Method for detecting opening degree of Eustachian tube based on stimulation signal

Abstract

The invention discloses a stimulus signal-based Eustachian tube opening degree detection method, which comprises the following steps of: placing a nose adapter with a built-in earphone loudspeaker into a nostril of a subject, and placing an ear canal adapter with a built-in in-ear loudspeaker into an external auditory canal on the same side of the subject, the earphone loudspeaker plays a stimulation signal capable of effectively distinguishing the open state and the closed state of the Eustachian tube; the design method of the stimulation signal comprises the following steps: designing the stimulation signal with specific frequency and characteristics by respectively measuring and calculating Eustachian tube transfer functions in a resting state and a swallowing action; when the subject is in a resting state and in a swallowing action, sound signals received by the in-ear loudspeaker are respectively collected through the sound sensing terminal; and processing the sound signal to obtain a relative sound power level, and inputting the relative sound power level into a pre-established and trained degree detection model to obtain a numerical value representing the opening degree of the Eustachian tube of the subject. The detection accuracy of the open function of the Eustachian tube is improved, and new auxiliary judgment intermediate data is provided for various Eustachian tube dysfunctions.

Description

Technical Field

[0001] The present invention belongs to the technical field of Eustachian tube opening function detection, and in particular relates to a method for detecting the degree of Eustachian tube opening based on a stimulation signal, which designs a stimulation signal with a specific frequency and characteristic and automatically identifies a received signal. Background Art

[0002] The Eustachian tube is a tube connecting the middle ear and the nasopharynx. It consists of a posterior and lateral 1 / 3 bone part and anterior and medial 2 / 3 cartilage part, which are connected at the isthmus. The most important function of the Eustachian tube is to maintain the balance of air pressure inside and outside the middle ear. When the normal opening function of the Eustachian tube is impaired, it may cause negative pressure in the tympanic cavity, which may then develop into secretory otitis media. How to detect the opening function of the Eustachian tube in real time and accurately is very important for the evaluation of the efficacy of secretory otitis media, the prognosis evaluation after tympanoplasty, and the selection of the timing of tube removal after myringotomy and tube placement. As early as 1869, Politzer first proposed the concept of Eustachian tube function acoustic measurement. In the experiment, a special headset was used to transmit a sound source signal to one side of the subject's anterior nostril, and a microphone was placed in the external auditory canal to receive the sound signal. The Eustachian tube was actively opened through the subject's swallowing action or yawning. In theory, a well-functioning Eustachian tube opening can lead to an increase in the sound power level of the external auditory canal sound source signal. The acoustic test method can directly and dynamically reflect the opening and closing status of the Eustachian tube in real time without applying additional pressure to the middle ear. It is an examination under physiological conditions, and the test results are basically not affected by the thickness of the tympanic membrane and whether the tympanic membrane is intact. It can be tested when the tympanic membrane is intact, and it can also be used to test the Eustachian tube function after tympanic membrane perforation or tympanic tube placement. In addition, the Eustachian tube function acoustic test method is non-invasive and inexpensive, and will not cause pain to the person being examined, so it is particularly suitable for the study of Eustachian tube function in children.

[0003] However, the traditional Eustachian tube function acoustic test method has deficiencies from the selection of sound source to the judgment conditions of the normal value of the received signal. These defects also lead to the failure of this method to serve the clinic. First, there is no recognized optimization design method for the test sound source frequency and time domain characteristics of the Eustachian tube function acoustic test method at home and abroad. Although the narrowband noise of 7-8kHz is the most commonly selected experimental frequency so far, each laboratory chooses the "best" method based on the experience and preferences of predecessors. Some experiments choose to use broadband white noise, some choose narrowband white noise, some use their own set "perfect sequence", and some use pure tone signals. Secondly, in previous studies, the quantitative index of the sound power level of the external auditory canal sound source signal rising by 5dB is usually used as the criterion for the opening of the Eustachian tube. Due to the differences between individuals and the non-steady-state characteristics of physiological activities, such an evaluation index is likely to be not robust enough. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the prior art, and a method for detecting the degree of Eustachian tube opening based on a stimulation signal is proposed. The present invention proposes how to design a more suitable method for testing the sound source signal, and uses a machine learning method to detect the degree of Eustachian tube opening of the received signal, so that the Eustachian tube function acoustic measurement system can be improved, and a new method for Eustachian tube function detection is provided. A further purpose of the present invention is to provide the above method, and combine the test results with other existing Eustachian tube function pressure examination methods such as acoustic impedance tympanogram, acoustic impedance peak pressure point change and other technical means to provide richer and more comprehensive intermediate data.

[0005] In order to achieve the above object, the present invention proposes a method for detecting the degree of Eustachian tube opening based on a stimulation signal, comprising:

[0006] A nasal adapter with a built-in earphone speaker is placed in the nostril of the subject, and an ear canal adapter with a built-in in-ear speaker is placed in the external auditory canal on the same side of the subject, and the earphone speaker plays a stimulation signal that can effectively distinguish between the open and closed states of the Eustachian tube; the stimulation signal is designed by measuring and calculating the transfer function of the Eustachian tube in a resting state and during swallowing action, respectively, to design a stimulation signal with specific frequency and characteristics;

[0007] When the subject is at rest and swallowing, the acoustic signal received by the in-ear speaker is collected by the acoustic sensor terminal;

[0008] The received sound signal is processed to obtain a relative sound power level, which is input into a pre-established and trained degree detection model to obtain a numerical value representing the degree of opening of the subject's Eustachian tube.

[0009] Preferably, the stimulation signal design method includes:

[0010] The nasal adapter with built-in earphone speakers is placed in the nostril of the subject, and the ear canal adapter with built-in in-ear speakers is placed in the external auditory canal on the same side of the subject. The earphone speakers play a short-duration pulse signal x(t). The sound signal y(t) received by the external auditory canal is collected in the resting state and during swallowing, and the corresponding Eustachian tube transfer function h(t) from the nostril to the ear canal is calculated:

[0011]

[0012] Where, t represents the time domain, FFT() and IFF() represent Fourier transform and inverse Fourier transform respectively;

[0013] The Eustachian tube transfer function h(t) at rest and during swallowing was analyzed to determine that a fixed-duration pure tone line spectrum signal of a specific frequency between 5.5kHz and 6.5kHz can effectively distinguish between the open and closed states of the Eustachian tube.

[0014] Preferably, the step of processing the received sound signal to obtain the relative sound power level comprises:

[0015] Calculate the average sound power level in the 6-10kHz frequency band within each 128 / 3000 second interval to obtain the strongest moment of the impact sound signal accompanying each swallowing action of the subject;

[0016] Obtain the maximum sound power level of the line spectrum signal corresponding to a window width of 128 / 3000 seconds measured within the range from 0.5 seconds before to 1.0 seconds after the strongest moment of the impact sound signal, as well as the average sound power level of the corresponding window except for the strongest moment of the impact sound signal;

[0017] The relative sound power level is calculated by calculating the difference between the maximum sound power level and the average sound power level.

[0018] Preferably, the degree detection model is a linear regression model or a support vector regression model.

[0019] Preferably, the method further comprises a step of training a degree detection model, comprising:

[0020] Experiments were conducted on patients with secretory otitis media and healthy people, and multiple groups of received signal samples with different labels were collected. The labels were numerical values ​​indicating the degree of patency of the Eustachian tube.

[0021] The received signal samples are divided into a training set and a test set, and the normalized training set data are input into the linear regression model y = β 0 +β 1 x, where x is the difference in sound power level and y is the degree of patency of the Eustachian tube. The model parameter β is estimated using the least squares method. 0 and β 1 , add regularization terms to prevent overfitting, use cross-validation method to train, adjust parameters, and use test set data for verification to obtain the corresponding degree of Eustachian tube opening. Use mean square error to evaluate the linear regression model to obtain a trained degree detection model.

[0022] Preferably, the method further comprises a step of training a degree detection model, comprising:

[0023] Experiments were conducted on patients with secretory otitis media and healthy people, and multiple groups of received signal samples with different labels were collected. The labels were numerical values ​​indicating the degree of patency of the Eustachian tube.

[0024] The received signal samples are divided into training set and test set. The training set data are input into the support vector regression model in sequence. The training process is transformed into a convex optimization problem. The goal is to find the weight vector ω and the bias vector b, as well as an interval boundary to minimize the regularization loss of the model. The key parameters of the model include: regularization parameter C, control of the penalty degree ε of the model for the error term, definition of the tolerance error between the predicted value and the actual value, and kernel function parameters. The kernel function selects the RBF kernel. The best parameter combination is found by grid search combined with cross-validation, and the test set data is used for verification to obtain the corresponding degree of Eustachian tube opening. The support vector regression model is evaluated using the mean square error to obtain a trained degree detection model.

[0025] Preferably, the experiment is conducted on patients with secretory otitis media and healthy people, and multiple groups of received signal samples with different labels are collected, including:

[0026] Step 1) inserting a nasal adapter with a built-in earphone speaker into the left nostril of the subject, inserting an ear canal adapter with a built-in in-ear speaker into the external auditory canal on the same side of the subject, and playing a sound source signal of a specific audio and a fixed duration through the earphone speaker; during the playing of each sound source signal, the subject cooperates to complete a swallowing action;

[0027] Repeat the above steps 3 times to obtain 3 sets of data;

[0028] Step 2) inserting a nasal adapter with a built-in earphone speaker into the right nostril of the subject, inserting an ear canal adapter with a built-in in-ear speaker into the external auditory canal on the same side of the subject, and playing a sound source signal of a specific audio and a fixed duration through the earphone speaker; during the playing of each sound source signal, the subject cooperates to complete a swallowing action;

[0029] Repeat the above steps 3 times to obtain 3 sets of data.

[0030] Compared with the prior art, the advantages of the present invention are:

[0031] 1. The method of the present invention can be used for the evaluation of the patency test of the Eustachian tube, providing intermediate data for the diagnosis of the Eustachian tube function, so as to facilitate in-depth research on the Eustachian tube.

[0032] In the present invention, the strongest moment of the impact sound signal accompanying each swallowing action is located by calculating the average sound power level in the 6-10kHz frequency band within each 128 / 3000 second interval, and this is used as the central moment to extract, track and conduct quantitative statistical analysis on the time-domain sound power level changes of the sound source pure tone signal. The definition of the swallowing center moment in the present invention has not been mentioned in the literature. Especially when the acoustic measurement method is applied to young children, it cannot be ensured that they have completed the swallowing action according to the instructions, and they can only rely on drinking water to induce swallowing. This irregular and unscheduled swallowing action requires a clear definition of the swallowing center moment for identification. Subsequently, by calculating the maximum relative sound power level corresponding to the swallowing moment and the average relative sound power level at other moments, the difference in relative sound power levels is obtained as the input of the machine learning model, such as a linear model linear regression model or a nonlinear model support vector regression model, and a numerical value representing the degree of Eustachian tube opening is calculated.

[0033] 2. The present invention considers the transfer function of the transmission link of the audio signal from the nose to the ear when the acoustic measurement method is used. Through different experiments under resting state and swallowing action, the transfer function of the link when the Eustachian tube is open and closed is calculated respectively, and the results are statistically analyzed, and the reasonable frequency band of the Eustachian tube function acoustic measurement method is proposed to be 5.5kHz-6.5kHz. Later, we use the pure tone signal of 6kHz in this frequency band for testing. In addition, because this method only involves the process of sound transmission from nose to ear, it can directly and dynamically reflect the opening and closing state of the Eustachian tube in real time without applying additional pressure to the middle ear. It is an inspection under physiological state, and the inspection result is basically not affected by the thickness of the tympanic membrane and whether the tympanic membrane is intact. It can be tested in the case of intact tympanic membrane, and the Eustachian tube function after tympanic membrane perforation or tympanic tube placement can also be tested. The whole process is non-invasive and cheap, and will not cause pain to the person being examined, so it is particularly suitable for the study of children's Eustachian tube function, and provides a new method for obtaining intermediate data of Eustachian tube function for clinical detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of stimulation signal design and receiving signal identification for Eustachian tube function acoustic measurement method. DETAILED DESCRIPTION

[0035] The present invention studies the corresponding physiological acoustic characteristic parameters of the audio signal from the anterior nasal cavity to the external auditory canal through digital audio signal analysis technology, measures and calculates the transfer function of the audio signal from the nostril to the auditory canal, and designs a stimulation signal with a specific frequency and characteristics based on the characteristics of the transfer function. The designed stimulation signal is used in the Eustachian tube opening function test, and a machine learning method is applied to use the difference in the relative sound power level of the ear canal received signal as input to obtain a numerical value representing the degree of Eustachian tube opening. The present invention is expected to improve the detection accuracy of the Eustachian tube opening function, provide new intermediate data for auxiliary judgment for the diagnosis, treatment and prognosis evaluation of various Eustachian tube dysfunction diseases, especially secretory otitis media in children, and assist clinical decision-making.

[0036] More specifically, the present invention is a method for checking the patency of the Eustachian tube using an acoustic signal of a specific frequency and characteristic, which comprises the steps of:

[0037] 1. The input end is the headphone speaker in the nose, which plays the specific audio signal of the sound source. The output terminal is an in-ear speaker. The sound sensor terminal is a 4mm diameter miniature microphone, which collects and stores the sound signals transmitted to the external auditory canal. The input / output audio signals are digitally converted through an external sound card to communicate with the computer.

[0038] 2. Input a short-time pulse signal, and calculate the acoustic transfer function of the transmission channel from the nose to the ear in the resting state and with the swallowing action, respectively, to obtain the impulse response function and the corresponding frequency response function. Among them, the frequency range of 5.5kHz-6.5kHz can effectively distinguish the open and closed states of the Eustachian tube.

[0039] 3. By analyzing the transfer function of the Eustachian tube in the open and closed states, a 6kHz stimulation signal is designed for the Eustachian tube opening function test. A 6kHz 5-second pure tone line spectrum signal is selected as the test sound source signal, repeated three times, with a 1-second interval between each playback.

[0040] 4. During the playback of each pure tone spectrum signal, the subject is required to complete a swallowing action (dry swallowing), complete three swallowing actions in total, and complete one experiment in both ears. Each subject obtains six sets of data. Experiments were conducted on subjects with secretory otitis media and healthy people to obtain multiple sets of data.

[0041] 5. After receiving the received signal, calculate the average sound power level in the 6-10kHz frequency band within each 128 / 3000 second interval of the received signal to locate the strongest moment of the impact sound signal accompanying each swallowing action, and obtain the maximum sound power level corresponding to the 128 / 3000 second window width of the line spectrum signal measured within the range of 0.5 seconds before and 1.0 seconds after this central moment.

[0042] 6. Calculate the difference between the maximum sound power level at the moment when the swallowing impact signal is strongest and the average sound power level excluding the swallowing moment, and obtain the relative sound power level as the input feature. Use the trained machine learning model to obtain the numerical value representing the degree of Eustachian tube opening.

[0043] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0044] Example 1

[0045] Embodiment 1 of the present invention proposes a method for detecting the degree of opening of the Eustachian tube based on a stimulation signal, comprising the following steps:

[0046] 1. Place the earphone speaker in the memory foam nose adapter into the left anterior nostril of the subject (this is the output end), place the sound sensor terminal in the external auditory canal adapter, which is a 4mm diameter micro-microphone, into the left external auditory canal (this is the input end), and use the external sound card to digitally convert the input / output audio signals to communicate with the computer. Play the specific audio signal of the sound source and output it to the earphone speaker in the memory foam nose adapter, while the micro-microphone in the external auditory canal adapter will collect the sound signal transmitted to the external auditory canal and input it into the computer, and simultaneously store it in a lossless digital audio format.

[0047] 2. The earphone speaker at the nostril plays a short-time pulse signal, and collects the sound signal received by the external auditory canal in a resting state. The original pulse signal is combined with the sound signal received in the resting state to calculate the transfer function in the closed state of the Eustachian tube. Continue to play the pulse signal, and complete the swallowing action during the pulse signal playback, and collect the sound signal received by the external auditory canal during the entire swallowing action. The transfer function in the open state of the Eustachian tube is calculated by the sound source signal and the sound signal received during the swallowing action. The sound signal output response y(t) received at the external auditory canal is equal to the convolution of the short-time pulse input excitation x(t) played at the speaker at the earphone and the pulse response h(t) through the Eustachian tube system:

[0048] y(t)=x(t)*h(t) (1)

[0049] The system frequency impulse response is obtained by calculating the output spectrum divided by the input spectrum and then taking the inverse Fourier transform.

[0050]

[0051] 3. A comparative analysis was performed on the transfer functions calculated under the closed and open states of the Eustachian tube. The results showed that there was a significant difference between the two transfer functions in the frequency range of 5.5kHz-6.5kHz. Therefore, it was proposed that the stimulation signal at 5.5kHz-6.5kHz can effectively distinguish between the open and closed states of the Eustachian tube. In this frequency band, a 6kHz pure tone signal was designed and selected as the stimulation sound source for experimental testing.

[0052] 4. Select a 5-second 6kHz pure tone line spectrum signal repeated three times as the test stimulus signal, with an interval of 1 second between pure tone signals. Play the sound source-specific audio signal through the headphone speaker at the left nostril. During the playback of each pure tone line spectrum signal, the subject is required to cooperate in completing a swallowing action (dry swallowing), and the sound signal transmitted to the ear is collected through the miniature microphone in the left external auditory canal. Place the headphone speaker at the right nostril and the microphone at the right external auditory canal, and repeat the above experiment. Complete the above experiment once on each ear, and obtain three repeated data on one side, and collect six sets of data for each subject. Experiments were conducted in patients with secretory otitis media and healthy people, and multiple sets of received signal samples with different labels were collected.

[0053] 5. In order to extract and track the time-domain sound power level changes of the test pure tone signal, and thus train the machine learning model, this study takes the strongest moment of the swallowing impact signal as the center (in this study, the strongest moment of the impact sound signal accompanying each swallowing action is located by calculating the average sound power level in the 6-10kHz frequency band within each 128 / 3000 second interval), and obtains the maximum sound power level of the line spectrum signal corresponding to the 128 / 3000 second window width within the range of 0.5 seconds before and 1.0 seconds after this center moment, as well as the average sound power level of the corresponding window except this center moment. Finally, the relative sound power level is calculated as the difference between the two, which is used as the feature set input of the machine learning model.

[0054] 6. Divide the collected multiple groups of ear canal reception data into training sets and test sets. Use the cross-validation method to train the linear regression model with the training set, and use the test set to verify the performance of the model. For each sample in the data set, calculate the relative sound power level as a feature, normalize the data and input it. The target variable of the linear regression model is a numerical value indicating the degree of Eustachian tube opening. The linear regression model is simple and easy to understand, which is conducive to understanding the impact of features on the target variable. Construct a simple linear regression model in the form of y = β 0 +β 1 x, where x is the difference in sound power level and y is the degree of patency of the Eustachian tube, the least squares method is used to estimate the model parameter β 0 and β 1, add regularization terms to prevent overfitting. After the model training is completed, for each sample in the test set, the trained model is used to obtain the corresponding Eustachian tube opening degree, and the model is evaluated using the mean square error. For the newly input ear canal reception data, the corresponding relative sound power level is calculated, and the corresponding value representing the Eustachian tube opening degree is calculated using the trained linear regression model.

[0055] Example 2

[0056] Embodiment 2 of the present invention proposes a method for detecting the degree of Eustachian tube opening based on a stimulation signal, which is different from Embodiment 1 in that a support vector regression (SVR) model is used. The method comprises the following steps:

[0057] 1. Place the earphone speaker in the memory foam nose adapter into the left anterior nostril of the subject (this is the output end), place the sound sensor terminal in the external auditory canal adapter, which is a 4mm diameter micro-microphone, into the left external auditory canal (this is the input end), and use the external sound card to digitally convert the input / output audio signals to communicate with the computer. Play the specific audio signal of the sound source and output it to the earphone speaker in the memory foam nose adapter, while the micro-microphone in the external auditory canal adapter will collect the sound signal transmitted to the external auditory canal and input it into the computer, and simultaneously store it in a lossless digital audio format.

[0058] 2. The earphone speaker at the nostril plays a short pulse signal, and collects the sound signal received by the external auditory canal in a resting state. The original pulse signal is combined with the sound signal received in the resting state to calculate the transfer function in the closed state of the Eustachian tube. Continue to play the pulse signal, and complete the swallowing action during the pulse signal playback, and collect the sound signal received by the external auditory canal during the entire swallowing action. The transfer function in the open state of the Eustachian tube is calculated by the sound source signal and the sound signal received during the swallowing action.

[0059] 3. A comparative analysis was performed on the transfer functions calculated under the closed and open states of the Eustachian tube. The results showed that there was a significant difference between the two transfer functions in the frequency range of 5.5kHz-6.5kHz. Therefore, it was proposed that the stimulation signal at 5.5kHz-6.5kHz can effectively distinguish between the open and closed states of the Eustachian tube. In this frequency band, a 6kHz pure tone signal was designed and selected as the stimulation sound source for experimental testing.

[0060] 4. Select a 5-second 6kHz pure tone line spectrum signal repeated three times as the test stimulus signal, with an interval of 1 second between pure tone signals. Play the sound source-specific audio signal through the headphone speaker at the left nostril. During the playback of each pure tone line spectrum signal, the subject is required to cooperate in completing a swallowing action (dry swallowing), and the sound signal transmitted to the ear is collected through the miniature microphone in the left external auditory canal. Place the headphone speaker at the right nostril and the microphone at the right external auditory canal, and repeat the above experiment. Complete the above experiment once on each ear, and obtain three repeated data on one side, and collect six sets of data for each subject. Experiments were conducted in patients with secretory otitis media and healthy people, and multiple sets of received signal samples with different labels were collected.

[0061] 5. In order to extract and track the changes in the time-domain sound power level of the test pure tone signal, and thus train the machine learning model, this study takes the strongest moment of the swallowing impact signal as the center (in this study, the strongest moment of the impact sound signal accompanying each swallowing action is located by calculating the average sound power level in the 6-10kHz frequency band within each 128 / 3000 second interval). The maximum sound power level of the line spectrum signal corresponding to the 128 / 3000 second window width measured within the range from 0.5 seconds before to 1.0 seconds after this center moment is obtained, as well as the average sound power level of the corresponding window excluding this center moment. Finally, the relative sound power level is calculated as the difference between the two, and this is used as the feature set input of the machine learning model.

[0062] 6. The collected multiple sets of ear canal reception data are divided into training sets and test sets. The training set is used to train the support vector regression (SVR) model, and the model performance is verified on the test set. The feature of each sample is the calculated relative sound power level, and the target variable is the numerical value indicating the degree of Eustachian tube opening. Unlike linear regression, support vector regression is a nonlinear model that can better capture the complex relationship in the data and map the data to a high-dimensional space through kernel techniques. The training process of the support vector regression model can be transformed into a convex optimization problem. The goal is to find the weight vector ω and the bias vector b, as well as an interval boundary to minimize the regularization loss of the model. The key parameters of the model include the regularization parameter C, which controls the degree of penalty ε for the error term of the model, defines the tolerance error between the predicted value and the actual value, and the kernel function parameters. The RBF kernel is selected as the kernel function, and the best parameter combination is found by grid search combined with cross validation. The mean square error is used to evaluate the model performance. After the support vector regression model is trained, the relative sound power level is calculated for the new ear canal reception data for input to obtain the numerical value indicating the degree of Eustachian tube opening.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention is described in detail with reference to the embodiments, it should be understood by those skilled in the art that any modification or equivalent replacement of the technical solutions of the present invention does not depart from the spirit and scope of the technical solutions of the present invention and should be included in the scope of the claims of the present invention.

Claims

1. A method for detecting the degree of eustachian tube opening based on a stimulation signal, comprising: A nasal adapter with a built-in earphone speaker is placed in the nostril of the subject, and an ear canal adapter with a built-in in-ear speaker is placed in the external auditory canal on the same side of the subject, and the earphone speaker plays a stimulation signal that can effectively distinguish between the open and closed states of the Eustachian tube; the stimulation signal is designed by measuring and calculating the transfer function of the Eustachian tube in a resting state and during swallowing action, respectively, to design a stimulation signal with specific frequency and characteristics; When the subject is at rest and swallowing, the acoustic signal received by the in-ear speaker is collected by the acoustic sensor terminal; The received sound signal is processed to obtain a relative sound power level, which is input into a pre-established and trained degree detection model to obtain a numerical value representing the degree of opening of the subject's Eustachian tube.

2. The method for detecting the degree of Eustachian tube opening based on stimulation signals according to claim 1, characterized in that: The design method of the stimulation signal comprises: The nasal adapter with built-in earphone speakers is placed in the nostril of the subject, and the ear canal adapter with built-in in-ear speakers is placed in the external auditory canal on the same side of the subject. The earphone speakers play a short-duration pulse signal x(t). The sound signal y(t) received by the external auditory canal is collected in the resting state and during swallowing, and the corresponding Eustachian tube transfer function h(t) from the nostril to the ear canal is calculated: Where, t represents the time domain, FFT() and IFF() represent Fourier transform and inverse Fourier transform respectively; The Eustachian tube transfer function h(t) at rest and during swallowing was analyzed to determine that a fixed-duration pure tone line spectrum signal of a specific frequency between 5.5kHz and 6.5kHz can effectively distinguish between the open and closed states of the Eustachian tube.

3. The method for detecting the patency of the Eustachian tube based on a stimulation signal according to claim 1, characterized in that: The step of processing the received sound signal to obtain a relative sound power level includes: Calculate the average sound power level in the 6-10kHz frequency band within each 128 / 3000 second interval to obtain the strongest moment of the impact sound signal accompanying each swallowing action of the subject; Obtain the maximum sound power level of the line spectrum signal corresponding to a window width of 128 / 3000 seconds measured within the range from 0.5 seconds before to 1.0 seconds after the strongest moment of the impact sound signal, as well as the average sound power level of the corresponding window except for the strongest moment of the impact sound signal; The relative sound power level is calculated by calculating the difference between the maximum sound power level and the average sound power level.

4. The method for detecting the degree of Eustachian tube opening based on stimulation signals according to claim 1, characterized in that: The degree detection model is a linear regression model or a support vector regression model.

5. The method for detecting the patency of the Eustachian tube based on a stimulation signal according to claim 4, characterized in that: The method also includes a step of training a degree detection model, including: Experiments were conducted on patients with secretory otitis media and healthy people, and multiple groups of received signal samples with different labels were collected. The labels were numerical values ​​indicating the degree of patency of the Eustachian tube. The received signal samples are divided into a training set and a test set. The normalized training set data are input into the linear regression model y=β0+β1x in sequence, where x is the difference in sound power levels and y is the degree of Eustachian tube opening. The model parameters β0 and β1 are estimated using the least squares method, and a regularization term is added to prevent overfitting. The cross-validation method is used for training and parameter adjustment. The test set data is used for verification to obtain the corresponding degree of Eustachian tube opening. The linear regression model is evaluated using the mean square error to obtain a trained degree detection model.

6. The method for detecting the patency of the Eustachian tube based on a stimulation signal according to claim 4, characterized in that: The method also includes a step of training a degree detection model, including: Experiments were conducted on patients with secretory otitis media and healthy people, and multiple groups of received signal samples with different labels were collected. The labels were numerical values ​​indicating the degree of patency of the Eustachian tube. The received signal samples are divided into training set and test set. The training set data are input into the support vector regression model in sequence. The training process is transformed into a convex optimization problem. The goal is to find the weight vector ω and the bias vector b, as well as an interval boundary to minimize the regularization loss of the model. The key parameters of the model include: regularization parameter C, control of the penalty degree ε of the model for the error term, definition of the tolerance error between the predicted value and the actual value, and kernel function parameters. The kernel function selects the RBF kernel. The best parameter combination is found by grid search combined with cross-validation, and the test set data is used for verification to obtain the corresponding degree of Eustachian tube opening. The support vector regression model is evaluated using the mean square error to obtain a trained degree detection model.

7. The method for detecting the patency of the Eustachian tube based on stimulation signals according to claim 5 or 6, characterized in that: The experiment was conducted on patients with secretory otitis media and healthy people, and multiple groups of received signal samples with different labels were collected, including: Step 1) inserting a nasal adapter with a built-in earphone speaker into the left nostril of the subject, inserting an ear canal adapter with a built-in in-ear speaker into the external auditory canal on the same side of the subject, and playing a sound source signal of a specific audio and a fixed duration through the earphone speaker; during the playing of each sound source signal, the subject cooperates to complete a swallowing action; Repeat the above steps 3 times to obtain 3 sets of data; Step 2) inserting a nasal adapter with a built-in earphone speaker into the right nostril of the subject, inserting an ear canal adapter with a built-in in-ear speaker into the external auditory canal on the same side of the subject, and playing a sound source signal of a specific audio and a fixed duration through the earphone speaker; during the playing of each sound source signal, the subject cooperates to complete a swallowing action; Repeat the above steps 3 times to obtain 3 sets of data.

Images