Negative pressure closed drainage device for preventing infection of orthopedic postoperative wound

Abstract

The application discloses a negative pressure closed drainage device for preventing infection of a postoperative wound in orthopedics, which comprises a negative pressure pump for generating negative pressure; a drainage tube for connecting the wound and the negative pressure pump to draw out secretions of the wound by the negative pressure pump; a sealing film for covering the wound to prevent air and bacteria from entering; a filter for filtering the secretions of the wound to prevent backflow and pollution; a camera for collecting a digital image of the wound; an infection early warning system for judging whether the wound is infected and automatically generating a wound infection early warning prompt; and a voice player for playing the wound infection early warning prompt. Thus, the wound infection condition can be monitored in real time, and an early warning can be timely given, so that medical staff can make timely and effective disposal, thereby improving the wound healing rate and reducing the infection risk.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent drainage devices, and more particularly, to a negative pressure closed drainage device for preventing postoperative wound infection in orthopedics. BACKGROUND

[0002] With the wide application of orthopedic surgery, postoperative infection has become a common problem that plagues medical staff and patients. Postoperative wound infection not only prolongs the recovery time of patients and increases the treatment cost, but also can lead to serious complications and even endanger the lives of patients. Therefore, how to effectively prevent and manage postoperative wound infection has become one of the urgent problems in the field of orthopedic surgery.

[0003] Traditional wound treatment methods, such as regular replacement of gauze, can absorb wound secretions, but cannot effectively prevent bacteria from entering the wound and cannot monitor the recovery of the wound in real time. Therefore, an optimized solution is expected. SUMMARY

[0004] To solve the above technical problems, the present application is proposed. The present application provides a negative pressure closed drainage device for preventing postoperative wound infection in orthopedics, which comprises: a negative pressure pump for generating negative pressure; a drainage tube for connecting a wound and the negative pressure pump to draw out secretions of the wound by the negative pressure pump; a sealing film for covering the wound to prevent air and bacteria from entering; a filter for filtering secretions of the wound to prevent backflow and contamination; a camera for collecting digital images of the wound; an infection early warning system for judging whether the wound is infected and automatically generating a wound infection early warning prompt; and a voice player for playing the wound infection early warning prompt. In this way, the wound infection situation can be monitored in real time, and timely warnings can be issued to assist medical staff to make timely and effective disposal, thereby improving the wound healing rate and reducing the risk of infection.

[0005] In a first aspect, a negative pressure closed drainage device for preventing postoperative wound infection in orthopedics is provided, which comprises:

[0006] a negative pressure pump for generating negative pressure;

[0007] a drainage tube for connecting a wound and the negative pressure pump to draw out secretions of the wound by the negative pressure pump;

[0008] a sealing film for covering the wound to prevent air and bacteria from entering;

[0009] a filter for filtering secretions of the wound to prevent backflow and contamination;

[0010] a camera for collecting digital images of the wound;

[0011] an infection early warning system for determining whether the wound is infected and automatically generating a wound infection early warning prompt;

[0012] a voice player for playing the wound infection early warning prompt.

[0013] Compared with the prior art, the negative pressure closed drainage device for postoperative wound infection prevention in orthopedics provided by the application fully utilizes the wound digital image collected by the camera, and combines intelligent image processing means to perform image analysis on the wound digital image, so as to extract the cross-depth correlation fusion features between the shallow features and the deep features in the wound digital image to depict the real-time state of the wound, thereby determining whether the wound is infected, and automatically generating an early warning prompt when an abnormal condition is found. In this way, the voice player can convey the infection early warning information to relevant medical staff in the form of playing a voice prompt. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0015] Figure 1 a block diagram of the negative pressure closed drainage device for postoperative wound infection prevention in orthopedics according to the embodiment of the application.

[0016] Figure 2 a block diagram of the infection early warning system in the negative pressure closed drainage device for postoperative wound infection prevention in orthopedics according to the embodiment of the application.

[0017] Figure 3 a flowchart of the negative pressure closed drainage method for postoperative wound infection prevention in orthopedics according to the embodiment of the application.

[0018] Figure 4 an application scenario diagram of the negative pressure closed drainage device for postoperative wound infection prevention in orthopedics according to the embodiment of the application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the application will be described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only some of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0020] Unless otherwise defined, all technical and scientific terms used in the embodiments of the application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0021] In the embodiments of the application, it should be noted that unless otherwise specified and limited, the term "connection" should be understood broadly, for example, it can be an electrical connection, or a connection between two elements, it can be directly connected, or indirectly connected through an intermediate medium. For those skilled in the art, the specific meaning of the above-mentioned term can be understood according to the specific circumstances.

[0022] It should be noted that the terms "first", "second", "third" involved in the embodiments of the application are only to distinguish similar objects, and do not represent a specific order of the objects. Understandably, "first", "second", "third" can be interchanged in specific order or sequence as allowed. It should be understood that the objects distinguished by "first", "second", "third" can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than those illustrated or described herein.

[0023] After introducing the basic principles of the application, the various non-limiting embodiments of the application will be specifically introduced with reference to the accompanying drawings.

[0024] With the widespread application of orthopedic surgery, postoperative wound infection has become a common problem that plagues medical staff and patients. This infection not only prolongs the patient's recovery time and increases the treatment cost, but also can lead to serious complications, and even endanger life.

[0025] Effective prevention of postoperative wound infection is crucial, involving the following key measures:

[0026] Preoperative preparation: Patients should be thoroughly evaluated and prepared before surgery, including controlling potential sources of infection, optimizing nutritional status, and quitting smoking.

[0027] Sterile surgical technique: The operating room should maintain a sterile environment, and strictly follow the principles of sterile surgery, including wearing sterile gloves, masks and surgical gowns.

[0028] Antibiotic prophylaxis: The use of antibiotics before and during surgery helps to prevent infection.

[0029] Wound care: The postoperative wound should be regularly cleaned and dressed to remove bacteria and promote healing.

[0030] Despite precautions, postoperative wound infections can still occur. In this case, timely and effective management is crucial: monitor the patient for signs of infection, such as fever, wound pain, and swelling; determine the pathogen causing the infection and determine the appropriate antibiotic; select the appropriate antibiotic for treatment based on culture results and antibiotic sensitivity tests; for abscesses or infected wounds, drainage may be required to drain fluid and bacteria; in some cases, surgical debridement may be required to remove infected tissue.

[0031] Traditional wound treatment methods, such as regular gauze changes, have certain drawbacks: regular gauze changes can absorb wound exudates, but do not form an effective barrier to prevent bacteria from entering the wound from the outside. Traditional wound treatment methods require regular removal of gauze to inspect the wound, which not only causes pain to the patient but also increases the risk of infection, and this method cannot monitor the progress of wound healing in real time, which is not conducive to timely detection and treatment of wound infection. Traditional wound treatment methods require frequent gauze changes, which can cause pain and discomfort to the patient, affecting the patient's rest and recovery. If not handled properly, gauze changes can easily contaminate the wound and increase the risk of infection. Traditional wound treatment methods cannot provide a suitable healing environment for the wound, which is not conducive to wound healing.

[0032] In recent years, advanced technologies have played an increasingly important role in preventing and managing postoperative wound infections: surgical instruments and implants are coated with antibacterial coatings to reduce bacterial adhesion and infection risk. The use of negative pressure devices to remove fluid and bacteria from the wound promotes healing. Biofilms are protective layers formed by bacteria that make it difficult for antibiotics to penetrate. Biofilm inhibitors can disrupt biofilms and improve the effectiveness of antibiotics.

[0033] In recent years, negative pressure wound therapy has been widely used in wound treatment. This technology creates a negative pressure environment at the wound site, promoting blood circulation and accelerating wound healing, while also removing wound exudates and reducing the likelihood of infection. However, existing negative pressure wound therapy devices generally lack real-time monitoring and early warning capabilities, making it difficult to detect and address wound infections in a timely manner. This means that healthcare professionals cannot immediately learn about changes in the wound, including the presence of signs of infection. This results in ineffective management of infection issues.

[0034] In one embodiment of the present application, Figure 1 A block diagram of a negative pressure wound therapy device for preventing postoperative wound infections in orthopedics according to an embodiment of the present application. As shown in Figure 1As shown, the negative pressure closed drainage device for preventing postoperative wound infection in orthopedics according to the embodiment of the present application 100 comprises: a negative pressure pump 1 for generating negative pressure; a drainage tube 2 for connecting the wound and the negative pressure pump 1 to use the negative pressure pump 1 to suck out the exudate of the wound; a sealing film 3 for covering the wound to prevent air and bacteria from entering; a filter 4 for filtering the exudate of the wound to prevent backflow and pollution; a camera 5 for collecting the digital wound image of the wound; an infection early warning system 6 for judging whether the wound is infected and automatically generating a wound infection early warning prompt; and a voice player 7 for playing the wound infection early warning prompt.

[0035] The use method is as follows:

[0036] After orthopedic surgery, the drainage tube is inserted into the wound and fixed in place; the sealing film is covered on the wound and the edge is sealed with tape or other means; the other end of the drainage tube is connected to the negative pressure pump, the switch is turned on, and the negative pressure parameters are adjusted; the filter is placed in a suitable position to collect the wound exudate; the sealing film, filter and drainage tube are replaced regularly according to the doctor's order, and the wound condition is observed.

[0037] The device has the following advantages: it can effectively prevent postoperative wound infection in orthopedics and reduce the risk of complications; it can shorten the hospitalization time and reduce the medical expenses; it can improve the patient's comfort and reduce pain and discomfort; it can adapt to different types and sizes of wounds and is easy and convenient to operate.

[0038] The infection early warning system makes full use of the digital wound image collected by the camera, and combines intelligent image processing means to analyze the digital wound image to extract the cross-depth correlation fusion features between the shallow features and the deep features in the digital wound image to depict the real-time state of the wound, so as to judge whether the wound is infected and automatically generate an early warning prompt when an abnormal condition is found. In this way, the voice player can convey the infection early warning information to the relevant medical staff through the voice prompt.

[0039] Figure 2 The block diagram of the infection early warning system in the negative pressure closed drainage device for preventing postoperative wound infection in orthopedics according to the embodiment of the present application is shown in the figure. As shown in the figure, Figure 2As shown, the image acquisition module 110 is configured to acquire the wound digital image collected by the camera of the negative pressure closed drainage device; the image data transmission module 120 is configured to transmit the wound digital image to the background server through the data transmission module; the multi-scale feature analysis module 130 is configured to perform image multi-scale feature extraction on the wound digital image to obtain a wound multi-scale feature map on the background server; the monitoring result generation module 140 is configured to determine a monitoring result based on the wound multi-scale feature map on the background server; and the response and early warning module 150 is configured to generate a wound infection early warning prompt to the negative pressure closed drainage device in response to the monitoring result indicating that the wound is infected on the background server.

[0040] The image acquisition module 110 ensures that the camera can clearly collect the wound image, avoids image blur or distortion, and provides a high-quality wound image as an accurate basis for subsequent analysis. The image data transmission module 120 uses a stable data transmission method to ensure the timeliness and integrity of image transmission, quickly and safely transmits the wound image to the background server for timely analysis. The multi-scale feature analysis module 130 uses a suitable image feature extraction algorithm to extract multi-scale features that can reflect the wound state, and through multi-scale feature analysis, fully describes the appearance, texture, and shape of the wound, providing a reliable basis for wound infection monitoring. The monitoring result generation module 140 establishes an accurate wound infection monitoring model that can accurately determine whether the wound is infected based on the multi-scale feature map, timely detects wound infection, and provides support for clinical decision-making. The response and early warning module 150 provides clear and timely early warning prompts and effectively reminds medical staff to take appropriate measures. When the wound is infected, the early warning is issued in a timely manner to prompt medical staff to intervene in a timely manner and prevent the infection from worsening.

[0041] Based on this, in the technical solution of the present application, the specific coding process of the infection early warning system includes: first, acquiring the wound digital image collected by the camera of the negative pressure closed drainage device; and transmitting the wound digital image to the background server through the data transmission module. Here, the wound digital image can be collected by the camera to realize real-time monitoring of the wound. In this way, the wound digital image is transmitted to the background server, and medical staff can remotely access the wound digital image while using the computing power of the background server to further analyze and process the wound digital image. In addition, transmitting the wound digital image to the background server through the data transmission module can also securely store and backup the wound digital image, avoid data loss or damage, and ensure the integrity and reliability of the image data.

[0042] In an embodiment of the present application, the multi-scale feature analysis module comprises: a deep-shallow feature mining unit configured to mine deep-shallow features of the wound digital image to obtain a wound shallow feature map and a wound deep feature map; and a deep-shallow feature fusion unit configured to fuse the wound shallow feature map and the wound deep feature map to obtain the wound multi-scale feature map.

[0043] In view of the fact that the wound digital image may contain various types of noise due to factors such as shooting conditions and equipment quality, which may affect the analysis of the wound state. In order to eliminate or alleviate these problems, in the technical solution of the present application, it is expected that the wound digital image is pre-processed on the background server to obtain a pre-processed wound digital image, so that the pre-processed wound digital image is clearer and easier to analyze and process.

[0044] In an embodiment of the present application, the deep-shallow feature mining unit comprises: an image preprocessing subunit configured to pre-process the wound digital image on the background server to obtain a pre-processed wound digital image; a shallow feature capturing subunit configured to pass the pre-processed wound digital image through an image shallow feature extractor based on a first convolutional neural network model to obtain the wound shallow feature map; and a deep feature capturing subunit configured to pass the wound shallow feature map through an image deep feature extractor based on a second convolutional neural network model to obtain the wound deep feature map, wherein the second convolutional neural network model and the first convolutional neural network model are cascaded.

[0045] Then, at the background server, the pre-processed wound digital image is inputted into an image shallow feature extractor based on a first convolutional neural network model to obtain a wound shallow feature map; meanwhile, at the background server, the wound shallow feature map is inputted into an image deep feature extractor based on a second convolutional neural network model to obtain a wound deep feature map, wherein the second convolutional neural network model and the first convolutional neural network model are cascaded. Here, by extracting the shallow features of the pre-processed wound digital image, the edge information and texture information about the wound can be captured, which are of great significance to the description of the size and shape of the wound and can help to preliminarily understand the surface features and basic appearance of the wound. By extracting the deep features of the pre-processed wound digital image, more advanced and abstract features can be further extracted from the shallow features to capture more complex patterns and semantic information in the image, such as the healing progress and infection signs of the wound. It is worth mentioning that the cascading of the second convolutional neural network model and the first convolutional neural network model can facilitate the transmission of information from the first convolutional neural network model to the second convolutional neural network model, so that the deep feature extractor can learn more abstract feature representations from the shallow features, thereby enabling the model to better understand and represent the real-time state of the wound.

[0046] In a specific embodiment of the present application, the deep-shallow feature fusion unit comprises a feature enhancement subunit configured to input the wound shallow feature map and the wound deep feature map into a feature map enhancer based on a reparameterization network at the background server to obtain an enhanced wound shallow feature map and an enhanced wound deep feature map; and a cross-depth correlation fusion subunit configured to input the enhanced wound shallow feature map and the enhanced wound deep feature map into a cross-depth feature correlation analyzer based on a class attention mechanism at the background server to obtain the wound multi-scale feature map.

[0047] Although the problem of noise in wound digital images can be alleviated to some extent by image preprocessing, the structure and characteristics of wounds can be complex, and the wound morphology can vary significantly between different individuals and different parts, which can also cause some features to be sparse or missing in the images. Therefore, in the technical solution of the present application, the background server inputs the wound shallow feature map and the wound deep feature map into a feature map enhancer based on a reparameterization network to learn the representation of more representative and highly abstracted features in the wound shallow feature map and the wound deep feature map, respectively, thereby obtaining an enhanced wound shallow feature map and an enhanced wound deep feature map. Specifically, the feature map enhancer based on the reparameterization network guides and normalizes the feature distribution by introducing a prior distribution, such as a Gaussian random distribution. In this way, by introducing the reparameterization technique, the wound shallow feature map and the wound deep feature map are made to have randomness after reconstruction, thereby cleverly achieving the effect of data augmentation in a high-dimensional feature space, so that the subsequent model can make full use of all available information.

[0048] In one specific embodiment of the present application, the feature enhancement subunit is configured to process the wound shallow feature map according to the following reparameterization formula to obtain the enhanced wound shallow feature map:

[0049]

[0050] wherein, is the mean of the wound shallow feature map, is the variance of the wound shallow feature map, ∈ θ is the θth value randomly sampled from the Gaussian distribution, is the θth feature value in the enhanced wound shallow feature map.

[0051] Further, the enhanced shallow wound feature map and the enhanced deep wound feature map are input into a cross-depth feature association analyzer based on a class attention mechanism to obtain a wound multi-scale feature map. That is, the cross-depth feature association analyzer based on the class attention mechanism fuses feature information with different emphases and different depths, and introduces an attention mechanism to enhance the attention to important features. Specifically, the specific implementation process of the cross-depth feature association analyzer based on the class attention mechanism is to perform global average pooling on the enhanced deep wound feature map as high-level features in the channel dimension. At this time, the high-dimensional feature map is compressed into a vector representation, and the vector representation has a global receptive field of the enhanced deep wound feature map. Then, the vector representation is fused with the enhanced shallow wound feature map as attention information to guide the shallow information in the enhanced shallow wound feature map to restore semantic class information, so as to obtain the wound multi-scale feature map. In this way, the wound multi-scale feature map has more rich feature representation.

[0052] In an embodiment of the present application, the cross-depth association fusion subunit is configured to: perform global average pooling on the enhanced deep wound feature map along the channel dimension to obtain an attention feature vector; pass the attention feature vector through a fully connected layer to obtain an attention encoding feature vector; perform weighted multiplication on the enhanced shallow wound feature map with each feature value in the attention encoding feature vector as a weight to obtain an attention adjusted feature map; and perform position-wise addition processing on the attention adjusted feature map and the enhanced deep wound feature map to obtain the wound multi-scale feature map.

[0053] Then, the wound multi-scale feature map is input into a wound infection monitor based on a classifier on the background server to obtain a monitoring result, the monitoring result being used to indicate whether the wound is infected; and in response to the monitoring result indicating that the wound is infected, a wound infection early warning prompt is generated to the negative pressure closed drainage device.

[0054] In an embodiment of the present application, the monitoring result generation module is configured to: input the wound multi-scale feature map into a wound infection monitor based on a classifier on the background server to obtain the monitoring result, the monitoring result being used to indicate whether the wound is infected.

[0055] In an embodiment of the present application, the negative pressure closed drainage device for preventing postoperative wound infection in orthopedics further comprises a training module for training the image shallow feature extractor based on the first convolutional neural network model, the image deep feature extractor based on the second convolutional neural network model, the feature map enhancer based on the reparameterization network, the cross-depth feature correlation analyzer based on the class attention mechanism, and the wound infection monitor based on the classifier. The training module comprises: a training data acquisition unit configured to acquire training data, the training data comprising training wound digital images and true values of whether the wounds are infected; a training data transmission unit configured to transmit the training wound digital images to a background server through a data transmission module; a training image preprocessing unit configured to perform image preprocessing on the training wound digital images to obtain training preprocessed wound digital images at the background server; a training image shallow feature extraction unit configured to obtain training wound shallow feature maps by inputting the training preprocessed wound digital images to the image shallow feature extractor based on the first convolutional neural network model at the background server; a training image deep feature extraction unit configured to obtain training wound deep feature maps by inputting the training wound shallow feature maps to the image deep feature extractor based on the second convolutional neural network model at the background server, wherein the second convolutional neural network model and the first convolutional neural network model are cascaded; a training feature map enhancement unit configured to input the training wound shallow feature maps and the training wound deep feature maps into the feature map enhancer based on the reparameterization network to obtain training enhanced wound shallow feature maps and training enhanced wound deep feature maps at the background server; a training feature correlation analysis unit configured to input the training enhanced wound shallow feature maps and the training enhanced wound deep feature maps into the cross-depth feature correlation analyzer based on the class attention mechanism to obtain training wound multi-scale feature maps at the background server; a training classification unit configured to obtain a classification loss function value by inputting the training wound multi-scale feature maps into the wound infection monitor based on the classifier at the background server; and a training unit configured to train the image shallow feature extractor based on the first convolutional neural network model, the image deep feature extractor based on the second convolutional neural network model, the feature map enhancer based on the reparameterization network, the cross-depth feature correlation analyzer based on the class attention mechanism, and the wound infection monitor based on the classifier using the classification loss function value, wherein in each iteration of the training, a training wound multi-scale feature vector obtained by unfolding the training wound multi-scale feature maps is aggregated and optimized.

[0056] In the technical solution, the training enhanced shallow wound feature map and the training enhanced deep wound feature map respectively express different scale and different depth enhanced image semantic features of the training pre-processed wound digital image based on the first convolutional neural network model and the second convolutional neural network model, and after the training enhanced shallow wound feature map and the training enhanced deep wound feature map are input into the cross-depth feature association analyzer based on the class attention mechanism, the classification probability of the training enhanced shallow wound feature map and the training enhanced deep wound feature map can be weighted and enhanced, which not only emphasizes the image semantic features with more significant class probability representation, but also further increases the image semantic feature expression difference between the training enhanced shallow wound feature map and the training enhanced deep wound feature map in scale and depth. In this way, the training wound multi-scale feature map may have a problem of insufficient aggregation of cross-scale and cross-depth image semantic feature distribution, which affects the convergence speed of the training wound multi-scale feature map when performing classification regression, thereby affecting the training speed of the model.

[0057] Therefore, the applicant of the present application preferably performs feature aggregation optimization on the training wound multi-scale feature vector, denoted as V, obtained after the training wound multi-scale feature map is unfolded when the training wound multi-scale feature vector is classified and iteratively trained by the classifier, and the optimization is specifically represented as: in each iteration of the training, the training wound multi-scale feature vector obtained after the training wound multi-scale feature map is unfolded is aggregated and optimized to obtain an optimized training wound multi-scale feature vector; wherein the optimization formula is:

[0058]

[0059] wherein, is the square of the 1-norm of the training wound multi-scale feature vector V obtained after the training wound multi-scale feature map is unfolded, ||V||2 -1 / 2 is the square root of the 2-norm of the training wound multi-scale feature vector V, L is the length of the training wound multi-scale feature vector V, and ε is a scaling hyperparameter, V is the training wound multi-scale feature vector, v' i is the feature value of the optimized training wound multi-scale feature vector, v i is the feature value of the training wound multi-scale feature vector, and log represents a logarithmic function with base 2.

[0060] Here, the low-rank structure based on the norm of the training wound multi-scale feature vector V is expressed as the voting cluster of the aggregation of the eigenvalue set of the training wound multi-scale feature vector V, each eigenvalue of the training wound multi-scale feature vector V is subjected to canonicalization voting with respect to the information perception framework, to map the eigenvalues belonging to the same regression class to a similar set of local canonical coordinates through the regression representation of the direction and scale of the aggregation of the feature distribution, thereby improving the aggregation effect of the feature set of the training wound multi-scale feature vector V, improving the classification regression effect of the training wound multi-scale feature vector V through the classifier, that is, improving the training efficiency of the classifier and the accuracy of the classification result.

[0061] In summary, the negative pressure closed drainage device 100 for postoperative wound infection prevention in orthopedics according to the embodiments of the present application is illustrated, which fully utilizes the wound digital image collected by the camera and combines intelligent image processing means to perform image analysis on the wound digital image, so as to extract the cross-depth association and fusion features between the shallow features and the deep features in the wound digital image to depict the real-time state of the wound, thereby judging whether the wound is infected, and automatically generating a warning prompt when an abnormal condition is found. In this way, the voice player can convey the infection warning information to the relevant medical staff by playing a voice prompt.

[0062] In an embodiment of the present application, Figure 3 The flowchart of the negative pressure closed drainage method for postoperative wound infection prevention in orthopedics according to the embodiments of the present application is shown in FIG. 2. As shown in FIG. 2, the negative pressure closed drainage method for postoperative wound infection prevention in orthopedics includes the following steps. Figure 3 210, using a negative pressure pump to generate negative pressure; 220, using a drainage tube to connect the wound and the negative pressure pump to use the negative pressure pump to extract the exudate of the wound; 230, using a sealing film to cover the wound to prevent air and bacteria from entering; 240, using a filter to filter the exudate of the wound to prevent backflow and contamination; 250, using a camera to collect wound digital images of the wound; 260, using an infection warning system to judge whether the wound is infected and automatically generate a wound infection warning prompt; 270, using a voice player to play the wound infection warning prompt.

[0063] Those skilled in the art can understand that the specific operations of each step in the above-mentioned negative pressure closed drainage method for postoperative wound infection prevention in orthopedics have been described in detail above with reference to the description of the negative pressure closed drainage device for postoperative wound infection prevention in orthopedics in Figures 1 to 2 Therefore, the repeated description thereof will be omitted.

[0064] Figure 4 The application scenario diagram of the negative pressure closed drainage device for postoperative wound infection prevention in orthopedics according to the embodiments of the present application is shown in FIG. 1.Figure 4 As shown in the application scenario, first, the wound digital image collected by the camera of the negative pressure closed drainage device is acquired (for example, as shown in C) in the above figure; then, the acquired wound digital image is input into the server in which the negative pressure closed drainage algorithm for orthopedic postoperative wound infection prevention is deployed (for example, as shown in S) in the above figure, wherein the server can process the wound digital image based on the negative pressure closed drainage algorithm for orthopedic postoperative wound infection prevention to generate the wound infection early warning prompt to the negative pressure closed drainage device. Figure 4 Figure 4

[0065] The above describes the basic principles of the present application in combination with specific embodiments, but it should be pointed out that the advantages, advantages, effects and the like mentioned in the present application are only examples and not limitations, and these advantages, advantages, effects and the like cannot be considered as the must-have of each embodiment of the present application. In addition, the above specific details are only for the purpose of example and for the purpose of understanding, and the above details do not limit the present application to the must-use specific details.

[0066] The above description of the disclosed aspects is provided so that any person skilled in the art can make or use the present application. Various modifications to these aspects will be apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0067] Finally, it should also be noted that in this paper, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or terminal device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or terminal device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or terminal device including the element.

[0068] The above description has been given for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain modifications, alterations, changes, additions and sub-combinations thereof.​​

Claims

1. A negative pressure closed drainage device for preventing infection of orthopedic postoperative wound, characterized in that, The application relates to a negative pressure wound therapy system, comprising: a negative pressure pump for generating negative pressure; a drainage tube for connecting a wound and the negative pressure pump to draw exudates of the wound by the negative pressure pump; a sealing film for covering the wound to prevent air and bacteria from entering; a filter for filtering the exudates of the wound to prevent backflow and contamination; a camera for capturing a digital wound image of the wound; an infection early warning system for determining whether the wound is infected and automatically generating a wound infection early warning prompt; a voice player for playing the wound infection early warning prompt; wherein the infection early warning system comprises: an image acquisition module for acquiring the digital wound image captured by the camera of the negative pressure wound therapy device; an image data transmission module for transmitting the digital wound image to a background server through a data transmission module; a multi-scale feature analysis module for performing image multi-scale feature extraction on the digital wound image to obtain a wound multi-scale feature map in the background server; a monitoring result generation module for determining a monitoring result based on the wound multi-scale feature map in the background server; a response and early warning module for generating the wound infection early warning prompt to the negative pressure wound therapy device in response to the monitoring result being that the wound is infected in the background server; the multi-scale feature analysis module comprises: a deep and shallow feature mining unit for performing deep and shallow feature mining on the digital wound image to obtain a wound shallow feature map and a wound deep feature map in the background server; a deep and shallow feature fusion unit for fusing the wound shallow feature map and the wound deep feature map to obtain the wound multi-scale feature map in the background server; the deep and shallow feature fusion unit comprises: a feature enhancement subunit for inputting the wound shallow feature map and the wound deep feature map into a feature map enhancer based on a reparameterization network to obtain an enhanced wound shallow feature map and an enhanced wound deep feature map in the background server; a cross-depth correlation fusion subunit for inputting the enhanced wound shallow feature map and the enhanced wound deep feature map into a cross-depth feature correlation analyzer based on an attention-like mechanism to obtain the wound multi-scale feature map in the background server; the feature enhancement subunit is configured to: process the wound shallow feature map by the following reparameterization formula to obtain the enhanced wound shallow feature map; wherein the reparameterization formula is: in, It is the mean value of the superficial feature map of the wound. It is the variance of the superficial feature map of the wound. The first sample obtained from a Gaussian distribution One value, It is the first in the enhanced superficial wound feature map. Each feature value.

2. The negative pressure wound closure device for preventing infection of postoperative wound in orthopedics of claim 1, wherein, the deep and shallow feature mining unit comprises: an image preprocessing subunit for performing image preprocessing on the digital wound image to obtain a preprocessed digital wound image in the background server; a shallow feature capturing subunit for inputting the preprocessed digital wound image into an image shallow feature extractor based on a first convolutional neural network model to obtain the wound shallow feature map in the background server; a deep feature capturing subunit, configured to, at the background server, obtain the wound deep feature map by inputting the wound shallow feature map into an image deep feature extractor based on a second convolutional neural network model, wherein the second convolutional neural network model and the first convolutional neural network model are cascaded.

3. The negative pressure wound closure device for preventing infection of postoperative wound in orthopedics of claim 2, wherein, the cross-depth feature association fusion subunit is configured to: perform global average pooling on the enhanced wound deep feature map along the channel dimension to obtain an attention feature vector; input the attention feature vector into a fully connected layer to obtain an attention encoded feature vector; perform weighted multiplication on the enhanced wound shallow feature map by using each feature value in the attention encoded feature vector as a weight to obtain an attention adjusted feature map; perform position-wise addition processing on the attention adjusted feature map and the enhanced wound deep feature map to obtain the wound multi-scale feature map.

4. The negative pressure wound closure device for preventing infection of postoperative wound in orthopedics of claim 3, wherein, the monitoring result generation module is configured to: at the background server, obtain the monitoring result by inputting the wound multi-scale feature map into a wound infection monitor based on a classifier, wherein the monitoring result is used to indicate whether the wound is infected.

5. The negative pressure wound closure device for preventing infection of postoperative wound in orthopedics of claim 4, wherein, The training module is further configured to train the image shallow feature extractor based on the first convolutional neural network model, the image deep feature extractor based on the second convolutional neural network model, the feature map enhancer based on the reparameterization network, the cross-depth feature association analyzer based on the class attention mechanism, and the wound infection monitor based on the classifier.

6. The negative pressure wound closure device for preventing infection of postoperative wound in orthopedics of claim 5, wherein, The training module comprises: a training data acquisition unit configured to acquire training data, wherein the training data comprises training wound digital images and true values of whether the wound is infected; a training data transmission unit configured to transmit the training wound digital images to the background server through a data transmission module; a training image preprocessing unit configured to perform image preprocessing on the training wound digital images to obtain training preprocessed wound digital images at the background server; a training image shallow feature extraction unit configured to obtain training wound shallow feature maps by inputting the training preprocessed wound digital images into the image shallow feature extractor based on the first convolutional neural network model at the background server; a training image deep feature extraction unit configured to obtain training wound deep feature maps by inputting the training wound shallow feature maps into the image deep feature extractor based on the second convolutional neural network model at the background server, wherein the second convolutional neural network model and the first convolutional neural network model are cascaded; a training feature map enhancement unit configured to input the training wound shallow feature maps and the training wound deep feature maps into the feature map enhancer based on the reparameterization network to obtain training enhanced wound shallow feature maps and training enhanced wound deep feature maps at the background server; a training feature association analysis unit configured to input the training enhanced wound shallow feature maps and the training enhanced wound deep feature maps into the cross-depth feature association analyzer based on the class attention mechanism to obtain training wound multi-scale feature maps at the background server; and a training classification unit, configured to pass the training wound multi-scale feature map through the classifier-based wound infection monitor to obtain a classification loss function value at the background server; a training unit, configured to train the first convolutional neural network model-based image shallow layer feature extractor, the second convolutional neural network model-based image deep layer feature extractor, the reparameterization network-based feature map enhancer, the class attention mechanism-based cross-depth feature correlation analyzer and the classifier-based wound infection monitor according to the classification loss function value, wherein, in each round of iteration of the training, a training wound multi-scale feature vector obtained by unfolding the training wound multi-scale feature map is aggregated and optimized; in each round of iteration of the training, when the training wound multi-scale feature vector obtained by unfolding the training wound multi-scale feature map is classified and regressed by the classifier, the training wound multi-scale feature vector is aggregated and optimized, and the aggregation and optimization of the training wound multi-scale feature vector is specifically represented as: in each round of iteration of the training, the training wound multi-scale feature vector obtained by unfolding the training wound multi-scale feature map is aggregated and optimized according to the following optimization formula to obtain an optimized training wound multi-scale feature vector; wherein, the optimization formula is: in, It is the training wound multi-scale feature vector obtained after expanding the training wound multi-scale feature map. The square of the 1 norm, It is the training wound multi-scale feature vector The reciprocal of the square root of the 2-norm, It is the training wound multi-scale feature vector The length, and It's a scaling hyperparameter. It is the multi-scale feature vector of the training wound. These are the feature values ​​of the optimized trained multi-scale feature vector of the wound. These are the feature values ​​of the multi-scale feature vector of the training wound. This represents the logarithmic function with base 2.

Images