Radiotherapy patient skin injury real-time detection record management system based on image processing

Abstract

The invention provides a radiotherapy patient skin injury real-time detection record management system based on image processing, and belongs to the technical field of radiotherapy patient skin injury real-time detection. Comprising a multispectral image acquisition module for imaging the skin of a patient at the same visual angle by adopting different spectrums to obtain three spectral images including near-infrared imaging, infrared imaging and ultraviolet imaging; a first identification module is used for fusing three spectral images to obtain an image data set, and a second identification module is used for constructing an epidermis injury coefficient # imgabs0 #, a dermis injury coefficient # imgabs1 # and a subcutaneous injury coefficient # imgabs2 # of the ith patient and evaluating the epidermis injury coefficient # imgabs0 #, the dermis injury coefficient # imgabs1 # and the subcutaneous injury coefficient # imgabs2 # respectively. A skin injury evaluation model is constructed, the comprehensive injury area ratio is calculated, the classification of injury levels is realized, data support is provided for a personalized treatment scheme, and a powerful auxiliary diagnosis tool is provided for a medical institution.

Description

Technical Field

[0001] The present invention relates to the technical field of real-time detection of skin lesions in radiotherapy patients, and in particular to a real-time detection record management system for skin lesions in radiotherapy patients based on image processing. Background Art

[0002] With the widespread use of radiotherapy (RT) in cancer treatment, RT patients often experience skin damage during the treatment process. These injuries usually manifest as symptoms such as erythema, peeling, and ulcers, which seriously affect the patient's quality of life. The assessment and management of skin damage is a crucial part of the RT treatment process. Timely and accurate detection and treatment of skin damage not only helps to alleviate the patient's pain, but also effectively prevent the occurrence of complications. RT kills cancer cells through high-energy radiation, but due to the penetration and range of its radiation, normal skin tissue is also inevitably damaged. The degree of RT-induced skin damage is usually related to factors such as the radiation dose, treatment site, and patient's physical condition. According to different treatment doses and treatment methods, skin damage caused by RT can usually be divided into different levels, specifically manifested as erythema, dryness, peeling, blisters, ulcers, etc.

[0003] The skin is divided into three major layers: the epidermis, dermis, and subcutaneous layer. When the epidermis is damaged, it will gradually affect the dermis and subcutaneous layer as time passes after radiotherapy. Existing assessment technologies mostly focus on damage analysis at a single layer, while ignoring the comprehensive damage analysis of different skin layers (epidermis, dermis, subcutaneous layer). In addition, most current technologies cannot monitor skin damage changes in radiotherapy patients in real time, cannot provide real-time damage feedback during radiotherapy, and cannot record patient skin detection in real time. Summary of the Invention

[0004] In order to overcome the above deficiencies, the present invention provides a real-time detection and record management system for skin lesions in radiotherapy patients based on image processing, which overcomes the above technical problems or at least partially solves the above problems.

[0005] The present invention is achieved in that: The present invention provides a real-time detection and record management system for skin damage in radiotherapy patients based on image processing, comprising: The multispectral image acquisition module uses different spectra to image the patient's skin at the same viewing angle, obtaining three spectral images: near-infrared imaging, infrared imaging, and ultraviolet imaging. The three spectral images are then divided into different location areas to obtain several target areas. The first recognition module is used to fuse the three spectral images to obtain an image data set, use neural network technology to build a skin damage assessment model, extract relevant damage area features from the image data set, and construct the comprehensive damage area ratio of the i-th patient. , and evaluate, and output the classified damage level; the classified damage level includes: first-level normal damage level, second-level abnormal damage level and third-level abnormal damage level; The second recognition module is used to identify the i-th patient with secondary abnormal injury level and tertiary abnormal injury level, summarize them into a treatment demand group, extract the skin injury features in the image data set of the treatment demand group through the skin injury assessment model, conduct in-depth analysis to obtain the injury depth, dermal inflammatory response factors, subcutaneous inflammatory factors and subcutaneous oxygenation, and construct the epidermal injury coefficient of the i-th patient. , dermal damage coefficient and subcutaneous layer damage coefficient , and evaluate them separately, outputting the corresponding damage results.

[0006] In a preferred solution, the multispectral image acquisition module includes a segmentation unit and an edge detection unit; The segmentation unit is used to divide the image data set into different position areas on the same patient target areas, m represents the total number of target areas; target areas include arms, abdomen, back and legs; The edge detection unit is used to detect and identify the number N of skin edge contour point pixels of each target area of the same patient using the Sobel operator, and obtain the injury area according to the number N of skin edge contour point pixels.

[0007] In a preferred embodiment, the first recognition module is used to fuse the three spectral images to obtain an image data set, and use a convolutional neural network to construct a skin damage assessment model as a data operation prediction; During the current monitoring period, three spectral images are acquired and the abnormal skin monitoring algorithm is used to calculate the comprehensive damage area ratio of the i-th patient. ; Abnormal skin monitoring algorithm includes the following steps: First, use image denoising algorithm and image enhancement algorithm to preprocess each frame of the extracted image; Second, use a deep learning model to process each pre-processed image frame and extract a high-dimensional feature vector containing the damage area characteristics; 3. Using convolutional neural networks, construct an initial convolutional neural network model, and test and train the initial convolutional neural network model with image data sets. The trained initial convolutional neural network model is used as a skin damage assessment model. At the same time, the intermediate layer output of the skin damage assessment model is used as a feature vector to identify relevant damage area features. The skin damage assessment model is trained and tested based on the obtained relevant damage area features. The trained skin damage assessment model is used as data for running predictions to calculate the comprehensive damage area ratio of the i-th patient. ; In a preferred solution, the first identification module further includes a damage area feature summary unit and an analysis unit; The lesion area feature summary unit is used to extract features from the image dataset of the i-th patient using threshold segmentation and deep learning segmentation technology, compare the color, texture, and spectral reflectance characteristics of normal skin and damaged skin, extract the boundary of the damaged area, perform calibration, and sum the pixel points of the damaged area using pixel counting to obtain the lesion area of the i-th patient. .

[0008] In a preferred embodiment, the analysis unit includes a skin lesion area subunit and a comparison subunit; The skin injury area subunit is used to calculate the comprehensive injury area ratio of the i-th patient according to the injury area of the i-th patient after dimensionless processing using the following formula: ; Where m represents the total number of target areas, represents the injury area of the i-th patient, represents the total body area of the i-th patient.

[0009] In a preferred embodiment, the comparison subunit is used to preset a first standard threshold value V1 and a second standard threshold value V2, and V1>V2, and compare the comprehensive injury area ratio of the i-th patient Compare and evaluate with the first standard threshold V1 and the second standard threshold V2 to obtain and output the corresponding classification damage level; When the comprehensive injury area of the i-th patient is Greater than or equal to the first standard threshold V1, that is When ≥V1, it means that the patient's skin damage is classified as Grade 3 abnormal damage, and there is a risk of immune infection, which requires further suppression; When the second standard threshold V2 is less than or equal to the comprehensive injury area ratio of the i-th patient , and is less than the first standard threshold V1, that is, When , it means that the skin injury area of the current patient i belongs to the second-level abnormal injury level; at this time, there is an immune infection risk, but it is lower than the risk of the third-level abnormal injury level; When the comprehensive injury area of the i-th patient is Less than the second standard threshold V2, that is When the value is less than V2, it indicates that the skin injury area of the current patient i is within the expected range and is in a recoverable state. It is marked as a normal injury level 1 and prompts debridement and disinfection, basic nursing care, and avoidance of overtreatment. The graded treatment plan improves the accuracy and efficiency of treatment, reduces unnecessary drug use, and reduces the burden on patients.

[0010] In a preferred embodiment, when the i-th patient is identified as having a second-level abnormal injury grade and a third-level abnormal injury grade, they are grouped into a healing need group. The image dataset in the healing need group is extracted using the skin injury assessment model to construct a second recognition module. The second recognition module includes an epidermis detection and analysis unit, a dermis detection and analysis unit, and a subcutaneous layer detection and analysis unit. The epidermal detection and analysis unit is used to associate the epidermal layer damage level of the patient's skin with epidermal self-healing, and the epidermal detection and analysis unit includes a skin self-healing subunit and an extraction unit. The skin self-healing subunit is used to identify the damaged area and the time t after the patient's radiotherapy through the image data set in the healing demand group. After dimensionless processing, the skin tissue self-healing coefficient of the i-th patient is calculated by the following formula : Where d represents the severity of skin damage, ranging from 0 to 1, and k represents the recovery rate of skin tissue. The range of skin recovery rate k has been recorded through existing medical research and clinical experiments. is represented by an exponential decay function, t represents the time after radiotherapy, It is expressed as the initial value of the recovery rate, which is obtained based on clinical data, medical research or experiments. It is expressed as a decay constant, with a value of 0-1; The extraction unit is used to extract the damaged area and skin tissue self-healing coefficient of the i-th patient After dimensionless processing, the epidermal damage coefficient of the i-th patient is obtained by the following formula : Where, S represents the influencing factor of the skin type of the i-th patient, which is obtained by looking up the table, and C represents the influencing factor of the radiotherapy dose of the i-th patient, which is selected based on clinical experience; When the epidermal damage coefficient of the i-th patient Compared with the preset epidermal damage threshold B, when the epidermal damage coefficient of the i-th patient When the value is less than the preset epidermal damage threshold B, it means that the epidermal damage level of the current i-th patient is the fourth abnormal damage level; When the epidermal damage coefficient of the i-th patient When ≥ the preset epidermal damage threshold B, it means that the epidermal damage level of the current i-th patient is the fifth level of abnormal damage.

[0011] In a preferred embodiment, the dermis detection and analysis unit is used to correlate the damage depth with the inflammatory response factor of the dermis layer, and obtain the time after radiotherapy by a timer. After dimensionless processing, the dermis damage coefficient of the i-th patient is obtained by the following formula: : Where, Indicates the depth of damage, represents the inflammatory response factor of the dermis, t represents the time after radiotherapy, represents the time decay factor, which is derived from clinical data. The coefficient indicating the influence of injury depth on recovery is obtained through clinical trial data; Dermis damage coefficient of patient i Compare and evaluate with the threshold value W for assessing dermal damage; When the dermis damage coefficient of the i-th patient When the threshold value W for evaluating dermal damage is less than , it means that the dermal damage level of the current i-th patient is level 6 abnormal damage level; When the dermis damage coefficient of the i-th patient When the dermis damage threshold W is greater than or equal to the evaluation threshold, it means that the abnormal level of dermis damage in the current i-th patient is the seventh abnormal damage level.

[0012] In a preferred embodiment, the subcutaneous layer detection and analysis unit is used to obtain the subcutaneous layer damage coefficient of the i-th patient through the following formula after dimensionless processing based on the correlation between the subcutaneous layer inflammatory factors and the subcutaneous layer oxygenation degree: : Where, It is expressed as the normalized subcutaneous inflammatory factor, with a value of 0-1. It is expressed as normalized subcutaneous oxygenation, with a value of 0-1. Expressed as the subcutaneous inflammation correction constant, Expressed as the correction constant for subcutaneous oxygenation.

[0013] In a preferred embodiment, the subcutaneous layer injury coefficient of the i-th patient is Compare and evaluate with the preset subcutaneous layer damage threshold G; When the subcutaneous layer damage coefficient of the i-th patient When the threshold value G of subcutaneous layer damage is less than the preset threshold value G, it means that the abnormal level of subcutaneous layer damage in the current i-th patient is level eight abnormal damage level; When the subcutaneous layer damage coefficient of the i-th patient When the value is greater than or equal to the preset subcutaneous layer injury threshold G, it indicates that the abnormal level of subcutaneous layer damage in the current i-th patient is the ninth abnormal level of damage.

[0014] The present invention provides a real-time detection and record management system for skin damage in radiotherapy patients based on image processing, which has the following beneficial effects: 1. By adopting three spectral imaging technologies, namely near-infrared, infrared and ultraviolet, it can detect damage information at different layers of the skin, effectively identify early minor damage, avoid the limitations of traditional visible light imaging, and improve the comprehensiveness and accuracy of detection. In addition, a skin damage assessment model is constructed through neural network technology, which can automatically extract damage features from image data sets and calculate the comprehensive damage area ratio to achieve damage level classification (first, second and third degree damage).

[0015] 2. By employing a multi-level injury analysis approach, the system further refines injury assessment for patients with grade II and III injuries, including key parameters such as injury depth, dermal inflammatory response, subcutaneous inflammatory factors, and oxygenation. By calculating the epidermal, dermal, and subcutaneous damage coefficients, it achieves precise grading and provides data support for personalized treatment plans, providing medical institutions with a powerful auxiliary diagnostic tool with broad clinical application value and promotion prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 This is a system block diagram provided by the present invention. DETAILED DESCRIPTION

[0018] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 making creative efforts shall fall within the scope of protection of the present invention.

[0019] Example 1, with reference to Figure 1 The present invention provides a technical solution: a real-time detection and record management system for skin damage in radiotherapy patients based on image processing, comprising: The multispectral image acquisition module uses different spectra to image the patient's skin at the same viewing angle, obtaining three spectral images: near-infrared imaging, infrared imaging, and ultraviolet imaging. The three spectral images are then divided into different location areas to obtain several target areas. The first recognition module is used to fuse the three spectral images to obtain an image data set, use neural network technology to build a skin damage assessment model, extract relevant damage area features from the image data set, and construct the comprehensive damage area ratio of the i-th patient. , and evaluate, and output the classified damage level; the classified damage level includes: first-level normal damage level, second-level abnormal damage level and third-level abnormal damage level; The second recognition module is used to identify the i-th patient with secondary abnormal injury level and tertiary abnormal injury level, summarize them into a treatment demand group, extract the skin injury features in the image data set of the treatment demand group through the skin injury assessment model, conduct in-depth analysis to obtain the injury depth, dermal inflammatory response factors, subcutaneous inflammatory factors and subcutaneous oxygenation, and construct the epidermal injury coefficient of the i-th patient. , dermal damage coefficient and subcutaneous layer damage coefficient , and evaluate them separately, outputting the corresponding damage results.

[0020] In this embodiment, three spectral imaging technologies, namely near-infrared, infrared and ultraviolet, are used to detect damage information at different layers of the skin, effectively identify early-stage minor damage, avoid the limitations of traditional visible light imaging, and improve the comprehensiveness and accuracy of detection. A skin damage assessment model is constructed through neural network technology, which can automatically extract damage features from image data sets and calculate the comprehensive damage area ratio to achieve damage level classification (first, second, and third degree damage).

[0021] This model continuously learns and optimizes, improving assessment accuracy while reducing manual intervention and enabling intelligent injury detection. Using a multi-level injury analysis approach, it further refines injury assessment for patients with secondary and tertiary injuries, including key parameters such as injury depth, dermal inflammatory response, subcutaneous inflammatory factors, and oxygenation. By calculating the epidermal, dermal, and subcutaneous injury coefficients, it achieves precise grading and provides data support for personalized treatment plans, offering medical institutions a powerful auxiliary diagnostic tool with broad clinical application value and widespread application prospects.

[0022] Example 2: This example is an explanation of Example 1. Please refer to Figure 1 ,Specifically, the multispectral image acquisition module includes a segmentation unit and an edge detection unit; The segmentation unit is used to divide the image data set into different position areas on the same patient target areas, m represents the total number of target areas; target areas include arms, abdomen, back and legs; The edge detection unit is used to detect and identify the number N of skin edge contour point pixels of each target area of the same patient using the Sobel operator, and obtain the injury area according to the number N of skin edge contour point pixels.

[0023] In this embodiment, the skin of the same patient is divided into multiple target areas such as arms, abdomen, back, and legs through a segmentation unit, making skin damage assessment more targeted and avoiding errors caused by overall assessment. This method can adapt to individual differences, realize personalized damage detection, and improve the clinical applicability of the system. By using near-infrared imaging, infrared imaging, and ultraviolet imaging, deep information of skin damage can be obtained across spectral levels, and hidden damage that is difficult to detect with traditional visible light can be effectively identified. By combining the imaging characteristics of different spectra, the comprehensiveness and accuracy of skin damage detection are improved, providing accurate data support for skin management after radiotherapy. By using an edge detection unit, the Sobel operator is used to accurately detect the edge contour of the skin damage area, and the number of pixels N of the skin edge contour points is calculated, thereby extracting the damage area with high precision. By combining multispectral image segmentation and edge detection, the damage area can be calculated quickly and accurately, reducing manual measurement errors. The system can collect multispectral skin damage data in real time, provide doctors with objective and quantitative damage area assessment results, and reduce subjective judgment errors.

[0024] Example 3, this example is the explanation in Example 1, please refer to Figure 1 ,Specifically, the first recognition module is used to fuse the three spectral ,images to obtain an image dataset, and use a convolutional neural network to ,build a skin damage assessment model as data operation prediction; During the current monitoring period, three spectral images are acquired and the abnormal skin monitoring algorithm is used to calculate the comprehensive damage area ratio of the i-th patient. ; Abnormal skin monitoring algorithm includes the following steps: First, use image denoising algorithm and image enhancement algorithm to preprocess each frame of the extracted image; Second, use a deep learning model to process each pre-processed image frame and extract a high-dimensional feature vector containing the damage area characteristics; 3. Using convolutional neural networks, construct an initial convolutional neural network model, and test and train the initial convolutional neural network model with image data sets. The trained initial convolutional neural network model is used as a skin damage assessment model. At the same time, the intermediate layer output of the skin damage assessment model is used as a feature vector to identify relevant damage area features. The skin damage assessment model is trained and tested based on the obtained relevant damage area features. The trained skin damage assessment model is used as data for running predictions to calculate the comprehensive damage area ratio of the i-th patient. .

[0025] In this embodiment, an image denoising algorithm is used to eliminate noise during the imaging process, improve image clarity, and ensure that the edge features of the damaged area are clearer. An image enhancement algorithm is used to enhance the texture, contrast, and color features of the damaged area, making the damaged area more distinct and improving the accuracy of abnormal skin identification. Near-infrared imaging, infrared imaging, and ultraviolet imaging are used to fuse data to obtain more comprehensive skin damage information. Near-infrared imaging: can detect blood flow and deep skin damage; Near-infrared imaging: can detect blood flow and deep skin damage; Ultraviolet imaging: can detect the degree of skin barrier damage.

[0026] By fusing three types of spectral data, the accuracy and comprehensiveness of skin damage detection are improved.

[0027] In addition, a deep learning model is used to automatically extract high-dimensional damage feature vectors from multispectral images, reducing manual intervention and improving computational efficiency. A CNN initial model is constructed and trained using image datasets to optimize the accuracy of skin damage detection. By outputting features through the intermediate layers of CNN, damage area features are accurately extracted, thereby improving the recognition accuracy of abnormal skin areas.

[0028] Example 4: This example is an explanation of Example 1. Figure 1 ,Specifically, the first recognition module also includes a damage area feature ,summarization unit and an analysis unit; The lesion area feature summary unit is used to extract features from the image dataset of the i-th patient using threshold segmentation and deep learning segmentation technology, compare the color, texture, and spectral reflectance characteristics of normal skin and damaged skin, extract the boundary of the damaged area, perform calibration, and sum the pixel points of the damaged area using pixel counting to obtain the lesion area of the i-th patient. .

[0029] In this embodiment, threshold segmentation technology is used to distinguish normal skin from damaged areas by setting parameters such as color, brightness, and spectral reflectance to achieve basic segmentation. In combination with deep learning segmentation technology (such as U-Net, Mask R-CNN, etc.), a large number of labeled data sets are used to train the model, enabling it to intelligently identify skin damaged areas of varying degrees, thereby improving segmentation accuracy and robustness. By comparing the color, texture, and spectral reflectance of normal skin and damaged skin, the extraction boundary of the damaged area is further optimized to reduce misjudgment. The boundary of the damaged area is accurately calibrated using edge detection algorithms (such as Sobel and Canny) to ensure the complete extraction of the morphological information of the damaged area. Morphological processing (dilation, erosion, and closing operations) is combined to optimize the connectivity of the damaged area, reduce noise interference, and ensure the stability of the detection results. The number of pixels in the damaged area is counted by summing the pixels, and the actual area is converted based on the image resolution to accurately calculate the damaged area of the i-th patient.

[0030] Example 5, this example is the explanation in Example 1, please refer to Figure 1 ,Specifically, the analysis unit includes a skin injury area subunit and a comparison subunit; The skin injury area subunit is used to calculate the comprehensive injury area ratio of the i-th patient according to the injury area of the i-th patient after dimensionless processing using the following formula: : Where m represents the total number of target areas, represents the injury area of the i-th patient, represents the total body area of the i-th patient.

[0031] The following is the comprehensive injury area ratio of the i-th patient Example table: In this embodiment, a mathematical formula is used to calculate the comprehensive damage area ratio to ensure that the damage to different target areas is accurate, and a more accurate damage assessment result is obtained. By calculating the damage to each target area by part + overall assessment, the degree of damage to each target area can be analyzed, and a more targeted treatment plan can be provided. By calculating the damage to each target area by part + overall assessment, the degree of damage to each target area can be analyzed, and a more targeted treatment plan can be provided.

[0032] Through dimensionless calculation, precise comprehensive damage area ratio assessment and dynamic monitoring and analysis, this invention improves the scientific nature, automation level and clinical application value of skin damage detection, and provides an efficient, precise and intelligent solution for skin management of radiotherapy patients.

[0033] Example 6, this example is the explanation in Example 1, please refer to Figure 1 Specifically, the comparison subunit is used to preset a first standard threshold value V1 and a second standard threshold value V2, and V1>V2, and compare the comprehensive injury area ratio of the i-th patient Compare and evaluate with the first standard threshold V1 and the second standard threshold V2 to obtain and output the corresponding classification damage level; When the comprehensive injury area of the i-th patient is Greater than or equal to the first standard threshold V1, that is When ≥V1, it means that the patient's skin damage is classified as Grade 3 abnormal damage, and there is a risk of immune infection, which requires further suppression; When the second standard threshold V2 is less than or equal to the comprehensive injury area ratio of the i-th patient , and is less than the first standard threshold V1, that is, When , it means that the skin injury area of the current patient i belongs to the second-level abnormal injury level; at this time, there is an immune infection risk, but it is lower than the risk of the third-level abnormal injury level; When the comprehensive injury area of the i-th patient is Less than the second standard threshold V2, that is When the value is less than V2, it indicates that the skin injury area of the current patient i is within the expected range and is in a recoverable state. It is marked as a normal injury level 1 and prompts debridement and disinfection, basic nursing care, and avoidance of overtreatment. The graded treatment plan improves the accuracy and efficiency of treatment, reduces unnecessary drug use, and reduces the burden on patients.

[0034] In this embodiment, a comprehensive damage area ratio is used for quantitative assessment, which makes the determination of the degree of skin damage more accurate and objective, reduces the error of human subjective judgment, and distinguishes the first level abnormal damage level, the second level abnormal damage level and the third level abnormal damage level through the setting of the first standard threshold value V1 and the second standard threshold value V2, thereby improving the scientific nature of the damage classification.

[0035] Example 7, this example is the explanation in Example 1, please refer to Figure 1 Specifically, when the i-th patient is identified as having a second-level abnormal injury level and a third-level abnormal injury level, they are summarized into a healing need group, and the image data set in the healing need group is extracted through the skin injury assessment model to construct a second recognition module, which includes an epidermis detection and analysis unit, a dermis detection and analysis unit, and a subcutaneous layer detection and analysis unit; The epidermal detection and analysis unit is used to associate the epidermal layer damage level of the patient's skin with epidermal self-healing, and the epidermal detection and analysis unit includes a skin self-healing subunit and an extraction unit. The skin self-healing subunit is used to identify the damaged area and the time t after the patient's radiotherapy through the image data set in the healing demand group. After dimensionless processing, the skin tissue self-healing coefficient of the i-th patient is calculated by the following formula : Where d represents the severity of skin damage, ranging from 0 to 1, and k represents the recovery rate of skin tissue. The range of skin recovery rate k has been recorded through existing medical research and clinical experiments. It is expressed as an exponential decay function, t is the time after radiotherapy, It is expressed as the initial value of the recovery rate, which is obtained based on clinical data, medical research or experiments. It is expressed as a decay constant, with a value of 0-1; Reference example for the value of skin damage severity d: 0: no damage; 0.1-0.4 (not included) minor injuries, such as redness and swelling; 0.4-0.7 (not included) moderate to high, such as blisters; 0.7-1 severe injury, such as ulceration; The recovery rate k of skin tissue refers to the following example: 0-0.5 (exclusive) slow recovery rate; 0.5-0.8 (not included) medium recovery rate; 0.8-1 fast recovery; decay constant Reference example: 0.1-0.4 (not included) slower decay; 0.4-0.7 (not included) medium attenuation; 0.7-1 decays relatively quickly; Recovery rate initial value Reference example: Healthy skin 0.8 (inclusive) -1; Skin after radiotherapy 0.4-0.8 (not included).

[0036] The following is the skin tissue self-healing coefficient of the i-th patient Example table: The extraction unit is used to extract the damaged area and skin tissue self-healing coefficient of the i-th patient After dimensionless processing, the epidermal damage coefficient of the i-th patient is obtained by the following formula : Where, S represents the skin type influencing factor of the i-th patient, which is obtained by looking up the table; C represents the radiotherapy dose influencing factor of the i-th patient, with a value range of 1.1-1.3, which is selected based on clinical experience; Reference example of the influencing factor S of the i-th patient's skin type: White skin 1.0-1.2; Brown skin 0.8-0.9; Reference example of the factor C affecting the radiotherapy dose of the i-th patient: Radiotherapy dose less than 30Gy (inclusive) 1.0-1.1 (exclusive); Radiotherapy dose 30 (not included) -50Gy1.1-1.2 (not included); The radiotherapy dose is greater than or equal to 50Gy1.2-1.3.

[0037] When the epidermal damage coefficient of the i-th patient Compare and evaluate with the preset epidermal damage threshold B, and when the epidermal damage coefficient of the i-th patient When the value is greater than or equal to the preset epidermal damage threshold B, it means that the epidermal damage level of the current i-th patient is at level 5 abnormal damage level and requires medication, which manifests as redness, swelling, and local pain. When the epidermal damage coefficient of the i-th patient When the value is less than the preset epidermal damage threshold B, it means that the epidermal damage level of the current i-th patient is the fourth abnormal damage level, which requires simple treatment and continued observation.

[0038] The following is the epidermal damage coefficient of patient i Example table: Group Damage area ratio% The skin tissue self-healing coefficient #timg# of the i-th patient The influencing factor S of the skin type of the i-th patient The factor C affecting the radiotherapy dose of the i-th patient Epidermal damage coefficient of patient i #timg# Preset threshold Evaluation results 1 4.29 1.48 1.0 1.0 1.93 1.5 Level 5 abnormal damage level 2 3.13 1.16 1.1 1.2 2.08 1.5 Level 5 abnormal damage level 3 4.59 1.65 0.9 1.0 2.15 1.5 Level 5 abnormal damage level 4 5.00 3.07 1.2 1.1 3.35 1.5 Level 5 abnormal damage level 5 3.53 1.15 1.0 1.0 1.41 1.5 Level 4 Abnormal Damage Level 6 3.88 1.76 1.1 1.3 3.43 1.5 Level 5 abnormal damage level 7 4.63 1.60 1.0 1.2 2.36 1.5 Level 5 abnormal damage level 8 4.16 2.00 1.1 1.0 2.92 1.5 Level 5 abnormal damage level 9 4.39 1.36 0.9 1.1 0.38 1.5 Level 4 Abnormal Damage Level In this embodiment, through injury grading assessment, the target areas of secondary and tertiary injuries are automatically summarized into healing need groups to ensure that high-risk patients receive key attention and precise treatment. Combined with the skin injury assessment model, the image data set in the healing need group is further analyzed to optimize personalized treatment plans. The skin self-healing subunit is used to calculate the skin tissue self-healing coefficient. Through dimensionless processing, the injury severity, injury area, recovery rate, radiotherapy time and other factors are integrated to provide a quantitative skin healing ability assessment. The exponential decay model is used to predict the skin recovery trend to help doctors adjust the treatment rhythm and optimize the patient's rehabilitation plan. The extraction unit combines the skin damaged area, skin type influencing factor, and radiotherapy dose influencing factor to calculate the epidermal damage coefficient, and analyze the skin damage condition in multiple dimensions.

[0039] Example 8, this example is the explanation in Example 1, please refer to Figure 1 Specifically, the dermis detection and analysis unit is used to correlate the damage depth with the inflammatory response factor of the dermis layer, and obtain the time after radiotherapy by the timer. After dimensionless processing, the dermis damage coefficient of the i-th patient is obtained by the following formula: : Where, Indicates the depth of damage, represents the inflammatory response factor of the dermis, t represents the time after radiotherapy, represents the time decay factor, which is derived from clinical data. The coefficient indicating the influence of injury depth on recovery is obtained through clinical trial data; Dermal inflammatory response factor Reference example for value selection: mild inflammation, with a value less than 0.5; moderate inflammation, with a value of 0.5 (inclusive)-0.7; severe inflammation, with a value greater than or equal to 0.7; Time decay factor Reference example for value selection: Radiotherapy duration is less than or equal to 10 days, 0.6-0.8 (inclusive); Radiotherapy duration 10-20 (inclusive) days, 0.3-0.6 (inclusive); Radiotherapy duration 20-30 days, 0.1-0.3 (inclusive); Influence coefficient of damage depth on recovery Reference example for value selection: The depth of the injury is less than 0.5 mm, 0.8-1.0; 0.5 (inclusive)-1.5mm, 0.5-0.8 (inclusive); 1.5 (inclusive)-3.5, 0.1-0.5 (inclusive).

[0040] Dermis damage coefficient of patient i Compare and evaluate with the threshold value W for assessing dermal damage; When the dermis damage coefficient of the i-th patient When the value is greater than or equal to the dermis damage threshold W, it means that the dermis damage abnormality level in the current i-th patient is level seven abnormal damage level; When the dermis damage coefficient of the i-th patient When the threshold value W for evaluating dermal damage is less than , it means that the abnormal level of dermal damage in the current i-th patient is level six abnormal damage level.

[0041] The following is the dermis damage coefficient of patient i Example table: Group Damage depth (mm) Dermal inflammatory response factor pg / mL Days after radiotherapy Time decay factor Damage depth recovery influence coefficient Dermis damage coefficient of patient i Evaluation threshold Evaluation results 1 2.0 0.8 7 0.67 0.5 0.53 0.5 Level 7 Abnormal Damage Level 2 1.5 0.7 10 0.33 0.6 0.2 0.5 Level 6 abnormal damage level 3 3.0 0.9 6 0.5 0.4 0.54 0.5 Level 7 Abnormal Damage Level 4 2.5 0.85 8 0.6 0.45 0.57 0.5 Level 7 Abnormal Damage Level 5 1.2 0.65 9 0.3 0.7 0.16 0.5 Level 6 abnormal damage level 6 3.5 0.95 4 0.55 0.35 0.64 0.5 Level 7 Abnormal Damage Level 7 2.8 0.88 5 0.45 0.42 0.47 0.5 Level 6 abnormal damage level 8 1.0 0.6 6 0.25 0.75 0.11 0.5 Level 6 abnormal damage level 9 3.2 0.92 7 0.52 0.38 0.58 0.5 Level 7 Abnormal Damage Level In this embodiment, the injury depth, dermal inflammatory response factor, and radiotherapy time are used as core parameters, combined with dimensionless processing to calculate the dermal damage coefficient, provide a digitized injury assessment, reduce subjective judgment errors, reflect the recovery trend after radiotherapy through the time attenuation factor, and combine the injury depth influence coefficient to provide a mathematical model that is more in line with the physiological mechanism of skin repair, optimize the prediction of skin injury recovery, and accurately divide the skin condition by comparing the dermal damage coefficient with the dermal damage threshold W for assessment.

[0042] Moreover, the radiotherapy time is recorded by a timer and combined with the dynamic time decay model to provide real-time assessment of dermal damage, support multiple review and optimize treatment plans.

[0043] Example 9, this example is the explanation in Example 1, please refer to Figure 1 Specifically, the subcutaneous layer detection and analysis unit is used to obtain the subcutaneous layer damage coefficient of the i-th patient through the following formula after dimensionless processing based on the correlation between the subcutaneous layer inflammatory factors and the subcutaneous layer oxygenation degree: : Where, It is expressed as the normalized subcutaneous inflammatory factor, with a value of 0-1. It is expressed as normalized subcutaneous oxygenation, with a value of 0-1. Expressed as the subcutaneous inflammation correction constant, Expressed as the correction constant for subcutaneous oxygenation.

[0044] Normalized subcutaneous inflammatory factors Reference example for value selection: mild inflammation, 0.0-0.4 (inclusive); moderate inflammation, 0.4-0.8 (inclusive); heavy inflammation, 0.8-1.0; Normalized subcutaneous oxygenation Value reference example: hypoxia, 0.0-0.4 (inclusive); Normal is low, 0.4-0.6 (inclusive); Normal, 0.6-0.8 (inclusive); superoxide, 0.8-1.0; Subcutaneous inflammation correction constant Reference example for values: Severe inflammation, 0.8-0.9; hypooxygenation, 1.0-1.3; severe inflammation and insufficient oxygenation, 0.8; Subcutaneous oxygenation correction constant Value reference example; Normal recovery, 1.0-1.1; Inflammation is more severe, 0.7-0.9; hypooxygenation, 1.0-1.2; severe inflammation and insufficient oxygenation, 0.8; Recovery is relatively fast, 1.2-1.3.

[0045] The subcutaneous layer damage coefficient of the i-th patient Compare and evaluate with the preset subcutaneous layer damage threshold G; When the subcutaneous layer damage coefficient of the i-th patient When the value is greater than or equal to the preset subcutaneous layer injury threshold G, it means that the abnormal level of subcutaneous layer damage in the current i-th patient is the ninth abnormal level of damage; When the subcutaneous layer damage coefficient of the i-th patient When < the preset subcutaneous layer injury threshold G, it means that the subcutaneous layer abnormality level in the current i-th patient is the eighth level of abnormal injury level.

[0046] The following is the subcutaneous layer damage coefficient of patient i Example table: Group Normalized subcutaneous inflammatory factors pg / mL Normalized subcutaneous oxygenation % Subcutaneous inflammation correction constant Subcutaneous oxygenation correction constant The subcutaneous layer damage coefficient of the i-th patient Evaluation threshold Evaluation results 1 0.6 0.8 1.2 0.9 1.32 1.0 Level 9 Abnormal Damage Level 2 0.4 0.7 1.1 0.8 0.92 1.0 Level 8 abnormal damage level 3 0.7 0.5 1.3 0.7 1.36 1.0 Level 9 Abnormal Damage Level 4 0.5 0.6 1.15 0.8 1.05 1.0 Level 9 Abnormal Damage Level 5 0.3 0.9 1.0 0.9 0.95 1.0 Level 8 abnormal damage level 6 0.8 0.4 1.25 0.85 1.38 1.0 Level 9 Abnormal Damage Level 7 0.6 0.7 1.05 0.95 1.11 1.0 Level 9 Abnormal Damage Level 8 0.5 0.6 1.1 0.8 1.0 1.0 Level 9 Abnormal Damage Level 9 0.7 0.5 1.2 0.9 1.26 1.0 Level 9 Abnormal Damage Level In this embodiment, subcutaneous inflammatory factors and subcutaneous oxygenation are used as key indicators, combined with dimensionless normalization processing, to calculate the subcutaneous injury coefficient, provide a digitized injury assessment, reduce the doctor's subjective judgment error, and through normalization processing (value range of 0-1), ensure the comparability of the subcutaneous state of different patients, improve the standardization of injury assessment, and the interaction between subcutaneous inflammatory factors and oxygenation can reflect the severity of injury and repair capacity of subcutaneous tissue, provide a more comprehensive injury assessment, optimize rehabilitation treatment plans, and combine individual differences of patients (skin type, radiotherapy dose, recovery speed, etc.) to provide personalized treatment plans and optimize intervention strategies for subcutaneous tissue repair. Through dynamic monitoring of subcutaneous oxygenation, the effectiveness of drugs or rehabilitation treatments can be evaluated, and the treatment plan can be adjusted in time to accelerate the patient's recovery process.

[0047] The above formulas are obtained by collecting a large amount of data and performing software simulation, and a formula close to the actual value is selected. The coefficients in the formula are set by those skilled in the art according to actual conditions. The above is only a preferred specific implementation method of the present invention, but the protection scope of the present invention is not limited to this. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, can make equivalent replacements or changes based on the technical solution and inventive concept of the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A real-time detection and record management system for skin damage in radiotherapy patients based on image processing, characterized in that: include: The multispectral image acquisition module uses different spectra to image the patient's skin at the same viewing angle, obtaining three spectral images: near-infrared imaging, infrared imaging, and ultraviolet imaging. The three spectral images are then divided into different location areas to obtain several target areas. The first recognition module is used to fuse the three spectral images to obtain an image data set, use neural network technology to build a skin damage assessment model, extract relevant damage area features from the image data set, and construct the comprehensive damage area ratio of the i-th patient. , and evaluate, output classification damage level; The classification injury levels include: level one normal injury level, level two abnormal injury level, and level three abnormal injury level; The second recognition module is used to identify the i-th patient with secondary abnormal injury level and tertiary abnormal injury level, summarize them into a treatment demand group, extract the skin injury features in the image data set of the treatment demand group through the skin injury assessment model, conduct in-depth analysis to obtain the injury depth, dermal inflammatory response factors, subcutaneous inflammatory factors and subcutaneous oxygenation, and construct the epidermal injury coefficient of the i-th patient. , dermal damage coefficient and subcutaneous damage coefficient , and evaluate them separately, outputting the corresponding damage results.

2. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 1 is characterized in that: The multispectral image acquisition module includes a segmentation unit and an edge detection unit; The segmentation unit is used to divide the image data set into different position areas on the same patient target areas, m represents the total number of target areas; target areas include arms, abdomen, back and legs; The edge detection unit is used to detect and identify the number N of skin edge contour point pixels of each target area of the same patient using the Sobel operator, and obtain the injury area according to the number N of skin edge contour point pixels.

3. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 2 is characterized in that: The first recognition module is used to fuse the three spectral images to obtain an image data set, and use a convolutional neural network to build a skin damage assessment model as a data operation prediction; During the current monitoring period, three spectral images are acquired and the abnormal skin monitoring algorithm is used to calculate the comprehensive damage area ratio of the i-th patient. ; Abnormal skin monitoring algorithm includes the following steps: First, use image denoising algorithm and image enhancement algorithm to preprocess each frame of the extracted image; Second, use a deep learning model to process each pre-processed image frame and extract a high-dimensional feature vector containing the damage area characteristics; 3. Using convolutional neural networks, construct an initial convolutional neural network model, and test and train the initial convolutional neural network model with image data sets, and use the trained initial convolutional neural network model as a skin damage assessment model. At the same time, use the intermediate layer output of the skin damage assessment model as a feature vector to identify relevant damage area features, and train and test the skin damage assessment model through the obtained relevant damage area features. Use the trained skin damage assessment model as data for prediction to calculate the comprehensive damage area ratio of the i-th patient. .

4. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 3 is characterized in that: The first recognition module further includes a damage area feature summary unit and an analysis unit; The lesion area feature summary unit is used to extract features from the image dataset of the i-th patient using threshold segmentation and deep learning segmentation technology, compare the color, texture, and spectral reflectance characteristics of normal skin and damaged skin, extract the boundary of the damaged area, perform calibration, and sum the pixel points of the damaged area using pixel counting to obtain the lesion area of the i-th patient. .

5. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 4 is characterized in that: The analysis unit includes a skin injury area subunit and a comparison subunit; The skin injury area subunit is used to calculate the comprehensive injury area ratio of the i-th patient according to the injury area of the i-th patient after dimensionless processing using the following formula: : Where m represents the total number of target areas, represents the injury area of the i-th patient, represents the total body area of the i-th patient.

6. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 5, characterized in that: The comparison subunit is used to preset a first standard threshold value V1 and a second standard threshold value V2, and V1>V2, and compare the comprehensive injury area ratio of the i-th patient Compare and evaluate with the first standard threshold V1 and the second standard threshold V2 to obtain and output the corresponding classification damage level; When the comprehensive injury area of the i-th patient is Greater than or equal to the first standard threshold V1, that is When ≥V1, it means that the patient's skin damage is classified as Grade 3 abnormal damage, and there is a risk of immune infection, which requires further suppression; When the second standard threshold V2 is less than or equal to the comprehensive injury area ratio of the i-th patient , and is less than the first standard threshold V1, that is, When , it means that the skin injury area of the current i-th patient belongs to the second-level abnormal injury level; At this point, there is a risk of infection with immunity, but it is lower than the risk of level three abnormal damage; When the comprehensive injury area of the i-th patient is Less than the second standard threshold V2, that is When <V2, it means that the skin injury area of the current i-th patient is within expectations and is in a recoverable state, and is marked as the first normal injury level.

7. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 6, characterized in that: When the i-th patient is identified as having a second-level abnormal injury grade or a third-level abnormal injury grade, the patient is grouped into a treatment need group. The image data set in the treatment need group is extracted using the skin injury assessment model to construct a second recognition module, which includes an epidermis detection and analysis unit, a dermis detection and analysis unit, and a subcutaneous layer detection and analysis unit. The epidermal detection and analysis unit includes a skin self-healing subunit and an extraction unit. The skin self-healing subunit is used to identify the association between the damaged area and the time t after the patient's radiotherapy through the image data set in the healing demand group. After dimensionless processing, the skin tissue self-healing coefficient of the i-th patient is calculated using the following formula : Where d represents the severity of skin damage, ranging from 0 to 1, and k represents the recovery rate of skin tissue. The range of skin recovery rate k has been recorded through existing medical research and clinical experiments. It is expressed as an exponential decay function, t is the time after radiotherapy, It is expressed as the initial value of the recovery rate, which is obtained based on clinical data, medical research or experiments. It is expressed as a decay constant, with a value of 0-1; The extraction unit is used to extract the damaged area and skin tissue self-healing coefficient of the i-th patient After dimensionless processing, the epidermal damage coefficient of the i-th patient is obtained by the following formula : Where S represents the influencing factor of the skin type of the i-th patient, which is obtained by looking up the table, and C represents the influencing factor of the radiotherapy dose of the i-th patient, which is selected based on clinical experience; The epidermal damage coefficient of the i-th patient Compared with the preset epidermal damage threshold B, when the epidermal damage coefficient of the i-th patient When the value is less than the preset epidermal damage threshold B, it means that the epidermal damage level of the current i-th patient is the fourth abnormal damage level; When the epidermal damage coefficient of the i-th patient When ≥ the preset epidermal damage threshold B, it means that the epidermal damage level of the current i-th patient is the fifth level of abnormal damage.

8. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 7, characterized in that: The dermis detection and analysis unit is used to correlate the damage depth with the inflammatory response factor of the dermis layer, and obtain the time after radiotherapy through a timer. After dimensionless processing, the dermis damage coefficient of the i-th patient is obtained through the following formula : Where, Indicates the depth of damage, represents the inflammatory response factor of the dermis, t represents the time after radiotherapy, represents the time decay factor, which is derived from clinical data. The coefficient indicating the influence of injury depth on recovery is obtained through clinical trial data; The dermis damage coefficient of the i-th patient Compare and evaluate with the threshold value W for assessing dermal damage; When the dermis damage coefficient of the i-th patient When the threshold value W for evaluating dermal damage is less than , it means that the dermal damage level of the current i-th patient is level 6 abnormal damage level; When the dermis damage coefficient of the i-th patient When the dermis damage threshold W is greater than or equal to the evaluation threshold, it means that the abnormal level of dermis damage in the current i-th patient is the seventh abnormal damage level.

9. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 8, characterized in that: The subcutaneous layer detection and analysis unit is used to obtain the subcutaneous layer damage coefficient of the i-th patient through the following formula after dimensionless processing based on the correlation between subcutaneous inflammatory factors and subcutaneous oxygenation. : Where, It is expressed as the normalized subcutaneous inflammatory factor, with a value of 0-1. It is expressed as normalized subcutaneous oxygenation, with a value of 0-1. Expressed as the subcutaneous inflammation correction constant, Expressed as the correction constant for subcutaneous oxygenation.

10. The real-time detection and record management system for skin damage in radiotherapy patients based on image processing according to claim 9, characterized in that: The subcutaneous layer damage coefficient of the i-th patient Compare and evaluate with the preset subcutaneous layer damage threshold G; When the subcutaneous layer damage coefficient of the i-th patient When the threshold value G of subcutaneous layer damage is less than the preset threshold value G, it means that the abnormal level of subcutaneous layer damage in the current i-th patient is level eight abnormal damage level; When the subcutaneous layer damage coefficient of the i-th patient When the value is greater than or equal to the preset subcutaneous layer injury threshold G, it indicates that the abnormal level of subcutaneous layer damage in the current i-th patient is the ninth abnormal level of damage.

Images