Intelligent wound measuring and recording system and method

Abstract

The invention relates to the technical field of medical treatment, and particularly discloses an intelligent wound measuring and recording system and method.The method comprises the steps that multi-dimensional image data and three-dimensional depth information of a wound area are synchronously collected, hierarchical space division and nonlinear feature coding are adopted for preprocessing the multi-dimensional image data, and wound basic information is generated; constructing a wound measurement model based on the basic information of the wound, extracting wound features with space-time relevance, and extracting a wound contour of multi-dimensional image data by using an improved U-Net architecture in combination with an HSV color space and morphological gradient method; according to the method, the multi-dimensional image data and the three-dimensional depth information are synchronously acquired, and the advanced image processing technology is combined, so that the accurate extraction of the wound contour is realized, and the problem that wound details cannot be accurately identified and measured in a traditional method is solved; based on an ICP registration algorithm and a three-dimensional point cloud reconstruction technology, submillimeter-level wound volume measurement is realized, and the wound volume evaluation precision is improved.

Description

Technical Field

[0001] The present invention belongs to the field of medical technology, and in particular relates to an intelligent wound measurement and recording system and method. Background Art

[0002] In clinical medicine, wound assessment and monitoring are crucial for managing treatment outcomes and the healing process. Currently, wound assessment relies primarily on medical staff using specialized paper rulers to take images of wounds and then measure wound parameters. This method suffers from low measurement accuracy and efficiency. Traditional wound measurement typically only provides two-dimensional area, failing to achieve high-precision three-dimensional wound volume assessment. This makes it particularly difficult to track dynamic changes in the wound during healing, especially during real-time.

[0003] With the advancement of medical imaging and computer vision technologies, three-dimensional imaging and image processing technologies have been introduced in an attempt to provide more accurate wound assessment and dynamic monitoring. However, existing wound measurement technologies still face some challenges. Traditional technologies mostly rely on static images, which makes it difficult to capture the three-dimensional changes of wounds and cannot fully assess the depth and volume of wounds, resulting in inaccurate measurement results. Although three-dimensional imaging technology has been applied, existing technologies have not yet been able to achieve accurate real-time monitoring during the wound healing process, especially when faced with cases with more complex wound changes, and still cannot achieve ideal tracking effects. Secondly, although dynamic learning methods based on time series provide new directions for monitoring wound healing, these methods have poor adaptability and cannot effectively handle data differences between different patients and different types of wounds.

[0004] Therefore, it is necessary to propose an intelligent wound measurement and recording system and method to solve the problems of low efficiency and lack of dynamic monitoring in wound measurement of traditional wound image measurement methods in the prior art.

[0005] The above information disclosed in this background technology is only for enhancing understanding of the background technology of the present invention and therefore it may contain information that does not constitute the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] The object of the present invention is to provide an intelligent wound measurement and recording system and method to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] An intelligent wound measurement and recording method, comprising:

[0009] Synchronously collecting multi-dimensional image data and three-dimensional depth information of the wound area, and preprocessing the multi-dimensional image data using hierarchical spatial partitioning and nonlinear feature coding to generate basic wound information;

[0010] A wound measurement model is constructed based on the basic wound information to extract wound features with spatiotemporal correlation, and the wound contour is extracted from the multi-dimensional image data using an improved U-Net architecture combined with HSV color space and morphological gradient method;

[0011] Based on the wound contour and combined with the three-dimensional depth information, the sub-millimeter level wound volume measurement is performed using the ICP registration algorithm and three-dimensional point cloud reconstruction technology to obtain wound volume data and extract wound volume parameters;

[0012] The basic wound information, the wound characteristics, the wound contour and the wound volume parameters are integrated to generate a structured report according to predefined medical standards. The structured report includes wound area, depth, volume, healing stage classification and infection risk index.

[0013] Preferably, the pretreatment step comprises:

[0014] A hierarchical regular hexagonal network was used to spatially divide the wound area, where each high-level regular hexagonal unit was divided into seven low-level regular hexagonal and six regular triangle subunits;

[0015] The data is compressed and denoised using nonlinear feature coding technology to obtain the compressed multi-dimensional image data; wherein the coding rules of the nonlinear feature coding technology are based on a multi-parameter combination of wound location, area and healing stage, namely:

[0016] Φ=F(α·P pos +β·P area +γ·S stage )

[0017] Where F is the nonlinear mapping function, P pos is the wound location parameter, P area is the wound area parameter, S stage is the wound healing stage parameter, and α, β, and γ are dynamic weight coefficients.

[0018] Preferably, the wound measurement model includes the following structure:

[0019] A temporal convolutional network module with dynamically adjusted dilation factors is introduced to capture the local temporal correlation of wound features through deformable convolution kernels and channel attention modules.

[0020] A bidirectional recurrent neural network module with integrated attention mechanism is used to update the hidden state through a gated adaptive forgetting mechanism to capture the long-range spatiotemporal dependencies in the wound healing process;

[0021] The update formula is:

[0022] h t =σ(W h ·h t-1 +W x ·x t +b)Θtanh(W hh ·h t-1 +W hx ·x t +b h )

[0023] Where h t is the hidden state at the current moment, σ is the sigmoid function, W h 、W x 、W hh 、W hx is the weight matrix, b and b h is the bias term;

[0024] Feature fusion layer based on dynamic weighted formula, where the weight coefficients are generated by the channel attention module:

[0025]

[0026] Where w k is the weight coefficient of the kth feature channel, z k is the feature significance index of the corresponding channel, W k and b k are the learnable parameter matrices and bias terms.

[0027] Preferably, the method of extracting the wound contour from the multi-dimensional image data by using an improved U-Net architecture combined with HSV color space and morphological gradient method includes:

[0028] Based on the spatiotemporal correlation of the wound features, performing spatiotemporal alignment and feature enhancement on the multidimensional image data to obtain the enhanced multidimensional image;

[0029] Converting the enhanced multidimensional image from the RGB color space to the HSV color space to obtain the multidimensional image in HSV color;

[0030] Based on the multidimensional image of HSV color, extracting the wound candidate area by HSV color space segmentation according to the hue, saturation and brightness characteristics of the wound area;

[0031] Optimizing the wound candidate region by using a morphological gradient method to eliminate noise, fill small holes, and smooth the boundaries to obtain the optimized wound candidate region;

[0032] Perform feature extraction and semantic segmentation on the optimized wound candidate area by introducing an attention mechanism and a U-Net architecture with depthwise separable convolution, and output the wound contour of the wound candidate area;

[0033] The wound contour of the wound candidate area is post-processed to output the final wound contour.

[0034] Preferably, the sub-millimeter level wound volume measurement is performed based on the wound contour in combination with three-dimensional depth information through an ICP registration algorithm and three-dimensional point cloud reconstruction technology to obtain wound volume data and extract wound volume parameters, including:

[0035] Mapping each pixel point on the wound contour to a corresponding position in three-dimensional space using a coordinate mapping algorithm to form preliminary point cloud data of the wound surface;

[0036] Point cloud registration is performed using the ICP algorithm based on preselected matching points, and the registration error threshold is set to 0.1 mm;

[0037] A multi-scale point cloud registration algorithm is introduced, and the following formula is used to perform multiple point cloud registrations on the wound contour at different scales to obtain the registered wound surface point cloud data;

[0038] Multi-scale point cloud registration formula:

[0039]

[0040] Where, ICP r is the ICP registration result at the rth scale, is the weight corresponding to each scale;

[0041] The α-shape algorithm is used to perform three-dimensional reconstruction on the registered wound surface point cloud data to generate a triangular mesh model of the wound surface;

[0042] Based on the triangular mesh model, the wound volume was calculated using the following volume formula, and the wound volume-related parameters were extracted:

[0043]

[0044] Where V is the wound volume, x j 、y j 、x j is the three-dimensional coordinate of the jth point in the point cloud, and m is the number of wound volume point clouds.

[0045] Preferably, the method further comprises:

[0046] Feeding the structured report back to the clinical system and binding it to the patient's biometrics, using an incremental learning framework combined with an elastic weight solidification algorithm to smooth the impact between new and old data;

[0047] Establish a wound feature evolution map and continuously update the pathological feature database using online learning mechanisms;

[0048] Set the dynamic memory playback buffer. The buffer size is calculated as follows:

[0049]

[0050] Where N is the current number of samples.

[0051] An intelligent wound measurement and recording system, comprising:

[0052] An image acquisition module, used to obtain multi-dimensional image data and three-dimensional depth information of the wound area;

[0053] A preprocessing module, configured to preprocess the multi-dimensional image data and three-dimensional depth information using hierarchical space partitioning and nonlinear feature encoding technology to generate basic wound information;

[0054] A wound measurement module is used to construct a wound measurement model based on the basic wound information, extract wound features with spatiotemporal correlation, and extract the wound contour of the multi-dimensional image data using an improved U-Net architecture combined with HSV color space and morphological gradient method;

[0055] A volume measurement module is used to measure the wound volume at the sub-millimeter level based on the wound contour and in combination with the three-dimensional depth information, using an ICP registration algorithm and three-dimensional point cloud reconstruction technology to obtain wound volume data and extract wound volume parameters;

[0056] a report generation module, configured to integrate the basic wound information, the wound characteristics, the wound contour, and the wound volume parameter to generate a structured report according to predefined medical standards, wherein the structured report includes wound area, depth, volume, healing stage classification, and infection risk index;

[0057] The feedback and learning module is used to feed the structured report back to the clinical system and bind it to the patient's biometrics, and adopt an incremental learning framework combined with an elastic weight solidification algorithm to smooth the impact between new and old data.

[0058] Preferably, the system further comprises:

[0059] A wound classification module uses ResNet-152 as the base network for feature extraction and improves classification accuracy by initializing weights of a pre-trained model on a large medical image dataset.

[0060] The wound healing prediction module is used to analyze time series measurement data through the LSTM network to predict the changing trends of wound area, volume and healing stage.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] The present invention achieves accurate extraction of wound contours by synchronously collecting multi-dimensional image data and three-dimensional depth information, and combines it with advanced image processing technology, thus solving the problem of the inability of traditional methods to accurately identify and measure wound details. It utilizes an ICP-based registration algorithm and three-dimensional point cloud reconstruction technology to achieve sub-millimeter wound volume measurement, improving the accuracy of wound volume assessment and enabling real-time tracking of wound changes. In addition, an incremental learning framework and an elastic weight solidification algorithm are used to smooth the impact between new and old data, resulting in high adaptability and robustness in different patients and different wound types. Structured reports are automatically generated according to medical standards to comprehensively assess the area, depth, volume, healing stage, and infection risk of the wound, and feedback is provided to the clinical system to provide doctors with accurate wound management and treatment decision support, thereby promoting the personalization and refinement of wound treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of the intelligent wound measurement and recording method of the present invention;

[0064] Figure 2 This is a framework diagram of the intelligent wound measurement and recording system of the present invention. DETAILED DESCRIPTION

[0065] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0066] Example 1:

[0067] See also Figure 1 As shown, an intelligent wound measurement and recording method includes:

[0068] Synchronously collect multi-dimensional image data and three-dimensional depth information of the wound area, and use hierarchical spatial division and nonlinear feature encoding to pre-process the multi-dimensional image data to generate basic wound information;

[0069] The preprocessing steps include:

[0070] A hierarchical regular hexagonal network was used to spatially divide the wound area, where each high-level regular hexagonal unit was divided into seven low-level regular hexagonal and six regular triangle subunits;

[0071] Nonlinear feature coding technology is used to compress and denoise data to obtain compressed multi-dimensional image data; the encoding rules of the nonlinear feature coding technology are based on a multi-parameter combination of wound location, area and healing stage.

[0072] Furthermore, by simultaneously acquiring multi-dimensional image data and three-dimensional depth information, combined with hierarchical spatial partitioning and nonlinear feature coding preprocessing, the accuracy and efficiency of wound assessment were effectively improved. The hierarchical regular hexagonal network optimized the spatial partitioning of the wound area, making the details of each area more precise; while the nonlinear feature coding technology retained important wound information while compressing and denoising the data, ensuring high-quality basic wound data can be obtained even in complex environments, providing strong support for the accurate assessment of wound morphology, area, and healing status.

[0073] A wound measurement model is constructed based on basic wound information to extract wound features with spatiotemporal correlation. An improved U-Net architecture is then combined with the HSV color space and morphological gradient method to extract wound contours from multi-dimensional image data.

[0074] The wound measurement model includes the following structures:

[0075] A temporal convolutional network module with dynamically adjusted dilation factors is introduced to capture the local temporal correlation of wound features through deformable convolution kernels and channel attention modules.

[0076] A bidirectional recurrent neural network module with integrated attention mechanism is used to update the hidden state through a gated adaptive forgetting mechanism to capture the long-range spatiotemporal dependencies in the wound healing process;

[0077] A feature fusion layer based on a dynamic weighted formula, where the weight coefficients are generated by the channel attention module;

[0078] Based on the spatiotemporal correlation of wound features, the multi-dimensional image data is spatiotemporally aligned and feature enhanced to obtain an enhanced multi-dimensional image.

[0079] Convert the enhanced multidimensional image from RGB color space to HSV color space to obtain a multidimensional image with HSV colors;

[0080] Based on the multi-dimensional image of HSV color, the wound candidate area is extracted by HSV color space segmentation according to the hue, saturation and brightness characteristics of the wound area;

[0081] The wound candidate region is optimized by using the morphological gradient method to eliminate noise, fill small holes and smooth the boundaries to obtain the optimized wound candidate region;

[0082] By introducing the attention mechanism and the U-Net architecture with depthwise separable convolution, the optimized wound candidate area is subjected to feature extraction and semantic segmentation, and the wound contour of the wound candidate area is output;

[0083] The wound contour of the wound candidate area is post-processed to output the final wound contour.

[0084] Furthermore, this wound measurement model effectively captures the spatiotemporal correlation of wound features by introducing a time-domain convolutional network with a dynamically adjusted expansion factor, a bidirectional recurrent neural network with an integrated attention mechanism, and a feature fusion layer based on a dynamic weighting formula, thereby improving the accuracy of feature extraction and semantic segmentation during the wound healing process. By optimizing wound candidate regions through HSV color space segmentation and morphological gradient methods, further noise removal and boundary smoothing are achieved, and combined with a deep separable convolutional U-Net architecture, wound contours are accurately extracted. Ultimately, high-precision wound assessment and dynamic tracking are achieved, providing strong support for monitoring and decision-making during wound treatment.

[0085] Based on the wound contour and combined with 3D depth information, the ICP registration algorithm and 3D point cloud reconstruction technology are used to measure the sub-millimeter wound volume, obtain wound volume data, and extract wound volume parameters.

[0086] A coordinate mapping algorithm is used to map each pixel on the wound contour to a corresponding position in three-dimensional space to form preliminary point cloud data of the wound surface;

[0087] Point cloud registration is performed using the ICP algorithm based on preselected matching points, and the registration error threshold is set to 0.1 mm;

[0088] A multi-scale point cloud registration algorithm is introduced to perform multiple point cloud registrations of the wound contour at different scales to obtain the registered wound surface point cloud data.

[0089] The α-shape algorithm is used to perform three-dimensional reconstruction on the registered wound surface point cloud data to generate a triangular mesh model of the wound surface;

[0090] Based on the triangular mesh model, the wound volume was calculated using the volume formula and wound volume-related parameters were extracted.

[0091] Furthermore, by combining wound contour and 3D depth information, the ICP registration algorithm achieves high-precision submillimeter wound volume measurement. A coordinate mapping algorithm maps the wound contour pixels into 3D space, generating preliminary point cloud data. Multi-scale point cloud registration is then employed to reduce registration errors and ensure the accuracy of the point cloud data. The registered point cloud is then reconstructed in 3D using the α-shape algorithm, generating a precise triangular mesh model. Ultimately, wound volume and related parameters are calculated based on this model, providing reliable technical support for accurate wound assessment and monitoring.

[0092] Comprehensive wound basic information, wound characteristics, wound contour and wound volume parameters are used to generate a structured report based on predefined medical standards. The structured report includes wound area, depth, volume, healing stage classification and infection risk index;

[0093] Feed structured reports back to the clinical system and bind them to the patient's biometrics. An incremental learning framework combined with an elastic weight solidification algorithm is used to smooth the impact between new and old data.

[0094] Establish a wound feature evolution map and continuously update the pathological feature database using online learning mechanisms;

[0095] Set the dynamic memory playback buffer. The buffer size is calculated.

[0096] Furthermore, by combining wound measurement and analysis results, a structured report is generated, covering information such as wound area, depth, volume, healing stage, and infection risk, providing accurate clinical assessments. This structured report is fed back to the clinical system in real time and linked to the patient's biometrics. An incremental learning framework and elastic weight consolidation algorithm effectively smooth the impact of new and old data, ensuring data consistency and reliability. Simultaneously, a wound feature evolution map is established, and an online learning mechanism is used to continuously update the pathology feature database, enabling dynamic monitoring and intelligent management. This helps to enhance the personalization of wound treatment and the long-term tracking of wound efficacy.

[0097] Example 2:

[0098] See also Figure 2 As shown, an intelligent wound measurement and recording system includes:

[0099] An image acquisition module, used to obtain multi-dimensional image data and three-dimensional depth information of the wound area;

[0100] A preprocessing module is used to preprocess multi-dimensional image data and three-dimensional depth information using hierarchical space partitioning and nonlinear feature encoding technology to generate basic wound information;

[0101] The wound measurement module is used to build a wound measurement model based on basic wound information, extract wound features with spatiotemporal correlation, and use an improved U-Net architecture combined with the HSV color space and morphological gradient method to extract the wound contour from multi-dimensional image data;

[0102] The volume measurement module is used to measure wound volume at the sub-millimeter level based on wound contours, 3D depth information, ICP registration algorithm, and 3D point cloud reconstruction technology to obtain wound volume data and extract wound volume parameters;

[0103] A report generation module is used to integrate basic wound information, wound characteristics, wound contours, and wound volume parameters to generate a structured report based on predefined medical standards. The structured report includes wound area, depth, volume, healing stage classification, and infection risk index;

[0104] The feedback and learning module is used to feed structured reports back to the clinical system and bind them to the patient's biometrics. It uses an incremental learning framework combined with an elastic weight solidification algorithm to smooth the impact between new and old data.

[0105] A wound classification module uses ResNet-152 as the base network for feature extraction and improves classification accuracy by initializing weights of a pre-trained model on a large medical image dataset.

[0106] The wound healing prediction module is used to analyze time series measurement data through the LSTM network to predict the changing trends of wound area, volume and healing stage.

[0107] Application example of intelligent wound measurement and recording system (diabetic foot ulcer assessment scenario)

[0108] In a hospital's endocrinology ward, a nurse was required to perform weekly assessments of a diabetic patient's foot ulcers. Traditional manual measurement presented challenges such as low efficiency, high infection risk, and an inability to quantify volume changes. This patented system was then used to perform fully automated measurement and dynamic analysis.

[0109] 1. Application Process Description

[0110] (1) Equipment operation: The nurse uses a handheld intelligent measuring instrument (integrated with a high-resolution camera, a 3D depth sensor, and a ring-shaped LED fill light) to scan the wound at a distance of 10 to 15 cm.

[0111] Data acquisition: ① Multi-dimensional images: Collect 5 sets of wound images from different angles (including visible light and near-infrared bands); ② Three-dimensional depth information: Obtain depth data with an accuracy of 0.05mm through the ToF sensor; ③ Environmental data: Automatically record room temperature, patient body temperature and measurement time.

[0112] (2) Hierarchical spatial division: The system divides the wound area into multi-level regular hexagonal grids (the minimum unit is 0.5mm 2 ), identifying irregular edge regions.

[0113] Nonlinear feature encoding: Dynamic fusion of wound location (the third metatarsal area of the plantar foot), area (initial 8.7 cm 2 ), exudate color (yellow-green) and other parameters to generate a standardized feature vector.

[0114] Output: 3D point cloud model after denoising (containing 12,800 valid data points).

[0115] (3) HSV color space optimization: The system converts the image into HSV space and preliminarily segments the wound area based on the color difference between necrotic tissue (H = 30°) and healthy skin (H = 15°).

[0116] Morphological gradient processing: Use a 3×3 circular kernel for opening operation to eliminate the interference of foot hair and fill small holes in the leakage area.

[0117] (4) Improved U-Net segmentation: Reduce the computational effort through depthwise separable convolution and achieve sub-pixel edge fitting at the arch bend (error < 0.2 mm).

[0118] (5) ICP registration process: 200 sets of feature points (including vascular bifurcation markers) from this week and last week were matched, and the registration error was 0.08 mm after three scale optimizations.

[0119] Volume calculation: Apply the α-shape algorithm to generate a triangular mesh model (containing 5432 facets), and calculate the current volume to be 4.15cm 3 (A decrease of 9.3% from the previous week).

[0120] Depth monitoring: The maximum depth was detected to decrease from 7.2 mm to 6.5 mm, indicating the growth of granulation tissue.

[0121] 2. Structured Report Content

[0122] parameter Numerical Medical interpretation area <![CDATA[7.8cm 2 ]]> 10.2% smaller than baseline volume <![CDATA[4.15cm 3 ]]> Weekly change rate meets the standard (>5%) Risk of infection Moderate (rating 6.2 / 10) Bacterial culture is recommended Healing stage Granulation stage (grade III) Can transition to restorative dressing

[0123] System linkage: Automatically push warning information to the attending physician's mobile phone APP (triggered when the volume change rate is <3% in the week); historical data comparison chart shows that the area reduction trend within 6 weeks is in line with expectations (R 2 =0.93). The system stores the measurement data in a dynamic memory buffer (retaining the most recent 50 similar cases) and updates the classification model weights using an elastic weight consolidation algorithm. Based on the LSTM model, the volume is expected to drop to 1.2 cm in 6 weeks based on the current trend. 3 (82% confidence level).

[0124] 3. Comparison of application effects

[0125]

[0126]

[0127] Clinical Value: 1. Precise Monitoring: Submillimeter volume change seeding revealed three cases of potential healing arrest two weeks in advance; 2. Infection Early Warning: Quantitative analysis of exudate color reduced antibiotic usage by 35%; 3. Standardized Management: Structured reports are automatically connected to the electronic medical record system, improving medical work efficiency by 60%.

[0128] Example 3:

[0129] An embodiment of the present invention further provides a computer-readable storage medium storing a program for the intelligent wound measurement and recording system described above. When executed by a processor, the program implements the various processes of the aforementioned embodiments of the wound measurement and recording system and achieves the same technical effects. To avoid repetition, the details are omitted here. The computer-readable storage medium may be, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0130] In the description of this specification, the reference terms "one embodiment", "some embodiments", "examples", "specific examples" or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0131] The drawings of the embodiments disclosed in the present invention only involve structures related to the embodiments disclosed in the present invention. Other structures may refer to common designs. In the absence of conflicts, the same embodiment and different embodiments of the present invention may be combined with each other.

[0132] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0133] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent wound measurement and recording method, characterized in that: include: Synchronously collecting multi-dimensional image data and three-dimensional depth information of the wound area, and preprocessing the multi-dimensional image data using hierarchical spatial partitioning and nonlinear feature coding to generate basic wound information; A wound measurement model is constructed based on the basic wound information to extract wound features with spatiotemporal correlation, and the wound contour is extracted from the multi-dimensional image data using an improved U-Net architecture combined with HSV color space and morphological gradient method; Based on the wound contour and combined with the three-dimensional depth information, the sub-millimeter level wound volume measurement is performed using the ICP registration algorithm and three-dimensional point cloud reconstruction technology to obtain wound volume data and extract wound volume parameters; The basic wound information, the wound characteristics, the wound contour and the wound volume parameters are integrated to generate a structured report according to predefined medical standards. The structured report includes wound area, depth, volume, healing stage classification and infection risk index.

2. The intelligent wound measurement and recording method according to claim 1, characterized in that: The pre-processing step comprises: A hierarchical regular hexagonal network was used to spatially divide the wound area, where each high-level regular hexagonal unit was divided into seven low-level regular hexagonal and six regular triangle subunits; The data is compressed and denoised using nonlinear feature coding technology to obtain the compressed multi-dimensional image data; wherein the coding rules of the nonlinear feature coding technology are based on a multi-parameter combination of wound location, area and healing stage, namely: Φ=F(α·P pos +β·P area +γ·S stage ) Where F is the nonlinear mapping function, P pos is the wound location parameter, P area is the wound area parameter, S stage is the wound healing stage parameter, and α, β, and γ are dynamic weight coefficients.

3. The intelligent wound measurement and recording method according to claim 2, characterized in that: The wound measurement model includes the following structure: A temporal convolutional network module with dynamically adjusted dilation factors is introduced to capture the local temporal correlation of wound features through deformable convolution kernels and channel attention modules. A bidirectional recurrent neural network module with integrated attention mechanism is used to update the hidden state through a gated adaptive forgetting mechanism to capture the long-range spatiotemporal dependencies in the wound healing process; The update formula is: h t =σ(W h ·h t-1 +W x ·x t +b)Θtanh(W hh ·h t-1 +W hx ·x t +b h ) Where h t is the hidden state at the current moment, σ is the sigmoid function, W h 、W x 、W hh 、W hx is the weight matrix, b and b h is the bias term; Feature fusion layer based on dynamic weighted formula, where the weight coefficients are generated by the channel attention module: Where w k is the weight coefficient of the kth feature channel, z k is the feature significance index of the corresponding channel, W k and b k are the learnable parameter matrices and bias terms.

4. The intelligent wound measurement and recording method according to claim 3, characterized in that: The method of extracting the wound contour from the multi-dimensional image data by using an improved U-Net architecture combined with the HSV color space and the morphological gradient method includes: Based on the spatiotemporal correlation of the wound features, performing spatiotemporal alignment and feature enhancement on the multidimensional image data to obtain the enhanced multidimensional image; Converting the enhanced multidimensional image from the RGB color space to the HSV color space to obtain the multidimensional image in HSV color; Based on the multidimensional image of HSV color, extracting the wound candidate area by HSV color space segmentation according to the hue, saturation and brightness characteristics of the wound area; Optimizing the wound candidate region by using a morphological gradient method to eliminate noise, fill small holes, and smooth the boundaries to obtain the optimized wound candidate region; Perform feature extraction and semantic segmentation on the optimized wound candidate area by introducing an attention mechanism and a U-Net architecture with depthwise separable convolution, and output the wound contour of the wound candidate area; The wound contour of the wound candidate area is post-processed to output the final wound contour.

5. The intelligent wound measurement and recording method according to claim 4, characterized in that: Based on the wound contour, combined with the three-dimensional depth information, the sub-millimeter level wound volume measurement is performed through the ICP registration algorithm and the three-dimensional point cloud reconstruction technology to obtain wound volume data and extract wound volume parameters, including: Mapping each pixel on the wound contour to a corresponding position in three-dimensional space using a coordinate mapping algorithm to form preliminary point cloud data of the wound surface; Point cloud registration is performed using the ICP algorithm based on preselected matching points, and the registration error threshold is set to 0.1 mm; A multi-scale point cloud registration algorithm is introduced, and the following formula is used to perform multiple point cloud registrations on the wound contour at different scales to obtain the registered wound surface point cloud data; Multi-scale point cloud registration formula: Where, ICP r is the ICP registration result at the rth scale, is the weight corresponding to each scale; The α-shape algorithm is used to perform three-dimensional reconstruction on the registered wound surface point cloud data to generate a triangular mesh model of the wound surface; Based on the triangular mesh model, the wound volume was calculated using the following volume formula, and the wound volume-related parameters were extracted: Where V is the wound volume, x j 、y j 、x j is the three-dimensional coordinate of the jth point in the point cloud, and m is the number of wound volume point clouds.

6. The intelligent wound measurement and recording method according to claim 5, characterized in that: The method further comprises: Feeding the structured report back to the clinical system and binding it to the patient's biometrics, using an incremental learning framework combined with an elastic weight solidification algorithm to smooth the impact between new and old data; Establish a wound feature evolution map and continuously update the pathological feature database using online learning mechanisms; Set the dynamic memory playback buffer. The buffer size is calculated as follows: Where N is the current number of samples.

7. An intelligent wound measurement and recording system, characterized in that: include: An image acquisition module, used to obtain multi-dimensional image data and three-dimensional depth information of the wound area; A preprocessing module, configured to preprocess the multi-dimensional image data and three-dimensional depth information using hierarchical space partitioning and nonlinear feature encoding technology to generate basic wound information; A wound measurement module is used to construct a wound measurement model based on the basic wound information, extract wound features with spatiotemporal correlation, and extract the wound contour of the multi-dimensional image data using an improved U-Net architecture combined with HSV color space and morphological gradient method; A volume measurement module is used to measure the wound volume at the sub-millimeter level based on the wound contour and in combination with the three-dimensional depth information, using an ICP registration algorithm and three-dimensional point cloud reconstruction technology to obtain wound volume data and extract wound volume parameters; a report generation module, configured to integrate the basic wound information, the wound characteristics, the wound contour, and the wound volume parameter to generate a structured report according to predefined medical standards, wherein the structured report includes wound area, depth, volume, healing stage classification, and infection risk index; The feedback and learning module is used to feed the structured report back to the clinical system and bind it to the patient's biometrics, and adopt an incremental learning framework combined with an elastic weight solidification algorithm to smooth the impact between new and old data.

8. The intelligent wound measurement and recording system according to claim 7, characterized in that: The system further comprises: A wound classification module uses ResNet-152 as the base network for feature extraction and improves classification accuracy by initializing weights of a pre-trained model on a large medical image dataset. The wound healing prediction module is used to analyze time series measurement data through the LSTM network to predict the changing trends of wound area, volume and healing stage.

Images