WO2024021534A1 - 基于人工智能的气道评估终端 - Google Patents

基于人工智能的气道评估终端 Download PDF

Info

Publication number
WO2024021534A1
WO2024021534A1 PCT/CN2023/070918 CN2023070918W WO2024021534A1 WO 2024021534 A1 WO2024021534 A1 WO 2024021534A1 CN 2023070918 W CN2023070918 W CN 2023070918W WO 2024021534 A1 WO2024021534 A1 WO 2024021534A1
Authority
WO
WIPO (PCT)
Prior art keywords
assessment
airway
data
oral
facial
Prior art date
Application number
PCT/CN2023/070918
Other languages
English (en)
French (fr)
Inventor
李文献
韩园
赵柏杨
王轶湛
Original Assignee
复旦大学附属眼耳鼻喉科医院
上海兰甲医疗科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 复旦大学附属眼耳鼻喉科医院, 上海兰甲医疗科技有限公司 filed Critical 复旦大学附属眼耳鼻喉科医院
Publication of WO2024021534A1 publication Critical patent/WO2024021534A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
    • A61B5/087—Measuring breath flow
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H15/00—ICT specially adapted for medical reports, e.g. generation or transmission thereof
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Definitions

  • the present invention relates to the field of artificial intelligence medical treatment, and in particular to an airway assessment terminal based on artificial intelligence.
  • Anesthesia is the process of using drugs to maintain a patient in a painless, unconscious, and muscle-relaxed state to facilitate the performance of invasive procedures such as surgery. After anesthesia, the patient is in a state of respiratory depression or cessation. If the existing difficult airway is not diagnosed in advance, the patient is very likely to suffocate and die due to the inability to establish an airway for ventilation. Through airway specialist evaluation, most difficult airways can be diagnosed in time before anesthesia, thereby avoiding the occurrence of the above critical situations. Therefore, accurate and rigorous airway assessment for each patient is an important guarantee to avoid the occurrence of unexpected difficult airway.
  • the purpose of the present invention is to provide an airway assessment terminal based on artificial intelligence to solve the above technical problems in the prior art.
  • the present invention provides an airway assessment terminal based on artificial intelligence.
  • the terminal includes: a facial assessment module for obtaining a facial model based on a three-dimensional facial model constructed from the collected facial data of a target object.
  • an oral assessment module is used to obtain oral key information for airway assessment based on the collected oral image data of the target object under set oral assessment actions information, and obtain the corresponding oral assessment results
  • the dynamic physical assessment module is used to obtain dynamic physical key information for airway assessment based on the captured image data of the target object under the set physical assessment action, and obtain the corresponding Dynamic physical assessment results
  • a respiratory assessment module for obtaining respiratory assessment results for airway assessment based on the collected respiratory sound wave data of the target object
  • a glottis assessment module for obtaining the throat anatomy of the target object based on the collected Structural image data obtains key information about the glottis for airway assessment and obtains corresponding glottis assessment results
  • a basic information entry module is used to enter basic information about the target object; a report generation module connects each evaluation module and the basic The information entry module is used to generate a vocal tract assessment report corresponding to the target object based on the facial assessment results, oral assessment results, dynamic
  • obtaining facial key feature information for airway assessment based on a three-dimensional facial model constructed from collected facial data of the target object, and obtaining corresponding facial assessment results includes: obtaining facial features based on infrared points Facial data of the target object collected by array technology; wherein the facial data includes: real distances corresponding to multiple facial landmark points; a distance-based three-dimensional facial model is constructed based on the facial data; based on the three-dimensional facial model, Capture multiple difficult airway judgment feature points and calculate key facial information for airway assessment; wherein the key facial information includes: the true distance between each difficult airway judgment feature point; according to the facial Key Information Perform airway assessment to obtain facial assessment results.
  • obtaining key oral information for airway assessment based on the collected oral image data of the target object under set oral assessment actions, and obtaining the corresponding oral assessment results includes: obtaining The collected oral image data of the target object under the set oral assessment action; wherein the set oral assessment action includes: mouth opening action and mouth closing action; based on the oral key information recognition model, obtained according to the oral image data Key oral information for airway assessment; wherein, the key oral information includes: the maximum pixel distance between the upper and lower mouth when opening and closing the mouth, and the pixel distance between the left and right opening and closing of the mouth; airway assessment is performed based on the key oral information, To obtain oral assessment results.
  • obtaining dynamic physical key information for airway assessment based on the captured image data of the target object under a set physical assessment action, and obtaining the corresponding dynamic physical assessment results includes: Obtain the captured image data of the target object under a set physical assessment action; wherein the set physical assessment action includes: a left and right neck rotation action and a neck pitching action; and a calibration point in the image data and Multiple reference points are positioned, and dynamic physique key information is obtained based on the fixed position between the calibration point and each reference point, and the relative position change of the calibration point; wherein the dynamic physique key information includes: neck The maximum angle of left and right rotation and the maximum angle of neck pitch movement; conduct airway assessment based on the key information of dynamic physique to obtain dynamic physique assessment results.
  • the calibration point is the tip of the nose point.
  • obtaining a respiratory assessment result for airway assessment based on the collected respiratory sound wave data of the target object includes: acquiring and collecting the breathing of the target object under a set respiratory assessment action. Sound wave data; wherein, the respiratory assessment action includes: exhalation action and inhalation action; identify abnormal waveform data in the breath sound wave data, and obtain the respiratory assessment result.
  • obtaining the key information of the glottis for airway assessment based on the collected image data of the throat anatomy of the target object, and obtaining the corresponding glottis evaluation results includes: obtaining the collected target object The image data of the anatomical structure of the throat; identify the glottis in the image data of the anatomical structure of the throat to obtain image data of the glottis corresponding to the complete exposure of the glottis; extract key information of the glottis from the image data of the glottis, and obtain glottis assessment results.
  • the data warehouse module includes: a data receiving unit, used to receive data from each evaluation module and the basic information entry module; a data processing unit, connected to the data receiving unit, used to process the received data The data are cleaned and organized according to their corresponding data types for uploading to the visual data center for visual display; the data storage unit is connected to the data receiving unit and is used to store the data of each evaluation module and the basic information entry module; The data analysis unit is connected to the data storage unit and is used to analyze and compare the stored data to obtain the weight coefficient ratio of each data type in the airway assessment.
  • generating a vocal tract assessment report corresponding to the target object based on the facial assessment results, oral assessment results, dynamic physical assessment results, respiratory assessment results, glottis assessment results and basic information includes: : Obtain corresponding vocal tract assessment results based on the facial assessment results, oral assessment results, dynamic physical assessment results, respiratory assessment results, glottis assessment results and basic information, and generate a vocal tract assessment report of the target object.
  • the basic information includes: age, height, weight, BMI, smoking history, airway surgery history, radiotherapy history, dysphagia, ringing, airway compression, and tracheotomy history.
  • the present invention is an airway assessment terminal based on artificial intelligence, which has the following beneficial effects: the present invention performs facial assessment through 3D reconstruction of collected facial data; and mathematicalizes the visual environment of the oral cavity through oral recognition technology. Analysis and machine learning for oral assessment; dynamic physical assessment through the overall dynamic capture of the human head; and lung function assessment through audio frequency analysis and machine learning of respiratory airflow, airflow changes generated throughout the breathing process ; Carry out glottis assessment by performing glottis recognition on the collected image data of the throat anatomy of the target object; Finally, a vocal tract assessment report is generated based on the comprehensive judgment of the above assessment data and basic information, and based on the existing medical field for airway The evaluation judgment criteria are digitally analyzed to provide comprehensive airway evaluation recommendations for anesthesiologists.
  • the present invention greatly saves the time of airway assessment and improves the accuracy of difficult airway diagnosis. It not only solves the current problem of shortage of anesthesiology practitioners, but also provides the possibility of accurate diagnosis of difficult airway.
  • Figure 1 shows a schematic structural diagram of an artificial intelligence-based airway assessment terminal in an embodiment of the present invention.
  • Figure 2 shows a schematic waveform diagram of a respiratory curve in an embodiment of the present invention.
  • Figure 3 shows a schematic diagram of modeling in TensorFlow in an embodiment of the present invention.
  • Figure 4 shows a schematic diagram of model gradient descent in an embodiment of the present invention.
  • Figure 5 shows a schematic flowchart of an artificial intelligence-based airway assessment method in an embodiment of the present invention.
  • first, second and third mentioned herein are used to describe various parts, components, regions, layers and/or segments, but are not limited thereto. These terms are only used to distinguish one part, component, region, layer or section from another part, component, region, layer or section. Therefore, a first part, component, region, layer or section described below can be referred to as a second part, component, region, layer or section without departing from the scope of the invention.
  • A, B or C or "A, B and/or C” means "any of the following: A; B; C; A and B; A and C; B and C; A, B and C” . Exceptions to this definition occur only when combinations of elements, functions, or operations are inherently mutually exclusive in some manner.
  • Specific components of airway assessment typically include visual inspection and physical examination of the face, neck, mouth, and interior of the nose, both statically and with command movements.
  • visual inspection and doctor-patient dynamics combined with physical examination can effectively identify and determine the type, cause and degree of difficulty of the difficult airway.
  • the patient’s mouth opening, tooth occlusion, beard density, Mallampati grade and other indicators need to be judged by visual inspection of the head, face, neck and oral cavity, and the thyroid-mental distance, neck circumference, neck circumference, etc.
  • the range of motion of the head and neck can be measured through a series of anatomical landmark distance measurements to comprehensively diagnose the airway condition. Therefore, airway assessment not only takes up a lot of working time of anesthesiologists, but also often misses important examination items and data due to cumbersome steps, which may lead to missed diagnosis and misdiagnosis of difficult airway.
  • the present invention provides an airway assessment terminal based on artificial intelligence, which performs facial assessment through 3D reconstruction of collected facial data; and performs oral assessment through mathematical analysis and machine learning of the visual environment of the oral cavity through oral recognition technology; Dynamic physical assessment is carried out through the overall dynamic capture of the human head; through audio frequency analysis and machine learning of the respiratory airflow, the lung function is evaluated on the airflow changes generated during the entire breathing process; through the collected throat of the target object The glottis is evaluated based on the glottis recognition based on the image data of the anatomical structure. Finally, a vocal tract assessment report is generated through the comprehensive judgment of the above assessment data and basic information, and digital analysis is performed based on the existing judgment standards for airway assessment in the medical field.
  • the present invention greatly saves the time of airway assessment and improves the accuracy of difficult airway diagnosis. It not only solves the current problem of shortage of anesthesiology practitioners, but also provides the possibility of accurate diagnosis of difficult airway.
  • Figure 1 shows a schematic structural diagram of an artificial intelligence-based airway assessment terminal in an embodiment of the present invention.
  • the terminal includes:
  • the facial assessment module 11 is used to obtain key facial information for airway assessment based on a three-dimensional facial model constructed from the collected facial data of the target object, and obtain corresponding facial assessment results;
  • the oral assessment module 12 is configured to obtain key oral information for airway assessment based on the collected oral image data of the target object under set oral assessment actions, and obtain corresponding oral assessment results;
  • the dynamic physical assessment module 13 is configured to obtain key dynamic physical information for airway assessment based on the captured image data of the target object under the set physical assessment action, and obtain the corresponding dynamic physical assessment results;
  • Respiratory assessment module 14 configured to obtain respiratory assessment results for airway assessment based on the collected respiratory sound wave data of the target object
  • the glottis assessment module 15 is used to obtain key glottis information for airway assessment based on the collected image data of the throat anatomy of the target object, and obtain corresponding glottis assessment results;
  • the basic information entry module 16 is used to enter the basic information of the target object
  • the report generation module 17 is connected to the facial assessment module 11, the oral assessment module 12, the dynamic physical assessment module 13, the respiratory assessment module 14, the glottis assessment module 15, and the basic information entry module 16, and is used to based on the facial assessment results, oral
  • the assessment results, dynamic physical assessment results, respiratory assessment results, glottis assessment results and basic information generate a vocal tract assessment report corresponding to the target object;
  • the data warehouse module 18 is connected to the facial assessment module 11, oral assessment module 12, dynamic physical assessment module 13, respiratory assessment module 14, glottis assessment module 15, and basic information entry module 16, and is used to enter each assessment module and basic information. Module data is processed, stored and analyzed.
  • the facial assessment module 11 obtains key facial information for airway assessment based on a three-dimensional facial model constructed from the collected facial data of the target object, and obtains the corresponding facial assessment results including:
  • the facial data of the target object collected based on the infrared dot matrix technology includes: the real distance corresponding to multiple facial landmark points; specifically, the infrared dot matrix technology can collect each infrared sensor of the infrared dot matrix.
  • the actual time the point is projected onto the face of the target object is multiplied by the propagation speed of infrared light in the air, so as to accurately obtain the actual distance between each facial landmark point and the camera;
  • multiple difficult airway judgment feature points are captured, and key facial information for airway assessment is calculated; wherein the key facial information includes: between the difficult airway judgment feature points the real distance;
  • the facial assessment result may be whether the airway is difficult, the probability of the difficult airway, or the grade of the difficult airway.
  • infrared dot matrix sensor 30,000 infrared dots are projected onto the target face in an instant.
  • the infrared light from the face is accurately captured by an infrared camera and a computer begins to draw a depth map.
  • the sensor module of this device the distance to the patient's face is accurately calculated through the ToF (time of flight) sensor.
  • ToF time of flight
  • the oral assessment module 12 obtains key oral information for airway assessment based on the collected oral image data of the target object under set oral assessment actions, and obtains the corresponding oral assessment results including:
  • Obtain the collected oral image data of the target object under the set oral assessment action specifically, obtain the collected oral image data of the target object in the process of opening and closing the mouth;
  • key oral information for airway assessment is obtained according to the oral image data; wherein the key oral information includes: the maximum pixel distance between the upper and lower mouth opening and closing when the mouth is opened and closed, and the left and right opening and closing of the oral cavity. Pixel distance; specifically, the oral key information recognition model passes multiple oral image data marked with the position of the upper jaw to the lower jaw and the left and right positions of the oral cavity, as well as the corresponding maximum pixel distance between the upper and lower mouth opening and closing when the oral cavity is opened and closed, and the left and right opening and closing of the oral cavity. Pixel distance obtained through training.
  • airway assessment is performed based on the oral key information to obtain oral assessment results.
  • the results of the oral assessment can be whether the airway is difficult, the probability of the difficult airway, or the grade of the difficult airway.
  • obtaining key oral information for airway assessment based on the collected oral image data of the target subject under set oral assessment actions includes: learning the data values of the oral cavity through a computer and defining the pixel distance from the upper jaw to the lower jaw of the oral cavity. is: MouthT_Mouth_B; the pixel distance between the left and right positions of the mouth is defined as: MouthL_MouthR.
  • the computer captures the distance value of the oral image data in real time and creates two lists: List_D_MouthT_Mouth_B and List_D_MouthL_MouthR.
  • the distance is collected in real time, and the patient opens and closes his mouth under the guidance of the doctor.
  • the computer traverses the data captured in the list. Through the two values of Max and SubMax, the largest Max value in the entire list is filtered out, that is, the oral image data of the target object during the mouth opening and mouth closing movements.
  • the dynamic physique assessment module 13 obtains key dynamic physique information for airway assessment based on the captured image data of the target object under set physique assessment actions, and obtains the corresponding dynamic physique assessment results. include:
  • Obtain the captured image data of the target object under the set physical assessment action specifically, obtain the captured head image data of the target object while performing neck left and right rotation movements and neck pitching activities; what needs to be explained Yes, the left and right rotation of the neck needs to reach the maximum angle at the limit of the target's neck rotation, and the neck pitching movement needs to reach the maximum angle at the limit of the target's neck pitching.
  • the key information of the dynamic physique includes: the maximum angle of left and right rotation of the neck and the maximum angle of neck pitch rotation; it should be noted that the positions of each reference point and the calibration point are fixed, that is, no matter how the head moves, only the relative position Changes, the absolute positions of each reference point and calibration point will not change, and the accuracy of relative position changes can be ensured through absolute positions.
  • airway assessment is performed based on the dynamic physical key information, and dynamic physical assessment results are obtained.
  • the result of the dynamic physical assessment can be whether the patient has a difficult airway, the probability of a difficult airway, or the grade of the difficult airway based on the dynamic physical evaluation.
  • the correct position of the neck is located through artificial intelligence recognition.
  • the program uses the current position of the tip of the nose as the calibration point and takes one or more other feature points as the reference point.
  • the reference point is the middle of the left and right eyes.
  • the position is based on the two ends of the corners of the mouth; the positive left and right distance of the neck is set to a constant value ⁇ .
  • the program calculates and records the three-dimensional movement angle of the patient's head through feature points and inverse trigonometric function calculation formulas.
  • we will calculate the movement information of the Euler angles (pitch, roll, yaw) of the head on the three coordinate axes of x, y, and z respectively.
  • the vector can be obtained from the above formula:
  • the angle needs to be converted into quaternion.
  • the quaternion is expressed as:
  • the computer has completed the complete algorithm transformation of dynamic feature point tracking, quaternion grabbing, Euler angle conversion and radians to degrees. Based on the calculation of the maximum angle obtained by the computer, the patient's maximum left and right rotation angle of the neck and maximum movement angle of the spine's up and down pitch angle can be obtained in the data list.
  • the respiratory assessment module 14 obtains respiratory assessment results for airway assessment based on the collected respiratory sound wave data of the target object, including:
  • Identify abnormal waveform data in the breath sound wave data and obtain respiratory assessment results; specifically, identify irregular waveforms in the breath sound wave data that are not smooth, for example, fault waveforms, protruding waveforms including sharp waveforms, resonance waveforms, and Waveforms such as echo waveforms; the respiratory assessment result may be whether the respiratory tract is determined to be a difficult airway, or the probability of a difficult airway, or the difficult airway judgment level.
  • the respiratory sound wave data of the lungs and upper respiratory tract is intercepted through an electronic stethoscope.
  • the subtle vibrations produced when air flows through the trachea and lungs are digitally recorded and visualized in the form of wavelength and frequency.
  • the computer also uses TensorFlow to learn and analyze the models generated when the patient "exhales” and "inhales", and calculates the curvature coefficient that best fits the curve change. Compare and judge possible dyspnea in the trachea and lungs during breathing through coefficient changes.
  • the model form of audio in a specific environment is defined through the microwave changes of the curve.
  • the audio changes generated during breathing are mainly concentrated in the high-frequency part.
  • the waveform of the breathing curve will be more obviously abnormal than that of normal people, that is, the trend corresponding to the curve does not match the curve expressed by the polynomial.
  • the glottis assessment module 15 obtains key glottis information for airway assessment based on the collected image data of the throat anatomy of the target object, and obtains the corresponding glottis assessment results including:
  • video laryngoscope video laryngoscope
  • the glottis in the image data of the anatomical structure of the throat is identified to obtain the glottis image data corresponding to the complete exposure of the glottis; specifically, through the convolutional neural network algorithm, the photos of the glottis structure are learned to achieve For automatic recognition of glottis. As soon as the glottis is discovered during inspection of the lens, the relevant identified parts will be photographed and saved to obtain glottis image data corresponding to the complete exposure of the glottis;
  • the glottis assessment result may be whether the airway is difficult to be determined through breathing, or the probability of a difficult airway, or the difficult airway determination level.
  • the key information of the glottis includes: the presence or absence of laryngeal neoplasms, laryngoscope exposure classification, glottis stenosis, subglottic stenosis, supraglottic stenosis and other information.
  • a video laryngoscope is used to collect image data of the throat anatomy of the target subject, by inserting a special laryngoscope with a monitor and a camera into a mirror body with a 90-degree angle at both ends, and arching the dorsal side of the mirror body toward the upper jaw.
  • a special laryngoscope with a monitor and a camera When inserted into the oral cavity, it can be used for throat examination.
  • the camera at the lens end can be pointed vertically at the throat and glottis. It can then follow the natural anatomical curvature to clearly display and collect various images in the oral cavity and throat. anatomical structure.
  • the basic information of the target object entered by the basic information entry module 16 includes: age, height, weight, BMI, smoking history, airway surgery history, radiotherapy history, dysphagia, thundering, and airway compression. And tracheotomy history; among them, each basic information can use different data types, as shown in the following table;
  • Table 1 Comparison table of basic information data types
  • the report generation module 17 generates a vocal tract assessment report corresponding to the target object based on the facial assessment results, oral assessment results, dynamic physical assessment results, respiratory assessment results, glottis assessment results and basic information. include:
  • the vocal tract assessment report includes: the vocal tract assessment results, facial assessment results, oral assessment results, dynamic physical assessment results, respiratory assessment results, glottis assessment results and basic information. It should be noted that the vocal tract evaluation result may be whether it is a difficult airway, or the probability of a difficult airway, or the difficult airway judgment level.
  • the vocal tract assessment report may also include the use of AI deep learning strategies to analyze the doctor's interpretation of the difficult airway classification, difficulty, and causes.
  • the data warehouse module 18 includes:
  • the data receiving unit is used to receive data from each evaluation module and the basic information entry module; including: facial key information and facial evaluation results from the facial evaluation module 11, oral key information and oral evaluation results from the oral evaluation module 12, Dynamic physical key information and dynamic physical assessment results from the dynamic physical assessment module 13, breath sound wave data and respiratory assessment results from the respiratory assessment module 14, glottal key information and glottal assessment results from the glottis assessment module 15 and from The basic information of the basic information entry module 16;
  • a data processing unit connected to the data receiving unit, is used to clean and organize the received data according to its corresponding data type for uploading to the visual data center for visual display; through sorting and cleaning, the program will Different types of data are processed accordingly. For example, BMI will be calculated based on basic physiological indicators such as age, height, and weight. After all data is finally cleaned and analyzed, all data is uploaded to the visual data center for display, ensuring that doctors can have a more complete real-time visualization of the data on the back end.
  • the data storage unit is connected to the data receiving unit and is used to store the data of each evaluation module and the basic information entry module; specifically, the facial key information and facial evaluation results from the facial evaluation module 11, the oral key information from the oral evaluation module information and oral assessment results, dynamic physical key information and dynamic physical assessment results from the dynamic physical assessment module, breath sound acoustic data and respiratory assessment results from the respiratory assessment module, glottal key information and glottis assessment from the glottis assessment module
  • the results and basic information from the basic information entry module are stored in the MySql database.
  • the data analysis unit is connected to the data storage unit and is used to analyze and compare the stored data to obtain the weight coefficient ratio of each data type in the airway assessment.
  • the program conducts centralized analysis and comparison of the collected data.
  • TensorFlow we use TensorFlow to learn the data and understand the linear relationship in the data. (Including but not limited to: age, height, weight, BMI, facial key information, oral key information, dynamic physique key information, breath sound wave data, glottis key information)
  • loss ie: loss function
  • the program will learn repeatedly, and by correcting the loss (ie: loss function), it can derive the coefficient value that best matches the data changes.
  • loss represents the error value between the computer predicted data y and the known data y_, that is: mean square error
  • the weight coefficient ratio of each data type in airway assessment Preferably, the facial assessment results, oral assessment results, dynamic physical assessment results, respiratory assessment results, glottis assessment results and the airway assessment weight of the basic information.
  • the present invention provides the following specific embodiments.
  • Embodiment 1 An airway assessment method using an artificial intelligence-based airway assessment terminal.
  • Figure 5 shows a schematic flow chart of the airway assessment method based on artificial intelligence; the method includes:
  • the patient is allowed to move the head and neck left and right and up and down, and dynamic data monitoring of the human body is carried out; that is, the head image data captured by the camera are used to track the dynamic feature points and capture the quaternion in turn.
  • the patient's maximum left and right rotation angle of the neck and maximum movement angle of the spine's up and down pitch angle can be obtained in the data list.
  • the visual environment of the oral cavity is mathematically analyzed and machine learned through oral recognition technology; that is, the distance is collected in real time through the camera, and the patient opens and closes his mouth under the guidance of the doctor.
  • the computer traverses the data captured in the list and selects the largest Max value in the entire list, that is: the maximum pixel distance between the upper and lower opening and closing of the mouth and the pixel distance between the left and right opening and closing of the mouth when the mouth is opened and closed. Record the corresponding coefficients and proportions;
  • the lung function is evaluated based on the changes in airflow generated during the entire breathing process; that is, the electronic stethoscope is used to intercept the breath sounds of the lungs and upper respiratory tract, and the computer also uses TensorFlow to analyze the patient's breath sounds.
  • the models generated when "exhaling” and “inhaling” are learned and analyzed, and the curvature coefficient that best fits the curve change is calculated. Compare and judge possible dyspnea in the trachea and lungs during breathing through coefficient changes.
  • the infrared dot matrix module can be used to collect the actual time each infrared point is projected onto the target face, and then multiply it by the propagation speed of infrared light in the air, thereby accurately The actual distance between the points in each facial landmark and the camera is obtained, thereby reconstructing a distance-based 3D facial feature model.
  • the machine After obtaining the above data, according to the traditional difficult airway judgment criteria, the machine combines all required points Grab and calculate the true distance between points.
  • a vocal tract evaluation report corresponding to the target object is generated based on the collected data.
  • This embodiment collects various static and dynamic physical data of the human body, applies infrared lattice technology, utilizes the visualization of the big data center and data learning in artificial intelligence algorithms, not only through artificial intelligence-related image processing technology and recognition technology Facial, oral, and dynamic physical examination data were collected and a detailed and comprehensive analysis of individual weighting factors was performed.
  • the algorithm is used to realize automatic collection and cooperate with the doctor's clinical operation to provide a comprehensive, intelligent and digital understanding and analysis of the airway assessment.
  • the airway assessment terminal based on artificial intelligence of the present invention performs facial assessment through 3D reconstruction of collected facial data; and performs oral assessment through mathematical analysis and machine learning of the visual environment of the oral cavity through oral recognition technology. ; Dynamic physical assessment through the overall dynamic capture of the human head; through audio frequency analysis and machine learning of respiratory airflow, lung function assessment of airflow changes generated during the entire breathing process; through the collection of target objects The throat anatomy image data is used to identify the glottis for glottis assessment; finally, a vocal tract assessment report is generated through the comprehensive judgment of the above assessment data and basic information, and digital analysis is performed based on the existing judgment standards for airway assessment in the medical field. This provides the anesthetist with comprehensive airway assessment recommendations.
  • the present invention greatly saves the time of airway assessment and improves the accuracy of difficult airway diagnosis. It not only solves the current problem of shortage of anesthesiology practitioners, but also provides the possibility of accurate diagnosis of difficult airway. Therefore, the present invention effectively overcomes various shortcomings in the prior art and has high industrial utilization value.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Surgery (AREA)
  • Physiology (AREA)
  • Pulmonology (AREA)
  • Molecular Biology (AREA)
  • Physics & Mathematics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Biophysics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Veterinary Medicine (AREA)
  • Primary Health Care (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

一种人工智能的气道评估终端，包括：面部评估模块（11），其通过采集的面部数据进行3D重建进行面部评估；口腔评估模块（12），其通过口腔识别技术对口腔的可视环境进行数学化分析和机器学习来进行口腔评估；动态体格评估模块（13），其通过对人体头部的整体动态捕捉来进行动态体格评估；呼吸评估模块（14），其通过对呼吸气流的音频频率分析和机器学习，对整个呼吸过程中产生的气流变化进行肺功能的评估；声门评估模块（15），其通过对采集的目标对象的咽喉部解剖结构图像数据进行声门识别来进行声门评估。通过以上评估数据以及基础信息的综合判断生成声道评估报告，并依据现有医学领域对于气道评估的判断标准进行数字化分析，从而为麻醉医生得出气道的综合评估建议。

Description

基于人工智能的气道评估终端 技术领域
本发明涉及人工智能医疗领域，特别是涉及一种基于人工智能的气道评估终端。
背景技术
麻醉是利用药物将患者维持在无痛、意识消失、肌肉松弛的状态下，以便于实施手术等创伤性操作的过程。麻醉后患者处于呼吸抑制或停止的状态，如已存在困难气道未被提前诊断，则患者极有可能因无法建立气道进行通气而窒息、死亡。通过气道专科评估，绝大多数的困难气道可在麻醉前及时诊断从而可避免上述危急情况的发生。因此，为每位患者进行精准和严格的气道评估是避免未预料困难气道发生的重要保障。
在现有的使用场景中，通常情况下医生还是借助目测和经验对病人的情况进行判断，但是往往存在主观性上的偏差。对于某些特殊情况的困难气道也很难通过单纯的外部目测来进行判断。另外，现在也有部分借助摄像头进行的人脸识别判断，但是依旧存在两大主要问题：1.单目摄像头无法采集三维的面部数据，所以也无法精确的计算出病人实际的面部各个特征位置的实际大小。2.现有技术中所使用的方法依然局限于对静态图片的分析，但是在实际实践中，整个头部的可活动度和脖颈的可活动度都是直接影响医生判断困难气道的重要依据。受限于以上问题，目前所能实现的依旧是人脸识别相关的模型比对，对于临床中所需要的真实数据和运动状态下病人相关关节和部位的活动情况和活动度多无法进行采集和测算。
发明内容
鉴于以上所述现有技术的缺点，本发明的目的在于提供一种基于人工智能的气道评估终端，用于解决用于解决现有技术中以上技术问题。
为实现上述目的及其他相关目的，本发明提供一种基于人工智能的气道评估终端，所述终端包括：面部评估模块，用于根据由采集的目标对象的面部数据构建的三维面部模型获得用于气道评估的面部关键信息，并获得对应的面部评估结果；口腔评估模块，用于根据采集的所述目标对象在设定口腔评估动作下的口腔图像数据获得用于气道评估的口腔关键信息，并获得对应的口腔评估结果；动态体格评估模块，用于根据捕捉的所述目标对象在设定体格评估动作下的图像数据获得用于气道评估的动态体格关键信息，并获得对应的动态体格评估结果；呼吸评估模块，用于基于采集的所述目标对象的呼吸音声波数据获得用于气道评估的呼吸评估结果；声门评估模块，用于基于采集的目标对象的咽喉部解剖结构图像数据获得用 于气道评估的声门关键信息，并获得对应的声门评估结果；基础信息录入模块，用于录入所述目标对象的基础信息；报告生成模块，连接各评估模块以及基础信息录入模块，用于基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息生成对应所述目标对象的声道评估报告；数仓模块，连接各评估模块以及基础信息录入模块，用于对各评估模块以及基础信息录入模块的数据进行处理、储存以及分析。
于本发明的一实施例中，所述根据由采集的目标对象的面部数据构建的三维面部模型获得用于气道评估的面部关键特征信息，并获得对应的面部评估结果包括：获取基于红外点阵技术采集的目标对象的面部数据；其中，所述面部数据包括：多个面部标志点分别对应的真实距离；基于所述面部数据构建基于距离的三维面部模型；基于所述三维面部模型，对多个困难气道判断特征点进行抓取，并计算用于气道评估的面部关键信息；其中，所述面部关键信息包括：各困难气道判断特征点之间的真实距离；根据所述面部关键信息进行气道评估，以获得面部评估结果。
于本发明的一实施例中，所述根据采集的所述目标对象在设定口腔评估动作下的口腔图像数据获得用于气道评估的口腔关键信息，并获得对应的口腔评估结果包括：获取采集的所述目标对象在设定口腔评估动作下的口腔图像数据；其中，所述设定口腔评估动作包括：张嘴动作以及闭嘴动作；基于口腔关键信息识别模型，根据所述口腔图像数据获得用于气道评估的口腔关键信息；其中，所述口腔关键信息包括：对应口腔张合时最大的嘴部上下张合像素距离以及口腔左右开合像素距离；根据所述口腔关键信息进行气道评估，以获得口腔评估结果。
于本发明的一实施例中，所述根据捕捉的所述目标对象在设定体格评估动作下的图像数据获得用于气道评估的动态体格关键信息，并获得对应的动态体格评估结果包括：获取捕捉的所述目标对象在设定体格评估动作下的图像数据；其中，所述设定体格评估动作包括：脖颈左右转动动作以及脖颈俯仰活动动作；对所述图像数据中的一标定点以及多个参考点进行定位，并基于所述标定点与各参考点之间的固定位置，以及所述标定点的相对位置变化量获得动态体格关键信息；其中，所述动态体格关键信息包括：脖颈左右转动最大角度以及脖颈俯仰活动最大角度；根据所述动态体格关键信息进行气道评估，获得动态体格评估结果。
于本发明的一实施例中，所述标定点为鼻尖点。
于本发明的一实施例中，所述基于采集的所述目标对象的呼吸音声波数据获得用于气道评估的呼吸评估结果包括：获取采集所述目标对象在设定呼吸评估动作下的呼吸音声波数据；其中，所述呼吸评估动作包括：呼气动作以及吸气动作；识别所述呼吸音声波数据中的异形 的波形数据，并获得呼吸评估结果。
于本发明的一实施例中，所述基于采集的目标对象的咽喉部解剖结构图像数据获得用于气道评估的声门关键信息，并获得对应的声门评估结果包括：获取采集的目标对象的咽喉部解剖结构图像数据；对所述咽喉部解剖结构图像数据中的声门进行识别，以获得对应声门完整暴露的声门图像数据；对所述声门图像数据提取声门关键信息，并获得声门评估结果。
于本发明的一实施例中，所述数仓模块包括：数据接收单元，用于接收各评估模块以及基础信息录入模块的数据；数据处理单元，连接所述数据接收单元，用于对接收到的数据分别按其对应的数据类型进行清洗以及整理，以供上传至可视化数据中心进行可视化展示；数据储存单元，连接所述数据接收单元，用于储存各评估模块以及基础信息录入模块的数据；数据分析单元，连接数据储存单元，用于对储存的数据进行了分析和比对，以获得各数据类型的数据在气道评估中的权重系数比。
于本发明的一实施例中，所述基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息生成对应所述目标对象的声道评估报告包括：基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息获得对应的声道评估结果，并生成所述目标对象的声道评估报告。
于本发明的一实施例中，所述基础信息包括：年龄、身高、体重、BMI、吸烟史、气道手术史、放疗史、吞咽困难、瑞鸣、气道压迫以及气切史。
如上所述，本发明是一种基于人工智能的气道评估终端，具有以下有益效果：本发明通过采集的面部数据进行3D重建进行面部评估；通过口腔识别技术对口腔的可视环境进行数学化分析和机器学习来进行口腔评估；通过对人体头部的整体动态捕捉来进行动态体格评估；通过对呼吸气流的音频频率分析和机器学习，对整个呼吸过程中产生的气流变化进行肺功能的评估；通过对采集的目标对象的咽喉部解剖结构图像数据进行声门识别来进行声门评估；最后通过以上评估数据以及基础信息的综合判断生成声道评估报告，并依据现有医学领域对于气道评估的判断标准进行数字化分析，从而为麻醉医生得出气道的综合评估建议。本发明大幅节约气道评估的时长、提高困难气道诊断的精准度，不仅可大大目前麻醉学从业人员紧缺的现实难题，还为精准诊断困难气道提供可能。
附图说明
图1显示为本发明一实施例中的基于人工智能的气道评估终端的结构示意图。
图2显示为本发明一实施例中的呼吸曲线的波形示意图。
图3显示为本发明一实施例中的在TensorFlow中建模的示意图。
图4显示为本发明一实施例中的模型梯度下降示意图。
图5显示为本发明一实施例中的基于人工智能的气道评估方法的流程示意图。
具体实施方式
以下通过特定的具体实例说明本发明的实施方式，本领域技术人员可由本说明书所揭露的内容轻易地了解本发明的其他优点与功效。本发明还可以通过另外不同的具体实施方式加以实施或应用，本说明书中的各项细节也可以基于不同观点与应用，在没有背离本发明的精神下进行各种修饰或改变。需说明的是，在不冲突的情况下，以下实施例及实施例中的特征可以相互组合。
需要说明的是，在下述描述中，参考附图，附图描述了本发明的若干实施例。应当理解，还可使用其他实施例，并且可以在不背离本发明的精神和范围的情况下进行机械组成、结构、电气以及操作上的改变。下面的详细描述不应该被认为是限制性的，并且本发明的实施例的范围仅由公布的专利的权利要求书所限定。这里使用的术语仅是为了描述特定实施例，而并非旨在限制本发明。空间相关的术语，例如“上”、“下”、“左”、“右”、“下面”、“下方”、““下部”、“上方”、“上部”等，可在文中使用以便于说明图中所示的一个元件或特征与另一元件或特征的关系。
在通篇说明书中，当说某部分与另一部分“连接”时，这不仅包括“直接连接”的情形，也包括在其中间把其它元件置于其间而“间接连接”的情形。另外，当说某种部分“包括”某种构成要素时，只要没有特别相反的记载，则并非将其它构成要素，排除在外，而是意味着可以还包括其它构成要素。
其中提到的第一、第二及第三等术语是为了说明多样的部分、成分、区域、层及/或段而使用的，但并非限定于此。这些术语只用于把某部分、成分、区域、层或段区别于其它部分、成分、区域、层或段。因此，以下叙述的第一部分、成分、区域、层或段在不超出本发明范围的范围内，可以言及到第二部分、成分、区域、层或段。
再者，如同在本文中所使用的，单数形式“一”、“一个”和“该”旨在也包括复数形式，除非上下文中有相反的指示。应当进一步理解，术语“包含”、“包括”表明存在所述的特征、操作、元件、组件、项目、种类、和/或组，但不排除一个或多个其他特征、操作、元件、组件、项目、种类、和/或组的存在、出现或添加。此处使用的术语“或”和“和/或”被解释为包括性的，或意味着任一个或任何组合。因此，“A、B或C”或者“A、B和/或C”意味着“以下任一个：A；B；C；A和B；A和C；B和C；A、B和C”。仅当元件、功 能或操作的组合在某些方式下内在地互相排斥时，才会出现该定义的例外。
气道评估的具体项目通常包括静态和指令动作下的面部、颈部、口腔和鼻内部视诊和体格检查。当上气道及其周围组织发生先天发育异常、后天创伤或疾病时，视诊和医患动态配合体格检查，可有效辨识和判断困难气道的类型、原因以及困难程度。然而，气道评估的项目众多、评估步骤也比较繁琐。例如，需通过头、面、颈部及口腔内的视诊判读患者的张口度、牙齿咬合关系、络腮胡密度、Mallampati分级等指标，通过医患配合的指令动作进行甲颏间距、颈围、头颈活动度等完成些列的解剖标志距离测量，综合诊断气道情况。因此，气道评估不仅需要占用麻醉医生大量的工作时间，也常常因为步骤繁琐而遗漏重要的检查项目及数据，从而可能导致困难气道漏诊和误诊。
因此，本发明提供一种基于人工智能的气道评估终端，通过采集的面部数据进行3D重建进行面部评估；通过口腔识别技术对口腔的可视环境进行数学化分析和机器学习来进行口腔评估；通过对人体头部的整体动态捕捉来进行动态体格评估；通过对呼吸气流的音频频率分析和机器学习，对整个呼吸过程中产生的气流变化进行肺功能的评估；通过对采集的目标对象的咽喉部解剖结构图像数据进行声门识别来进行声门评估；最后通过以上评估数据以及基础信息的综合判断生成声道评估报告，并依据现有医学领域对于气道评估的判断标准进行数字化分析，从而为麻醉医生得出气道的综合评估建议。本发明大幅节约气道评估的时长、提高困难气道诊断的精准度，不仅可大大目前麻醉学从业人员紧缺的现实难题，还为精准诊断困难气道提供可能。
下面以附图为参考，针对本发明的实施例进行详细说明，以便本发明所述技术领域的技术人员能够容易地实施。本发明可以以多种不同形态体现，并不限于此处说明的实施例。
如图1展示本发明实施例中的一种基于人工智能的气道评估终端的结构示意图。
所述终端包括：
面部评估模块11，用于根据由采集的目标对象的面部数据构建的三维面部模型获得用于气道评估的面部关键信息，并获得对应的面部评估结果；
口腔评估模块12，用于根据采集的所述目标对象在设定口腔评估动作下的口腔图像数据获得用于气道评估的口腔关键信息，并获得对应的口腔评估结果；
动态体格评估模块13，用于根据捕捉的所述目标对象在设定体格评估动作下的图像数据获得用于气道评估的动态体格关键信息，并获得对应的动态体格评估结果；
呼吸评估模块14，用于基于采集的所述目标对象的呼吸音声波数据获得用于气道评估的呼吸评估结果；
声门评估模块15，用于基于采集的目标对象的咽喉部解剖结构图像数据获得用于气道评估的声门关键信息，并获得对应的声门评估结果；
基础信息录入模块16，用于录入所述目标对象的基础信息；
报告生成模块17，连接面部评估模块11、口腔评估模块12、动态体格评估模块13、呼吸评估模块14、声门评估模块15、以及基础信息录入模块16，用于基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息生成对应所述目标对象的声道评估报告；
数仓模块18，连接面部评估模块11、口腔评估模块12、动态体格评估模块13、呼吸评估模块14、声门评估模块15、以及基础信息录入模块16，用于对各评估模块以及基础信息录入模块的数据进行处理、储存以及分析。
在一实施例中，面部评估模块11根据由采集的目标对象的面部数据构建的三维面部模型获得用于气道评估的面部关键信息，并获得对应的面部评估结果包括：
获取基于红外点阵技术采集的目标对象的面部数据；其中，所述面部数据包括：多个面部标志点分别对应的真实距离；具体的，红外点阵技术能够采集红外点阵传感器的每一个红外点投射到目标对象的面部的实际时间，再乘以红外光在空气中的传播速度，从而精准得出每一个面部标志点与摄像头的实际真实距离；
基于所述面部数据构建基于距离的三维面部模型；
基于所述三维面部模型，对多个困难气道判断特征点进行抓取，并计算用于气道评估的面部关键信息；其中，所述面部关键信息包括：各困难气道判断特征点之间的真实距离；
基于传统困难气道的判断标准，根据所述面部关键信息进行气道评估，以获得面部评估结果。其中，面部评估结果可以为通过面部判断为是否为困难气道，或是困难气道概率，又或是困难气道判断等级。
在现有的困难气道判断过程中，医生需要对面部一些特征做目测。目测存在标准难以统一、数据难以量化的两个明显问题。而当下使用的大部分人脸重建技术，都是基于单目摄像头的2D重建。2D重建在面部识别方面具有很好的效果，但是在实际的真实世界场景中，对于三维的位置距离测算就存在很大的局限性。而采用本方案不仅可以解决以上的缺陷，还不但能够将准确的真实数据反馈给医生，并且能够将面部的特征点进行采集和学习。
优选的，通过红外点阵传感器，在瞬间向目标人脸投射30000个红外点。通过红外摄像头精确捕捉面部的红外光通过计算机开始绘制深度图。同时在本设备的传感器模组中，通过ToF(飞行时间)传感器，对患者面部的距离进行精确的计算。通过正面的距离计算以及像 素点和摄像头成像原理公式，推导采集的人体体格数据中的面部数据。
在一实施例中，所述口腔评估模块12根据采集的所述目标对象在设定口腔评估动作下的口腔图像数据获得用于气道评估的口腔关键信息，并获得对应的口腔评估结果包括：
获取采集的所述目标对象在设定口腔评估动作下的口腔图像数据；具体的，获取采集的所述目标对象在做张嘴动作以及闭嘴动作的过程中的口腔图像数据；
基于口腔关键信息识别模型，根据所述口腔图像数据获得用于气道评估的口腔关键信息；其中，所述口腔关键信息包括：对应口腔张合时最大的嘴部上下张合像素距离以及口腔左右开合像素距离；具体的，所述口腔关键信息识别模型经过标记有口腔上颌到下颌位置以及口腔左右位置的多个口腔图像数据以及对应的口腔张合时最大的嘴部上下张合像素距离以及口腔左右开合像素距离训练获得。
基于传统困难气道的判断标准，根据所述口腔关键信息进行气道评估，以获得口腔评估结果。口腔评估结果可以为通过口腔判断为是否为困难气道，或是困难气道概率，又或是困难气道判断等级。
优选的，根据采集的所述目标对象在设定口腔评估动作下的口腔图像数据获得用于气道评估的口腔关键信息包括：通过计算机学习口腔的数据值，将口腔上颌到下颌的像素距离定义为：MouthT_Mouth_B；口腔左右位置的像素距离定义为：MouthL_MouthR。在动态的视频学习中，通过计算机实时抓取所述口腔图像数据距离值，建立两个列表：List_D_MouthT_Mou th_B和List_D_MouthL_MouthR。在整个实操过程当中，对距离做实时的采集，患者在医生的指导下进行张嘴、闭嘴的动作。在完成整个动作之后，计算机对列表中抓取到的数据进行遍历。通过Max和SubMax两个数值，筛选出整个列表当中最大的Max值，即所述目标对象在做张嘴动作以及闭嘴动作的过程中的口腔图像数据。
可选的，由于口腔环境对于性别以及年龄的限制，因此再获取口腔关键信息时，需要根据基于性别以及年龄对应设定的系数或比例去获得最终的口腔关键信息，并且在在获得口腔关键信息的同时，同步记录对应系数和/或比例。
在一实施例中，所述动态体格评估模块13根据捕捉的所述目标对象在设定体格评估动作下的图像数据获得用于气道评估的动态体格关键信息，并获得对应的动态体格评估结果包括：
获取捕捉的所述目标对象在设定体格评估动作下的图像数据；具体的，获取采集的所述目标对象在做脖颈左右转动动作以及脖颈俯仰活动的过程中的头部图像数据；需要说明的是，脖颈左右转动动作需达到目标对象脖颈左右转动自身极限下的最大角度，脖颈俯仰活动需达到目标对象脖颈俯仰活动自身极限下的最大角度。
对所述图像数据中的一标定点以及多个参考点进行定位，并基于所述标定点与各参考点之间的固定位置，以及所述标定点的相对位置变化量获得动态体格关键信息；其中，所述动态体格关键信息包括：脖颈左右转动最大角度以及脖颈俯仰转动最大角度；需要说明的是，各参考点与标定点的位置是固定的，即无论头部怎么动，仅是相对位置变化，各参考点与标定点的位置是绝对位置不会变化，可通过绝对位置保证相对位置变化的准确性。
基于传统困难气道的判断标准，根据所述动态体格关键信息进行气道评估，获得动态体格评估结果。动态体格评估结果可以为通过动态体格判断为是否为困难气道，或是困难气道概率，又或是困难气道判断等级。
优选的，通过人工智能的识别定位脖颈的正位置，在脖颈的转动过程当中，程序以当下鼻尖位置为标定点，取一或多个其他特征点作为参考点，例如，参考点为左右眼中间位置以嘴角两端位置；将脖颈的正向左右间距设定为常值α。取脖颈转动时X轴的变化量，计算在平面投射中标定点左右两侧X轴距离变化比例。通过这个比例，从而得出x平面中，180°活动角度内所产生的怕偏离转动角。程序通过特征点以及反三角函数的计算公式，对患者头部在三维的运动角度进行计算和记录。此处，我们将要计算头部的欧拉角(pitch、roll、yaw)分别在x、y、z三个坐标轴上的运动信息。
在平面中，我们使用的是单目摄像头进行捕获。在捕获上图中的特征点后，计算机持续对各个点中的数据进行追踪。通过范数计算，可以得出旋转角为(rotation angle)：
Figure PCTCN2023070918-appb-000001
由以上公式可以得出向量：
Figure PCTCN2023070918-appb-000002
在得出旋转角(rotation angle)后，需要对角度实现四元数的转化。在三维空间中，将四元数表达为：
q＝w+xi+yj+zk；   (3)
转换为向量表达式：
q＝((x,y,z)，w)＝(v,w)；   (4)
在实现旋转时，转变为：
Figure PCTCN2023070918-appb-000003
根据四元数，再次转换欧拉角：由式(5)可得出：
x＝sin(Y/2)sin(Z/2)cos(X/2)+cos(Y/2)cos(Z/2)sin(X/2)；   (6)
y＝sin(Y/2)cos(Z/2)cos(X/2)+cos(Y/2)sin(Z/2)sin(X/2)；   (7)
z＝cos(Y/2)sin(Z/2)cos(X/2)-sin(Y/2)cos(Z/2)sin(X/2)；   (8)
w＝cos(Y/2)cos(Z/2)cos(X/2)-sin(Y/2)sin(Z/2)sin(X/2)；   (9)
在得到以上欧拉角数据之后，再次转换欧拉角默认弧度为度数：
Pitch＝x/π*180；   (10)
Roll＝y/π*180；   (11)
Yaw＝z/π*180；   (12)
至此，计算机完成了对于动态特征点的追踪、四元数抓取、欧拉角转换和弧度到度数的完整算法转变。根据计算机得到的最大角度计算，能够在数据列表中获得患者在镜头追踪下脖颈左右的最大转动角以及脊柱上下俯仰角的最大运动角度。
在一实施例中，所述呼吸评估模块14基于采集的所述目标对象的呼吸音声波数据获得用于气道评估的呼吸评估结果包括：
获取采集所述目标对象在设定呼吸评估动作下的呼吸音声波数据；具体的，获取采集的所述目标对象在做呼气动作以及吸气动作过程中的呼吸音声波数据；
识别所述呼吸音声波数据中的异形的波形数据，并获得呼吸评估结果；具体的，识别呼吸音声波数据中不顺滑的异形波形，例如，断层波形，突起波形包括尖锐波形，共振波形以及回响波形等波形；所述呼吸评估结果可以为通过呼吸判断为是否为困难气道，或是困难气道概率，又或是困难气道判断等级。
优选的，针对不同人群，通过电子听诊器完成对肺部、上呼吸道的呼吸音声波数据截取。气流在气管和肺部流动时，所产生的细微震动，通过波长和频率的形式数字化的实现记录和可视化。对于波形的学习，计算机同样借助于TensorFlow对患者“呼”和“吸”时产生的模型进行学习和分析，计算最为符合曲线变化的曲率系数。通过系数变化对呼吸中可能存在的气管和肺部的呼吸困难进行比对判断。如图2所示，音频频率从0到4096的数字等分中，通过曲线的微波变化来界定音频在特定环境中的模型形态，呼吸时产生的音频变化主要集中在高频部分。而通常情况下，困难气道患者在呼吸时，呼吸曲线的波形相较于正常人会出现比较明显异形，即：曲线对应的趋势与多项式表达的曲线不相符。
在一实施例中，所述声门评估模块15基于采集的目标对象的咽喉部解剖结构图像数据获得用于气道评估的声门关键信息，并获得对应的声门评估结果包括：
获取采集的目标对象的咽喉部解剖结构图像数据；具体的，采用气道检查专用可视吼镜片配合图像处理器(可视喉镜)，将位于声门及周边的解剖结构进行充分的可视化暴露，并对数据进行采集。
对所述咽喉部解剖结构图像数据中的声门进行识别，以获得对应声门完整暴露的声门图像数据；具体的，通过卷积神经网络算法，将声门结构的照片进行学习，从而实现对于声门的自动识别。在检查镜片发现声门的第一时间，就会对相关的识别部位进行拍照保存，以获得对应声门完整暴露的声门图像数据；
对所述声门图像数据提取声门关键信息，并获得声门评估结果。所述声门评估结果可以为通过呼吸判断为是否为困难气道，或是困难气道概率，又或是困难气道判断等级。
其中，声门关键信息包括：有无喉新生物、喉镜暴露分级、声门狭窄情况、声门下狭窄情况、声门上狭窄情况等信息。
优选的，采用可视咽喉检查镜采集目标对象的咽喉部解剖结构图像数据，通过将带有显示器和摄像头的特殊喉镜插入两端呈90度的镜体，并将镜体弓背侧向上颚方向置入口腔，既可用于咽喉部检查，并在置入口腔后可使镜头端的摄像头垂直指向咽喉部、声门，便可顺着自然解剖弧度，清晰显示并采集口腔内、咽喉部的各种解剖结构。
可选的，所述基础信息录入模块16录入的所述目标对象的基础信息包括：年龄、身高、体重、BMI、吸烟史、气道手术史、放疗史、吞咽困难、瑞鸣、气道压迫以及气切史；其中，每种基础信息可以采用不同的数据类型，如下表所示；
表1：基础信息数据类型对照表
数据名称 数据类型
年龄 数字
身高 数字
体重 数字
BMI(身体质量指数) 数字
吸烟史 布尔值
气道手术史 文字
放疗史 文字
吞咽困难 布尔值
瑞鸣 布尔值
气道压迫 布尔值+文字
气切史 布尔值+文字
在一实施例中，所述报告生成模块17基于所述面部评估结果、口腔评估结果、动态体格 评估结果、呼吸评估结果、声门评估结果以及基础信息生成对应所述目标对象的声道评估报告包括：
基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息获得对应的声道评估结果，并生成所述目标对象的声道评估报告；
其中，声道评估报告包括：所述声道评估结果、面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息。需要说明的是，所述声道评估结果可以为是否为困难气道，或是困难气道概率，又或是困难气道判断等级。
优选的，声道评估报告还可包括采用AI深度学习的策略分析医生对困难气道分型、难度及原因的判读。
在一实施例中，所述数仓模块18包括：
数据接收单元，用于接收各评估模块以及基础信息录入模块的数据；包括：来自所述面部评估模块11的面部关键信息以及面部评估结果、来自口腔评估模块12的口腔关键信息以及口腔评估结果、来自动态体格评估模块13的动态体格关键信息以及动态体格评估结果、来自呼吸评估模块14的呼吸音声波数据以及呼吸评估结果、来自声门评估模块15的声门关键信息以及声门评估结果以及来自基础信息录入模块16的基础信息；
数据处理单元，连接所述数据接收单元，用于对接收到的数据分别按其对应的数据类型进行清洗以及整理，以供上传至可视化数据中心进行可视化展示；通过整理和清洗，程序将对所有不同类型的数据进行对应的处理，其中，例如将依据基础生理指标例如：年龄、身高、体重计算BMI。最终所有数据完成清洗和分析之后，所有数据上传至可视化数据中心进行展示，保障医生在后端可以对数据有较为完整的实时可视化效果。
数据储存单元，连接所述数据接收单元，用于储存各评估模块以及基础信息录入模块的数据；具体将来自所述面部评估模块11的面部关键信息以及面部评估结果、来自口腔评估模块的口腔关键信息以及口腔评估结果、来自动态体格评估模块的动态体格关键信息以及动态体格评估结果、来自呼吸评估模块的呼吸音声波数据以及呼吸评估结果、来自声门评估模块的声门关键信息以及声门评估结果以及来自基础信息录入模块的基础信息储存进MySql数据库。
数据分析单元，连接数据储存单元，用于对储存的数据进行了分析和比对，以获得各数据类型的数据在气道评估中的权重系数比。
优选的，在数据库中，程序将采集到的数据进行了集中的分析和比对。对于一些有较强相关因素的数据，我们使用TensorFlow进行数据的学习，了解数据中的线性关系。(包括但 不限于：年龄、身高、体重、BMI、面部关键信息、口腔关键信息、动态体格关键信息、呼吸音声波数据、声门关键信息)在TensorFlow中，我们对数据进行两两学习，建模如图3所示，对应梯度下降如图4；在上述流程中，程序将反复学习，通过对loss(即：损失函数)的修正，从而推导出最符合数据变化的系数值。其中loss表示电脑预测数据y与已知数据y_的误差值，即：均方误差
Figure PCTCN2023070918-appb-000004
通过上述的一系列计算，我们可以得出各个数据类型在气道评估中的权重系数比。优选的，所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息的气道评估权重。
为了更好的说明上述基于人工智能的气道评估终端，本发明提供以下具体实施例。
实施例1：一种应用基于人工智能的气道评估终端的气道评估方法。如图5所示为基于人工智能的气道评估方法流程示意图；所述方法包括：
基础信息的采集；即录入所述目标对象的基础信息；其中包含：1.年龄 2.身高 3.体重 4.BMI(身体质量指数) 5.吸烟史 6.气道手术史 7.放疗史 8.吞咽困难 9.瑞鸣 10.气道压迫 11.气切史；
通过对人体头部的整体动态捕捉，让患者进行头颈的左右活动和上下活动，进行人体体格的动态数据监测；即通过摄像头捕捉的头部图像数据依次进行动态特征点的追踪、四元数抓取、欧拉角转换和弧度到度数的完整算法转变。根据计算机得到的最大角度计算，能够在数据列表中获得患者在镜头追踪下脖颈左右的最大转动角以及脊柱上下俯仰角的最大运动角度。
通过口腔识别技术对口腔的可视环境进行数学化分析和机器学习；即通过摄像头对距离做实时的采集，患者在医生的指导下进行张嘴、闭嘴的动作。在完成整个动作之后，计算机对列表中抓取到的数据进行遍历，筛选出整个列表当中最大的Max值，即：口腔张合时最大的嘴部上下张合像素距离以及口腔左右开合像素距离。对相应系数以及比例进行记录；
通过对呼吸气流的音频频率分析和机器学习，对整个呼吸过程中产生的气流变化进行肺功能的评估；即通过电子听诊器完成对肺部、上呼吸道的呼吸音截取，计算机同样借助于TensorFlow对患者“呼”和“吸”时产生的模型进行学习和分析，计算最为符合曲线变化的曲率系数。通过系数变化对呼吸中可能存在的气管和肺部的呼吸困难进行比对判断。
通过红外3D点阵投射技术对面部进行数据采集和3D重建；即利用红外点阵模块能够采集每一个红外点投射到目标面部的实际时间，再乘以红外光在空气中的传播速度，从而精准 得出每一个面部标志物中的点与摄像头的实际真实距离，从而重建出一张基于距离的3D面部特征模型，获取以上数据之后，根据传统困难气道的判断标准，机器将所有需要的点进行抓取，并计算出点与点之间的真实距离。
在完成对于数据的采集之后，所有的数据将被进行二次处理。数据全部进入MySql数仓进行保存。通过整理和清洗，程序将对所有不同类型的数据进行对应的处理。最终所有数据完成清洗和分析之后，所有数据上传至可视化数据中心进行展示，保障医生在后端可以对数据有较为完整的实时可视化效果。在数据库中，程序将采集到的数据进行了集中的分析和比对。对于一些有较强相关因素的数据，我们使用TensorFlow进行数据的学习，了解数据中的线性关系。
基于采集的数据生成对应所述目标对象的声道评估报告。
本实施例通过对人体各类静态和动态的体格数据的采集，应用红外点阵技术，利用大数据中心的可视化和人工智能算法中的数据学习，不仅通过人工智能相关的图像处理技术和识别技术采集面部、口腔和动态的体格检查数据还对各个权重因素进行了详尽且全面的分析。利用算法实现自动化采集，配合医生的临床操作，对于气道的评估做出了全面的、智能化的、数字化的认识和分析。
综上所述，本发明的基于人工智能的气道评估终端，通过采集的面部数据进行3D重建进行面部评估；通过口腔识别技术对口腔的可视环境进行数学化分析和机器学习来进行口腔评估；通过对人体头部的整体动态捕捉来进行动态体格评估；通过对呼吸气流的音频频率分析和机器学习，对整个呼吸过程中产生的气流变化进行肺功能的评估；通过对采集的目标对象的咽喉部解剖结构图像数据进行声门识别来进行声门评估；最后通过以上评估数据以及基础信息的综合判断生成声道评估报告，并依据现有医学领域对于气道评估的判断标准进行数字化分析，从而为麻醉医生得出气道的综合评估建议。本发明大幅节约气道评估的时长、提高困难气道诊断的精准度，不仅可大大目前麻醉学从业人员紧缺的现实难题，还为精准诊断困难气道提供可能。所以，本发明有效克服了现有技术中的种种缺点而具高度产业利用价值。
上述实施例仅示例性说明本发明的原理及其功效，而非用于限制本发明。任何熟悉此技术的人士皆可在不违背本发明的精神及范畴下，对上述实施例进行修饰或改变。因此，但凡所属技术领域中具有通常知识者在未脱离本发明所揭示的精神与技术思想下所完成的一切等效修饰或改变，仍应由本发明的权利要求所涵盖。

Claims (10)

  1. 一种基于人工智能的气道评估终端，其特征在于，所述终端包括：
    面部评估模块，用于根据由采集的目标对象的面部数据构建的三维面部模型获得用于气道评估的面部关键信息，并获得对应的面部评估结果；
    口腔评估模块，用于根据采集的所述目标对象在设定口腔评估动作下的口腔图像数据获得用于气道评估的口腔关键信息，并获得对应的口腔评估结果；
    动态体格评估模块，用于根据捕捉的所述目标对象在设定体格评估动作下的图像数据获得用于气道评估的动态体格关键信息，并获得对应的动态体格评估结果；
    呼吸评估模块，用于基于采集的所述目标对象的呼吸音声波数据获得用于气道评估的呼吸评估结果；
    声门评估模块，用于基于采集的目标对象的咽喉部解剖结构图像数据获得用于气道评估的声门关键信息，并获得对应的声门评估结果；
    基础信息录入模块，用于录入所述目标对象的基础信息；
    报告生成模块，连接各评估模块以及基础信息录入模块，用于基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息生成对应所述目标对象的声道评估报告；
    数仓模块，连接各评估模块以及基础信息录入模块，用于对各评估模块以及基础信息录入模块的数据进行处理、储存以及分析。
  2. 根据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述根据由采集的目标对象的面部数据构建的三维面部模型获得用于气道评估的面部关键特征信息，并获得对应的面部评估结果包括：
    获取基于红外点阵技术采集的目标对象的面部数据；其中，所述面部数据包括：多个面部标志点分别对应的真实距离；
    基于所述面部数据构建基于距离的三维面部模型；
    基于所述三维面部模型，对多个困难气道判断特征点进行抓取，并计算用于气道评估的面部关键信息；其中，所述面部关键信息包括：各困难气道判断特征点之间的真实距离；
    根据所述面部关键信息进行气道评估，以获得面部评估结果。
  3. 根据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述根据采集的所述目标对象在设定口腔评估动作下的口腔图像数据获得用于气道评估的口腔关键信息，并获得对应的口腔评估结果包括：
    获取采集的所述目标对象在设定口腔评估动作下的口腔图像数据；其中，所述设定口 腔评估动作包括：张嘴动作以及闭嘴动作；
    基于口腔关键信息识别模型，根据所述口腔图像数据获得用于气道评估的口腔关键信息；其中，所述口腔关键信息包括：对应口腔张合时最大的嘴部上下张合像素距离以及口腔左右开合像素距离；
    根据所述口腔关键信息进行气道评估，以获得口腔评估结果。
  4. 根据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述根据捕捉的所述目标对象在设定体格评估动作下的图像数据获得用于气道评估的动态体格关键信息，并获得对应的动态体格评估结果包括：
    获取捕捉的所述目标对象在设定体格评估动作下的图像数据；其中，所述设定体格评估动作包括：脖颈左右转动动作以及脖颈俯仰活动动作；
    对所述图像数据中的一标定点以及多个参考点进行定位，并基于所述标定点与各参考点之间的固定位置，以及所述标定点的相对位置变化量获得动态体格关键信息；其中，所述动态体格关键信息包括：脖颈左右转动最大角度以及脖颈俯仰活动最大角度；
    根据所述动态体格关键信息进行气道评估，获得动态体格评估结果。
  5. 根据权利要求4中所述的基于人工智能的气道评估终端，其特征在于，所述标定点为鼻尖点。
  6. 根据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述基于采集的所述目标对象的呼吸音声波数据获得用于气道评估的呼吸评估结果包括：
    获取采集所述目标对象在设定呼吸评估动作下的呼吸音声波数据；其中，所述呼吸评估动作包括：呼气动作以及吸气动作；
    识别所述呼吸音声波数据中的异形的波形数据，并获得呼吸评估结果。
  7. 根据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述基于采集的目标对象的咽喉部解剖结构图像数据获得用于气道评估的声门关键信息，并获得对应的声门评估结果包括：
    获取采集的目标对象的咽喉部解剖结构图像数据；
    对所述咽喉部解剖结构图像数据中的声门进行识别，以获得对应声门完整暴露的声门图像数据；
    对所述声门图像数据提取声门关键信息，并获得声门评估结果。
  8. 根据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述数仓模块包括：
    数据接收单元，用于接收各评估模块以及基础信息录入模块的数据；
    数据处理单元，连接所述数据接收单元，用于对接收到的数据分别按其对应的数据类型进行清洗以及整理，以供上传至可视化数据中心进行可视化展示；
    数据储存单元，连接所述数据接收单元，用于储存各评估模块以及基础信息录入模块的数据；
    数据分析单元，连接数据储存单元，用于对储存的数据进行了分析和比对，以获得各数据类型的数据在气道评估中的权重系数比。
  9. 据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息生成对应所述目标对象的声道评估报告包括：
    基于所述面部评估结果、口腔评估结果、动态体格评估结果、呼吸评估结果、声门评估结果以及基础信息获得对应的声道评估结果，并生成所述目标对象的声道评估报告。
  10. 根据权利要求1中所述的基于人工智能的气道评估终端，其特征在于，所述基础信息包括：年龄、身高、体重、BMI、吸烟史、气道手术史、放疗史、吞咽困难、瑞鸣、气道压迫以及气切史。
PCT/CN2023/070918 2022-07-26 2023-01-06 基于人工智能的气道评估终端 WO2024021534A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210883918.X 2022-07-26
CN202210883918.XA CN115381429B (zh) 2022-07-26 2022-07-26 基于人工智能的气道评估终端

Publications (1)

Publication Number Publication Date
WO2024021534A1 true WO2024021534A1 (zh) 2024-02-01

Family

ID=84117652

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/070918 WO2024021534A1 (zh) 2022-07-26 2023-01-06 基于人工智能的气道评估终端

Country Status (2)

Country Link
CN (1) CN115381429B (zh)
WO (1) WO2024021534A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115381429B (zh) * 2022-07-26 2023-07-07 复旦大学附属眼耳鼻喉科医院 基于人工智能的气道评估终端
CN116630534A (zh) * 2023-05-06 2023-08-22 华中科技大学协和深圳医院 一种气道管理人工智能决策系统
CN117238509B (zh) * 2023-11-15 2024-02-27 首都医科大学宣武医院 一种基于普通摄像数据的困难气道评估系统及评估方法

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070173724A1 (en) * 2006-01-24 2007-07-26 Singh Gurdev D System and method for three-dimensional airway reconstruction, assessment and analysis
US8460215B2 (en) * 2009-06-08 2013-06-11 The Brigham And Women's Hospital, Inc. Systems and methods for predicting potentially difficult intubation of a subject
US20160278670A1 (en) * 2013-10-11 2016-09-29 Centre Hospitalier Universitaire Vaudois (Chuv) Difficult intubation or ventilation or extubation prediction system
CN108053470A (zh) * 2017-12-28 2018-05-18 昆明医科大学第附属医院 一种基于mri困难气道三维重建方法及应用
CN112370018A (zh) * 2020-11-10 2021-02-19 皖南医学院第一附属医院（皖南医学院弋矶山医院） 一种预测困难气道的计算机应用软件及气道管理数据系统
US20210196217A1 (en) * 2019-12-31 2021-07-01 Jerry Chi Hu Dynamic 3-d anatomical mapping and visualization
CN113069080A (zh) * 2021-03-22 2021-07-06 上海交通大学医学院附属第九人民医院 一种基于人工智能的困难气道评估方法及装置
CN113571088A (zh) * 2021-07-27 2021-10-29 上海交通大学医学院附属第九人民医院 一种基于深度学习声纹识别的困难气道评估方法及装置
CN115381429A (zh) * 2022-07-26 2022-11-25 复旦大学附属眼耳鼻喉科医院 基于人工智能的气道评估终端

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110584775A (zh) * 2018-06-13 2019-12-20 凯勋国际股份有限公司 气道模型生成系统及插管辅助系统
US20210034841A1 (en) * 2019-07-31 2021-02-04 Wake Forest University Health Sciences System and method for automated prediction of difficult airway management using images
CN113657458A (zh) * 2021-07-27 2021-11-16 浙江大华技术股份有限公司 气道分类方法及其装置、计算机可读存储介质
CN216793317U (zh) * 2021-11-15 2022-06-21 上海交通大学医学院附属第九人民医院 一种基于ai人脸特征识别的困难气道预测建模装置

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070173724A1 (en) * 2006-01-24 2007-07-26 Singh Gurdev D System and method for three-dimensional airway reconstruction, assessment and analysis
US8460215B2 (en) * 2009-06-08 2013-06-11 The Brigham And Women's Hospital, Inc. Systems and methods for predicting potentially difficult intubation of a subject
US20160278670A1 (en) * 2013-10-11 2016-09-29 Centre Hospitalier Universitaire Vaudois (Chuv) Difficult intubation or ventilation or extubation prediction system
CN108053470A (zh) * 2017-12-28 2018-05-18 昆明医科大学第附属医院 一种基于mri困难气道三维重建方法及应用
US20210196217A1 (en) * 2019-12-31 2021-07-01 Jerry Chi Hu Dynamic 3-d anatomical mapping and visualization
CN112370018A (zh) * 2020-11-10 2021-02-19 皖南医学院第一附属医院（皖南医学院弋矶山医院） 一种预测困难气道的计算机应用软件及气道管理数据系统
CN113069080A (zh) * 2021-03-22 2021-07-06 上海交通大学医学院附属第九人民医院 一种基于人工智能的困难气道评估方法及装置
CN113571088A (zh) * 2021-07-27 2021-10-29 上海交通大学医学院附属第九人民医院 一种基于深度学习声纹识别的困难气道评估方法及装置
CN115381429A (zh) * 2022-07-26 2022-11-25 复旦大学附属眼耳鼻喉科医院 基于人工智能的气道评估终端

Also Published As

Publication number Publication date
CN115381429A (zh) 2022-11-25
CN115381429B (zh) 2023-07-07

Similar Documents

Publication Publication Date Title
WO2024021534A1 (zh) 基于人工智能的气道评估终端
Cuendet et al. Facial image analysis for fully automatic prediction of difficult endotracheal intubation
EP3054844B1 (en) Difficult intubation or ventilation or extubation prediction system
WO2022100520A1 (zh) 一种预测困难气道的计算机应用软件及气道管理数据系统
KR20130026041A (ko) 의료 영상의 일부 정보를 활용한 장기 영상 생성 방법 및 장치
US11642043B2 (en) Diagnostic tool and method of use
CN114999646B (zh) 新生儿运动发育评估系统、方法、装置及存储介质
US11723614B2 (en) Dynamic 3-D anatomical mapping and visualization
Soleimani et al. Remote, depth-based lung function assessment
Hanif et al. Estimation of apnea-hypopnea index using deep learning on 3-D craniofacial scans
CN110473180A (zh) 胸部呼吸运动的识别方法、系统及存储介质
KR20220021900A (ko) 생체신호 분석 기반 코골이 분석과 질병 진단 예측 서비스를 제공하는 코골이 분석 서비스 시스템
US20240090795A1 (en) Methods for Pulmonary Function Testing With Machine Learning Analysis and Systems for Same
Li et al. How do you breathe-a non-contact monitoring method using depth data
Hu et al. Photographic analysis and machine learning for diagnostic prediction of adenoid hypertrophy
KR20220021989A (ko) 수면중 코골이 소리와 수면 자세 분석 기반 수면 자세와 코골이 상관관계 모니터링 방법 및 장치
Chinthala et al. An internet of medical things (IoMT) approach for remote assessment of head and neck cancer patients
US20240074724A1 (en) Monitoring airflow with b-mode ultrasound
CN117238509B (zh) 一种基于普通摄像数据的困难气道评估系统及评估方法
Stefunova et al. Multi-camera scanning system for collecting and annotating 3D models of the head and neck
US20210236050A1 (en) Dynamic anatomic data collection and modeling during sleep
Transue et al. Behavioral Analysis of Airway Deformation during Drug Induced Sleep Endoscopy
De Rosa et al. The Future of Artificial Intelligence Using Images and Clinical Assessment for Difficult Airway Management
Xi et al. Combining Artificial Intelligence and Automatic Analysis to Study of the Closure Characteristics of Healthy Human Vocal Cords during Phonation Using Synchronous Electroglottography and Laryngeal High-Speed Videoendoscopy
WO2023105529A1 (en) Method and system for automatic diagnosing predisposition to disease

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23844753

Country of ref document: EP

Kind code of ref document: A1