Visual identification-based anal fistula cleaning device and system

Abstract

The invention provides an anal fistula cleaning device and system based on visual identification, and relates to the technical field of medical instruments, the device comprises an endoscopic device used for establishing a first focus image and a second focus image; the image processing module is used for extracting image features including lesion features and lesion associated features; the interaction module is used for acquiring selection feedback of the user; and the cleaning guidance module is used for performing cleaning management on the user according to the cleaning scheme. The technical problems of poor positioning precision and poor user adaptability in the anal fistula cleaning process are solved, and the technical effects of reducing operation errors and improving the cleaning effect and the user experience are achieved by combining precise positioning of the endoscopic equipment, an image processing technology and a user feedback mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of medical appliances, and particularly to an anal fistula cleaning device and system based on visual identification. Background Art

[0002] As a common anorectal disease, anal fistula brings great physical pain and psychological pressure to patients. The treatment of anal fistula usually requires cleaning and nursing to prevent infection and promote wound healing. However, there are many technical challenges in the process of anal fistula cleaning, especially in the precise positioning of the lesion location, the matching of the cleaning posture, and the guidance of user operation. Traditional anal fistula cleaning methods mostly rely on doctors' experience and manual operation, which have certain operation errors, resulting in unstable cleaning effects, and may even cause discomfort and secondary infection to patients. During the cleaning process, accurate lesion positioning, reasonable cleaning postures, and personalized operation guidance are crucial for ensuring the cleaning effect. Existing anal fistula cleaning devices usually lack a real-time feedback mechanism and cannot be dynamically adjusted according to the patient's body posture, lesion characteristics, and actual conditions during the cleaning process, resulting in problems such as low cleaning efficiency and poor user compliance. Summary of the Invention

[0003] This application provides an anal fistula cleaning device and system based on visual identification, which is used to solve the technical problems of poor positioning accuracy and user adaptability during the anal fistula cleaning process. By combining the precise positioning of endoscopic equipment, image processing technology, and user feedback mechanism, the technical effects of reducing operation errors, improving cleaning effects, and user experience are achieved.

[0004] In view of the above problems, this application provides an anal fistula cleaning device and system based on visual identification.

[0005] The present application provides an anal fistula cleaning device based on a visual identifier. The device includes: an endoscope device, which is configured to, after receiving a start instruction, locate a lesion area according to an auxiliary positioning signal, automatically adjust the device position of the endoscope device, perform external image acquisition, establish a first lesion image, and when the endoscope device is placed in the fistula tract, perform internal image acquisition and establish a second lesion image; an image processing module, which is configured to perform image enhancement on the first lesion image and the second lesion image and then extract image features, where the image features include lesion features and lesion-related features; an interaction module, which is configured to display the cleaning holding postures in a general holding posture library to the user through a display, and perform display order sorting based on the adaptation matching degree between the user body posture, the lesion state and the holding posture library, and obtain the user's selection feedback; a cleaning guidance module, which is configured to input the lesion features, the lesion-related features and the selection feedback into a cleaning response model to generate a cleaning plan, where the cleaning plan includes a cleaning control plan and a user action guidance plan, and perform cleaning management of the user according to the cleaning plan.

[0006] The present application also provides an anal fistula cleaning system based on a visual identifier. The system is used for an anal fistula cleaning device based on a visual identifier. The system includes: a memory, which is configured to store executable instructions; and a processor, which is configured to execute the executable instructions stored in the memory.

[0007] One or more technical solutions provided in the present application have at least the following technical effects or advantages: After receiving a start instruction, locate a lesion area according to an auxiliary positioning signal, automatically adjust the device position of the endoscope device, perform external image acquisition, establish a first lesion image, and when the endoscope device is placed in the fistula tract, perform internal image acquisition and establish a second lesion image; perform image enhancement on the first lesion image and the second lesion image and then extract image features, where the image features include lesion features and lesion-related features; display the cleaning holding postures in a general holding posture library to the user through a display, and perform display order sorting based on the adaptation matching degree between the user body posture, the lesion state and the holding posture library, and obtain the user's selection feedback; input the lesion features, the lesion-related features and the selection feedback into a cleaning response model to generate a cleaning plan, where the cleaning plan includes a cleaning control plan and a user action guidance plan, and perform cleaning management of the user according to the cleaning plan. The technical effect of reducing operation errors, improving the cleaning effect and the user experience is achieved by combining the precise positioning of the endoscope device, image processing technology and the user feedback mechanism. Description of the Drawings

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0009] Figure 1 Schematic structural diagram of an anal fistula cleaning device based on visual identification provided by an embodiment of the present application.

[0010] Figure 2 Schematic flow diagram of a cleaning guidance module in an anal fistula cleaning device based on visual identification provided by an embodiment of the present application.

[0011] Explanation of reference numerals: endoscope device 11, image processing module 12, interaction module 13, cleaning guidance module 14. Detailed implementation manners

[0012] The present application provides an anal fistula cleaning device and system based on visual identification, which is used to solve the technical problems of poor positioning accuracy and poor user adaptability in the anal fistula cleaning process. By combining the precise positioning of the endoscope device, image processing technology and user feedback mechanism, the technical effects of reducing operation errors, improving the cleaning effect and user experience are achieved.

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0014] Embodiment 1, as Figure 1 shown, the present application provides an anal fistula cleaning device based on visual identification, and the device includes: An endoscope device 11, configured to, after receiving a start instruction, perform lesion area positioning according to an auxiliary positioning signal, automatically adjust the device position of the endoscope device 11, perform external image acquisition, establish a first lesion image, and when the endoscope device 11 is placed in the fistula tract, perform internal image acquisition and establish a second lesion image.

[0015] Specifically, after receiving the start instruction, the endoscopic device 11 determines the location of the lesion area according to the auxiliary positioning signal and automatically adjusts its relative position (at this time, the endoscopic device 11 is in the external environment) to ensure that the device is aligned with the target area. After the endoscopic device 11 is aligned with the lesion area, it will collect external images of the lesion area to generate the first lesion image, which is used to present the external situation of the lesion; when the endoscopic device 11 is placed inside the fistula, it will perform internal image collection to generate the second lesion image, which is used to present the internal structure and characteristics of the deeper lesion. This process relies on precise image collection and processing to ensure that the detailed information of the lesion can be comprehensively recorded, providing accurate data support for the subsequent cleaning plan design and treatment decision-making.

[0016] The image processing module 12 is used to enhance the first lesion image and the second lesion image and then extract image features, where the image features include lesion features and lesion-related features.

[0017] Specifically, the image processing module 12 enhances the first lesion image and the second lesion image through steps such as global image recognition, general enhancement parameter matching, and image attention recognition to improve the clarity and detail performance of the images, ensuring more accurate identification of the features of the lesion area. After image enhancement, the image processing module 12 analyzes the important features in the enhanced images and extracts key information to form image features. These extracted image features include basic lesion features such as the specific shape, size, and location of the lesion, and also include lesion-related features such as the relationship between the lesion and the surrounding tissues, such as the edge of the lesion and the density change of the surrounding tissues. The feature extraction process is mainly realized through technologies such as image segmentation, edge detection, and region growing algorithms. That is, edge detection algorithms (such as Canny or Sobel algorithms) are used to identify the edge of the lesion, the region growing method is used to extract the shape and size of the lesion, and then based on methods such as gray value and texture analysis, the relationship features between the lesion and the surrounding tissues are extracted. Through these processing steps, the detailed information of the lesion can be accurately obtained, providing a necessary basis for the subsequent cleaning plan design.

[0018] In a possible implementation, as Figure 2 shown, the image processing module 12 is further used for: Performing global image recognition on the first lesion image and the second lesion image to establish a global image recognition result; using the global image recognition result for general enhancement parameter matching, and performing initial enhancement on the first lesion image and the second lesion image according to the enhancement parameter matching result to establish an initial enhancement result; performing image attention recognition on the initial enhancement result, and performing additional enhancement based on the image attention recognition result to establish an additional enhancement result, and using the additional enhancement result as the enhanced image to perform feature extraction.

[0019] Optionally, first, global image recognition is performed on the first lesion image and the second lesion image to comprehensively evaluate the key features and overall structure in the images. The global image recognition process usually relies on a convolutional neural network (CNN) to identify the overall features of the lesion area, such as morphology, size, edge, position, etc., so as to generate global image recognition results. These results provide basic data for subsequent image enhancement and feature extraction. The convolutional neural network used for global image recognition is pre-trained with historical external lesion images, historical internal lesion images, and historical global image features. The training process includes steps such as forward propagation, loss calculation, backpropagation, and parameter update. Subsequently, based on the global image recognition results, the image processing module 12 will select a set of suitable general enhancement parameters. These general enhancement parameters include contrast enhancement, brightness adjustment, sharpening, color balance, noise suppression, etc. Usually, these enhancement parameters are selected through preset rules or machine learning models. For example, when using rules to determine parameters, if the error between the contrast of the global image recognition result and the standard contrast is not within the corresponding error tolerance range, the contrast parameter with the corresponding gradient will be selected according to the deviation from the error tolerance range. When using a machine learning model (such as a neural network) to determine parameters, the historical enhancement parameters corresponding to each historical global image feature will be extracted from the historical library, and through the same training process as above, it will gradually learn how to map the required enhancement parameters from the global image features. After that, according to these matched enhancement parameters, the initial enhancement of the first lesion image and the second lesion image is performed, aiming to optimize the visual effect of the images, make the lesion features more prominent, and thus establish the initial enhancement result. After the initial enhancement, image focus recognition is performed on the initial enhancement result, that is, the corresponding positions in the initial enhancement result are marked according to the preset ROI marking rules. This preset ROI (Region of Interest) marking rule is designed according to expert decisions, anatomical structures, and diagnostic requirements, and is used to guide the automatic recognition and marking of the areas that need to be focused on, such as features areas like anal fistula tracts, the opening parts of anal fistula tracts, and anal fistula openings. After the marking of the initial enhancement result is completed according to the ROI marking rules, the marked areas are segmented out through an image segmentation algorithm to form the image focus recognition result. Based on the obtained image focus recognition result, additional enhancement is also performed to obtain the additional enhancement result. The additional enhancement is mainly to perform further optimization processing on the focused areas. For example, through means such as local contrast enhancement, detail sharpening, and noise reduction, the details of these focused areas are made clearer. The method for obtaining the enhancement parameters of the additional enhancement is the same as above. Finally, the additional enhancement result is used as the enhanced image, and the aforementioned feature extraction is performed to extract the lesion features (such as shape, size, texture, etc.) in the enhanced image, providing accurate data support for the design of subsequent cleaning solutions.

[0020] The interaction module 13 is used to display the cleaning holding postures in the general holding posture library to the user through a display, and sort the display order based on the adaptation matching degree between the user's body posture, lesion state and the holding posture library, and obtain the selection feedback of the user.

[0021] Specifically, the function of the interaction module 13 is to extract the preset cleaning holding postures from the general holding posture library, and compare the reference body posture features, reference lesion state features corresponding to each cleaning holding posture with the user's body posture features (such as sitting posture, standing posture, bending degree, etc.) and lesion state features (such as the size, position, type, etc. of the lesion). Then, the cosine similarity is used to quantitatively distinguish the reference features and user features of the cleaning holding postures, and calculate the matching degree between the user's current state and each cleaning holding posture; Subsequently, the interaction module 13 will sort the cleaning holding postures in descending order of the matching degree, and present them to the user in sequence through the display for the user to view and select. The user can select the most suitable posture according to their own body posture and the specific situation of the lesion. The interaction module 13 will receive and record the selection feedback of the user for subsequent cleaning management and personalized guidance.

[0022] The cleaning guidance module 14 is used to input the lesion features, the lesion associated features and the selection feedback into a cleaning response model to generate a cleaning plan, where the cleaning plan includes a cleaning control plan and a user action guidance plan, and perform cleaning management on the user according to the cleaning plan.

[0023] Specifically, the main function of the cleaning guidance module 14 is to transmit the obtained lesion features, lesion associated features, and the selection feedback of the user during the interaction process to the cleaning response model. After receiving these data, the cleaning response model will, based on the lesion features and lesion associated features, generate a personalized cleaning control plan through the internal partition processing layer and cleaning parameter matching layer, and will also, based on the cleaning control plan and the selection feedback, generate a personalized user action guidance plan through the internal user action guidance layer; Subsequently, by associating the generated cleaning control plan and user action guidance plan, a cleaning plan for cleaning the user is generated. Among them, the cleaning control plan includes parameters such as appropriate cleaning liquid parameters, cleaning liquid flow rate, cleaning liquid injection pressure, etc., to ensure that the cleaning operation can be carried out accurately and efficiently. The user action guidance plan provides specific posture guidance and action suggestions to help the user maintain the correct posture during cleaning and avoid affecting the cleaning effect or causing discomfort due to improper operation; Finally, the cleaning guidance module 14 performs cleaning management on the user through this cleaning plan to ensure that the cleaning process meets personalized requirements and improves the cleaning effect and user experience.

[0024] In a possible implementation manner, the cleaning guidance module 14 is used for: Invoke the partition processing layer in the cleaning response model, and use the partition processing layer to perform partition analysis based on the lesion features and the lesion-associated features, and establish partition identifiers, where the partition identifiers include a high-inflammation partition identifier, a pus accumulation partition identifier, and a narrow fistula partition identifier; activate the cleaning parameter matching layer in the cleaning response model, synchronize the partition identifiers to the cleaning parameter matching layer, and then perform sequential cleaning parameter matching according to the cleaning priority to establish a sequential cleaning parameter matching result, where the sequential cleaning parameter matching result includes cleaning liquid parameters, pressure parameters, flow rate parameters, and nozzle angle parameters; organize the sequential cleaning parameter matching result into a cleaning control plan for output.

[0025] Optionally, in the cleaning response model, the cleaning response model transmits the received lesion features and lesion-associated features to the partition processing layer. This partition processing layer is mainly responsible for performing partition analysis on the lesion area based on the lesion features and lesion-associated features. The key to this process is to divide the lesion image or different regions of the lesion according to their features (such as the degree of inflammation, pus accumulation, the morphology of the lesion, etc.) so as to apply different cleaning strategies to different regions. The types of partition identifiers include, but are not limited to, high-inflammation partition identifiers, pus-accumulation partition identifiers, and narrow fistula partition identifiers. Among them, the high-inflammation partition identifier is used to identify the regions with relatively severe inflammation, usually requiring higher-intensity cleaning or higher-concentration drugs. The pus-accumulation partition is used to identify the regions with more pus accumulation, and these regions may require longer cleaning time or special liquid injection methods. The narrow fistula partition identifier is used to identify the regions with relatively narrow or irregular structures around the lesion. The cleaning operations in these regions may require more precise and accurate control to avoid over-stimulation. Through these partition identifiers, personalized cleaning control can be ensured for different lesion regions. To enable the partition processing layer to more accurately identify these regions, the cleaning guidance module 14 extracts historical lesion features and historical lesion-associated features with different identifiers from the historical database and inputs this feature data with identifiers into the initial partition processing layer. This initial partition processing layer can be constructed based on convolutional neural networks, deep neural networks, support vector machines, etc. If convolutional neural networks, deep neural networks, etc. are used, the construction process is the same as described above, also including steps such as forward propagation, loss calculation, backward propagation, and parameter update. If a support vector machine is used, it is to map the features by selecting an appropriate kernel function (such as a linear kernel, a radial basis function kernel RBF kernel, etc.), transform the data into a high-dimensional space, and then find the optimal hyperplane for classification by maximizing the margin to ensure that the partition identifiers for different regions (such as high inflammation, pus accumulation, narrow fistula, etc.) can be accurately distinguished. After completing the partition identification, the cleaning response model activates the cleaning parameter matching layer and synchronizes the partition identifiers. The main task of this layer is to match appropriate cleaning parameters based on the lesion partition identifiers. The cleaning priority of each partition may be different, so it is necessary to set the priority cleaning regions and secondary cleaning regions according to the characteristics of each region. The cleaning priority is determined according to the severity of the lesion, the affected range, and the requirements for the cleaning effect. For example, the high-inflammation region has a higher priority and needs to be cleaned first, while the pus-accumulation region may require long-term treatment, but because the urgency is slightly lower than that of the high-inflammation region, the priority is lower than that of the high-inflammation region;Subsequently, the lesion features and lesion-associated features corresponding to each partition identifier are transmitted to the cleaning parameter matching layer according to the priority of the partition identifiers. The cleaning parameter matching layer will match the corresponding cleaning parameters for each partition based on these features and the learned mapping relationship, and arrange them in order of priority to generate a time-series cleaning parameter matching result. This cleaning parameter matching layer can be constructed based on methods such as deep neural networks, convolutional neural networks, and multi-layer perceptrons. The construction method is the same as described above. The time-series cleaning parameter matching result generated by the cleaning parameter matching layer includes the cleaning liquid parameters, pressure parameters, flow rate parameters, spray head angle parameters, etc. for each partition. Among them, the cleaning liquid parameters refer to the type and concentration of the cleaning liquid to ensure the best cleaning effect for different regions. The pressure parameter refers to the cleaning pressure. For relatively sensitive regions such as narrow fistulas, the pressure should be set moderately to avoid damage. For regions with pus accumulation, a higher pressure may be required to ensure the complete removal of pus. The flow rate parameter refers to the flow rate of the cleaning liquid, and the spray head angle parameter refers to the angle at which the spray head sprays the cleaning liquid to ensure that the cleaning liquid can accurately cover each region. After that, the cleaning parameter matching layer will organize and package the time-series cleaning parameter matching result and output it in the form of a cleaning control plan to ensure the smooth progress of the entire cleaning process.

[0026] In a possible implementation manner, the cleaning guidance module 14 is further configured to: Activate the user action guidance layer in the cleaning response model, synchronize the cleaning control plan and the selection feedback to the user action guidance layer, and generate a mapped user action guidance plan; after temporally associating the user action guidance plan with the cleaning control plan, output it as a cleaning plan.

[0027] Optionally, after generating the cleaning control scheme, the cleaning response model activates the internal user action guidance layer and synchronizes the received selection feedback and the cleaning control scheme to the user action guidance layer. The selection feedback includes the cleaning holding posture selected by the user. The user action guidance layer includes a pre-constructed evaluation function for calculating the evaluation value of the user's current posture and determining whether the user's current posture meets the requirements of the cleaning holding posture. During the calculation process, the user action guidance layer also receives the user's current posture information passed in from the outside, including the user's opening and closing angle (such as the leg bending angle, etc.) and the actual cleaning control parameters used in the current posture (such as actual cleaning liquid parameters, actual flow parameters, etc.). Through the comprehensive calculation of these parameters by the evaluation function, the evaluation value of the user's current posture can be obtained. If a certain evaluation value is less than the preset evaluation value, the user action guidance layer will calculate the difference between the user's opening and closing angle and the opening and closing angle set in the cleaning holding posture, and arrange all the calculation results according to the time sequence relationship to form a user action guidance scheme; the cleaning response model forms a cleaning scheme and outputs it by associating the user action guidance scheme with the cleaning control scheme in time sequence, that is, combining the cleaning actions with the cleaning control parameters during the cleaning process according to time points, so as to ensure the synchronization of the user's actions with the parameter changes of the cleaning, improve the cleaning effect and avoid unnecessary errors.

[0028] In a possible implementation manner, the cleaning guidance module 14 is further configured to: The evaluation function of the user action guidance layer is as follows: Generate a mapped user action guidance scheme according to the calculation result of the evaluation function.

[0029] Optionally, the evaluation function of the user action guidance layer is specifically as follows: , where A s represents the evaluation value of the current user posture, F u represents the selection feedback, D p represents the deviation value between the cleaning parameters used in the user's current posture and the cleaning parameters in the cleaning control scheme, and the calculation formula of D p is specifically as follows: , where n is the number of cleaning parameters, i represents any one of the cleaning parameters, P actual,i is the actual use value of the cleaning parameter i in the current posture, such as the current cleaning pressure, flow rate, spray angle, etc., P optimal,i is the optimal value of the cleaning parameter i in the current posture action, that is, the cleaning parameter in the cleaning control scheme, and D m is the maximum allowable deviation value of the system for normalizing the deviation value, and w 1 , w 2They are the weight coefficients of the selection feedback and the cleaning parameter deviation value respectively, which are used to adjust their importance in the evaluation; the user action guidance layer will judge the validity of the current posture in the same way as described above according to the calculated evaluation value. If the requirements of the preset evaluation value are met, the user action guidance at the corresponding moment will be set to 0. Otherwise, the user action guidance at the corresponding moment will be set to the calculated angle deviation value; finally, according to the chronological relationship, all user action guidances will be summarized and sorted out to map the user action guidance scheme to ensure the efficiency and comfort of the cleaning operation.

[0030] In a possible implementation manner, the device further includes: A real-time feedback module, configured to display the user action guidance scheme on a display based on corresponding time nodes, collect user images through an image acquisition device, establish collected image data, generate a real-time guidance response according to the collected image data and the user action guidance scheme, and update the display on the display according to the real-time guidance response.

[0031] Optionally, the main function of the real-time feedback module is to display the user action guidance scheme to the user in real time and dynamically adjust the display content according to the user's actual operation. The real-time feedback module will display the specific user action guidance scheme on the display according to the time node. The guidance scheme will visually guide the user to complete the cleaning action in the form of graphics and animations. For example, it will display the cleaning posture, the process of posture adjustment, precautions, etc.; the real-time feedback module will render the user action guidance scheme to the display for posture guidance. During this process, it will analyze the content in the user action guidance scheme. If non-zero data appears in the user action guidance scheme, this data will be extracted to generate a guidance response. This guidance response includes the posture adjustment direction (determined by the positive or negative of the extracted data) and the posture adjustment amplitude (determined by the size of the extracted data). Then, according to this guidance response, corresponding graphics and animations will be rendered and displayed on the display. For example, if the new user action guidance scheme shows that the bending angle of the leg is small, at this time, an arrow will be used to render the process of posture change on the display, and this arrow is used to indicate the direction of leg change; after the guidance is completed, an image acquisition device (such as a camera) will be used to record the actual adjusted posture of the user in real time, and the collected actual adjusted posture will be identified for deviation from the theoretically adjusted posture in the user action guidance scheme, and the adjustment effectiveness will be judged according to the identification result. If the adjustment still does not meet the requirements, mobile guidance optimization will be performed, and then the optimization result will be synchronously updated to the display for display to ensure that the user can accurately complete the cleaning action according to the guidance, improving the accuracy of the cleaning operation and the user experience.

[0032] In a possible implementation manner, the real-time feedback module is used for: Perform co-location action deviation recognition on the collected image data and the user action guidance plan, and establish a deviation identifier; perform movement guidance optimization based on the deviation identifier, and establish a movement guidance optimization result; use the movement guidance optimization result as a real-time guidance response to update the display on the monitor.

[0033] Optionally, when performing co-location action deviation recognition on the collected image data and the user action guidance plan, compare the user's current image data with the theoretically adjusted posture at the corresponding time node in the user action guidance plan, analyze the angle between the user's actual posture and the theoretical posture, calculate the specific deviation value through difference calculation. If the deviation value exceeds the deviation threshold, a deviation identifier will be generated to clearly indicate the deviation of the user's action; according to the deviation identifier, the real-time feedback module will perform movement guidance optimization, that is, extract the corresponding deviation value from the deviation identifier, determine the direction and amplitude of the offset based on the deviation value, and then establish a movement guidance optimization result based on the direction and amplitude of the offset, including specific action adjustment instructions such as tilting 10 degrees to the left, etc.; afterwards, convert the generated movement guidance optimization result into a real-time guidance response and update the content of the monitor. The updated display content may intuitively display the adjustment suggestions in the form of graphics and animations to ensure that the user can adjust the action in real time during the cleaning process, thereby improving the accuracy of the operation and the cleaning effect.

[0034] In a possible implementation manner, the device further includes: An interaction feedback module, configured to perform real-time cleaning interaction feedback for the user, generate a feedback database, establish a scheme compensation for the cleaning scheme according to the feedback database, and perform user cleaning management according to the scheme compensation.

[0035] Optionally, the main function of the interaction feedback module is to collect the operation feedback of the user during the cleaning process in real time, generate relevant feedback data, and dynamically adjust the cleaning plan based on this data. Specifically, the interaction feedback module will obtain feedback content such as the operation information, action adjustment, and subjective evaluation of the cleaning effect of the user during the cleaning process through sensors, image acquisition devices, or manual input of the user. These data will be stored in a feedback database as the historical record of the cleaning process. Based on the feedback database, it is possible to analyze whether the user's cleaning operations conform to the guiding plan, and at the same time identify the problems or optimization requirements existing in the cleaning process. For example, it can be found that the user may frequently deviate in certain actions, or is not satisfied with the adjustment effect of a certain cleaning parameter. According to these analysis results, a plan compensation is generated to specifically adjust certain parameters in the cleaning plan (such as the concentration, pressure, flow rate of the cleaning liquid, or action guidance content) to make the plan more in line with the actual needs of the user. Finally, the adjusted cleaning plan will be applied to the user's cleaning management process to achieve personalized and precise cleaning guidance through dynamic optimization, ensuring that the cleaning process is more efficient and comfortable, and at the same time improving the user experience and cleaning effect.

[0036] In a possible implementation manner, the device further includes: A cleaning feedback module, configured to collect lesion images during the cleaning process, establish a time-sequential cleaning image, evaluate the cleaning effect according to the time-sequential cleaning image, generate a cleaning feedback using the cleaning effect evaluation result, and perform cleaning management of the user with the cleaning feedback.

[0037] Optionally, the cleaning feedback module collects images of the lesion area during the cleaning process, records the lesion status at each time node, and forms a set of cleaning image data with a time sequence. Through these time-sequential cleaning images, the changes in the lesion can be dynamically tracked, including the degree of inflammation reduction, the degree of pus removal before and after cleaning, and the cleaning effect of the lesion area, etc. According to the time-sequential cleaning images, the cleaning feedback module evaluates the cleaning effect. The evaluation process includes the comparative analysis of the lesion characteristics in the images. For example, indicators such as the coverage rate of the cleaned area, the change in the lesion edge, and the amount of residual pus are calculated. These indicators are obtained based on the cleaning feedback evaluation network, and the construction method of this cleaning feedback evaluation network is the same as that described above; subsequently, by comparing these evaluation indicators with the expected cleaning target, a quantitative score of the cleaning effect is obtained, and this score is used as the cleaning effect evaluation result; after that, the cleaning effect evaluation result will further generate a cleaning feedback. If some areas are not completely cleaned or the cleaning intensity is insufficient, the cleaning feedback will be used to adjust subsequent cleaning operations or guide the user's cleaning actions. For example, the cleaning liquid concentration, cleaning time, or cleaning actions can be adjusted according to the feedback suggestions; finally, the cleaning feedback module applies these feedback results to the user's cleaning management. Through dynamic optimization and adjustment, the effect and efficiency of the cleaning process are improved, ensuring that the lesion can be thoroughly cleaned, while enhancing the user experience and treatment effect.

[0038] Embodiment 2. Based on the same inventive concept as the anal fistula cleaning device based on visual identification in the foregoing embodiment, the present application also provides an anal fistula cleaning system based on visual identification, including a processor, a memory, an input device, and an output device; the number of processors in this system can be one or more, and the processor, memory, input device, and output device can be connected through a bus or other means.

[0039] As a computer-readable storage medium, the memory can be used to store software programs, computer-executable programs, and components, such as the program instructions / components corresponding to the anal fistula cleaning device based on visual identification in the embodiments of the present invention. The processor executes various functional applications and data processing of the anal fistula cleaning device based on visual identification by running the software programs, instructions, and components stored in the memory.

[0040] The anal fistula cleaning system based on visual identification provided by the embodiments of the present invention is used for an anal fistula cleaning device based on visual identification, and has corresponding functional components and beneficial effects of the anal fistula cleaning device based on visual identification.

[0041] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. In addition, the above specific embodiments of this specification have been described. Further, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0042] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0043] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An anal fistula cleaning device based on visual identification, characterized in that: The device comprises: The endoscope device is used to locate the lesion area according to the auxiliary positioning signal after receiving the start instruction, and automatically adjust the device position of the endoscope device, perform external image acquisition, and establish a first lesion image. When the endoscope device is placed in the fistula, perform internal image acquisition and establish a second lesion image; An image processing module, configured to extract image features after performing image enhancement on the first lesion image and the second lesion image, wherein the image features include lesion features and lesion-related features; An interactive module, used to display the cleaning holding postures in the general holding posture library to the user through a display, and to sort the display order based on the adaptability and matching degree between the user's body shape, lesion state and the holding posture library, so as to obtain the user's selection feedback; A cleaning guidance module, for inputting the lesion characteristics, the lesion-related characteristics and the selection feedback into a cleaning response model to generate a cleaning scheme, wherein the cleaning scheme includes a cleaning control scheme and a user action guidance scheme, and performing cleaning management of the user according to the cleaning scheme; The cleaning guidance module is used to call the partition processing layer in the cleaning response model, use the partition processing layer to perform partition analysis based on the lesion characteristics and the lesion association characteristics, and establish partition identification, wherein the partition identification includes a high inflammation partition identification, a pus accumulation partition identification, and a narrow fistula partition identification; Activate the cleaning parameter matching layer in the cleaning response model, synchronize the partition identifier to the cleaning parameter matching layer, perform sequential cleaning parameter matching according to the cleaning priority, and establish a sequential cleaning parameter matching result, wherein the sequential cleaning parameter matching result includes cleaning fluid parameters, pressure parameters, flow parameters, and nozzle angle parameters; The timing cleaning parameter matching results are sorted into a cleaning control solution output.

2. The anal fistula cleaning device based on visual marking according to claim 1, characterized in that: The cleaning guidance module is also used for: activating the user action guidance layer in the cleaning response model, synchronizing the cleaning control scheme and the selection feedback to the user action guidance layer, and generating a mapping user action guidance scheme; After the user action guidance scheme is associated with the cleaning control scheme in time sequence, the cleaning scheme is outputted.

3. The anal fistula cleaning device based on visual marking according to claim 2, characterized in that: The cleaning guidance module is also used for: The evaluation function of the user action guidance layer is as follows: ； Among them, A s Characterization evaluation value, F u Characterization selection feedback, D p Characterizes the deviation value between posture action and cleaning parameters, , where n is the number of cleaning parameters, i represents any cleaning parameter, P actual,i is the actual value of cleaning parameter i, P optimal,i is the optimal value of cleaning parameter i under the current posture action, D m is the maximum deviation allowed by the system, w1 and w2 are weight coefficients; A mapped user action guidance scheme is generated according to the calculation result of the evaluation function.

4. The anal fistula cleaning device based on visual marking according to claim 1, characterized in that: The device also includes: A real-time feedback module is used to display the user action guidance plan based on the display at the corresponding time node, and to collect user images through an image collection device to establish collected image data, generate a real-time guidance response based on the collected image data and the user action guidance plan, and update the display according to the real-time guidance response.

5. The anal fistula cleaning device based on visual marking according to claim 4, characterized in that: The real-time feedback module is used for: Performing same-position action deviation identification on the collected image data and the user action guidance scheme, and establishing a deviation mark; Performing mobile guidance optimization according to the deviation identifier and establishing a mobile guidance optimization result; The mobile guidance optimization result is used as a real-time guidance response to update the display.

6. The anal fistula cleaning device based on visual marking according to claim 1, characterized in that: The image processing module is also used for: Performing global image recognition on the first lesion image and the second lesion image to establish a global image recognition result; Using the global image recognition result to perform universal enhancement parameter matching, performing initial enhancement of the first lesion image and the second lesion image according to the enhancement parameter matching result, and establishing an initial enhancement result; Image focus recognition is performed on the initial enhancement result, additional enhancement is performed based on the image focus recognition result, an additional enhancement result is established, and feature extraction is performed using the additional enhancement result as an enhanced image.

7. The anal fistula cleaning device based on visual marking according to claim 1, characterized in that: The device also includes: The interactive feedback module is used to perform real-time interactive cleaning feedback of the user, generate a feedback database, establish a solution compensation for the cleaning solution according to the feedback database, and perform cleaning management of the user according to the solution compensation.

8. The anal fistula cleaning device based on visual marking according to claim 1, characterized in that: The device also includes: The cleaning feedback module is used to perform lesion image acquisition during the cleaning process, establish a time-series cleaning image, evaluate the cleaning effect based on the time-series cleaning image, generate cleaning feedback using the cleaning effect evaluation result, and use the cleaning feedback to manage the user's cleaning.

9. An anal fistula cleaning system based on visual identification, characterized in that: The system is used to execute the steps of the anal fistula cleaning device based on visual identification according to any one of claims 1 to 8, including: A memory for storing executable instructions; A processor is used to execute the executable instructions stored in the memory.

Images