Automated navigation method for tracheal intubation based on artificial intelligence analysis anatomical structure

Abstract

The invention provides a tracheal intubation automatic navigation method based on an artificial intelligence analysis anatomical structure, and relates to the technical field of tracheal navigation, and the tracheal intubation automatic navigation method comprises the following steps: carrying out image scanning on a user to obtain real-time airway image information; constructing a human airway image anatomical model, and performing anatomical structure analysis on the real-time airway image information according to the human airway image anatomical model to obtain anatomical structure data; connecting an electromagnetic navigation mechanical control system; the path planning subsystem analyzes the anatomical structure data to obtain a first intubation navigation path; and the path control subsystem synchronously generates intubation control parameters according to the first intubation navigation path, and controls a mechanical arm of intubation equipment to perform tracheal intubation on the user according to the intubation control parameters. The technical problem that in the prior art, due to the limitation that tracheal intubation operation depends on experience and skills of doctors, the intubation success rate is not high under the emergency condition can be solved, and the technical effect of improving the intubation success rate is achieved.

Description

Technical Field

[0001] The present application relates to the field of tracheal navigation technology, and in particular to an automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structures. Background Art

[0002] Endotracheal intubation is a common operation in surgical general anesthesia, clinical emergency and intensive care. It is the process of inserting an endotracheal tube into the trachea through the patient's glottis. It is crucial for maintaining the patient's airway patency and performing mechanical ventilation.

[0003] However, traditional endotracheal intubation relies heavily on the selection of intubation tools, the mastery of intubation tools and the doctor's experience and judgment. At the same time, most devices still have problems such as complex operation, insufficient navigation accuracy and poor adaptability to the patient's anatomical structure.

[0004] In summary, there are technical problems in the prior art that the tracheal intubation operation relies on the doctor's experience and skills, resulting in a low success rate of intubation in emergency situations, which further affects the reliability and success rate of tracheal intubation in emergency rescue and complex scenarios. Summary of the invention

[0005] The purpose of this application is to provide an automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structures, in order to solve the technical problem in the prior art that the tracheal intubation operation relies on the doctor's experience and skills, resulting in a low success rate of intubation in emergency situations, further affecting the reliability and success rate of tracheal intubation in emergency rescue and complex scenarios.

[0006] In view of the above problems, the present application provides an automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structure, including: obtaining real-time airway image information by performing image scanning on the user; constructing a human airway image anatomical model, and performing anatomical structure analysis on the real-time airway image information according to the human airway image anatomical model to obtain anatomical structure data; connecting an electromagnetic navigation mechanical control system, the electromagnetic navigation mechanical control system including a path planning subsystem and a path control subsystem, wherein the electromagnetic navigation mechanical control system is connected to the human airway image anatomical model; the path planning subsystem analyzes the anatomical structure data to obtain a first intubation navigation path; the path control subsystem synchronously generates intubation control parameters according to the first intubation navigation path, and controls the mechanical arm of the intubation equipment according to the intubation control parameters to perform tracheal intubation on the user.

[0007] In a possible implementation, the electromagnetic navigation mechanical control system also includes an electromagnetic navigation feedback module, and the method includes: the electromagnetic navigation feedback module includes a navigation magnet and a wireless transmission module; the electromagnetic navigation feedback module is used to monitor the intubation process in real time to obtain an electromagnetic positioning sensor data set; according to the first intubation navigation path, the electromagnetic positioning sensor data set is compared for deviation to obtain a deviation sensor data set, and the intubation control parameters are adjusted for real-time feedback using the deviation sensor data set.

[0008] In a possible implementation, the method also includes: performing anatomical structure analysis on the real-time airway image information according to the human airway image anatomical model, the method includes: constructing a human airway image anatomical model, wherein the human airway image anatomical model is obtained by acquiring case image data samples and training pixel features representing each anatomical structure in the case image data samples; the human airway image anatomical model is identified according to the pixel features of the real-time airway image information to obtain anatomical structure annotation results; based on the anatomical structure annotation results, the features of each anatomical structure are extracted as the anatomical structure data output.

[0009] In a possible implementation, the path planning subsystem analyzes the anatomical structure data to obtain a first intubation navigation path, and the method includes: the path planning subsystem includes a single-lumen tracheal path planning branch and a double-lumen tracheal path planning branch, and the double-lumen tracheal path planning branch includes left tracheal path planning and right tracheal path planning; feature recognition is performed on the anatomical structure data, and if it is a single-lumen trachea, the anatomical structure data is identified according to the single-lumen trachea path planning branch to obtain the first intubation navigation path; if it is a double-lumen trachea, the anatomical structure data is identified according to the double-lumen trachea path planning branch to obtain a left intubation navigation path and a right intubation navigation path, and the first intubation navigation path is output according to the left intubation navigation path and the right intubation navigation path.

[0010] In a possible implementation, it also includes: calling the path operation model in the path control subsystem to decompose the output first intubation navigation path to determine multiple key target positions; identifying multiple intubation control parameters corresponding to the multiple key target positions, performing operations based on the multiple intubation control parameters, and generating intubation control parameters.

[0011] In a possible implementation, the method for training a path operation model includes: initializing a convolutional neural network, wherein the convolutional neural network is composed of a plurality of convolutional layers, pooling layers, and fully connected layers; obtaining labeled anatomical structure data samples, intubation path samples, and target position labels for each path, wherein the labels of the target position labels include the deflection angle, advancement distance, and intubation depth of each key target position; using the labeled anatomical structure data samples, intubation path samples, and target position labels for each path as training data sets, performing cross-entropy loss supervised learning on the convolutional neural network, outputting a trained path operation model, and downloading the trained path operation model to the path manipulation subsystem.

[0012] In a possible implementation, the real-time airway image information is acquired by performing an image scan on the user, including an image data set starting from the user's nasal cavity level to the left and right main bronchi level.

[0013] In a possible implementation, after acquiring the real-time airway image information, the method further includes: performing image processing on the real-time airway image information, including noise removal, edge enhancement, and image contrast, to acquire processed real-time airway image information; performing anatomical structure analysis on the processed real-time airway image information according to the human airway image anatomical model to acquire anatomical structure data.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0015] Real-time airway image information is acquired by performing an image scan on the user; a human airway image anatomical model is constructed, and an anatomical structure analysis is performed on the real-time airway image information according to the human airway image anatomical model to acquire anatomical structure data; an electromagnetic navigation mechanical control system is connected, and the electromagnetic navigation mechanical control system includes a path planning subsystem and a path control subsystem, wherein the electromagnetic navigation mechanical control system is connected to the human airway image anatomical model; the path planning subsystem analyzes the anatomical structure data to acquire a first intubation navigation path; the path control subsystem synchronously generates intubation control parameters according to the first intubation navigation path, and controls the mechanical arm of the intubation equipment to perform tracheal intubation on the user according to the intubation control parameters. That is to say, by realizing the technical goal of automation and intelligence of the tracheal intubation process, the technical effect of improving the success rate of intubation, reducing dependence on the doctor's personal skills, and improving the efficiency of emergency rescue is achieved.

[0016] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0018] Figure 1 This is a flowchart of the automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structure in this application;

[0019] Figure 2 This is a schematic diagram of the process of performing anatomical structure analysis on the real-time airway image information in the automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structure in this application. DETAILED DESCRIPTION

[0020] This application provides an automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structures, which solves the technical problem that the tracheal intubation operation relies on the experience and skills of doctors, resulting in a low success rate of intubation in emergency situations, which further affects the reliability and success rate of tracheal intubation in emergency rescue and complex scenarios. The technical goal of achieving automation and intelligence of the tracheal intubation process is achieved, and the technical effects of improving the success rate of intubation, reducing dependence on doctors' personal skills, and improving the efficiency of emergency rescue are achieved.

[0021] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0022] Please see attached Figure 1The present application provides an automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structure, specifically including:

[0023] Step 1: Obtain real-time airway image information by performing an image scan on the user.

[0024] Specifically, the user is a patient who is to undergo endotracheal intubation. Through the collection of multimodal imaging data, real-time image scanning of the patient is performed to obtain accurate airway anatomical information, help display the anatomical details inside the airway, and provide an accurate reference for the endotracheal intubation process. For example, image scanning technology includes a combination of chest X-rays, CT, MRI, electronic laryngoscopes, and other high-resolution imaging equipment to provide clear layered airway images and display the shape and position of the trachea, bronchi, etc. Taking CT scanning as an example, by scanning from the nasal level to the left and right main bronchi in sequence, or scanning from the user's oral level to the left and right main bronchi level, the image data set can obtain complete airway internal anatomical information based on the scanning resolution preset by technicians in this field. Through the acquired image data, the airway path is identified in real time, and subsequent navigation operations are assisted, making the endotracheal intubation operation under different patient anatomical conditions more accurate and personalized.

[0025] Step 2: construct a human airway image anatomical model, perform anatomical structure analysis on the real-time airway image information according to the human airway image anatomical model, and obtain anatomical structure data.

[0026] Specifically, the real-time airway image information is processed, and the airway image data is used to dissect the airway structure to form a human airway image anatomical model containing anatomical features, which is used to identify different structures of the airway, such as the glottis, trachea, and bronchi, as anatomical structure data. The human airway image anatomical model is trained based on the anatomical feature data of the airway in the case sample, so that the human airway image anatomical model can accurately reflect the anatomical structure. In subsequent applications, the anatomical model can analyze the airway image information acquired in real time, help quickly determine the airway structure during intubation, and improve the accuracy and safety of the operation.

[0027] Step three: connecting an electromagnetic navigation mechanical manipulation system, wherein the electromagnetic navigation mechanical manipulation system includes a path planning subsystem and a path manipulation subsystem, wherein the electromagnetic navigation mechanical manipulation system is connected to the human airway imaging anatomical model.

[0028] Specifically, the electromagnetic navigation mechanical control system is connected so that the electromagnetic navigation mechanical control system can perform precise intubation operations according to the anatomical structure of the airway. The electromagnetic navigation mechanical control system includes a path planning subsystem and a path control subsystem. The path planning subsystem determines the most appropriate intubation path based on the anatomical data provided by the airway imaging anatomical model, such as adjusting the intubation angle and depth. The path control subsystem is responsible for executing specific actions in the path planning, and ensures the smooth completion of the intubation by adjusting the position, angle and depth of the mechanical equipment in real time. In addition, the electromagnetic navigation mechanical control system is connected to the human airway imaging anatomical model, so that the path planning and control can adapt to the changes in the patient's anatomical structure, thereby ensuring the accuracy and flexibility of the intubation operation.

[0029] Step 4: The path planning subsystem analyzes the anatomical structure data to obtain a first intubation navigation path.

[0030] Specifically, the path planning subsystem analyzes the anatomical data to identify features such as the shape, size, and position of the airway, thereby generating the first intubation navigation path. The anatomical data contains specific anatomical information about the airway, such as the width of the trachea, the angle of bifurcation, and the position of the openings of the left and right upper lobe bronchus and the distance from the carina, which helps to accurately plan the intubation path. By analyzing the anatomical data, the optimal insertion angle and insertion depth are obtained, thereby forming a suitable intubation navigation path to ensure that the intubation path conforms to the specific conditions of the patient's airway, thereby improving the success rate and safety of intubation.

[0031] Step 5: The path control subsystem synchronously generates intubation control parameters according to the first intubation navigation path, and controls the robotic arm of the intubation device to perform tracheal intubation on the user according to the intubation control parameters.

[0032] Specifically, the path control subsystem synchronously generates intubation control parameters according to the first intubation navigation path, such as information including intubation angle, depth, and speed, for guiding the robotic arm of the intubation device to perform the intubation operation. By following the intubation control parameters, the robotic arm can accurately and smoothly insert the intubation instrument into the patient's trachea, thereby improving the accuracy and safety of the operation.

[0033] The automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structure can achieve the technical goal of automation and intelligence of the tracheal intubation process, and achieve the technical effects of improving the success rate of intubation, reducing dependence on the doctor's personal skills, and improving the efficiency of emergency rescue.

[0034] Furthermore, the present application also includes:

[0035] The electromagnetic navigation feedback module includes a navigation magnet and a wireless transmission module; the electromagnetic navigation feedback module is used to monitor the intubation process in real time and obtain an electromagnetic positioning sensor data set; according to the first intubation navigation path, the electromagnetic positioning sensor data set is compared for deviation to obtain a deviation sensor data set, and the intubation control parameters are adjusted for real-time feedback using the deviation sensor data set.

[0036] Specifically, the electromagnetic navigation feedback module consists of a navigation magnet and a wireless transmission module. The navigation magnet is responsible for generating a stable electromagnetic signal during the intubation process, and the wireless transmission module transmits the positioning data to the electromagnetic navigation mechanical control system in real time. Through the electromagnetic navigation feedback module, the intubation process is monitored in real time to obtain an electromagnetic positioning sensor data set, including parameters such as instrument position, direction, and movement speed. For example, when the position of the intubation instrument deviates slightly, the angle and distance data of the deviation are immediately recorded.

[0037] Next, the acquired electromagnetic positioning sensing data set is compared with the first intubation navigation path to identify the deviation of the instrument and form a deviation sensing data set. The deviation sensing data set contains the real-time deviation of the intubation instrument relative to the planned path. For example, if the deviation is three millimeters, the data set will be recorded and used for further adjustment. Finally, the deviation sensing data set is used to make real-time feedback adjustments to the intubation control parameters to automatically correct the intubation direction, depth or speed to ensure that the operation is always consistent with the navigation path.

[0038] Through real-time monitoring and feedback adjustment of deviation data, precise control of intubation instruments can be achieved, thereby improving the success rate and safety of intubation.

[0039] Further, if Figure 2 As shown, this application also includes:

[0040] A human airway imaging anatomical model is constructed, wherein the human airway imaging anatomical model is obtained by acquiring case image data samples and training pixel features representing each anatomical structure in the case image data samples; the human airway imaging anatomical model is identified according to the pixel features of the real-time airway image information to obtain anatomical structure annotation results; and based on the anatomical structure annotation results, the features of each anatomical structure are extracted as the anatomical structure data output.

[0041] Specifically, case image data samples are obtained, such as medical records of patients who have undergone tracheal intubation in the past obtained through CT or MRI scans. The case image data samples can show the various anatomical structures of the airway in detail. The case image data samples contain pixel features of each anatomical part, such as the width of the trachea, etc. Through the training of pixel features, the human airway image anatomical model can learn and form an accurate anatomical structure representation. Among them, by dividing the pixel features into training data and verification data, the human airway image anatomical model is trained with the training data. When the output data of the human airway image anatomical model tends to converge, the human airway image anatomical model is verified with the verification data. When the output accuracy of the human airway image anatomical model meets the preset output accuracy threshold, the training of the human airway image anatomical model is completed.

[0042] After obtaining real-time airway image information, the pixel features of the real-time airway image information are identified through the human airway image anatomical model. Pixel features refer to the brightness and color values ​​of different parts of the image, which help the human airway image anatomical model to identify the various anatomical structures of the airway, such as the larynx, trachea and bronchi, etc., and can timely reflect the patient's anatomical condition and ensure accurate navigation during intubation.

[0043] Finally, based on the annotation results of the anatomical structures, the features of each anatomical structure are extracted and output as anatomical structure data to help achieve higher automation and accuracy in subsequent tracheal intubation operations.

[0044] By constructing a human airway imaging anatomical model, accurate identification and data output of the airway anatomical structure can be achieved, ensuring accurate understanding and identification of each anatomical structure during intubation, thereby improving the success rate and safety of tracheal intubation.

[0045] Furthermore, the present application also includes:

[0046] The path planning subsystem includes a single-lumen trachea path planning branch and a double-lumen trachea path planning branch, and the double-lumen trachea path planning branch includes a left trachea path planning and a right trachea path planning; feature recognition is performed on the anatomical structure data, and if it is a single-lumen trachea, the anatomical structure data is identified according to the single-lumen trachea path planning branch to obtain a first intubation navigation path; if it is a double-lumen trachea, the anatomical structure data is identified according to the double-lumen trachea path planning branch to obtain a left intubation navigation path and a right intubation navigation path, and the first intubation navigation path is output according to the left intubation navigation path and the right intubation navigation path.

[0047] Specifically, the path planning subsystem includes a single-lumen tracheal path planning branch and a double-lumen tracheal path planning branch, where the double-lumen branch is further divided into left tracheal path planning and right tracheal path planning. The single-lumen tracheal path planning branch is suitable for single-lumen tracheal intubation paths, such as the intubation path used in emergency intubation; the double-lumen path planning branch is suitable for situations where separate ventilation of the left and right main bronchi is required, such as in thoracic surgery where intubation to the left and right trachea is required.

[0048] The anatomical structure data is feature-recognized to determine whether the tracheal structure is a single-lumen trachea or a double-lumen trachea. If the anatomical structure is identified as a single-lumen trachea, the single-lumen tracheal path planning branch is used to determine the insertion angle and insertion depth of the instrument based on the anatomical data, and generate an adaptive first intubation navigation path to ensure tracheal patency.

[0049] If the anatomical structure is identified as a double-lumen trachea, the double-lumen trachea path planning branch is used to plan the path. Based on the anatomical features of the left and right tracheas, a left intubation navigation path and a right intubation navigation path are generated respectively. For example, the left trachea path planning includes the insertion angle and insertion depth of the left trachea, while the right trachea path includes the insertion angle and insertion depth of the right trachea. The path data of the left and right intubations are integrated and output to generate a first intubation navigation path so that the device can accurately locate the direction and depth of the double-lumen tracheal tube.

[0050] The appropriate path planning branch is selected according to the trachea type, and the navigation path suitable for single-lumen or double-lumen trachea is generated through feature recognition of anatomical structure data to ensure the accuracy and applicability of the intubation path.

[0051] Furthermore, the present application also includes:

[0052] The path operation model in the path control subsystem is called to decompose the output first intubation navigation path to determine multiple key target positions; multiple intubation control parameters corresponding to the multiple key target positions are identified, and operations are performed based on the multiple intubation control parameters to generate intubation control parameters.

[0053] Specifically, the first intubation navigation path is decomposed by the path operation model in the path manipulation subsystem, and the continuous intubation path is disassembled into several key target positions. Each target position represents an important node in the intubation process, such as a turning point or a depth control point, to ensure smooth intubation. For example, the path decomposition may include four target positions, including the starting point, the angle deflection curvature of the front end of the catheter at the root of the tongue, the catheter deflection angle at the entrance of the glottis and the middle position of the distance from the vocal cords on both sides, and the final intubation position.

[0054] Next, the corresponding intubation control parameters for each key target position are identified, including angle adjustment, speed control, and depth limit, to ensure accurate operation during intubation. For example, the control parameters at the starting point may be an angle of zero degrees and a speed of one centimeter per second, while the angle deflection of the tip of the catheter at the base of the tongue is adjusted to thirty degrees and the speed is increased to one centimeter per second.

[0055] In the process of generating intubation control parameters, the path control subsystem integrates the parameters of all key target positions and converts these control parameters into motion instructions for the robotic arm to ensure that the intubation instrument executes as planned along the entire path.

[0056] Through path decomposition and step-by-step generation of control parameters, the intubation path is carefully controlled, thereby helping each operation node during the intubation process to maintain accuracy and stability.

[0057] Furthermore, the present application also includes:

[0058] Initialize a convolutional neural network, which is composed of multiple convolutional layers, pooling layers and fully connected layers; obtain labeled anatomical structure data samples, intubation path samples and target position labels of each path, wherein the target position labels include the deflection angle, advance distance and intubation depth of each key target position; use the labeled anatomical structure data samples, intubation path samples and target position labels of each path as training data sets, perform cross-entropy loss supervised learning on the convolutional neural network, output the trained path operation model, and download the trained path operation model to the path manipulation subsystem.

[0059] Specifically, initializing a convolutional neural network means setting up a neural network model with multiple convolutional layers, pooling layers, and fully connected layers, where the convolutional layer is used to extract local features of the input data, the pooling layer performs dimensionality reduction on the features, and the fully connected layer further integrates the extracted features for output results.

[0060] Then, we obtain the labeled anatomical structure data samples, intubation path samples and target position labels for each path. Each target position label records the deflection angle, advancement distance and intubation depth of the key points in detail. For example, the deflection angle of a target position may be 20 degrees, the advancement distance is 3 centimeters, and the intubation depth is 20 centimeters. The label data provides the learning target for the neural network.

[0061] Next, the labeled anatomical structure data samples, intubation path samples, and target position labels are used as training data sets for cross-entropy loss supervised learning, and the model is optimized by minimizing the difference between the predicted output and the actual label. The convolutional neural network continuously learns how to identify the target position in different paths by gradually adjusting parameters. Finally, after the training is completed, a model with precise path operation capabilities is output and downloaded to the path manipulation subsystem for subsequent intubation path planning and control.

[0062] By initializing the convolutional neural network and collecting annotated data samples and labels for supervised learning, an optimized path operation model can be constructed, which helps to achieve automation and precise control of intubation operations.

[0063] Furthermore, the present application also includes:

[0064] The real-time airway image information is obtained by performing image scanning on the user, including an image data set starting from the user's nasal cavity level to the left and right main bronchial levels.

[0065] Specifically, an imaging device is used to scan the user's airway in real time to collect anatomical image data. Tracheal intubation includes oral intubation and nasal intubation. When performing oral intubation, the image scan starts from the user's oral level, and airway images are collected layer by layer until the left and right main bronchi of the tracheal branches are covered; when performing nasal intubation, the image scan starts from the user's nasal level, and airway images are collected layer by layer until the left and right main bronchi of the tracheal branches are covered, ensuring that the entire intubation path is covered. The scanning process collects image data of each layer to form an image data set, which is used to build a complete airway structure model of the user. For example, starting from the nasal cavity, the depth of each layer increases by one centimeter until the main bronchi are located, forming a detailed airway anatomical image.

[0066] By scanning the user's airway, complete image acquisition from the nasal cavity to the left and right main bronchi is achieved, generating a real-time image data set to provide support for the planning and control of subsequent intubation paths.

[0067] Furthermore, the present application also includes:

[0068] The real-time airway image information is subjected to image processing, including noise removal, edge enhancement and image contrast, to obtain processed real-time airway image information; and the processed real-time airway image information is subjected to anatomical structure analysis according to the human airway image anatomical model to obtain anatomical structure data.

[0069] Specifically, image processing of real-time airway image information refers to multi-step optimization of the collected airway image data in order to more clearly identify key anatomical features. First, noise removal is used to remove impurities or unnecessary details in the image to improve the clarity of the image. For example, interfering particles smaller than 0.1 cm can be filtered out to make the airway edges smoother. Next, edge enhancement processing is performed to make the boundaries of the anatomical structures in the image clearer, such as increasing the boundary line between the airway and the esophagus to twice the clarity of the original image. Finally, the image details are further optimized by adjusting the image contrast to enhance the resolution of each layer of the structure, such as increasing the image contrast by 30% to ensure the visibility of anatomical details.

[0070] Performing anatomical analysis on processed real-time airway image information based on the human airway image anatomical model refers to analyzing the processed image data using a pre-built airway anatomical model. The anatomical model matches each anatomical feature with the labeled structure in the model to identify key airway structures, such as the location and size of the trachea, left and right main bronchi, etc. The anatomical data generated by the analysis contains specific information about the real-time airway structure, such as the tracheal width of 1.2 cm, the angle between the left main bronchus and the midline of the trachea is 35 degrees, and the angle between the right main bronchus and the midline of the trachea is 25 degrees.

[0071] By optimizing the real-time airway images, clear images were obtained through noise removal, edge enhancement and contrast adjustment. The images were then analyzed and processed through the anatomical model to accurately identify the specific anatomical features of each part of the airway, and the anatomical structure data was output, providing basic data for further path planning and operational control.

[0072] In summary, the endotracheal intubation automated navigation method based on artificial intelligence analysis of anatomical structure provided in this application has the following technical effects:

[0073] Real-time airway image information is acquired by performing an image scan on the user; a human airway image anatomical model is constructed, and an anatomical structure analysis is performed on the real-time airway image information according to the human airway image anatomical model to acquire anatomical structure data; an electromagnetic navigation mechanical control system is connected, and the electromagnetic navigation mechanical control system includes a path planning subsystem and a path control subsystem, wherein the electromagnetic navigation mechanical control system is connected to the human airway image anatomical model; the path planning subsystem analyzes the anatomical structure data to acquire a first intubation navigation path; the path control subsystem synchronously generates intubation control parameters according to the first intubation navigation path, and controls the mechanical arm of the intubation equipment to perform tracheal intubation on the user according to the intubation control parameters. That is to say, by realizing the technical goal of automation and intelligence of the tracheal intubation process, the technical effect of improving the success rate of intubation, reducing dependence on the doctor's personal skills, and improving the efficiency of emergency rescue is achieved.

[0074] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0075] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. An automated navigation method for tracheal intubation based on artificial intelligence analysis of anatomical structure, characterized in that: Methods include: Obtain real-time airway image information by performing image scanning on the user; Constructing a human airway image anatomical model, and performing anatomical structure analysis on the real-time airway image information according to the human airway image anatomical model to obtain anatomical structure data; Connecting an electromagnetic navigation mechanical manipulation system, the electromagnetic navigation mechanical manipulation system comprising a path planning subsystem and a path manipulation subsystem, wherein the electromagnetic navigation mechanical manipulation system is connected to the human airway imaging anatomical model; The path planning subsystem analyzes the anatomical structure data to obtain a first intubation navigation path; The path control subsystem synchronously generates intubation control parameters according to the first intubation navigation path, and controls the robotic arm of the intubation device to perform tracheal intubation on the user according to the intubation control parameters.

2. The method according to claim 1, characterized in that The electromagnetic navigation mechanical control system further includes an electromagnetic navigation feedback module, and the method includes: The electromagnetic navigation feedback module includes a navigation magnet and a wireless transmission module; The electromagnetic navigation feedback module is used to monitor the intubation process in real time and obtain an electromagnetic positioning sensor data set; According to the first intubation navigation path, a deviation comparison is performed on the electromagnetic positioning sensor data set to obtain a deviation sensor data set, and the intubation control parameters are adjusted in real time based on the deviation sensor data set.

3. The method according to claim 1, characterized in that Performing anatomical structure analysis on the real-time airway image information according to the human airway image anatomical model, the method comprising: Constructing a human airway imaging anatomical model, wherein the human airway imaging anatomical model is obtained by acquiring case imaging data samples and training pixel features representing various anatomical structures in the case imaging data samples; The human airway image anatomical model is identified according to the pixel features of the real-time airway image information to obtain an anatomical structure annotation result; According to the anatomical structure annotation result, the features of each anatomical structure are extracted and output as the anatomical structure data.

4. The method according to claim 1, characterized in that The path planning subsystem analyzes the anatomical structure data to obtain a first intubation navigation path, the method comprising: The path planning subsystem includes a single-lumen tracheal path planning branch and a double-lumen tracheal path planning branch, and the double-lumen tracheal path planning branch includes a left tracheal path planning and a right tracheal path planning; Performing feature recognition on the anatomical structure data, and if it is a single-lumen trachea, recognizing the anatomical structure data according to the single-lumen trachea path planning branch, and acquiring a first intubation navigation path; If it is a double-lumen trachea, the anatomical structure data is identified according to the double-lumen trachea path planning branch, a left intubation navigation path and a right intubation navigation path are obtained, and a first intubation navigation path is output according to the left intubation navigation path and the right intubation navigation path.

5. The method according to claim 1, characterized in that The method also includes: Calling the path operation model in the path control subsystem to perform path decomposition on the output first intubation navigation path to determine a plurality of key target positions; A plurality of cannula control parameters corresponding to the plurality of key target locations are identified, and operations are performed based on the plurality of cannula control parameters to generate cannula control parameters.

6. The method according to claim 5, characterized in that Methods for training path operation models include: Initialize a convolutional neural network, wherein the convolutional neural network is composed of a plurality of convolutional layers, a pooling layer, and a fully connected layer; Obtain annotated anatomical structure data samples, intubation path samples, and target position labels for each path, wherein the target position labels include a deflection angle, an advancement distance, and an intubation depth for each key target position; The labeled anatomical structure data samples, intubation path samples and the target position labels of each path are used as training data sets, and the convolutional neural network is subjected to cross-entropy loss supervised learning, and the trained path operation model is output, and the trained path operation model is downloaded to the path manipulation subsystem.

7. The method according to claim 1, characterized in that, The real-time airway image information is obtained by performing image scanning on the user, including an image data set starting from the user's nasal cavity level to the left and right main bronchial levels.

8. The method according to claim 1, characterized in that After acquiring the real-time airway image information, the method further includes: Performing image processing on the real-time airway image information, including noise removal, edge enhancement and image contrast, to obtain processed real-time airway image information; The processed real-time airway image information is subjected to anatomical structure analysis according to the human airway image anatomical model to obtain anatomical structure data.

Images