Asthma detection method and portable asthma detection equipment

Abstract

The invention discloses an asthma detection method and portable asthma detection equipment. The method comprises the following steps: acquiring breathing sound of a user; denoising, segmenting, and converting into a spectrogram through short-time Fourier transform and logarithm transform; and inputting the spectrogram into a neural network model to obtain the wheeze and the surge. The detection equipment comprises a shell, a main control unit is arranged in the shell, and a contact microphone connected with the main control unit is embedded in the bottom of the shell; wherein the main control unit is used for controlling the contact microphone to collect breathing sound and identifying wheeze and surge sound by using the asthma detection method. The detection method is high in operation speed and good in real-time performance, can detect the asthma condition of the patient in time, and has a good distinguishing effect; meanwhile, the detection equipment is light, small, simple to operate and convenient to carry.

Description

Technical Field

[0001] The present invention relates to the technical field of asthma detection, and more particularly to an asthma detection method and a portable asthma detection device. Background Art

[0002] Asthma is a high-pitched, musical breath sound caused by airway narrowing or obstruction, which is the sound produced by local vibration caused by the eddy current generated when the air flow passes through the narrow bronchus or bronchiole.

[0003] As a common chronic airway inflammatory disease in children, its incidence shows an increasing trend. However, due to children, especially young children, being unable to accurately describe their symptoms and inducing factors, and the caregivers having weak ability to distinguish the condition, it is impossible to apply an intervention plan to prevent the progression of acute exacerbation early and effectively. Therefore, it is necessary to construct an asthma monitoring method for children to achieve early identification and timely intervention.

[0004] However, the current asthma monitoring methods for children mainly collect the exhalation information of patients through implanted devices, or detect the thermal imaging information of the environment where the patients are located, or use wrist breathing to judge asthma, with poor accuracy and inconvenience in carrying. For example:

[0005] 1) Implantable detection means, such as fixing an implantable respiratory sensor in the airway through minimally invasive surgery, not only has a large device volume and does not meet the daily carrying and use requirements, but also, although it can obtain accurate data, there are risks of foreign body rejection and poor compliance;

[0006] 2) Non-contact monitoring based on environmental thermal imaging relies on a high-resolution infrared camera and is usually only applicable to medical institutions; at the same time, it is easily interfered by environmental temperature fluctuations and has insufficient signal-to-noise ratio of the respiratory signal;

[0007] 3) The core algorithm of the wrist-worn respiratory monitoring device relies on multi-modal sensor fusion (such as accelerometers, gyroscopes), with a relatively high hardware circuit complexity, and indirectly calculates respiratory parameters through chest and abdomen movements, making it difficult to capture the characteristics of high-frequency wheezing sounds.

[0008] In addition, existing algorithms generally adopt complex network structures (such as deep residual networks or 3D convolutional networks), with a large number of model parameters and difficulty in realizing real-time data processing.

[0009] Therefore, how to solve this problem is an urgent need for those skilled in the art. Summary of the Invention

[0010] To at least partially solve the above technical problems, the present invention provides an asthma detection method and a portable asthma detection device for detecting children's asthma sounds.

[0011] To achieve the above object, the present invention adopts the following technical solutions:

[0012] In a first aspect, the present application provides an asthma detection method, the steps of which include:

[0013] Obtain the user's breath sound;

[0014] After denoising, segment it and convert it into a spectrogram through short-time Fourier transform and logarithmic transform;

[0015] Input the spectrogram into a neural network model to obtain wheezing sounds and stridorous breath sounds.

[0016] In an optional embodiment, wavelet threshold is used to filter and denoise the user's breath sound.

[0017] In an optional embodiment, the short-time Fourier transform calculation formula is:

[0018]

[0019] In the formula, t represents the current window center time point, f represents the frequency, x(τ) represents the value of the denoised and segmented breath signal at τ, h(τ - t) represents the time-shifted window function of the intercepted signal segment, centered at t, τ represents the integration time variable, and e -j2πfτ represents the complex exponential basis function of the Fourier transform.

[0020] In an optional embodiment, the logarithmic transform calculation formula is:

[0021]

[0022] In the formula, imf(i, j) represents the spectrogram obtained through STFT, i represents the time frame index, and j represents the frequency index.

[0023] In an optional embodiment, the neural network model includes: a plurality of feature extraction modules and a fully connected layer; each feature extraction module includes:

[0024] Basic unit: one or more consecutive convolutional layers;

[0025] Normalization unit: a batch normalization layer and a ReLU activation layer immediately following the convolutional layer;

[0026] Downsampling unit: an optional max pooling layer for reducing the resolution of the feature map.

[0027] In a second aspect, the present application discloses a portable asthma detection device, including a housing, a main control unit is built in the housing, and a contact microphone connected to the main control unit is embedded at the bottom; wherein, the main control unit is used to control the contact microphone to collect breath sounds and execute the asthma detection method described in any one of the above to identify wheezing sounds and stridorous breath sounds.

[0028] In an alternative embodiment, buttons and a display panel are arranged on the top of the housing. The buttons are used to achieve manual control of the detection, and the display panel is used to display the recognition result.

[0029] In an alternative embodiment, the main control unit is integrated on a circuit board, and the circuit board is connected with a rechargeable lithium battery, a TYPE-C interface, an SD card interface and a speaker.

[0030] In an alternative embodiment, an analog data processing module is also integrated on the circuit board, which is used to amplify, filter and perform analog-to-digital conversion on the collected breath sounds.

[0031] In a third aspect, the present application discloses a portable asthma detection device, including a housing, a main control unit and a wireless module are arranged inside the housing, and a contact microphone connected to the unit is embedded at the bottom; wherein, the main control unit is used to control the contact microphone to collect breath sounds and transmit them to a terminal through the wireless module, and the terminal executes the asthma detection method described in any one of the above to identify wheezing sounds and stridulous breath sounds.

[0032] The asthma detection method and the portable asthma detection device provided by the present invention can solve problems such as cumbersome implantation of current detection devices and inability to perform detection in daily life. Compared with the prior art, the advantages of the present application include:

[0033] (1) By denoising, segmenting, performing short-time Fourier transform and logarithmic transform on the user's breath sounds, the time-domain signal is converted into a spectrogram, which can more clearly show the frequency characteristics of the breath sounds; further, the specially designed lightweight neural network model of the present application has fewer model parameters, reduces the computational complexity, realizes real-time processing of breath sound data, and at the same time ensures a high detection accuracy, meeting the requirements of real-time asthma monitoring.

[0034] (2) The detection device provided by the present application is designed to be lightweight, small, simple to operate, easy to use, and convenient for users to carry out asthma detection anytime and anywhere. It is especially suitable for children with asthma who need long-term monitoring, and can effectively assist caregivers in judging the condition. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0036] Figure 1 It is a flowchart of the asthma detection method provided by the present invention;

[0037] Figure 2Schematic diagram of the neural network model structure provided by the present invention;

[0038] Figure 3 Schematic diagram of the structure of the portable asthma detection device provided by the present invention;

[0039] Figure 4 Three-dimensional perspective view of the portable asthma detection device provided by the present invention;

[0040] Figure 5 Bottom view of the portable asthma detection device provided by the present invention;

[0041] Figure 6 Top view of the portable asthma detection device provided by the present invention. Detailed implementation manners

[0042] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0043] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0044] Embodiment 1:

[0045] The embodiment of the present invention first discloses an asthma detection method, the purpose of which is to quickly and accurately detect and judge the asthma condition of children. The steps include,

[0046] Obtain the breathing sound of the user;

[0047] After denoising, segment it and convert it into a spectrogram through short-time Fourier transform and logarithmic transform;

[0048] Input the spectrogram into the neural network model to obtain wheezing sound and stridor.

[0049] In this embodiment, the asthma detection method mainly includes three parts: signal processing (denoising), feature extraction, and asthma recognition; the specific process refers to Figure 1 ;

[0050] In an alternative embodiment, wavelet thresholding is used for denoising. By setting an appropriate threshold, the energy of the effective information and the noise is separated, achieving the removal of high-frequency noise in the collected acoustic signal. Preferably, sym8 wavelet is selected, the wavelet decomposition level is 3, and the decomposition coefficients of the signal for each layer are calculated. A threshold is set for each layer of coefficients. Signals higher than this threshold are retained, and wavelet coefficients lower than this threshold are set to 0. Finally, wavelet reconstruction is performed, and the signal is restored according to the low-frequency coefficients and the high-frequency coefficients of each layer to obtain the denoised audio.

[0051] In an alternative embodiment, feature extraction includes: dividing the filtered breath sound into segments of 5 s in length, windowing the signal into frames by short-time Fourier transform (STFT), and then obtaining its DFT to get a spectrogram; The formula is expressed as:

[0052]

[0053] In the formula, t represents the central time point of the current window, f represents the frequency, x(τ) represents the value of the denoised segmented breath signal at τ, h(τ - t) represents the time-shifted window function for intercepting the signal segment, centered at t, τ represents the integration time variable, and e -j2πfτ represents the complex exponential basis function of the Fourier transform.

[0054] Since most of the power values corresponding to the pixels in the power spectrum are very low and difficult to distinguish after directly performing short-time Fourier transform on the signal to generate the power spectrum. Therefore, this application performs additional processing on the power spectrum to enhance the features in the figure. Specifically, through logarithmic transformation, the transformed logarithmic power spectrum contains more information than the original power spectrum. The logarithmic transformation calculation formula is:

[0055]

[0056] In the formula, imf(i, j) represents the spectrogram obtained by STFT, i represents the time frame index, and j represents the frequency index.

[0057] In an alternative embodiment, a neural network model is constructed for asthma recognition; including:

[0058] First, a simplified convolutional neural network is constructed. All convolutional layers in the initial part are replaced or deleted with pooling layers (which will not have any impact on the model results); in the middle part, some convolutional layers are replaced with pooling layers, using continuous 2*1 convolution + 1*2 pooling; in the last part, the fully connected layer is retained, and the number of convolutional layers and convolutional filters is reduced to reduce the parameters, and at least one 3*3 convolutional layer is retained to ensure that the test accuracy of the model is not affected.

[0059] The finally simplified neural network model includes: multiple feature extraction modules and a fully connected layer; each feature extraction module is composed of the following hierarchical combinations:

[0060] a. Basic unit: one or more consecutive convolutional layers;

[0061] b. Normalization unit: batch normalization layer and ReLU activation layer immediately following the convolutional layer;

[0062] c. Downsampling unit: optional max pooling layer for reducing the resolution of the feature map;

[0063] Then, apply the Softmax function to the output of the fully connected layer to obtain the classification result.

[0064] In one embodiment, the neural network model structure is as Figure 2 , specifically including:

[0065] Input layer for receiving grayscale images with a size of 1024×256×1;

[0066] Normalization layer for normalizing the input image;

[0067] First convolutional layer, using a 2×1 convolutional kernel, with an output channel number of 3 and a stride of [2,1];

[0068] Second convolutional layer, using a 5×3 convolutional kernel, with an output channel number of 4;

[0069] First batch normalization layer and first ReLU activation layer;

[0070] Third convolutional layer, using a 2×1 convolutional kernel, with an output channel number of 8 and a stride of [2,1];

[0071] First max pooling layer, with a pooling window size of 2×2 and a stride of 2;

[0072] Fourth convolutional layer, using a 5×5 convolutional kernel, with an output channel number of 10 and a padding of 1;

[0073] Fifth convolutional layer, using a 5×5 convolutional kernel, with an output channel number of 12 and a padding of 1;

[0074] Sixth convolutional layer, using a 5×5 convolutional kernel, with an output channel number of 14 and a padding of 1;

[0075] Second batch normalization layer and second ReLU activation layer;

[0076] Second max pooling layer, with a pooling window size of 2×2 and a stride of 2;

[0077] Seventh convolutional layer, using a 3×3 convolutional kernel, with an output channel number of 18 and a padding of 1;

[0078] Third batch normalization layer and third ReLU activation layer;

[0079] The third max pooling layer, with a pooling window size of 2×2 and a stride of 2;

[0080] The eighth convolutional layer, using a 3×3 convolutional kernel and having 30 output channels;

[0081] The fourth batch normalization layer and the fourth ReLU activation layer;

[0082] The fourth max pooling layer, with a pooling window size of 2×2 and a stride of 2;

[0083] The ninth convolutional layer, using a 5×5 convolutional kernel, having 56 output channels, and a padding of 1;

[0084] The tenth convolutional layer, using a 5×5 convolutional kernel, having 72 output channels, and a padding of 1;

[0085] The fifth batch normalization layer and the fifth ReLU activation layer;

[0086] The fifth max pooling layer, with a pooling window size of 2×2 and a stride of 2;

[0087] The eleventh convolutional layer, using a 5×5 convolutional kernel, having 120 output channels, and a padding of 1;

[0088] The twelfth convolutional layer, using a 3×3 convolutional kernel and having 300 output channels;

[0089] The sixth batch normalization layer and the sixth ReLU activation layer;

[0090] The dropout layer, with a dropout probability of 0.5;

[0091] The fully connected layer, with an output dimension of 4;

[0092] The Softmax layer, used to generate classification results.

[0093] Preferably, all max pooling layers use global max pooling or local max pooling, with local max pooling being preferred.

[0094] The constructed respiratory sound dataset containing wheezing and stridor sounds is used to train this network. After obtaining a high accuracy rate, this model is used as the final recognition model to distinguish wheezing and stridor sounds in respiratory sounds.

[0095] In this embodiment, the number of parameters of the neural network model is 14.3K. It not only has a fast operation speed and good real-time performance, and can realize the timely detection of the patient's asthma condition; but also has a good discrimination effect and good robustness through the recognition of asthma sounds by a simplified neural network.

[0096] Example 2:

[0097] This embodiment discloses a portable asthma detection device, including a shell, a main control unit built in the shell, and a contact microphone connected to the main control unit embedded in the bottom; wherein the main control unit is used to control the contact microphone to collect breathing sounds and execute any of the asthma detection methods described above to identify wheezing sounds and stridor sounds.

[0098] This application uses a contact microphone to collect the user's breathing sounds, avoiding the risk of foreign body rejection and compliance issues brought by implanted sensors, improving the comfort and acceptance of detection, and is especially suitable for special groups such as children.

[0099] In an optional embodiment, a button and a display panel are arranged on the top of the housing, the button is used to realize manual control of the detection, and the display panel is used to display the recognition result. Figures 3 - 6 ;

[0100] Furthermore, in this embodiment, the main control unit is integrated into the circuit board, and the circuit board is connected with a rechargeable lithium battery, a TYPE-C interface, an SD card interface, and a speaker. Among them, the TYPE-C interface, the SD card interface, and the speaker are embedded on the shell, the TYPE-C interface is used to charge the rechargeable lithium battery, the SD card interface is used to store the breathing sound signal collected by the contact microphone, and the speaker is used to alarm when asthma is detected.

[0101] When performing asthma detection, the collected breathing sound signals are first stored in the tablet's built-in memory card, and then the built-in machine learning model is used to identify the asthma condition and generate the recognition result. Finally, the breathing signal waveform and recognition result are displayed on the display panel.

[0102] In a preferred embodiment, an analog data processing module is also integrated on the circuit board for amplifying, filtering and performing analog-to-digital conversion on the collected respiratory sounds.

[0103] Embodiment 3:

[0104] In this embodiment, a main control unit and a wireless module are simultaneously placed inside the housing of the portable asthma detection device, wherein the main control unit is used to control the contact microphone to collect breathing sounds and transmit them to the terminal through the wireless module, and the terminal executes any of the above-mentioned asthma detection methods to identify wheezing sounds and wheezing sounds.

[0105] In this embodiment, the terminal may be a tablet device or a mobile phone device, and the wireless module is a Bluetooth module;

[0106] Among them, the asthma detection method exists in the form of an APP and includes functions related to the operation of the acquisition equipment, such as controlling the start and end of the acquisition and sound playback through a wireless module.

[0107] In the above embodiment,

[0108] The main control unit uses STM32 series chips;

[0109] The contact microphone model is CM-01B. This microphone relies on the mechanical vibration on the human body surface to collect signals, can significantly suppress background noise, can still clearly collect the target body sound in a noisy environment, and has good response from low frequency to medium frequency band, and can accurately capture physiological related sounds such as breathing sounds and asthma sounds;

[0110] The Bluetooth module is an HC-04 Bluetooth serial communication module, the working frequency band is 2.4GHz ISM, and the GFSK modulation method is adopted, which can realize Bluetooth transmission within 10m;

[0111] The rechargeable lithium battery is a rechargeable lithium battery with 1200mAh, a rated power of 4.4Wh, and an output voltage of 3.7V;

[0112] The terminal uses a Huawei HONOR Pad X8 tablet computer, with an asthma detection APP built-in. It can realize data transmission with the collection device through Bluetooth, perform data processing and analysis, and display and play the collected data and recognition results. In this embodiment, this APP can remotely control the device, including controlling the start of collection and setting the collection mode (normal mode or calibration mode);

[0113] Preferably, a buzzer and an LED lamp are also installed on the shell to be used as prompts for power on / off, battery power, and Bluetooth connection status.

[0114] The detection device of this application collects breathing sounds, calculates the sound waveform and uses machine learning methods to complete the recognition and judgment of asthma within a period of time. Through the Bluetooth collection device and the receiving tablet device, the conversion of the incoming sound signal to an electrical signal is realized. The whole is powered by a battery and interacts through Bluetooth. The Bluetooth collection device realizes the collection, storage, upload and recognition of sounds through button operations. The receiving tablet device can also control various operations of the collection device, and can display the collected signals and display the recognition results and other information.

[0115] The asthma detection device of this application realizes real-time and accurate monitoring of asthma through non-invasive and high-precision breathing sound analysis, combined with a lightweight neural network model and portable device design, overcomes the defects of the prior art in terms of invasiveness, accuracy, real-time performance and portability, and can be applied to the auxiliary condition judgment of children with asthma, providing an effective technical means for the early recognition and timely intervention of childhood asthma. Specifically, by placing this device at a specific position on the chest for a period of time for collection, and controlling the start and end of the collection process in the upper computer software, the auxiliary judgment of the children's asthma situation is realized, and the caregiver is assisted to confirm the child's state in real time.

[0116] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the apparatuses disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, reference can be made to the description in the method section.

[0117] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting asthma, characterized in that: Obtain user breathing sounds; After denoising, the image is segmented and transformed into a spectrogram through short-time Fourier transform and logarithmic transform; The spectrogram was input into the neural network model to obtain wheezing and stridor.

2. The asthma detection method according to claim 1, characterized in that: Wavelet threshold is used to filter and denoise the user's breathing sound.

3. The asthma detection method according to claim 1, characterized in that: The short-time Fourier transform calculation formula is: Where t represents the center time point of the current window, f represents the frequency, x(τ) represents the value of the respiratory signal at τ after denoising and segmentation, h(τ-t) represents the time-shifted window function of the intercepted signal segment, with the center at t, τ represents the integral time variable, and e -j2πfτ Represents the complex exponential basis function of the Fourier transform.

4. The asthma detection method according to claim 1, characterized in that: The logarithmic transformation calculation formula is: Where imf(i,j) represents the spectrum obtained by STFT, i represents the time frame index, and j represents the frequency index.

5. The asthma detection method according to claim 1, characterized in that: The neural network model includes: multiple feature extraction modules and fully connected layers; each feature extraction module includes: Basic unit: 1 or more consecutive convolutional layers; Normalization unit: Batch normalization layer and ReLU activation layer immediately after the convolution layer; Downsampling unit: An optional max pooling layer that reduces the feature map resolution.

6. A portable asthma detection device, characterized in that: It comprises a shell, a main control unit is built in the shell, and a contact microphone connected to the main control unit is embedded in the bottom; wherein the main control unit is used to control the contact microphone to collect breathing sounds and execute the asthma detection method described in any one of claims 1-5 to identify wheezing sounds and stridor sounds.

7. The portable asthma detection device according to claim 6, characterized in that: Buttons and a display panel are arranged on the top of the shell. The buttons are used to realize manual control of the detection, and the display panel is used to display the recognition results.

8. The portable asthma detection device according to claim 6, characterized in that: The main control unit is integrated into a circuit board, and the circuit board is connected with a rechargeable lithium battery, a TYPE-C interface, an SD card interface and a speaker.

9. The portable asthma detection device according to claim 8, characterized in that: The circuit board also integrates an analog data processing module for amplifying, filtering and analog-to-digital conversion of the collected respiratory sounds.

10. A portable asthma detection device, characterized in that: It comprises a shell, a main control unit and a wireless module are built in the shell, and a contact microphone connected to the unit is embedded at the bottom; wherein the main control unit is used to control the contact microphone to collect respiratory sounds and transmit them to a terminal through the wireless module, and the terminal executes the asthma detection method described in any one of claims 1-5 to identify wheezing sounds and stridor sounds.

Images