in-depth detection requirement for the depth of the chronic wound of the target patient is specifically determined in the following manner: comparing the preliminary depth level of the chronic wound image of the target patient with the minimum preliminary depth level corresponding to the in-depth detection requirement for the depth of the chronic wound stored in the cloud database; if the preliminary depth level of the chronic wound image of the target patient is higher than or equal to the minimum preliminary depth level corresponding to the in-depth detection requirement for the depth of the chronic wound, the in-depth detection requirement for the depth of the chronic wound of the target patient is recorded as a requirement requiring in-depth detection; otherwise, the in-depth detection requirement for the depth of the chronic wound of the target patient is recorded as a requirement not requiring in-depth detection.

[0078] The present invention analyzes the geometric feature vectors, texture feature vectors and color feature vectors of the chronic wound image of the target patient, thereby obtaining the preliminary depth level of the chronic wound image of the target patient and judging the in-depth detection requirement of the chronic wound depth of the target patient, which helps to have a more comprehensive understanding of the morphology, structure and color changes of the chronic wound, thereby obtaining the depth level of the chronic wound, and further helps to formulate a more personalized detection plan, which is beneficial to reduce unnecessary examinations and treatments, and reduce the economic burden and psychological pressure on patients.

[0079] The chronic wound depth detection module is used to determine the target patient's chronic wound depth detection method based on the target patient's preliminary depth level of the chronic wound image and basic medical record information, and perform real-time detection during the detection process to obtain the artifact existence index of the chronic wound site during the target patient's chronic wound depth detection process.

[0080] As a preferred feasibility example, the basic medical record information of the target patient includes an implant presence index at the chronic wound site.

[0081] It needs to be further explained that the specific method of obtaining the implant existence index in the chronic wound area of ​​the target patient is: the medical staff directly obtains from the target patient's medical record whether there is an implant in the chronic wound area of ​​the target patient during the in-depth detection of the chronic wound; if there is an implant in the chronic wound area of ​​the target patient, the implant existence index in the chronic wound area of ​​the target patient during the in-depth detection of the chronic wound is recorded as 1; otherwise, the implant existence index in the chronic wound area of ​​the target patient during the in-depth detection of the chronic wound is recorded as 0.

[0082] It needs to be further explained that the specific method of obtaining the artifact existence index of the chronic wound area of ​​the target patient in the process of in-depth detection of chronic wounds is: the medical staff directly observes from the chronic wound in-depth detection result image of the target patient. If there are artifacts in the chronic wound area of ​​the target patient in the process of in-depth detection of chronic wounds, the artifact existence index of the chronic wound area of ​​the target patient in the process of in-depth detection of chronic wounds is recorded as 1; otherwise, the artifact existence index of the chronic wound area of ​​the target patient in the process of in-depth detection of chronic wounds is recorded as 0.

[0083] It should be explained that artifacts are common phenomena in medical imaging, which may be caused by a variety of factors, such as radiation hardening effect, volume effect, interference from metal objects, patient movement during examination, improper scanning parameter settings, and equipment failure. The artifact presence index can quantify the degree of artifacts in the image, thereby reflecting the quality of the image. In the in-depth detection of chronic wounds, high-quality images are essential for accurate diagnosis. Artifacts may affect the interpretation of images and the accuracy of diagnosis. By evaluating the artifact presence index, doctors can determine whether the image is interfered with by artifacts, thereby avoiding misdiagnosis or missed diagnosis.

[0084] As a preferred feasibility example, the method for determining the depth detection method of chronic wounds of the target patient includes: matching the preliminary depth level of the chronic wound image of the target patient with the chronic wound depth detection methods corresponding to the preliminary depth levels stored in the cloud database to obtain the depth detection methods of the chronic wounds of the target patient.

[0085] It should be further explained that the in-depth detection methods of the chronic wounds include but are not limited to ultrasound examination, magnetic resonance imaging, CT scanning and OCT scanning.

[0086] The implant presence index of the chronic wound site of the target patient is extracted. If it is 1, the chronic wound in-depth detection methods in which implants may exist are screened from the chronic wound in-depth detection methods of the target patient, and they are recorded as the preferred chronic wound in-depth detection methods of the target patient. Further, a chronic wound in-depth detection method is randomly selected from them as the chronic wound in-depth detection method of the target patient.

[0087] If it is 0, a chronic wound surface depth detection method is randomly selected from various chronic wound surface depth detection methods for the target patient as the chronic wound surface depth detection method for the target patient.

[0088] The chronic wound depth analysis module is used to analyze the final depth level of the chronic wound images of the target patient.

[0089] As a preferred feasibility example, the final depth level of the chronic wound image of the target patient is specifically analyzed in the following manner: extracting the artifact existence index of the chronic wound part of the target patient in the process of in-depth detection of the chronic wound; if it is 1, it means that there are artifacts in the in-depth detection image of the chronic wound part of the target patient, and the in-depth detection correction method of the chronic wound of the target patient is recorded as image artifact correction; if it is 0, it means that there are no artifacts and implants in the chronic wound part of the target patient, and the in-depth detection correction method of the chronic wound of the target patient is not required and is recorded as no correction is required.

[0090] Based on the determined target patient's chronic wound depth detection method, the target patient's chronic wound site is detected to obtain detection result data of the target patient's chronic wound site, and further based on the target patient's chronic wound depth detection correction method, the detection result data of the target patient's chronic wound site is corrected to obtain the final depth level of the target patient's chronic wound image.

[0091] The cloud database is used to store the permitted area of ​​chronic wound images without affecting the normal life conditions of patients, store the hue mean, saturation mean and brightness mean of normal part images of target patients, store the minimum preliminary depth level corresponding to the in-depth detection requirements of the depth of chronic wounds, and store the in-depth detection methods of various chronic wounds corresponding to various preliminary depth levels.

[0092] The present invention determines the chronic wound depth detection method of the target patient, performs detection and correction, and analyzes the final depth level of the chronic wound image of the target patient. The artifact existence index obtained during the real-time detection process helps to identify and correct interference factors in the image, thereby improving the accuracy of diagnosis. Accurate chronic wound depth assessment provides an important basis for formulating personalized treatment plans, and precise wound assessment and personalized treatment plans can improve the treatment effect, thereby increasing patient satisfaction and trust.

[0093] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they shall all fall within the protection scope of the present invention.

Claims

1. A chronic wound depth analysis system based on image recognition, characterized by: include: The chronic wound preliminary detection module is used to record the patient whose chronic wound depth needs to be detected as the target patient, use professional camera equipment to shoot the target patient's wound site, obtain the chronic wound images of the target patient from various angles, and perform fusion processing on the images to obtain the chronic wound images of the target patient; A chronic wound preliminary analysis module is used to obtain geometric feature data, texture feature data and color feature data of the chronic wound image of the target patient, analyze the geometric feature vector, texture feature vector and color feature vector of the chronic wound image of the target patient, and obtain the preliminary depth level of the chronic wound image of the target patient based on the data, wherein the depth level includes shallow ulcer, half-thickness ulcer, full-thickness ulcer and deep ulcer; A chronic wound in-depth detection demand judgment module is used to judge the target patient's chronic wound depth in-depth detection demand. If it is a need for in-depth detection, the chronic wound in-depth detection module is executed; A chronic wound depth detection module is used to determine the target patient's chronic wound depth detection method based on the target patient's preliminary depth level of the chronic wound image and basic medical record information, and to perform real-time detection during the detection process to obtain an artifact presence index of the chronic wound site during the target patient's chronic wound depth detection process; A chronic wound in-depth analysis module, used to analyze the final depth level of chronic wound images of target patients; The cloud database is used to store the permitted area of ​​chronic wound images without affecting the normal life conditions of patients, store the hue mean, saturation mean and brightness mean of the normal part images of the target patients, store the minimum preliminary depth level corresponding to the in-depth detection requirements of the depth of chronic wounds, and store the in-depth detection methods of each chronic wound corresponding to each preliminary depth level.

2. The chronic wound depth analysis system based on image recognition according to claim 1, characterized in that: The specific method of acquiring the chronic wound surface image of the target patient includes: The chronic wound images of the target patient at each angle are segmented to obtain the sub-regions to which the chronic wound images of the target patient at each angle belong. Then, different fusion methods are selected according to the different texture features and color features of the sub-regions to which the chronic wound images of the target patient at each angle belong. The fused sub-regions to which the chronic wound images of the target patient belong are further obtained, and the fused sub-regions are reconstructed to obtain the chronic wound images of the target patient.

3. The chronic wound depth analysis system based on image recognition according to claim 1, characterized in that: The geometric feature data of the chronic wound surface image of the target patient includes area and fractal dimension; The texture feature data of the chronic wound image of the target patient includes contrast, energy, correlation and entropy; The color feature data of the chronic wound image of the target patient includes hue mean, hue variance, saturation mean, saturation variance, brightness mean and brightness variance; The basic medical record information of the target patient includes the implant presence index at the chronic wound site.

4. The chronic wound depth analysis system based on image recognition according to claim 3 is characterized by: The specific analysis method of the geometric feature vector of the chronic wound image of the target patient includes: The area and fractal dimension of the chronic wound image of the target patient are extracted and recorded as , , analyzing the geometric feature vectors of chronic wound images of target patients ,in The permitted area of ​​chronic wound images extracted from the cloud database without affecting the patient's normal life. , are the weight factors of the geometric eigenvectors corresponding to the set area and fractal dimension respectively.

5. The chronic wound depth analysis system based on image recognition according to claim 4, characterized in that: The texture feature vector of the chronic wound image of the target patient is specifically analyzed in the following manner: The contrast, energy, correlation and entropy of the chronic wound image of the target patient are extracted and recorded as , , , , analyzing the texture feature vector of the chronic wound image of the target patient ,in The weight coefficients of the texture feature vector corresponding to the set contrast, energy, correlation and entropy respectively.

6. The chronic wound depth analysis system based on image recognition according to claim 5, characterized in that: The specific analysis method of the color feature vector of the chronic wound image of the target patient includes: The hue mean, hue variance, saturation mean, saturation variance, brightness mean and brightness variance of the chronic wound image of the target patient are extracted and recorded as , analyzing the color feature vector of the chronic wound image of the target patient ,in are the mean hue, saturation, and brightness of the normal part images of the target patient extracted from the cloud database, respectively.

7. The chronic wound depth analysis system based on image recognition according to claim 6, characterized in that: The preliminary depth level of the chronic wound image of the target patient is obtained by: The geometric feature vector, texture feature vector and color feature vector of the chronic wound image of the target patient are recorded as input vectors, which are input into the chronic wound learning model, based on which the preliminary depth level of the chronic wound image of the target patient can be directly obtained.

8. The chronic wound depth analysis system based on image recognition according to claim 1, characterized in that: The in-depth detection requirements for the depth of chronic wounds of the target patients include: The preliminary depth level of the chronic wound image of the target patient is compared with the minimum preliminary depth level corresponding to the in-depth detection requirement of the chronic wound depth stored in the cloud database; if the preliminary depth level of the chronic wound image of the target patient is higher than or equal to the minimum preliminary depth level corresponding to the in-depth detection requirement of the chronic wound depth, the in-depth detection requirement of the chronic wound depth of the target patient is recorded as requiring in-depth detection requirement; otherwise, the in-depth detection requirement of the chronic wound depth of the target patient is recorded as not requiring in-depth detection requirement.

9. The chronic wound depth analysis system based on image recognition according to claim 3, characterized in that: The method for determining the in-depth detection of chronic wounds of target patients specifically includes: Matching the preliminary depth level of the chronic wound image of the target patient with each chronic wound depth detection method corresponding to each preliminary depth level stored in the cloud database to obtain each chronic wound depth detection method of the target patient; Extracting the implant presence index of the chronic wound site of the target patient, if it is 1, screening various chronic wound in-depth detection methods in which implants may exist from various chronic wound in-depth detection methods of the target patient, recording them as various preferred chronic wound in-depth detection methods of the target patient, and further randomly selecting a chronic wound in-depth detection method from them as the chronic wound in-depth detection method of the target patient; If it is 0, a chronic wound surface depth detection method is randomly selected from various chronic wound surface depth detection methods for the target patient as the chronic wound surface depth detection method for the target patient.

10. The chronic wound depth analysis system based on image recognition according to claim 9, characterized in that: The final depth level of the chronic wound image of the target patient is specifically analyzed in the following manner: Extract the artifact existence index of the chronic wound part of the target patient during the in-depth detection of the chronic wound. If it is 1, it means that there are artifacts in the in-depth detection image of the chronic wound part of the target patient, and the in-depth detection correction method of the chronic wound part of the target patient is recorded as image artifact correction. If it is 0, it means that there are no artifacts and implants in the chronic wound part of the target patient, and the in-depth detection correction method of the chronic wound part of the target patient is not required, and it is recorded as no correction is required; Based on the determined target patient's chronic wound depth detection method, the target patient's chronic wound site is detected to obtain detection result data of the target patient's chronic wound site, and further based on the target patient's chronic wound depth detection correction method, the detection result data of the target patient's chronic wound site is corrected to obtain the final depth level of the target patient's chronic wound image.

Images