Detection flushing method and system based on deep infectious wound sinus tract

Abstract

The invention relates to a detection flushing method and system based on a deep infectious wound sinus tract, and belongs to the technical field of medical intelligent detection instruments. The method comprises the steps that a sinus tract internal image signal is obtained through a miniature endoscope connected with a probe of a detection flushing device, and the sinus tract internal image signal is transmitted to an external display device in real time; the terminal application of the external display device receives the sinus tract internal image signal and converts the sinus tract internal image signal into recognizable image information, an infected area and potential necrotic tissue in the sinus tract are recognized in a self-adaptive mode based on the image information, and flushing parameters are set; the terminal application transmits the flushing parameters to a wireless receiver of the detection flushing device, and the control module generates a control instruction according to the received flushing parameters; and flushing the sinus tract by adjusting the flushing pressure and the flushing flow according to the control instruction. Accurate detection and efficient flushing of the deep infectious wound sinus tract are achieved, and the pertinence and effectiveness of flushing are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of medical intelligent detection instruments, and particularly relates to a detection and flushing method and system based on a deep infectious wound sinus. BACKGROUND

[0002] Traditional sinus treatment highly depends on the visual judgment and manual probe palpation of doctors, and lacks objective quantitative tools. The manual flushing process has significant technical shortcomings: the actual pressure fluctuation range may exceed the safety threshold by relying on the injection speed of the syringe to adjust the pressure; the open flushing system cannot maintain a constant flow rate, resulting in uneven distribution of the flushing liquid and insufficient coverage of deep infection foci; there is a lack of real-time temperature feedback mechanism, and when the flushing liquid temperature deviates from the ideal interval, it may damage the newly formed granulation tissue;

[0003] The sterilization dead angle of the reusable metal probe leads to an increase in the infection rate, especially for the poor prevention and control effect of drug-resistant bacteria; the single debridement takes a long time and requires multiple follow-up visits, significantly increasing the consumption of medical resources. At present, although some automatic flushing devices have been proposed, most of these devices have single functions and lack intelligent identification and control functions, which cannot meet the complex processing needs of deep infectious wound sinuses. For example, although the current mainstream devices have pressure / temperature adjustment functions, the flushing parameter settings rely on preset programs and cannot be dynamically adjusted according to real-time images (such as endoscopic images). SUMMARY

[0004] To solve the above problems in the prior art, the present application provides a detection and flushing method and system based on a deep infectious wound sinus;

[0005] The purpose of the present application can be achieved by the following technical solutions:

[0006] S1: Obtain the internal image signal of the sinus through the micro endoscope connected to the probe in the detection and flushing device, and transmit the image signal to the external display device in real time through the wireless transceiver module integrated in the microcontroller;

[0007] S2: The external display device receives the internal image signal of the sinus through the wireless communication module of the terminal application and converts it into computer recognizable image information, and the terminal application adaptively identifies the infected area and potential necrotic tissue in the sinus based on the image information and sets the flushing parameters;

[0008] S3: The terminal application feeds back the flushing parameters to the wireless transceiver module of the detection and flushing device through the wireless communication module, and the microcontroller generates a flushing control instruction according to the flushing parameters;

[0009] S4: The micro motor and the flow regulation module adaptively control the flushing pressure and flow rate of the flushing pipeline according to the flushing control instruction.

[0010] Specifically, the probe flushing device in S1 includes a probe, a micro endoscope, a microcontroller, a flushing button, a switch button, a power module, a micro motor, a flow regulation module, a flushing channel, a liquid storage cavity, and a flushing liquid buffer cavity. One end of the probe is connected to the micro endoscope, and the other end is used for inserting into the interior of the sinus for depth detection. The micro endoscope integrates an LED illumination module for providing illumination inside the sinus. The microcontroller integrates a wireless transceiver module for transmitting image signals inside the sinus to an external display device in real time and receiving flushing parameters transmitted by the external display device. The microcontroller is electrically connected to the micro motor and the flow regulation module. The flushing button is connected to the micro motor for start-stop control. The switch button is electrically connected to the power module for turning on or off the entire probe flushing device. The flushing channel penetrates through the probe, the flushing liquid buffer cavity, and the liquid storage cavity in sequence. The micro motor adjusts the pressure of the flushing liquid buffer cavity to control the flushing flow, and the flow regulation module adjusts the flushing liquid flowing from the liquid storage cavity into the flushing liquid buffer cavity through the flushing channel.

[0011] Specifically, the microcontroller generates a PWM duty cycle control signal and a flow control analog signal according to the received flushing parameters and transmits them to the micro motor and the flow regulation module via a drive circuit. The micro motor adjusts the output pressure by changing the duty cycle through the PWM duty cycle signal, and the flow regulation module controls the flushing flow through the flow control analog signal.

[0012] Specifically, the flushing button triggers the micro motor to start and push the stored flushing liquid into the flushing channel. When the capacitive sensor detects that the flushing liquid buffer cavity is empty, it sends a feedback signal to the microcontroller.

[0013] Specifically, the flow regulation module generates a proportional voltage to drive the stroke length of the liquid pump through a D / A converter according to the flow control analog signal.

[0014] After the flushing button is triggered, the microcontroller sends a flow control analog signal and a liquid pump start signal to the flow regulation module through the received feedback signal.

[0015] Specifically, the adaptive identification method in S2 is as follows:

[0016] S201: Determine the segmentation threshold of each pixel by dynamically calculating the gray scale statistics of the local area of the image information, dynamically segment the gray scale difference between the enhanced lesion and the normal tissue according to the segmentation threshold, and extract and analyze the target positioning area according to the segmentation result.

[0017] S202: extract morphological features from the analysis target positioning area, construct a combined classifier based on the kernel function outputs of different feature subsets in the morphological features, and optimize the classifier parameters according to the minimum classification error;

[0018] S203: output the confidence of the infection area and the potential necrotic tissue based on the target positioning area through the trained classifier, mark the target positioning area with a confidence greater than a preset threshold on the image display interface of the terminal application to realize automatic identification of the infection area and the potential necrotic tissue, and feed back the identification result to the terminal application to set corresponding flushing parameters.

[0019] Specifically, the dynamic segmentation reinforcement method in S201 is:

[0020] Calculate the local mean and variance according to the field of each pixel, perform brightness screening through the local mean, and select the area smaller than the global mean as a dark area candidate area; perform contrast screening through the variance, and perform image enhancement processing on the low-contrast area.

[0021] Specifically, the morphological features in S202 include high eccentricity features, low compactness features, low intrinsic degree features, and structural heterogeneity features; the kernel function adaptation strategy of different feature subsets in the morphological features is:

[0022] According to the high eccentricity features and the low compactness features, a shape complexity subset is formed, and a radial basis kernel is used to capture the nonlinear boundary; according to the low intrinsic degree features, a topological structure subset is formed, and a polynomial kernel is used to enhance the convexity difference; according to the structural heterogeneity features, a tissue heterogeneity subset is formed, and a linear kernel is used to improve the classification efficiency; different kernel functions are combined through weight coefficients to optimize the classifier parameters.

[0023] Specifically, the objective function of the minimum classification error optimization in S203 is set as:

[0024]

[0025] Where λ is the Lagrange multiplier of the support vector machine, n is the total number of training samples, y i represents the true label of the i-th sample, x i , represents the predicted value of the i-th sample, α is the regularization coefficient, and K is the kernel function.

[0026] A detection and flushing system based on deep infected wound sinus, comprising: a detection communication module, a terminal application identification unit, a feedback control module, and a flushing execution unit;

[0027] The detection communication module is used for acquiring image signals inside the sinus tract through a micro endoscope connected with a probe in the flushing device, and transmitting the image signals to an external display device in real time through a wireless transceiver module integrated with a microcontroller;

[0028] The terminal application recognition unit is used for receiving the image signals inside the sinus tract through a wireless communication module of a terminal application and converting the image signals into computer recognizable image information, and the terminal application is used for adaptively recognizing an infected area and potential necrotic tissue inside the sinus tract based on the image information and setting flushing parameters;

[0029] The feedback control module is used for feeding back the flushing parameters to the wireless transceiver module of the detection flushing device through the wireless communication module of the terminal application, and the microcontroller is used for generating a flushing control instruction according to the flushing parameters;

[0030] The flushing execution unit is used for adaptively controlling flushing pressure and flushing flow of a flushing pipeline according to the flushing control instruction through a micro motor and a flow regulating module.

[0031] The present application has the following advantages:

[0032] The detection flushing system provided by the present application realizes real-time monitoring and intelligent processing of the internal condition of a deep infective wound sinus tract. The system can not only adaptively recognize an infected area and potential necrotic tissue inside the sinus tract, but also dynamically adjust flushing parameters according to the recognition result, thereby ensuring the safety and effectiveness of the flushing process. Meanwhile, through precise control of the micro motor and the flow regulating module, adaptive adjustment of the flushing pressure and the flushing flow is realized, thereby avoiding problems such as pressure fluctuation and uneven flow that may occur in the traditional flushing process. In addition, the system of the present application also has a wireless communication function, which can realize real-time data transmission and interaction with an external display device, thereby improving the convenience and flexibility of operation. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to facilitate understanding of those skilled in the art, the present application will be further described below with reference to the accompanying drawings.

[0034] Fig. 1 A flowchart of a detection flushing method based on a deep infective wound sinus tract according to the present application;

[0035] Fig. 2 A structure block diagram of a detection flushing device according to the present application.

[0036] In the figure: 1, probe; 2, flushing channel; 3, micro endoscope; 4, flushing button; 5, switch button; 6, micro motor; 7, flushing liquid buffer cavity; 8, flow regulating module; 9, power module; 10, microcontroller; 11, liquid storage cavity; 12, capacitance sensor. DETAILED DESCRIPTION

[0037] To further clarify the technical means and effects taken by the present application to achieve the predetermined inventive purpose, the specific embodiments, structures, features and effects thereof according to the present application are described in detail below in conjunction with the drawings and preferred embodiments.

[0038] Please refer to Figs. 1-2 A detection and flushing method based on deep infectious wound sinus, characterized in that, comprising:

[0039] S1: obtaining the image signal inside the sinus through the micro endoscope connected to the probe in the detection and flushing device, and transmitting the image signal to the external display device in real time through the wireless transceiver module integrated in the microcontroller;

[0040] S2: the external display device receives the image signal inside the sinus through the wireless communication module of the terminal application and converts it into image information recognizable by the computer, and the terminal application adaptively identifies the infected area and potential necrotic tissue inside the sinus based on the image information and sets the flushing parameters;

[0041] S3: the terminal application feeds back the flushing parameters to the wireless transceiver module of the detection and flushing device through the wireless communication module, and the microcontroller generates flushing control instructions according to the flushing parameters;

[0042] S4: the micro motor and the flow regulating module adaptively control the flushing pressure and flow of the flushing pipeline according to the flushing control instructions.

[0043] Specifically, the detection and flushing device in S1 includes a probe, a micro endoscope, a microcontroller, a flushing button, a switch button, a power module, a micro motor, a flow regulating module, a flushing channel, a liquid storage cavity, and a flushing liquid buffer cavity. One end of the probe is connected to the micro endoscope, and the other end is used for inserting into the interior of the sinus for deep detection. The micro endoscope integrates an LED illumination module to provide illumination for the interior of the sinus. The wireless transceiver module integrated in the microcontroller is used to transmit the image signal inside the sinus to the external display device in real time and receive the flushing parameters transmitted by the external display device. The microcontroller is electrically connected to the micro motor and the flow regulating module. The flushing button is connected to the micro motor for start-stop control. The switch button is electrically connected to the power module for turning on or off the entire detection and flushing device. The flushing channel passes through the probe, the flushing liquid buffer cavity, and the liquid storage cavity in sequence. The micro motor adjusts the pressure of the flushing liquid buffer cavity to control the flushing flow, and the flow regulating module adjusts the flushing liquid flowing from the liquid storage cavity into the flushing liquid buffer cavity through the flushing channel.

[0044] Specifically, the microcontroller generates a PWM duty cycle control signal and a flow control analog signal according to the received flushing parameters, and transmits them to the micro motor and the flow regulating module via the drive circuit, the micro motor adjusts the duty cycle to change the output pressure through the PWM duty cycle signal, and the flow regulating module controls the flushing flow through the flow control analog signal.

[0045] Specifically, the flushing button triggers the micro motor to start and push the stored flushing liquid into the flushing channel; when the capacitive sensor detects that the flushing liquid buffer cavity is empty, it sends a feedback signal to the microcontroller.

[0046] Specifically, the flow regulating module generates a proportional voltage to drive the stroke length of the liquid pump through a D / A converter according to the flow control analog signal;

[0047] After the flushing button is triggered, the microcontroller sends a flow control analog signal and a liquid pump start signal to the flow regulating module through the received feedback signal.

[0048] Specifically, the adaptive identification method in S2 is:

[0049] S201: Determine the segmentation threshold of each pixel by dynamically calculating the gray scale statistics of the local area of the image information, dynamically segment the gray difference between the enhanced lesion and normal tissue according to the segmentation threshold, and extract the target positioning area according to the segmentation result;

[0050] S202: Extract morphological features from the analysis target positioning area, construct a combined classifier based on the kernel function output of different feature subsets in the morphological features, and optimize the classifier parameters according to the minimum classification error;

[0051] S203: Train the optimized classifier, the classifier output after training is the confidence of the infection area and potential necrotic tissue based on the target positioning area, mark the target positioning area with a confidence greater than a preset threshold on the image display interface of the terminal application to realize automatic identification of the infection area and potential necrotic tissue, and feed back the identification result to the terminal application to set the corresponding flushing parameters.

[0052] Specifically, the dynamic segmentation enhancement method in S201 is:

[0053] Calculate the local mean and variance according to the field of each pixel; perform brightness screening through the local mean, and select the region with a brightness less than the global mean as a dark region candidate area; perform contrast screening through the variance, and perform image enhancement processing on the low-contrast region.

[0054] In this embodiment, for the field Sxy of each pixel (x, y), the local mean and variance are calculated:

[0055]

[0056] where N is the number of pixels in the field, rs,t is the pixel gray value, mxy is the local mean, and the variance; the local mean reflects the brightness, and the variance measures the contrast;

[0057] The brightness screening and contrast screening are performed through threshold determination; the dynamic segmentation is realized by using the dyn_threshold function of Halcon, by comparing the original image with the threshold image after smoothing (Gaussian filtering result), and combining the offset amount Offset to control the segmentation sensitivity.

[0058] Specifically, the morphological features in S202 include a high eccentricity feature, a low compactness feature, a low intrinsic feature, and a structural heterogeneity feature; the kernel function adaptation strategies of different feature subsets in the morphological features are as follows:

[0059] The high eccentricity feature and the low compactness feature constitute a shape complexity subset, and a radial basis kernel is used to capture the nonlinear boundary; the low intrinsic feature constitutes a topological structure subset, and a polynomial kernel is used to enhance the convexity difference; the structural heterogeneity feature constitutes a tissue heterogeneity subset, and a linear kernel is used to improve the classification efficiency; different kernel functions are combined through a weight coefficient to optimize the classifier parameters.

[0060] In this embodiment, the infection area (such as inflammation or tumor) often presents an irregular growth pattern, and the difference between the long axis and the short axis is significant, resulting in an increased eccentricity. For example, the infiltrative growth of the edge of a malignant tumor destroys the symmetry; the eccentricity describes the deviation of the target shape from the ideal ellipse, and the higher the eccentricity, the more complex the shape:

[0061] where a represents the length of the long axis of the identified shape in the target area, and b represents the length of the short axis of the identified shape in the target area; the value range is 0 (circular) to 1 (linear). The infection area (such as a malignant tumor) has a significant difference between the long axis and the short axis due to infiltrative growth, resulting in an increased eccentricity and a deviation from symmetry.

[0062] The low compactness feature reflects the tortuosity of the boundary of the target area, and the lower the compactness, the more complex the boundary:

[0063] where P is the total number of boundary area pixels of the identified shape in the target area, and A is the total number of internal pixels of the identified shape in the target area. The larger the value, the more complex the shape (the more tortuous the boundary). The necrotic tissue has increased perimeter and slow area growth due to liquefaction or fibrosis, resulting in reduced compactness and low value.

[0064] Low intrinsic degree feature is used to measure the difference between the target area and its convex hull, which reflects the complexity of the topological structure of the region:

[0065] Where A is the total number of internal pixels of the identified shape in the target area, and Aconvex represents the total number of internal pixels of the most convex polygon that completely encloses the target identified shape. The value closer to 1 indicates that the shape is more convex, and the lower the value, the more concave or hole. The necrotic area is caused by internal collapse or cavity, resulting in a convex area much larger than the actual area, and the intrinsic degree is reduced.

[0066] The structural heterogeneity feature focuses on the uniformity of the distribution of pixel gray value or color in the target area, and is used to describe the heterogeneity of the tissue, based on the entropy (Entropy) calculation of the gray level co-occurrence matrix (GLCM):

[0067] Where p(i,j) represents the joint probability of gray values i and j in the gray level co-occurrence matrix under a certain direction and distance; N represents the number of gray levels.

[0068] These morphological features together provide an important basis for the identification of infected areas and potential necrotic tissue.

[0069] Specifically, the objective function of the minimum classification error optimization in S203 is set as:

[0070]

[0071] Where λ is the Lagrange multiplier of the support vector machine, n is the total number of training samples, y i represents the true label of the i-th sample, x i , represents the predicted value of the i-th sample, α is the regularization coefficient, and K is the kernel function.

[0072] A detection and flushing system based on deep infected wound sinus, comprising: a detection communication module, a terminal application identification unit, a feedback control module, and a flushing execution unit;

[0073] The detection communication module is used to obtain the internal image signal of the sinus through the micro endoscope connected to the probe in the detection and flushing device, and to transmit the image signal to the external display device in real time through the wireless transceiver module integrated with the microcontroller;

[0074] The terminal application identification unit is used for the external display device to receive the internal image signal of the sinus through the wireless communication module of the terminal application and convert it into computer recognizable image information, and the terminal application is used to adaptively identify the infected area and potential necrotic tissue in the sinus based on the image information, and set the flushing parameters;

[0075] The feedback control module feeds back the flushing parameters to the wireless transceiver module of the detection flushing device through the wireless communication module according to the terminal application, and the microcontroller generates a flushing control instruction according to the flushing parameters;

[0076] The flushing execution unit performs self-adaptive control on the flushing pressure and the flushing flow of the flushing pipeline through the micro motor and the flow regulation module according to the flushing control instruction.

[0077] In the embodiment, the microcontroller is used as the core control unit of the device, establishes a physical layer connection with the wireless transceiver module through the UART / SPI high-speed communication protocol, and receives the flushing parameters (pressure value, flow value and pulse mode) issued by the external terminal in real time; a parsing algorithm based on a finite state machine (FSM) is used to convert the digital signal into two types of driving instructions:

[0078] The piezoelectric pressure control instruction: the target pressure value (0-50kPa) is mapped into a 0-5V analog voltage signal, which is output through a 16-bit DAC module;

[0079] The flow control analog signal: the target flow value is mapped into a 0-10V analog voltage signal, which is output through a 12-bit DAC module. The flow regulation module adjusts the flow of the flushing liquid through the proportional electromagnetic valve according to the received flow control analog signal.

[0080] After the flushing key is triggered, a preset anti-impact start sequence (pressure ramp-up, flow gradual opening) is immediately started; the on-off key adopts a long-press protection mechanism (effective for 3 seconds), which avoids abnormal power-off of the system caused by accidental touch.

[0081] The microcontroller is also used to distribute the power of the power module to ensure stable power supply of each component.

[0082] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to make equivalent embodiments with equivalent changes, without departing from the scope of the technical solution of the present application. Any simple modification, equivalent change and modification of the above embodiments made according to the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A method for detecting and flushing sinus tracts in deep infected wounds, characterized in that: include: S1: Acquire internal sinus image signals through a micro-endoscope connected to a probe in a detection and flushing device, and transmit the image signals in real time to an external display device through a wireless transceiver module integrated in a microcontroller; S2: The external display device receives the internal image signal of the sinus tract through the wireless communication module of the terminal application and converts it into computer-recognizable image information. The terminal application adaptively identifies the infected area and potential necrotic tissue in the sinus tract based on the image information and sets flushing parameters. S3: The terminal application feeds back the flushing parameters to the wireless transceiver module of the detection and flushing device through the wireless communication module, and the microcontroller generates a flushing control instruction according to the flushing parameters; S4: Adaptively controlling the flushing pressure and flushing flow of the flushing pipeline according to the flushing control instruction through the micro motor and the flow regulating module.

2. The method according to claim 1, characterized in that The detection and flushing device in S1 includes a probe, a micro endoscope, a microcontroller, a flushing button, a switch button, a power module, a micro motor, a flow regulation module, a flushing channel, a liquid storage cavity, and a flushing liquid buffer cavity; one end of the probe is connected to the micro endoscope, and the other end is used to insert into the sinus tract for deep detection; the micro endoscope integrates an LED lighting module for providing light inside the sinus tract; the wireless transceiver module integrated in the microcontroller is used to transmit the internal image signal of the sinus tract to an external display device in real time, and receive the flushing parameters transmitted by the external display device; the microcontroller is electrically connected to the micro motor and the flow regulation module; the flushing button is connected to the micro motor for start and stop control; the switch button is electrically connected to the power module for turning on or off the entire detection and flushing device; the flushing channel passes through the probe, the flushing liquid buffer cavity, and the liquid storage cavity in sequence; the micro motor adjusts the pressure of the flushing liquid buffer cavity to control the flushing flow, and the flow regulation module adjusts the flushing liquid from the liquid storage cavity to the flushing liquid buffer cavity through the flushing channel.

3. The method according to claim 2, characterized in that The microcontroller generates a PWM duty cycle control signal and a flow control analog signal based on the received flushing parameters, and transmits them to the micro motor and the flow regulation module via the drive circuit. The micro motor adjusts the duty cycle through the PWM duty cycle signal to change the output pressure, and the flow regulation module controls the flushing flow through the flow control analog signal.

4. The method according to claim 2, characterized in that The flushing button triggers the micro motor to start, pushing the stored flushing liquid into the flushing channel; when the capacitive sensor detects that the flushing liquid buffer cavity is empty, a feedback signal is sent to the microcontroller.

5. The method according to claim 2, characterized in that The flow regulating module generates a proportional voltage to drive the stroke length of the liquid pump via a D / A converter according to the flow control analog signal; After the flush button is triggered, the microcontroller sends a flow control analog signal and a liquid pump start signal to the flow regulation module through the received feedback signal.

6. The method according to claim 1, characterized in that The adaptive recognition method in S2 is: S201: Determine a segmentation threshold for each pixel by dynamically calculating the grayscale statistics of a local area of ​​the image information, dynamically segment the grayscale difference between the enhanced lesion and normal tissue according to the segmentation threshold; and extract and analyze the target positioning area based on the segmentation result; S202: extracting morphological features from the analysis target positioning area, constructing a combined classifier based on kernel function outputs of different feature subsets in the morphological features, and optimizing classifier parameters according to minimum classification error; S203: The optimized classifier is trained. After training, the classifier outputs the confidence of the infected area and the potential necrotic tissue based on the target positioning area. The target positioning area with a confidence greater than a preset threshold is marked on the image display interface of the terminal application to achieve automatic recognition of the infected area and the potential necrotic tissue, and the recognition result is fed back to the terminal application to set the corresponding flushing parameters.

7. The method according to claim 6, characterized in that The dynamic segmentation enhancement method described in S201 is: The local mean and variance are calculated based on the area of ​​each pixel; the brightness is screened by the local mean, and the area with a value smaller than the global mean is selected as the candidate dark area; the contrast is screened by the variance, and the image enhancement processing is performed on the low-contrast area.

8. The method according to claim 6, characterized in that The morphological features in S202 include high eccentricity features, low compactness features, low intrinsic degree features, and structural heterogeneity features; the kernel function adaptation strategy for different feature subsets in the morphological features is: A shape complexity subset is formed based on the high eccentricity feature and low compactness feature, and the radial basis kernel is used to capture nonlinear boundaries; a topological structure subset is formed based on the low intrinsic degree feature, and the convexity difference is enhanced by the polynomial kernel; a tissue heterogeneity subset is formed based on the structural heterogeneity feature, and the linear kernel is used to improve the classification efficiency; different kernel functions are combined by weight coefficients to optimize the classifier parameters.

9. The method according to claim 6, characterized in that The objective function of the minimum classification error optimization in S203 is set to: ， Among them, λ is the Lagrange multiplier of the support vector machine, n is the total number of training samples, y i represents the true label of the i-th sample, x i for, represents the predicted value of the i-th sample, α is the regularization coefficient, and K is the kernel function.

10. A detection and flushing system based on deep infectious wound sinus tracts, used to perform the method according to any one of claims 1 to 9, characterized in that: Including: detection communication module, terminal application identification unit, feedback control module, flushing execution unit; The detection and communication module is used to obtain image signals inside the sinus tract through a micro-endoscope connected to a probe in the detection and flushing device, and transmit the image signals in real time to an external display device through a wireless transceiver module integrated in the microcontroller; The terminal application recognition unit is used for the external display device to receive the internal image signal of the sinus tract through the wireless communication module of the terminal application and convert it into computer-recognizable image information. The terminal application adaptively identifies the infected area and potential necrotic tissue in the sinus tract based on the image information and sets the flushing parameters; The feedback control module feeds back the flushing parameters to the wireless transceiver module of the detection and flushing device through the wireless communication module according to the terminal application, and the microcontroller generates a flushing control instruction according to the flushing parameters; The flushing execution unit performs adaptive control of the flushing pressure and flushing flow of the flushing pipeline according to the flushing control instruction through the micro motor and the flow regulating module.

Images