A method for detecting the opening shape of the Eustachian tube by using a Eustachian tube acoustic measurement method

Abstract

The present invention discloses a method for detecting the open morphology of the Eustachian tube by using the acoustic measurement method of the Eustachian tube, comprising: using a Eustachian tube acoustic measurement device, a vibration sensor device and an imaging device together, performing acoustic measurement detection when the vibration sensor is attached to the throat, and performing imaging recording when the vibration sensor detects the vibration of the throat. The vibration sensor is used to mark the swallowing moment, and the acoustic transfer function of the Eustachian tube opening obtained by the acoustic measurement method is calculated and matched with the morphology of the Eustachian tube opening recorded by the imaging device, and a calculation model for inversely deducing the open morphology of the Eustachian tube based on the acoustic transfer function of the acoustic measurement method is established. The advantages of the present invention are: providing a new method for detecting the open morphology of the Eustachian tube, which is conducive to the evaluation and research of the Eustachian tube function.

Description

Technical Field

[0001] The present invention belongs to the field of clinical auxiliary organ function examination, and specifically relates to the examination of Eustachian tube function. It is a method for calculating the opening shape of the Eustachian tube during swallowing by using the Eustachian tube function acoustic measurement method, combining a vibration sensor to obtain the time corresponding to the swallowing action, and recording the shape of the Eustachian tube when it is open through imaging technology. The image data of the Eustachian tube when it is open guides the acoustic measurement method to detect the opening shape of the Eustachian tube. Background Art

[0002] The Eustachian tube is a channel connecting the middle ear and the nasopharynx. It is composed of the outer 1 / 3 bone part and the inner 2 / 3 cartilaginous part. The connecting part between the two is called the narrow part. The functions of the Eustachian tube include balancing the pressure between the atmosphere and the middle ear, protecting the middle ear, and draining secretions from the middle ear to the nasopharynx. It is closed at rest and opens during movements such as swallowing and yawning. The principle of the Eustachian tube sonography method is to use the principle that the sound is transmitted through the air in the tube cavity at the moment of the Eustachian tube opening during swallowing, emit a sound source signal in the nasal cavity, and receive the sound transmitted by the opening of the Eustachian tube in the external auditory canal on the same side. The sonography method can directly and real-timely reflect the opening and closing state of the Eustachian tube and record the duration of the Eustachian tube opening. The whole process is non-invasive and inexpensive, and it is an examination under physiological conditions. High-resolution electronic computed tomography provides clear images in the axial and coronal positions of the bony Eustachian tube, while magnetic resonance imaging is a method to display the cartilaginous part of the Eustachian tube. It is particularly effective in displaying the anatomy of the Eustachian tube region through weighted phase continuous scanning. Imaging equipment can record and display the bony and cartilaginous parts of the Eustachian tube. Temporal bone high-resolution equipment can clearly display the tympanic opening of the bony part of the Eustachian tube and the bilateral pharyngeal openings on the axial plane. The Eustachian tube cartilage, tensor veli palatini, levator veli palatini and their surrounding fascia of the cartilaginous part can also be clearly displayed. In recent years, computer-assisted three-dimensional reconstruction has been used to study thin-layer specimens of the Eustachian tube in cadavers, and three-dimensional stereoscopic images of the Eustachian tube and surrounding tissues have been established, but the results are still somewhat different from the three-dimensional reconstruction of living tissues. Summary of the invention

[0003] The purpose of the present invention is to provide a method for detecting the open morphology of the Eustachian tube by using the acoustic measurement method of the Eustachian tube, wherein the Eustachian tube transfer function is calculated by the acoustic measurement method, and a bone conduction vibration sensor is attached to the larynx to detect the swallowing action, and the morphology of the Eustachian tube when it is open is observed by an imaging device at the moment of swallowing. The key physiological parameters of the open morphology of the Eustachian tube are detected by using the acoustic transfer function of the Eustachian tube in the open morphology calculated by the acoustic measurement method, and the image data of the open morphology of the Eustachian tube recorded by the imaging device is used as a guide to deepen the application of the acoustic measurement method and the study of the open morphology and function of the Eustachian tube.

[0004] The specific technical solution for achieving the purpose of the present invention is:

[0005] A method for detecting the opening morphology of the Eustachian tube by using a Eustachian tube acoustic detection method comprises the following specific steps:

[0006] Step 1: Use the Eustachian tube sonication device, place the earphone speaker at the front nostril on one side, seal it with memory foam, and play broadband white noise in the speaker; place the miniature microphone or microphone probe at the external auditory canal on the same side and seal it, receive the sound signal transmitted to the external auditory canal and perform analog-to-digital conversion, and input it into the computer as the sound measurement method received data for storage;

[0007] Step 2: Attach a vibration sensor to the throat to collect throat vibration data, complete a swallowing action, and when the vibration sensor detects throat vibration, start the imaging scanning device to start scanning the Eustachian tube morphology; when the vibration sensor detects that the throat vibration stops, that is, the swallowing action ends, stop scanning the Eustachian tube morphology, and record the imaging morphology data of the Eustachian tube in an open state;

[0008] Step 3: Time-align the laryngeal vibration data collected by the vibration sensor in step 2 with the sound signal received at the external auditory canal by the acoustic measurement method in step 1, and mark the swallowing moment by the laryngeal vibration data. The start of laryngeal vibration is the start of swallowing, and the end of laryngeal vibration is marked as the end of swallowing. The sound signal of the white noise in the nasal cavity transmitted to the external auditory canal through the open Eustachian tube after swallowing is intercepted from the received sound data obtained by the acoustic measurement method;

[0009] Step 4: After swallowing, the Eustachian tube opens actively, and the sound is transmitted from the nasal cavity to the external auditory canal through the Eustachian tube. The sound source signal x(t) played at the nostril and the sound signal y(t) intercepted in step 4 are used to calculate the Eustachian tube sound transfer function. The two sound signals are Fourier transformed to determine the frequency components of the transmitted signal. The cross spectrum C of the sound signal of the external auditory canal and the sound source signal of the nostril is obtained. xy (e jω ) divided by the autospectrum P of the nasal sound source signal xx (e jω ) to calculate the transfer function H1(e jω ), this transfer function is regarded as the response function of the Eustachian tube system to the input sound source signal in the frequency domain;

[0010]

[0011] Among them, X(e jω ) and Y(e jω ) are the Fourier transforms of the input signal x(t) and the output received signal y(t), respectively. jω ) is Y(e jω )'s complex conjugate;

[0012] In addition, the data of the acoustic measurement method when the swallowing action is not completed are collected to calculate the transfer function H2 (e j ω ) ; Dividing the two in the frequency domain, we get the frequency domain representation of the acoustic transfer function of the Eustachian tube in the open state:

[0013] H(e jω )=H1(e jω ) / H2(e jω );

[0014] Step 5: Conduct measurements on multiple real subjects, including people with normal Eustachian tube function and people with Eustachian tube dysfunction, and conduct measurements of Eustachian tube imaging morphology and Eustachian tube sonometry. Calculate the Eustachian tube acoustic transfer function from the sonometry data, and establish a database corresponding to the imaging scan morphology and acoustic transfer function of different people in the open Eustachian tube state;

[0015] Step 6: Using the database described in step 5, a computational model for inversely inferring the open morphology of the Eustachian tube from the acoustic transfer function of the Eustachian tube is established, wherein the input of the computational model is the acoustic transfer function of the Eustachian tube calculated based on the Eustachian tube acoustic measurement method, and the output is the key physiological parameters representing the open morphology of the Eustachian tube, wherein the key physiological parameters representing the open morphology of the Eustachian tube are given by the imaging record of the Eustachian tube; the database is divided into a training set and a test set, wherein the training set is used to train the computational model, and the test set is used to evaluate the model; there are two ways to establish the computational model, one is to construct the computational model based on a numerical calculation method, and the other is to construct the computational model using a deep learning network structure;

[0016] Furthermore, the key physiological parameters include the inner diameter of the Eustachian tube when it is open, i.e., the inner diameter of the cartilaginous part of the Eustachian tube, the inner diameter of the bony part, the inner diameter and length of the narrow part at the connection, the height difference between the tube and the horizontal line.

[0017] Furthermore, the computational model is constructed based on the numerical calculation method: according to the Eustachian tube morphology recorded by the image, the preliminary key physiological parameters of the Eustachian tube are determined, and preliminary modeling is performed; the morphology of the Eustachian tube is modeled as a finite element model or a one-dimensional transmission line, its inner diameter and length are defined, the Eustachian tube is divided into many small units, and the sound pressure or flow velocity boundary conditions at both ends of the Eustachian tube are set according to actual conditions. It is considered that the propagation mode of sound waves in the Eustachian tube is the same as the propagation mode of sound waves in the air. The one-dimensional wave equation for the propagation of sound waves in the Eustachian tube is:

[0018]

[0019] Where p is the sound pressure, t is the time, x is the displacement along the length of the Eustachian tube, and c is the speed of sound, that is, the speed of sound waves propagating in the Eustachian tube. represents the second-order partial derivative of the sound pressure p with respect to time t, which describes the rate at which the sound wave changes with time. It represents the second-order partial derivative of the sound pressure p with respect to the displacement x, and it describes the rate of change of the sound wave in space. The discrete system equations are solved by numerical methods to obtain the acoustic transfer function; the error between the acoustic transfer function obtained in the finite element model or the one-dimensional transmission line and the Eustachian tube acoustic transfer function calculated by the acoustic measurement method in the database is calculated, and the key physiological parameters of the Eustachian tube morphology in the computational model to be established are adjusted by the optimization method so that the acoustic transfer function output by it is close to the Eustachian tube acoustic transfer function obtained by experimental measurement, and the computational model is established by numerical calculation method.

[0020] Furthermore, the network structure using deep learning is used to construct a computational model: the numerical relationship between the acoustic transfer function of the Eustachian tube and the physiological parameters under the open state of the Eustachian tube is trained, the imaging data under the open state of the Eustachian tube is used as a reference, the acoustic transfer function under the open state of the Eustachian tube is input, the acoustic transfer function is standardized, and the normalized input is input into the network; the key physiological parameters of the Eustachian tube are analyzed and extracted from the imaging data of the open morphology of the Eustachian tube as the target label of the model; the convolution layer and the long short-term memory layer are used together to extract the input features, the one-dimensional convolution layer extracts the local features in the acoustic transfer function, the pooling layer reduces the spatial dimension of the features, extracts the key features and reduces the computational complexity, the long short-term memory layer processes the sequence data and captures the dynamic features in the time series; the fully connected layer is used to process the transformed features, and the features are integrated and transformed in the deep layer of the network; the linear layer generates the final prediction result according to the changed features, and the model output is the key physiological parameters of the Eustachian tube.

[0021] Beneficial effects of the present invention:

[0022] 1) The present invention can be used to detect the open morphology of the Eustachian tube, and provide a new method for evaluating the function of the Eustachian tube; the present invention uses the acoustic measurement method to play white noise at the nostril and collect the received signal at the external auditory canal, calculate the acoustic transfer function when the Eustachian tube is open, and calculate the key physiological parameters representing the open morphology of the Eustachian tube through the acoustic transfer function, which can be used to quantify the degree of the Eustachian tube opening, evaluate the function of the Eustachian tube, and divide the degree of obstruction of the Eustachian tube opening function more clearly. In addition, the present invention only involves the process of transmitting sound from the nostril to the external auditory canal, does not involve the process of pressurizing the nostril or the external auditory canal, and is not constrained by the physiological condition of the subject. The whole process is non-invasive and rapid, and has good clinical application prospects.

[0023] 2) The present invention attaches a vibration sensor to the throat to collect throat vibrations at the time of swallowing, aligns the data collected by the vibration sensor with the receiving data of the external auditory canal after swallowing collected by the acoustic measurement method in time, and marks the swallowing moment on the receiving data of the acoustic measurement method. When the subject is a patient with otitis media with water accumulation in the ear, or when the subject is young and swallowing is relatively irregular, it can assist in locating the swallowing moment and mark the opening of the Eustachian tube within a short time after swallowing. The collected external auditory canal receiving signal is valid sound data. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the present invention;

[0025] Figure 2 This is a deep learning network structure diagram. DETAILED DESCRIPTION

[0026] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0027] Example 1

[0028] See also Figure 1 A method for detecting the opening shape of the Eustachian tube by using the Eustachian tube acoustic detection method of this embodiment includes:

[0029] Step 1: Use the Eustachian tube sound measurement device to place the earphone speaker in the memory foam nose adapter into the front nostril on one side of the subject. This is the sound output end, and broadband white noise will be played as the sound source signal. Place the miniature microphone in the external auditory canal adapter into the external auditory canal on the same side of the subject, where the sound signal is received and collected. It is the sound input end, and wear earmuffs to soundproof the external noise. Connect the output speaker device and the input microphone device through an external sound card and communicate with the computer software. Play and record sound signals of different channels at the same time through multi-track files. The software generates broadband white noise and outputs it to the earphone speaker wrapped in memory foam and placed at the nostril. The microphone in the ear canal on the same side records the sound signal transmitted to the external auditory canal and inputs it into an audio track in the computer, and stores it in digital audio format simultaneously.

[0030] Step 2: After attaching the vibration sensor to the throat and completing the swallowing action, when the throat starts to vibrate and the swallowing action begins, the imaging scanning device is started to scan the morphology of the Eustachian tube. When the vibration sensor detects that the throat stops vibrating and the swallowing action ends, the scanning of the Eustachian tube morphology is stopped, and the laryngeal vibration sensor is used to indicate the start and stop of the imaging device, and the imaging morphology data of the Eustachian tube in an open state is recorded.

[0031] Step 3: Attach a vibration sensor to the throat and connect it to the computer using an external sound card. Complete a swallowing action and synchronously store the signals received by the vibration sensor from the beginning to the end of the swallowing action. By collecting the vibration of the throat during swallowing, the beginning and end of the swallowing action can be detected. The signal received by the vibration sensor is aligned in time with the signal received by the external auditory canal after the swallowing by the acoustic measurement method, the swallowing moment is marked, and the time period after the swallowing moment when the Eustachian tube actively opens and the external auditory canal receives the sound signal transmitted from the Eustachian tube is selected.

[0032] Step 4: After completing the swallowing action, the Eustachian tube opens actively, and the broadband white noise signal output at the anterior nares is transmitted to the external auditory canal through the Eustachian tube, and an increase in the decibel number of the sound signal is detected in the external auditory canal. By attaching a vibration sensor to the throat for synchronous recording, the moment when the swallowing action occurs is located, and the period of time when the sound signal received at the external auditory canal after the swallowing action is completed is intercepted. It is believed that the sound signal is a white noise signal played at the nostrils and transmitted through the open Eustachian tube channel. The Eustachian tube sound transfer function is calculated using the white noise sound source signal x(t) played at the nostrils and the effective received signal y(t) collected at the external auditory canal. The two sound signals are Fourier transformed, and the cross spectrum C of the external auditory canal received signal and the nostril sound source signal is obtained. xy (e jω ) divided by the autospectrum P of the nasal sound source signal xx (e jω ) to calculate the transfer function H1(e jω ), this transfer function can be regarded as the response function of the Eustachian tube system to the input sound source signal in the frequency domain.

[0033]

[0034] Among them, X(e jω ) and Y(e jω ) are the Fourier transforms of the input signal x(t) and the output received signal y(t), respectively. jω ) is Y(e jω ) is the complex conjugate of .

[0035] Step 5: In a resting state without swallowing, play white noise in the nasal cavity, collect the received signal in the external auditory canal, and calculate the sound transfer function H2 (e jω ), compared with the open state, divided in the frequency domain, and calculated the acoustic transfer function H(e) in the frequency domain when the Eustachian tube is in the open state jω ).

[0036] H(e jω )=H1(e jω ) / H2(ejω )

[0037] Further measurements are conducted on multiple real subjects. When selecting experimental subjects, both people with normal Eustachian tube function and people with impaired Eustachian tube function can be included. The Eustachian tube morphology and the Eustachian tube acoustic transfer function are measured when the Eustachian tube is actively open, and a database corresponding to the imaging morphological data and transfer function of the Eustachian tube of different populations when the Eustachian tube is open is established.

[0038] Step six: Use numerical calculation methods to establish the corresponding Eustachian tube acoustic transfer function calculation model when the Eustachian tube is open, and use the above database to modify and optimize the established calculation model. The key physiological parameters of the Eustachian tube morphology include the inner diameter and length of the Eustachian tube when it is open. According to the Eustachian tube morphology recorded in the image, the preliminary key physiological parameters of the Eustachian tube are determined, and preliminary modeling is performed. The morphology of the Eustachian tube is modeled as a finite element model or a one-dimensional transmission line, and its inner diameter and length are defined. The Eustachian tube is divided into many small units, and the boundary conditions such as sound pressure or flow velocity at both ends of the Eustachian tube are set according to actual conditions. It is assumed that the propagation mode of sound waves in the Eustachian tube is the same as the propagation mode of sound waves in the air. The one-dimensional wave equation for the propagation of sound waves in the Eustachian tube is:

[0039]

[0040] Where p is the sound pressure, t is the time, x is the length of the Eustachian tube, and c is the speed of sound, that is, the speed of sound waves propagating in the Eustachian tube. represents the second-order partial derivative of the sound pressure p with respect to time t, which describes the rate at which the sound wave changes with time. It represents the second-order partial derivative of the sound pressure p with respect to the displacement x, which describes the rate of change of the sound wave in space. The discrete system equations are solved by numerical methods to obtain the acoustic transfer function. The error between the acoustic transfer function obtained in the finite element method or one-dimensional transmission line and the Eustachian tube acoustic transfer function calculated in the database according to the acoustic measurement method is calculated. The key physiological parameters of the Eustachian tube morphology in the computational model to be established are adjusted by the optimization method so that the acoustic transfer function output by it is as close as possible to the Eustachian tube acoustic transfer function obtained by experimental measurement. The computational model of the Eustachian tube is established by numerical calculation methods. The inner diameter, length and other physiological parameters of the Eustachian tube are inversely deduced by the acoustic transfer function of the Eustachian tube, and the opening morphology of the Eustachian tube is calculated.

[0041] Example 2

[0042] A method for detecting the patency of the Eustachian tube by using the Eustachian tube acoustic detection method of this embodiment includes:

[0043] Step 1: Place the earphone speaker in the memory foam nose adapter into the front nostril on one side of the subject, which is the output end of the sound, and play broadband white noise as the sound source signal. Place the miniature microphone in the external auditory canal adapter into the external auditory canal on the same side of the subject, where the sound signal is received and collected, which is the input end of the sound, and wear earmuffs to soundproof the external noise. The output microphone device and the input microphone device are connected through an external sound card and communicate with the computer. The computer generates broadband white noise and outputs it to the earphone speaker wrapped in memory foam and placed in the nostril. The sound signal collected by the microphone in the ear canal on the same side is transmitted from the nostril sound source through the open Eustachian tube to the external auditory canal and input into the computer, and is stored in digital audio format simultaneously. The output signal corresponds to the input signal and waits for subsequent processing.

[0044] Step 2: Stick a vibration sensor on the throat and connect it to the computer using an external sound card. Complete a swallowing action and synchronously store the signals received by the vibration sensor from the beginning to the end of the swallowing action. By collecting the vibration of the throat during swallowing, the beginning and end of the swallowing action can be detected. The signal received by the vibration sensor is aligned in time with the signal received by the external auditory canal after the swallowing by the acoustic measurement method, the swallowing moment is marked, and the time period after the swallowing moment when the Eustachian tube actively opens and the external auditory canal receives the sound signal transmitted from the Eustachian tube is selected. The sound signal received in this intercepted time period is further used to calculate the sound transfer function when the Eustachian tube is open.

[0045] Step 3: After attaching the vibration sensor to the throat and completing the swallowing action, when the throat starts to vibrate and the swallowing action begins, the imaging scanning device is started to scan the morphology of the Eustachian tube. When the vibration sensor detects that the throat stops vibrating and the swallowing action ends, the scanning of the Eustachian tube morphology is stopped, and the laryngeal vibration sensor is used to indicate the start and stop of the imaging device, and the imaging morphology data of the Eustachian tube in an open state is recorded.

[0046] Step 4: After completing the swallowing action, the Eustachian tube opens actively, and the broadband white noise signal output at the anterior nares is transmitted to the external auditory canal through the Eustachian tube, and an increase in the decibel number of the sound signal is detected in the external auditory canal. By attaching a vibration sensor to the throat for synchronous recording, the moment when the swallowing action occurs is located, and the period of time when the sound signal received at the external auditory canal after the swallowing action is completed is intercepted. It is believed that the sound signal is a white noise signal played at the nostrils and transmitted through the open Eustachian tube channel. The Eustachian tube sound transfer function is calculated using the white noise sound source signal x(t) played at the nostrils and the effective received signal y(t) collected at the external auditory canal. The two sound signals are Fourier transformed, and the cross spectrum C of the external auditory canal received signal and the nostril sound source signal is obtained. xy (e jω ) divided by the autospectrum P of the nasal sound source signalxx (e jω ) to calculate the transfer function H1(e jω ), this transfer function can be regarded as the response function of the Eustachian tube system to the input sound source signal in the frequency domain.

[0047]

[0048] Among them, X(e jω ) and Y(e jω ) are the Fourier transforms of the input signal x(t) and the output received signal y(t), respectively. * (e jω ) is Y(e jω ) is the complex conjugate of .

[0049] Step 5: In a resting state without swallowing, play white noise in the nasal cavity, collect the received signal in the external auditory canal, and calculate the sound transfer function H2 (e jω ), compared with the open state, divided in the frequency domain, and calculated the acoustic transfer function H(e) in the frequency domain when the Eustachian tube is in the open state jω ).

[0050] H(e jω )=H1(e jω ) / H2(e jω )

[0051] Further measurements are conducted on multiple real subjects. When selecting experimental subjects, both people with normal Eustachian tube function and people with impaired Eustachian tube function can be included. The Eustachian tube morphology and the Eustachian tube acoustic transfer function are measured when the Eustachian tube is actively open, and a database corresponding to the imaging morphological data and transfer function of the Eustachian tube of different populations when the Eustachian tube is open is established.

[0052] Step 6: Use a deep learning network structure to build a computational model, train the numerical relationship between the acoustic transfer function of the Eustachian tube and the physiological parameters under the open state of the Eustachian tube, use the imaging data of the open state of the Eustachian tube as a reference, input the acoustic transfer function under the open state of the Eustachian tube, output the physiological parameters of the Eustachian tube, standardize the acoustic transfer function, and input it into the network after normalization. Analyze and extract the key physiological parameters of the Eustachian tube such as the inner diameter and length from the imaging data of the open morphology of the Eustachian tube as the target label of the model. The network structure is as follows: Figure 2As shown in the figure, the acoustic transfer function of the Eustachian tube is input, and the convolution layer and the long short-term memory layer are used to extract the input features. The one-dimensional convolution layer extracts local features in the acoustic transfer function. The pooling layer reduces the spatial dimension of the features, extracts key features and reduces the computational complexity. The long short-term memory layer processes sequence data and captures dynamic features in the time series. The fully connected layer is used to process the transformed features, and the features are integrated and transformed in the deep layer of the network. The linear layer generates the final prediction results based on the changed features. The model outputs the key physiological parameters of the Eustachian tube and detects the inner diameter and length of the Eustachian tube. The mean square error is used as the loss function during training, and the stochastic gradient descent is used as the optimizer for weight update during model training. The data set is divided into a training set, a validation set, and a test set. The training set is used for model training, the validation set is used for model parameter adjustment, and the test set is used for model evaluation. After the model training is completed, the acoustic transfer function under the new Eustachian tube open state is calculated using the acoustic measurement method as input to detect the key physiological parameters under the open morphology of the Eustachian tube.

Claims

1. A method for detecting the opening morphology of the Eustachian tube by using the Eustachian tube acoustic detection method, characterized in that: The specific steps include: Step 1: Use the Eustachian tube sonication device to place the earphone speaker at the front nostril on one side, seal it with memory foam, and play broadband white noise in the speaker; Place a miniature microphone or microphone probe in the external auditory canal on the same side and seal it, receive the sound signal transmitted to the external auditory canal and perform analog-to-digital conversion, and input it into a computer for storage as the received data of the acoustic measurement method; Step 2: Attach a vibration sensor to the throat to collect throat vibration data, complete a swallowing action, and when the vibration sensor detects throat vibration, start the imaging scanning device to start scanning the Eustachian tube morphology; when the vibration sensor detects that the throat vibration stops, that is, the swallowing action ends, stop scanning the Eustachian tube morphology, and record the imaging morphology data of the Eustachian tube in an open state; Step 3: Time-align the laryngeal vibration data collected by the vibration sensor in step 2 with the sound signal received at the external auditory canal by the acoustic measurement method in step 1, and mark the swallowing moment by the laryngeal vibration data. The start of laryngeal vibration is the start of swallowing, and the end of laryngeal vibration is marked as the end of swallowing. The sound signal of the white noise in the nasal cavity transmitted to the external auditory canal through the open Eustachian tube after swallowing is intercepted from the received sound data obtained by the acoustic measurement method; Step 4: After swallowing, the Eustachian tube opens actively, and the sound is transmitted from the nasal cavity to the external auditory canal through the Eustachian tube. The sound source signal x(t) played at the nostril and the sound signal y(t) intercepted in step 4 are used to calculate the Eustachian tube sound transfer function. The two sound signals are Fourier transformed to determine the frequency components of the transmitted signal. The cross spectrum C of the sound signal of the external auditory canal and the sound source signal of the nostril is obtained. xy (e jω ) divided by the autospectrum P of the nasal sound source signal xx (e jω ) to calculate the transfer function H1(e jω ), this transfer function is regarded as the response function of the Eustachian tube system to the input sound source signal in the frequency domain; Among them, X(e jω ) and Y(e jω ) are the Fourier transforms of the input signal x(t) and the output received signal y(t), respectively. * (e jω ) is Y(e jω )'s complex conjugate; In addition, the data of the acoustic measurement method when the swallowing action is not completed are collected to calculate the transfer function H2 (e jω ) ; Dividing the two in the frequency domain, we get the frequency domain representation of the acoustic transfer function of the Eustachian tube in the open state: H(e jω )=H1(and jω ) / H2(e jω ); Step 5: Conduct measurements on multiple real subjects, including people with normal Eustachian tube function and people with Eustachian tube dysfunction, and conduct measurements of Eustachian tube imaging morphology and Eustachian tube sonometry. Calculate the Eustachian tube acoustic transfer function from the sonometry data, and establish a database corresponding to the imaging scan morphology and acoustic transfer function of different people in the open Eustachian tube state; Step six: Use the database described in step five to establish a computational model for inverting the Eustachian tube acoustic transfer function to infer the Eustachian tube opening morphology, the input of the computational model is the Eustachian tube acoustic transfer function calculated based on the Eustachian tube sonic measurement method, and the output is the key physiological parameters representing the Eustachian tube opening morphology, wherein the key physiological parameters representing the Eustachian tube opening morphology are given by the imaging records of the Eustachian tube; divide the database into a training set and a test set, wherein the training set is used to train the computational model, and the test set is used to evaluate the model; there are two ways to establish the computational model, one is to construct the computational model based on the numerical calculation method, and the other is to construct the computational model using a deep learning network structure.

2. The method according to claim 1, characterized in that: The key physiological parameters include the inner diameter of the Eustachian tube when it is open, namely the inner diameter of the cartilaginous part of the Eustachian tube, the inner diameter of the bony part, the inner diameter and length of the narrow part at the connection, the height difference between the upper and lower parts of the tube, and the angle with the horizontal line.

3. The method according to claim 1, characterized in that: The computational model is constructed based on the numerical calculation method: according to the Eustachian tube morphology recorded by the image, the preliminary key physiological parameters of the Eustachian tube are determined, and preliminary modeling is performed; the morphology of the Eustachian tube is modeled as a finite element model or a one-dimensional transmission line, and its inner diameter and length are defined. The Eustachian tube is divided into many small units, and the sound pressure or flow velocity boundary conditions at both ends of the Eustachian tube are set according to actual conditions. It is considered that the propagation mode of sound waves in the Eustachian tube is the same as the propagation mode of sound waves in the air. The one-dimensional wave equation for the propagation of sound waves in the Eustachian tube is: Where p is the sound pressure, t is the time, x is the length of the Eustachian tube, and c is the speed of sound, that is, the speed of sound waves propagating in the Eustachian tube. represents the second-order partial derivative of the sound pressure p with respect to time t, describing the rate at which the sound wave changes with time. The second-order partial derivative of the sound pressure p with respect to the displacement x describes the rate of change of the sound wave in space; the discrete system equations are solved by numerical methods to obtain the acoustic transfer function; the error between the acoustic transfer function obtained in the finite element model or the one-dimensional transmission line and the Eustachian tube acoustic transfer function calculated in the database according to the acoustic measurement method is calculated, and the key physiological parameters of the Eustachian tube morphology in the computational model to be established are adjusted by the optimization method so that the acoustic transfer function output by it is close to the Eustachian tube acoustic transfer function obtained by experimental measurement, and the computational model is established by numerical calculation method.

4. The method according to claim 1, characterized in that The computational model is constructed by using a deep learning network structure: the numerical relationship between the acoustic transfer function of the Eustachian tube and the physiological parameters under the open state of the Eustachian tube is trained, the imaging data under the open state of the Eustachian tube is used as a reference, the acoustic transfer function under the open state of the Eustachian tube is input, the acoustic transfer function is standardized, and the normalized data is input into the network; The key physiological parameters of the Eustachian tube are analyzed and extracted from the imaging data of the Eustachian tube's open morphology as the target label of the model; the convolution layer and the long short-term memory layer are used together to extract the input features. The one-dimensional convolution layer extracts the local features in the acoustic transfer function. The pooling layer reduces the spatial dimension of the features, extracts the key features and reduces the computational complexity. The long short-term memory layer processes the sequence data and captures the dynamic features in the time series. The fully connected layer is used to process the transformed features and integrates and transforms the features in the deep layer of the network. The final prediction results are generated based on the changed features, and the model output is the key physiological parameters of the Eustachian tube.

Images