CN112274174A - Intelligent electronic auscultation control system, method, storage medium and electronic stethoscope - Google Patents
Intelligent electronic auscultation control system, method, storage medium and electronic stethoscope Download PDFInfo
- Publication number
- CN112274174A CN112274174A CN202011151260.0A CN202011151260A CN112274174A CN 112274174 A CN112274174 A CN 112274174A CN 202011151260 A CN202011151260 A CN 202011151260A CN 112274174 A CN112274174 A CN 112274174A
- Authority
- CN
- China
- Prior art keywords
- data
- heart sound
- signal
- cardiac cycle
- heart
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000002555 auscultation Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000003860 storage Methods 0.000 title claims abstract description 8
- 230000000747 cardiac effect Effects 0.000 claims abstract description 64
- 230000011218 segmentation Effects 0.000 claims abstract description 50
- 238000013135 deep learning Methods 0.000 claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 32
- 238000000605 extraction Methods 0.000 claims abstract description 32
- 238000012545 processing Methods 0.000 claims abstract description 28
- 230000006870 function Effects 0.000 claims abstract description 26
- 238000013528 artificial neural network Methods 0.000 claims abstract description 18
- 238000013480 data collection Methods 0.000 claims abstract description 6
- 238000005070 sampling Methods 0.000 claims description 26
- 230000005236 sound signal Effects 0.000 claims description 26
- 238000001228 spectrum Methods 0.000 claims description 10
- 238000012549 training Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 230000003750 conditioning effect Effects 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 9
- 230000003595 spectral effect Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 230000002861 ventricular Effects 0.000 claims description 8
- 230000005284 excitation Effects 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 230000008602 contraction Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 5
- 239000008280 blood Substances 0.000 claims description 4
- 210000004369 blood Anatomy 0.000 claims description 4
- 101100127891 Caenorhabditis elegans let-4 gene Proteins 0.000 claims description 3
- 230000003321 amplification Effects 0.000 claims description 3
- 210000001765 aortic valve Anatomy 0.000 claims description 3
- 230000001746 atrial effect Effects 0.000 claims description 3
- 230000002146 bilateral effect Effects 0.000 claims description 3
- 230000017531 blood circulation Effects 0.000 claims description 3
- 210000000988 bone and bone Anatomy 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 210000003205 muscle Anatomy 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 3
- 230000000737 periodic effect Effects 0.000 claims description 3
- 210000003102 pulmonary valve Anatomy 0.000 claims description 3
- 238000005316 response function Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 2
- 239000000284 extract Substances 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 12
- 238000011160 research Methods 0.000 description 12
- 230000008901 benefit Effects 0.000 description 8
- 238000011161 development Methods 0.000 description 7
- 230000018109 developmental process Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 230000004044 response Effects 0.000 description 5
- 230000007547 defect Effects 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 208000024172 Cardiovascular disease Diseases 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 2
- 230000001575 pathological effect Effects 0.000 description 2
- 230000001737 promoting effect Effects 0.000 description 2
- 208000031229 Cardiomyopathies Diseases 0.000 description 1
- 208000009525 Myocarditis Diseases 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 208000029078 coronary artery disease Diseases 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 210000005069 ears Anatomy 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 208000019622 heart disease Diseases 0.000 description 1
- 208000018578 heart valve disease Diseases 0.000 description 1
- 238000013095 identification testing Methods 0.000 description 1
- 238000011534 incubation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B7/00—Instruments for auscultation
- A61B7/02—Stethoscopes
- A61B7/04—Electric stethoscopes
Abstract
The invention belongs to the technical field of stethoscopes, and discloses an intelligent electronic auscultation control system, a method, a storage medium and an electronic stethoscope.A hardware circuit is started, a corresponding serial port is selected and opened after the operation is stable, and a real-time display function of heart sound waveforms is started; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface; after the collection is finished, stopping data collection, calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the collected data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished; after extraction is finished, the MFCC features are sent to a DBN deep learning neural network for intelligent classification and identification of heart sounds, after identification is finished, the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound identification and classification result are respectively displayed in a human-computer interface, and data are stored in TXT and WAV formats. The invention can provide the functions of heart sound acquisition and intelligent identification and provide data support for intelligent medical treatment.
Description
Technical Field
The invention belongs to the technical field of stethoscopes, and particularly relates to an intelligent electronic auscultation control system, an intelligent electronic auscultation control method, a storage medium and an electronic stethoscope.
Background
The cardiovascular diseases have long incubation period and sudden onset, and seriously harm the national health. The traditional cardiovascular disease screening mostly depends on conventional medical treatment modes such as electrocardiogram, blood routine and CT, and patients in less developed areas cannot obtain accurate diagnosis at the first time due to unbalanced medical development and uneven medical conditions in China areas.
In order to solve the problem of regional medical level difference, the best method is to introduce intelligent medical equipment, and the medical staff can make more accurate judgment and treatment on the state of illness of the patient with the assistance of the intelligent medical equipment only by having basic professional ability. With the progress of digital technology, especially the rapid development of deep learning and big data technology in recent years, it is possible to use digital means to process and identify physiological signals of human body, so as to assist the traditional medical treatment to diagnose cardiovascular diseases of patients. Deep learning represents a great performance advantage in image recognition, voice recognition, and the like, but requires an extremely large amount of data as a support. Compared with the traditional images and natural languages, the heart sound data resources are extremely deficient, the heart sound data which can be used on the internet is few, most of the electronic stethoscopes on the market at present are only used as digital improved versions of the traditional stethoscopes, and sufficient data support cannot be provided for heart sound identification research based on deep learning, so that an intelligent processing electronic stethoscope based on deep learning is urgently needed to be provided, not only can complete the work of heart sound collection, processing, feature extraction and the like, but also can be used for realizing intelligent classification and identification of pathological heart sounds by further combining the deep learning while completing basic functions, accumulating data for the large data research of the heart sound identification, promoting intelligent medical related research, finally solving the problems of unbalanced regional medical level, dependence on artificial diagnosis and treatment and the like, and providing intelligent and high-quality medical services for patients.
Through the above analysis, the problems and defects of the prior art are as follows:
(1) the existing electronic stethoscope can not provide enough data support for deep learning-based heart sound recognition research, can not finish the work of heart sound acquisition, processing, feature extraction and the like, and can not solve the problems of unbalanced regional medical level, dependence on artificial diagnosis and treatment and the like
The difficulty in solving the above problems and defects is: most of electronic stethoscopes on the market are only electronic upgrade versions of traditional physical stethoscopes, can only solve the problems of poor low-frequency response and the like of the traditional physical stethoscopes, cannot combine the heart sound signals of patients with big data technology and machine learning algorithm, and cannot form a set of complete heart sound processing system for collecting front-end data and storing and analyzing back-end data. Due to the lack of the heart sound processing system, the heart sound data cannot be effectively shared on the internet in the form of big data, and related research based on heart sound identification, particularly the heart sound identification research based on deep learning, is hindered.
The significance of solving the problems and the defects is as follows: the intelligent processing electronic stethoscope provided by the invention can effectively solve the defects and further promote the development of an intelligent medical technology taking a big data technology and deep learning as core technologies, thereby effectively reducing the problem of uneven medical resource distribution caused by unbalanced economic development and asynchronous regional development in all parts of the country at present, including uneven medical equipment and uneven doctor level. By using and popularizing the invention, the most advanced medical equipment is not needed to be purchased in regions with laggard medical development, the most elegant doctors are recruited, local patients can enjoy the best medical service, the medical service experience of the local patients is improved, and the dividend of sharing economic and technical development of people in China is promoted.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides an intelligent electronic auscultation control system, an intelligent electronic auscultation control method, a storage medium and an electronic stethoscope.
The invention is realized in such a way that an intelligent electronic auscultation control method comprises the following steps:
starting a hardware circuit, selecting and opening a corresponding serial port after the operation is stable, and simultaneously starting a real-time display function of the heart sound waveform; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface;
after the collection is finished, stopping data collection, calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the collected data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished;
after extraction is finished, the MFCC features are sent to a DBN (digital broadcast network) for classification and identification, after identification is finished, the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound identification and classification result are respectively displayed in a human-computer interface, and data are stored in TXT and WAV formats.
Furthermore, the intelligent electronic auscultation control method adopts a piezoelectric sensor and a high-precision ADC to form a heart sound data acquisition module so as to ensure the high-efficiency control of a hardware circuit and the high-precision acquisition of heart sounds; after data acquisition is finished, sending the data to an upper computer through a serial port, and processing the data in the upper computer; the data processing part adopts MATLAB to program and uses MATLAB GUI to compile a test interface; and displaying the real-time heart sound waveform, the single heart cycle segmentation result, the MFCC feature extraction result and the heart sound identification result in a test interface.
Furthermore, the intelligent electronic auscultation control method uses a high-precision piezoelectric sensor to collect original heart sound signals, and the original signals are weak electric signals due to small amplitude of the original signals output by the piezoelectric sensor, and are amplified by an amplifying circuit; a voltage follower, an anti-aliasing filter and a forward amplifying circuit are adopted to form a signal conditioning circuit, the original signal is processed, the voltage follower is used for increasing the input resistance and reducing the output resistance, and then the signal is sent to the anti-aliasing filter; finally, the signal is accessed into a forward amplifying circuit, and the amplification factor and the direct current bias are adjusted to enable the amplitude value to meet the requirement of an analog-to-digital conversion chip; after the signal conditioning is finished, discrete acquisition is carried out by using an analog-to-digital conversion chip, so that an analog signal is converted into a digital signal which can be identified by the MCU, and finally, data is packed and is sent to an upper computer through a serial port of the MCU.
Further, after the upper computer receives the data, effective data needs to be extracted from the data packet, whether the received data is a frame header with 2 bytes is detected, if yes, the subsequently received data is stored in a corresponding data buffer area until a frame tail is detected; if the received data is not the frame header, continuing to detect and waiting for frame header data; after receiving a data packet, sequentially integrating the data in the data buffer into 18-bit original data; intercepting the middle stable part of the original signal as data of a single-heart-cycle segmentation algorithm, selecting the middle point of the original signal, and selecting 35000 sampling points from the middle point to two sides respectively as sample data; optionally, 5-layer decomposition filtering is performed using the sym5 wavelet basis to filter out high frequency noise in the signal.
Further, the intelligent electronic auscultation control method calls an algorithm to process the data after signal preprocessing is completed, firstly, frequency information in sample signals needs to be calculated, then, the number of sampling points included in a single cardiac cycle is calculated according to the sampling rate and the signal frequency, finally, the maximum value in the sample, namely the peak point of a certain cardiac cycle is selected, the number of sample points in the single cycle is taken, the single cardiac cycle of the sample signals is extracted, and a heart sound function model H (t):
h (t) in the model is the response function of the heart sound generation system, h1S1, the ventricular wall vibration caused by ventricular contraction, reentry of blood by impact with atrioventricular valves, and the vibration caused by closing of atrioventricular valves, h2(x-n1) S2, vibration signal caused by sudden closure of aortic and pulmonary valves at the beginning of ventricular diastole, h3(x-n2) Is S3, due to blood flow impinging on the ventricles, h4(x-n3) For S4, S4 was generated with a lower frequency associated with atrial contractions, where ni(i-1, 2, 3) represents the delay time of the corresponding heart sound in the single cardiac cycle, and ω (t) is the echo generated by muscle, bone, fat and other noises; a (x + T) is an excitation signal of the heart sound generating system, namely when A (x) is effective, the heart beat generates a heart sound H (T), T is a period, and the period T or the frequency F of A (x) can be calculated according to the sampling frequency Fs and the period TCounting, completing single heart cycle segmentation, performing MFCC feature extraction on single heart cycle heart sound signals after the single heart cycle heart sound signals are completed, normalizing the extracted feature values, sending the normalized feature values into a pre-trained DBN neural network for recognition, and finally completing the functions of recognizing and classifying the heart sounds.
Further comprising:
1) firstly, removing a direct current component in a sample signal to obtain a non-direct current bias heart sound signal, and performing Hilbert transform on the non-direct current bias heart sound signal to obtain an upper envelope of the signal;
2) FFT transformation is respectively carried out on the original signal and the envelope signal, the frequency spectrum of the original signal and the envelope signal is observed, most high-frequency heart sound information in the original signal can be filtered out by processing that the Hilbert transform is used for extracting the envelope on the heart sound signal, and the periodic signal A (x) and a part of low-frequency heart sound information of the lower heart beat are reserved;
3) after the original heart sound signal is subjected to Hilbert transform, low-frequency components with higher modulus appear in frequency components of an envelope signal, the spectral line is a frequency domain representation of an excitation signal A (x), and the spectral line is extracted by adopting a low-frequency filtering mode to obtain the frequency and the heart rate F of A (x);
4) after obtaining the signal A (x), the frequency of the signal A (x) can be calculated by the following formula, wherein F is the frequency of A (x), Fs is the sampling frequency, N is the number of FFT points, and P is the number of points corresponding to the maximum value in the negative half axis of the envelope bilateral spectrum, namely the FFT point corresponding to the maximum amplitude spectral line in the negative half axis of the envelope signal spectrum;
5) in the system, Fs is 5KHz, N is 70K, F is 1.4286Hz, and the number of sampling points contained in a single period is calculated through the following formula after F is obtained;
6) in the sample, Point is 3500, after the number of sample points in a single cycle is calculated, the time domain maximum Point in the sample, namely the S1 peak Point of a certain cardiac cycle, is selected, and the front and rear thresholds of the single cycle are set according to the following formula, namely the single cardiac cycle can be intercepted, wherein H is a single heart sound period function, L is the position of the maximum Point in the sample, Point is the number of sampling points, and α and β are extraction thresholds;
H＝(L-α(Point)；L+β(Point)；
7) after the single cardiac cycle is divided, considering the sample signal as a stable signal, and taking the MFCC characteristic parameter as a default extracted characteristic parameter;
8) after the MFCC features are extracted, performing normalization processing on 12 MFCC parameters, projecting the MFCC parameters into a [0, 1] interval, inputting the MFCC parameters into a pre-trained deep learning network as input to perform recognition and classification, outputting a classification result through a softmax classifier, and finally completing the functions of recognizing and classifying the heart sounds, wherein the recognition result is displayed in a heart sound recognition result text box on the left side of a human-computer interface;
9) in the upper computer software, after all algorithms are operated and result data are normal, original heart sound signals, single heart cycle signals, MFCC characteristic parameters and heart sound recognition and classification results are stored in the upper computer in TXT and WAV formats.
Furthermore, in the DBN network of the intelligent electronic auscultation control method, connection weights of each layer are trained by using a CD algorithm, and after network training is completed, the connection weights can be used as a hidden layer of a BP neural network to realize construction of a deep belief network, and the specific operation steps are as follows:
1) initializing a multi-layer deep learning network, and training the network by using a CD algorithm to obtain a weight of the deep learning network;
2) constructing a neural network with the number of hidden layers and the number of nodes consistent with that of the deep learning network, and taking the weight of the deep learning network obtained by training as the weight of the neural network;
3) and (4) finely adjusting the neural network by using a BP algorithm to finally obtain an available deep trust network.
It is another object of the present invention to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
starting a hardware circuit, selecting and opening a corresponding serial port after the operation is stable, and simultaneously starting a real-time display function of the heart sound waveform; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface;
after the collection is finished, stopping data collection, calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the collected data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished;
after extraction is finished, the MFCC features are sent to a DBN (digital broadcast network) for recognition, after recognition is finished, the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound recognition classification result are respectively displayed in a human-computer interface, and data are stored in TXT and WAV formats.
Another object of the present invention is to provide an intelligent electronic auscultation control system implementing the intelligent electronic auscultation control method, the intelligent electronic auscultation control system comprising:
the data acquisition module is used for starting a hardware circuit, selecting and opening a corresponding serial port after the hardware circuit is stable in operation, and simultaneously starting a real-time display function of the heart sound waveform; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface;
the data preprocessing module is used for calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the acquired data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished;
and the result output module is used for sending the MFCC features into the DBN for recognition after extraction is finished, respectively displaying the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound recognition classification result in a human-computer interface after recognition is finished, and storing the data in TXT and WAV formats.
Another object of the present invention is to provide an intelligent electronic stethoscope equipped with the intelligent electronic auscultation control system, the intelligent electronic stethoscope comprising: the device comprises a piezoelectric sensor, an amplifying circuit, a voltage follower, an anti-aliasing filter, a forward amplifying circuit, an analog-to-digital conversion chip and an upper computer.
By combining all the technical schemes, the invention has the advantages and positive effects that: in the invention, on the hardware level, the sound pickup effect of an air conduction type sound pickup device such as a condenser microphone and a contact conduction type sound pickup device such as a piezoelectric sensor is tested, the advantages and disadvantages of the air conduction type sound pickup device and the contact conduction type sound pickup device are comprehensively compared, the piezoelectric sensor with higher signal-to-noise ratio is selected as the sound pickup device of the invention, and a signal conditioning circuit is used for assisting, so that the external environment noise is well isolated, and the excellent sound pickup effect is obtained. Meanwhile, a plurality of AD sampling chips are investigated, parameters such as INL, DNL, SNR, price and the like are comprehensively compared, and AD7606 is selected as a digital-to-analog conversion chip.
In the aspect of an algorithm, because a piezoelectric sensor is used as a pickup device, the acquired heart sound signal waveform has obvious difference with the signal waveform acquired by a capacitance microphone, and the specific expression is that the signal is more continuous and the segmentation is not obvious. After the characteristic extraction is finished, the characteristic value is used as input and is sent to a DBN deep trust network for identification, and the obtained classification result can be used for assisting a doctor to diagnose and classify and identify common heart diseases such as coronary heart disease, valvular heart disease, cardiomyopathy, myocarditis and the like.
In the software level, in consideration of the test and use requirements, the invention adopts MATLAB GUI to compile a human-computer interface, and has the advantages of simple operation, concise interface and strong usability.
Compared with the existing electronic stethoscope, the invention not only can provide the functions of heart sound collection and identification, solves the problem of poor low-frequency response of the traditional stethoscope, but also can provide data support for heart sound identification research, especially for intelligent medical treatment of big data by adopting the heart sound identification research of deep learning, can better solve the problem of lack of the existing heart sound database, and has positive significance for the heart sound identification research by using the deep learning.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments of the present application will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained from the drawings without creative efforts.
Fig. 1 is a flowchart of an intelligent electronic auscultation control method provided by an embodiment of the invention.
Fig. 2 is a schematic structural diagram of an intelligent electronic auscultation control system provided by an embodiment of the invention;
in fig. 2: 1. a data acquisition module; 2. a data preprocessing module; 3. and a result output module.
Fig. 3 is a schematic structural diagram of an intelligent electronic auscultation control system provided by an embodiment of the present invention;
FIG. 4 is a schematic diagram of the core portion of STM32 provided by an embodiment of the present invention.
Fig. 5 is a schematic diagram of a power circuit provided by an embodiment of the invention.
Fig. 6 is a schematic diagram of a signal preprocessing and acquisition part provided by an embodiment of the invention.
Fig. 7 is a circuit for conditioning signals according to an embodiment of the present invention.
Fig. 8 is a flowchart of an implementation of the intelligent electronic auscultation control method according to the embodiment of the present invention.
Fig. 9 is a screenshot of the operation of the human-computer interface provided by the embodiment of the present invention.
FIG. 10 is a diagram illustrating the single cardiac cycle segmentation result provided by the embodiment of the present invention.
Fig. 11 is a diagram illustrating a result of the operation of the human-machine interface according to the embodiment of the present invention.
Fig. 12 is a diagram of a deep belief network architecture provided by an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In view of the problems in the prior art, the present invention provides an intelligent electronic auscultation control system, method, storage medium and electronic stethoscope, and the present invention is described in detail with reference to the accompanying drawings.
As shown in fig. 1, the intelligent electronic auscultation control method provided by the invention comprises the following steps:
s101: starting a hardware circuit, selecting and opening a corresponding serial port after the operation is stable, and simultaneously starting a real-time display function of the heart sound waveform; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface, and the data acquisition is recommended for more than 30 seconds;
s102: after the collection is finished, stopping data collection, calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the collected data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished;
s103: after extraction is finished, the MFCC features are sent to a DBN (digital broadcast network) for recognition, after recognition is finished, the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound recognition classification result are respectively displayed in a human-computer interface, and data are stored in TXT and WAV formats.
The intelligent electronic auscultation control method provided by the invention can be implemented by adopting other steps by persons skilled in the art, and the intelligent electronic auscultation control method provided by the invention in fig. 1 is only one specific embodiment.
As shown in fig. 2, the intelligent electronic auscultation control system provided by the invention comprises:
the data acquisition module 1 is used for starting a hardware circuit, selecting and opening a corresponding serial port after the operation is stable, and simultaneously starting a real-time display function of heart sound waveforms; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface;
the data preprocessing module 2 is used for calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the acquired data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished;
and the result output module 3 is used for sending the MFCC features into the DBN for recognition after extraction is finished, respectively displaying the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound recognition classification result in a human-computer interface after recognition is finished, and storing the data in TXT and WAV formats.
The technical solution of the present invention is further described below with reference to the accompanying drawings.
As shown in fig. 3, in the structural block diagram of the intelligent electronic auscultation control system provided by the present invention, a hardware circuit part uses STM32F103ZET6 as a core, and a piezoelectric sensor and a high-precision ADC are used to form a heart sound data acquisition module, so as to ensure high-efficiency control and high-precision acquisition of heart sounds of the hardware circuit, fig. 4 is a schematic diagram of the core part of STM32 provided by the embodiment of the present invention, fig. 5 is a schematic diagram of a power supply circuit provided by the embodiment of the present invention, and fig. 6 is a schematic diagram of a signal preprocessing and acquisition part provided by the embodiment of the present invention. After data acquisition is finished, the data are sent to an upper computer through a serial port and processed in the upper computer. The data processing part was programmed with MATLAB and a test interface was written using MATLAB GUI. And displaying the real-time heart sound waveform, the single heart cycle segmentation result, the MFCC feature extraction result and the heart sound identification result in a test interface.
The system comprises a signal conditioning circuit (see figure 7) which is composed of a voltage follower, an anti-aliasing filter and a forward amplifying circuit and is used for processing the original signal, wherein the voltage follower is used for increasing an input resistor and reducing an output resistor, then the signal is sent into the anti-aliasing filter to filter high-frequency noise of the signal so as to ensure the purity of the heart sound signal and avoid the occurrence of frequency spectrum aliasing phenomenon during discrete acquisition, and finally the signal is accessed into the forward amplifying circuit to adjust the amplification factor and direct current offset so that the amplitude of the signal meets the requirements of an analog-to-digital conversion chip. After the signal conditioning is finished, discrete acquisition is carried out by using an analog-to-digital conversion chip, so that an analog signal is converted into a digital signal which can be identified by the MCU, and finally, data is packed and is sent to an upper computer through a serial port of the MCU.
The piezoelectric sensor is a CM-01B piezoelectric sensor, has the advantages of high accuracy, high sensitivity, high signal-to-noise ratio, excellent frequency response characteristic and the like, is internally integrated with a signal conditioning circuit, and effectively inhibits environmental noise because an output signal is only changed along with relative displacement after being placed stably.
The analog-to-digital converter is AD7606, the sampling resolution of the AD7606 is 16bit, 95.5dB SNR, -107dB THD, ± 0.5LSB INL, ± 0.5LSB DNL quantization error, the input buffer of 1M Ω analog input impedance and the sampling rate as high as 200KHz all make it very suitable to be the digital-to-analog converter of the invention, and can adopt the parallel communication, can provide the extremely accurate heart sound data fast.
FIG. 8 is a software flow chart of the intelligent electronic auscultation control method of the present invention, after the system is running, a hardware circuit is started first, after the operation is stabilized, a corresponding serial port is selected and opened, and simultaneously, a real-time display function of heart sound waveforms is started, at this time, received heart sound data is displayed in real time in a graphic frame at the upper end of a human-computer interface (see FIG. 9), data acquisition is recommended for more than 30 seconds, after the acquisition is finished, the data acquisition is stopped by clicking a button for closing the serial port, a cardiac cycle segmentation algorithm is called, the acquired data is subjected to single cardiac cycle segmentation processing, MFCC characteristic parameters of the cardiac cycle are extracted after the segmentation is finished, MFCC characteristics are sent to a DBN network for identification, after the identification is finished, the single cardiac cycle segmentation result, the MFCC characteristic parameter extraction result and the heart sound identification classification result are respectively displayed in the human-computer interface, and the data is stored in TXT and WAV formats.
Specifically, after receiving data, an upper computer firstly needs to extract effective data from a data packet, the data packet related in the invention is 204 bytes long, wherein the data packet comprises a frame header of 2 bytes, a frame tail of 2 bytes, and 200 bytes of effective data, and the effective data are arranged in a form of high 8 bits, low 8 bits, high 8 bits and low 8 bits, firstly, whether the received data is the frame header of 2 bytes is detected, and if so, subsequently received data are stored in a corresponding data buffer area until the frame tail is detected; if the received data is not the frame header, continuing to detect and waiting for the frame header data. After receiving a data packet, the data in the data buffer is sequentially integrated into 18 bits of original data. In the actual acquisition process, the process of contact and separation of the acquisition terminal and the skin exists, and the signal interference is large in the process, so that the middle stable part of the original signal needs to be intercepted and used as data of a single heart cycle segmentation algorithm. Firstly, the midpoint of an original signal is selected, and 35000 sampling points are respectively selected from the midpoint to the two sides as sample data. In the invention, a sym5 wavelet base is selected to carry out 5-layer decomposition filtering, compared with an FIR low-pass filter designed by a Kaiser window, the filter has better retention effect of a high-frequency-band non-noise part signal, can effectively filter the high-frequency noise in the signal, and furthest ensures the integrity of the signal.
And calling an algorithm to process the data after the signal preprocessing is finished. Firstly, calculating frequency information in a sample signal, then calculating the number of sampling points contained in a single cardiac cycle according to the sampling rate and the signal frequency, finally selecting the maximum value in the sample, namely the peak value point of a certain cardiac cycle, and taking the number of single-cycle sample points to extract the single cardiac cycle of the sample signal, in order to solve the problem, the invention provides a heart sound function model H (t) based on a heart sound generation mechanism:
h (t) in the model is the response function of the heart sound generation system, h1S1, the ventricular wall vibration caused by ventricular contraction, reentry of blood by impact with atrioventricular valves, and the vibration caused by closing of atrioventricular valves, h2(x-n1) Is S2, is a ventricleVibration signal caused by sudden closure of aortic and pulmonary valves at the beginning of diastole, h3(x-n2) Is S3, due to blood flow impinging on the ventricles, h4(x-n3) For S4, S4 was generated with a lower frequency associated with atrial contractions, where niAnd (i-1, 2 and 3) represents the delay time of the corresponding heart sound in a single cardiac cycle, and ω (t) represents the echo generated by muscle, bone, fat and the like and other noises. A (x + T) is an excitation signal of a heart sound generation system, namely when A (x) is effective, the heart beat generates a heart sound H (T) once, and T is a period, so that the number of sampling points in a single period can be calculated according to sampling frequency Fs and the period T (frequency F) by only calculating the period T or the frequency F of A (x), so that the single heart cycle division is completed, MFCC feature extraction is carried out on the single heart cycle heart sound signal after the single heart cycle heart sound signal division is completed, the extracted feature value is normalized and then is sent to a pre-trained DBN neural network for recognition, and finally the functions of recognizing and classifying the heart sound are completed.
Specifically, the method comprises the following steps:
1) firstly, removing a direct current component in a sample signal to obtain a heart sound signal without direct current bias, and performing Hilbert transform on the heart sound signal without direct current bias to obtain an upper envelope of the signal.
2) FFT transformation is respectively carried out on the original signal and the envelope signal, the frequency spectrum is observed, most high-frequency heart sound information in the original signal can be filtered out by processing that the upper envelope of the heart sound signal is extracted through Hilbert transformation, and the periodic signal A (x) and a part of low-frequency heart sound information of the lower heart beat are reserved.
3) After the original heart sound signal is subjected to Hilbert transform, low-frequency components with higher modulus values appear in frequency components of the envelope signal, and theoretical analysis shows that the spectral line is the frequency domain representation of the excitation signal A (x), so that the frequency and the heart rate F of the excitation signal A (x) can be obtained only by extracting the spectral line in a low-pass filtering mode.
4) After the signal A (x) is obtained, the frequency of the signal A (x) can be calculated through the following formula, wherein F is the frequency of the signal A (x), Fs is the sampling frequency, N is the number of FFT points, and P is the number of points corresponding to the maximum value in the negative half axis of the envelope bilateral spectrum, namely the FFT point corresponding to the maximum amplitude spectral line in the negative half axis of the envelope signal spectrum.
5) In the system, Fs is 5KHz, N is 70K, and F is 1.4286Hz in the sample, and after F is obtained, the number of sampling points Point included in a single cycle can be calculated by the following formula.
6) In this sample, Point is 3500. After the number of sample points in a single cycle is calculated, a time domain maximum value Point in the sample, namely an S1 peak Point of a certain cardiac cycle, is selected, and a threshold before and after the single cycle is set according to the following formula, so that the single cardiac cycle can be intercepted, wherein H is a single heart sound period function, L is the position of the maximum value Point in the sample, Point is the number of sampling points, and alpha and beta are extraction thresholds. In the present system
A better result can be achieved and the segmentation result is shown in figure 8.
H＝(L-α(Point)；L+β(Point)；
7) After the single cardiac cycle is divided, the sample signal can be considered as a stable signal. In the mainstream heart sound identification research at present, the MFCC characteristics of heart sound signals are mostly used as parameters to perform deep learning training, so that the MFCC characteristic parameters are used as default extracted characteristic parameters by the system.
8) After the MFCC features are extracted, the 12 MFCC parameters are subjected to normalization processing, projected into a [0, 1] interval, then used as input and sent into a pre-trained deep learning network for recognition and classification, the classification result is output through a softmax classifier, finally, the heart sound recognition and classification functions are completed, and the recognition result is displayed in a heart sound recognition result text box on the left side of a human-computer interface (see FIG. 11).
9) In the upper computer software, after all algorithms are operated and result data are normal, original heart sound signals, single heart cycle signals, MFCC characteristic parameters and heart sound recognition and classification results are stored in the upper computer in TXT and WAV formats.
The deep learning network is a DBN (distributed binary network) deep trust network, the deep trust network is the combination of a deep learning algorithm and a BP (Back propagation) algorithm and is also the combination of supervised learning and unsupervised learning, the DBN network has the self-organization of the BP neural network, no requirement is required for input data, and the characteristics of deeper data can be theoretically learned by increasing the number of hidden layers, so that the performance of the system is improved. The deep trust network solves the problem that the convergence speed of the BP neural network is too low, and the network learning efficiency is greatly improved.
The deep trust network of the invention takes a limit Boltzmann machine RBM as a basic unit (refer to fig. 12), realizes a basic framework of deep learning by RBM units stacked layer by layer, in the network, each layer of connection weight is trained by using a CD algorithm, and after the network training is finished, the connection weight can be used as a hidden layer of a BP neural network, thereby realizing the construction of the deep trust network, and the specific operation steps are as follows:
1) initializing a multi-layer deep learning network, and training the network by using a CD algorithm to obtain a weight of the deep learning network;
2) constructing a neural network with the number of hidden layers and the number of nodes consistent with that of the deep learning network, and taking the weight of the deep learning network obtained in the first training step as the weight of the neural network;
3) and (4) finely adjusting the neural network by using a BP algorithm to finally obtain an available deep trust network.
In the deep learning heart sound identification module, the MFCC is adopted as a characteristic parameter, the MFCC has obvious advantages in the process of heart sound identification, and the MFCC enables audio to better accord with the auditory characteristic of human ears and can better represent the frequency response characteristic of a sound production organ by mapping the audio from a standard frequency domain to an MEL scale frequency domain. The frequency response characteristic of the heart sound generation model can be well reflected in the heart sound recognition of deep learning, and the recognition accuracy is improved. In the conventional mainstream heart sound identification research based on deep learning, MFCC is mostly adopted as a characteristic parameter, and in the pathological heart sound identification test, MFCC is used as the characteristic parameter and is sent to a DBN deep learning network for heart sound identification, so that a good effect is achieved, and the comprehensive identification rate can reach 80.6%.
Through the cooperative work of the software and the hardware, the method can better convert the original heart sound data into the single cardiac cycle heart sound data and extract the MFCC characteristic parameters in the single cardiac cycle heart sound data to carry out heart sound identification based on deep learning, has the advantages of high precision, high accuracy, independence on synchronous electrocardiosignals and the like, realizes intelligent diagnosis aiming at the heart sound by combining the deep learning while completing the basic function of the electronic stethoscope, increases the practicability of a heart sound identification system, and has positive significance for promoting intelligent medical research.
In the description of the present invention, "a plurality" means two or more unless otherwise specified; the terms "upper", "lower", "left", "right", "inner", "outer", "front", "rear", "head", "tail", and the like, indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, are only for convenience in describing and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed in a particular orientation, and be operated, and thus, should not be construed as limiting the invention. Furthermore, the terms "first," "second," "third," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
It should be noted that the embodiments of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or specially designed hardware. Those skilled in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such code being provided on a carrier medium such as a disk, CD-or DVD-ROM, programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier, for example. The apparatus and its modules of the present invention may be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of hardware circuits and software, e.g., firmware.
The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. An intelligent electronic auscultation control method is characterized by comprising the following steps:
starting a hardware circuit, selecting and opening a corresponding serial port after the operation is stable, and simultaneously starting a real-time display function of the heart sound waveform; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface;
after the collection is finished, stopping data collection, calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the collected data, and extracting the feature parameters of the Mel frequency cepstrum coefficient MFCC of the cardiac cycle after the segmentation is finished;
after extraction is finished, the MFCC features are sent into a deep trust network DBN for recognition, after recognition is finished, the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound recognition classification result are respectively displayed in a human-computer interface, and data are stored in TXT and WAV formats.
2. The intelligent electronic auscultation control method according to claim 1, wherein the intelligent electronic auscultation control method adopts a piezoelectric sensor and a high-precision ADC to form a heart sound data acquisition module so as to ensure efficient control of hardware circuits and high-precision acquisition of heart sounds; after data acquisition is finished, sending the data to an upper computer through a serial port, and processing the data in the upper computer; the data processing part adopts MATLAB to program and uses MATLAB GUI to compile a test interface; and displaying the real-time heart sound waveform, the single heart cycle segmentation result, the MFCC feature extraction result and the heart sound identification result in a test interface.
3. The intelligent electronic auscultation control method according to claim 1, wherein the intelligent electronic auscultation control method uses a high-precision piezoelectric sensor to collect original heart sound signals, and the original signals are amplified by an amplifying circuit because the original signals output by the piezoelectric sensor are weak electrical signals with small amplitude; a voltage follower, an anti-aliasing filter and a forward amplifying circuit are adopted to form a signal conditioning circuit, the original signal is processed, the voltage follower is used for increasing the input resistance and reducing the output resistance, and then the signal is sent to the anti-aliasing filter; finally, the signal is accessed into a forward amplifying circuit, and the amplification factor and the direct current bias are adjusted to enable the amplitude value to meet the requirement of an analog-to-digital conversion chip; after the signal conditioning is finished, discrete acquisition is carried out by using an analog-to-digital conversion chip, so that an analog signal is converted into a digital signal which can be identified by the MCU, and finally, data is packed and is sent to an upper computer through a serial port of the MCU.
4. The intelligent electronic auscultation control method according to claim 1, wherein after the host computer receives the data, it first extracts valid data from the data packet, detects whether the received data is a 2-byte header, and if so, stores the subsequently received data in the corresponding data buffer until the end of the frame is detected; if the received data is not the frame header, continuing to detect and waiting for frame header data; after receiving a data packet, sequentially integrating the data in the data buffer into 18-bit original data; intercepting the middle stable part of the original signal as data of a single-heart-cycle segmentation algorithm, selecting the middle point of the original signal, and selecting 35000 sampling points from the middle point to two sides respectively as sample data; optionally, 5-layer decomposition filtering is performed using the sym5 wavelet basis to filter out high frequency noise in the signal.
5. The intelligent electronic auscultation control method according to claim 4, wherein after the signal preprocessing of the intelligent electronic auscultation control method is completed, an algorithm is called to process the data, first, frequency information in the sample signal is calculated, then, the number of sampling points included in a single cardiac cycle is calculated according to the sampling rate and the signal frequency, finally, the maximum value in the sample, that is, the peak point of a certain cardiac cycle is selected, the number of sample points in the single cycle is taken, the single cardiac cycle of the sample signal is extracted, and the heart sound function model h (t):
h (t) in the model is the response function of the heart sound generation system, h1S1, the ventricular wall vibration caused by ventricular contraction, reentry of blood by impact with atrioventricular valves, and the vibration caused by closing of atrioventricular valves, h2(x-n1) S2, vibration signal caused by sudden closure of aortic and pulmonary valves at the beginning of ventricular diastole, h3(x-n2) Is S3, due to blood flow impinging on the ventricles, h4(x-n3) For S4, S4 was generated with a lower frequency associated with atrial contractions, where ni(i-1, 2, 3) represents the delay time of the corresponding heart sound in the single cardiac cycle, and ω (t) is the echo generated by muscle, bone, fat and other noises; a (x + T) is an excitation signal of a heart sound generation system, namely when A (x) is effective, the heart beat generates a heart sound H (T), T is a period, the period T or the frequency F of A (x) is calculated, namely the number of sampling points in a single period can be calculated according to the sampling frequency Fs and the period T, the division of the single heart cycle is completed, MFCC feature extraction is carried out on the single heart cycle heart sound signal after the division is completed, the extracted feature value is normalized and then is sent into a pre-trained DBN neural network for recognition, and finally the functions of recognizing and classifying the heart sound are completed.
6. The intelligent electronic auscultation control method of claim 5, further comprising:
1) firstly, removing a direct current component in a sample signal to obtain a non-direct current bias heart sound signal, and performing Hilbert transform on the non-direct current bias heart sound signal to obtain an upper envelope of the signal;
2) FFT transformation is respectively carried out on the original signal and the envelope signal on the original signal, the frequency spectrum of the original signal is observed, most high-frequency heart sound information in the original signal can be filtered out by processing that Hilbert transformation is used for extracting the envelope on the heart sound signal, and the periodic signal A (x) of the lower heart beat and a part of low-frequency heart sound information are reserved;
3) after the original heart sound signal is subjected to Hilbert transform, low-frequency components with higher modulus appear in frequency components of an envelope signal, the spectral line is a frequency domain representation of an excitation signal A (x), the spectral line is extracted in a low-pass filtering mode, and the frequency and the heart rate F of the A (x) can be obtained through calculation;
4) after obtaining the signal A (x), the frequency of the signal A (x) can be calculated by the following formula, wherein F is the frequency of A (x), Fs is the sampling frequency, N is the number of FFT points, and P is the number of points corresponding to the maximum value in the negative half axis of the envelope bilateral spectrum, namely the FFT point corresponding to the maximum amplitude spectral line in the negative half axis of the envelope signal spectrum;
5) in the system, Fs is 5KHz, N is 70K, F is 1.4286Hz, and the number of sampling points contained in a single period is calculated through the following formula after F is obtained;
6) in the sample, Point is 3500, after the number of sample points in a single cycle is calculated, the time domain maximum Point in the sample, namely the S1 peak Point of a certain cardiac cycle, is selected, and the front and rear thresholds of the single cycle are set according to the following formula, namely the single cardiac cycle can be intercepted, wherein H is a single heart sound period function, L is the position of the maximum Point in the sample, Point is the number of sampling points, and α and β are extraction thresholds;
H＝(L-α(Point)；L+β(Ponit))；
7) after the single cardiac cycle is divided, considering the sample signal as a stable signal, and taking the MFCC characteristic parameter as a default extracted characteristic parameter;
8) after MFCC feature extraction is completed, performing normalization processing on 12 MFCC parameters, projecting the MFCC parameters into a [0, 1] interval, inputting the MFCC parameters into a pre-trained deep learning DBN (direct binary learning) network for recognition and classification as input, outputting a classification result through a softmax classifier, finally completing the functions of recognizing and classifying heart sounds, and displaying the recognition result in a heart sound recognition result text box on the left side of a human-computer interface;
9) in the upper computer software, after all algorithms are operated and result data are normal, original heart sound signals, single heart cycle signals, MFCC characteristic parameters and heart sound recognition and classification results are stored in the upper computer in TXT and WAV formats.
7. The intelligent electronic auscultation control method according to claim 1, wherein in the DBN network of the intelligent electronic auscultation control method, connection weights of each layer are trained by using a CD algorithm, and after network training is completed, the network can be used as a hidden layer of a BP neural network to realize construction of a deep belief network, and the specific operation steps are as follows:
1) initializing a multi-layer deep learning network, and training the network by using a CD algorithm to obtain a weight of the deep learning network;
2) constructing a neural network with the number of hidden layers and the number of nodes consistent with that of the deep learning network, and taking the weight of the deep learning network obtained by training as the weight of the neural network;
3) and (4) finely adjusting the neural network by using a BP algorithm to finally obtain an available deep trust network.
8. A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
starting a hardware circuit, selecting and opening a corresponding serial port after the operation is stable, and simultaneously starting a real-time display function of the heart sound waveform; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface;
after the collection is finished, stopping data collection, calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the collected data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished;
after extraction is finished, the MFCC features are sent to a DBN (digital broadcast network) for recognition, after recognition is finished, the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound recognition classification result are respectively displayed in a human-computer interface, and data are stored in TXT and WAV formats.
9. An intelligent electronic auscultation control system for implementing the intelligent electronic auscultation control method according to any one of claims 1 to 7, wherein the intelligent electronic auscultation control system comprises:
the data acquisition module is used for starting a hardware circuit, selecting and opening a corresponding serial port after the hardware circuit is stable in operation, and simultaneously starting a real-time display function of the heart sound waveform; the received heart sound data can be displayed in real time in a graphic frame at the upper end of the human-computer interface;
the data preprocessing module is used for calling a cardiac cycle segmentation algorithm, carrying out single cardiac cycle segmentation processing on the acquired data, and extracting MFCC characteristic parameters of the cardiac cycle after the segmentation is finished;
and the result output module is used for sending the MFCC features into the DBN for recognition after extraction is finished, respectively displaying the single cardiac cycle segmentation result, the MFCC feature parameter extraction result and the heart sound recognition classification result in a human-computer interface after recognition is finished, and storing the data in TXT and WAV formats.
10. An intelligent electronic stethoscope equipped with the intelligent electronic auscultation control system of claim 9, wherein said intelligent electronic stethoscope comprises: the device comprises a piezoelectric sensor, an amplifying circuit, a voltage follower, an anti-aliasing filter, a forward amplifying circuit, an analog-to-digital conversion chip and an upper computer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011151260.0A CN112274174A (en) | 2020-10-25 | 2020-10-25 | Intelligent electronic auscultation control system, method, storage medium and electronic stethoscope |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011151260.0A CN112274174A (en) | 2020-10-25 | 2020-10-25 | Intelligent electronic auscultation control system, method, storage medium and electronic stethoscope |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112274174A true CN112274174A (en) | 2021-01-29 |
Family
ID=74424880
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011151260.0A Pending CN112274174A (en) | 2020-10-25 | 2020-10-25 | Intelligent electronic auscultation control system, method, storage medium and electronic stethoscope |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112274174A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112971802A (en) * | 2021-02-08 | 2021-06-18 | 中北大学 | Heart sound signal detection method and system based on deep learning model |
| CN113679410A (en) * | 2021-07-08 | 2021-11-23 | 北京中医药大学东直门医院 | Pulmonary nodule patient uses sharing stethoscope diagnostic system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6415033B1 (en) * | 1999-09-15 | 2002-07-02 | Ilife Systems, Inc. | Physiological condition monitors utilizing very low frequency acoustic signals |
| CN1938941A (en) * | 2004-02-09 | 2007-03-28 | 音频专用集成电路公司 | Digital microphone |
| US20110167989A1 (en) * | 2010-01-08 | 2011-07-14 | Samsung Electronics Co., Ltd. | Method and apparatus for detecting pitch period of input signal |
| CN102697520A (en) * | 2012-05-08 | 2012-10-03 | 天津沃康科技有限公司 | Electronic stethoscope based on intelligent distinguishing function |
| CN103778432A (en) * | 2014-01-08 | 2014-05-07 | 南京邮电大学 | Human being and vehicle classification method based on deep belief net |
| CN104757992A (en) * | 2015-03-16 | 2015-07-08 | 广东工业大学 | Cardiac sound diagnostic system based on depth confidence network and diagnostic method |
-
2020
- 2020-10-25 CN CN202011151260.0A patent/CN112274174A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6415033B1 (en) * | 1999-09-15 | 2002-07-02 | Ilife Systems, Inc. | Physiological condition monitors utilizing very low frequency acoustic signals |
| CN1938941A (en) * | 2004-02-09 | 2007-03-28 | 音频专用集成电路公司 | Digital microphone |
| US20110167989A1 (en) * | 2010-01-08 | 2011-07-14 | Samsung Electronics Co., Ltd. | Method and apparatus for detecting pitch period of input signal |
| CN102697520A (en) * | 2012-05-08 | 2012-10-03 | 天津沃康科技有限公司 | Electronic stethoscope based on intelligent distinguishing function |
| CN103778432A (en) * | 2014-01-08 | 2014-05-07 | 南京邮电大学 | Human being and vehicle classification method based on deep belief net |
| CN104757992A (en) * | 2015-03-16 | 2015-07-08 | 广东工业大学 | Cardiac sound diagnostic system based on depth confidence network and diagnostic method |
Non-Patent Citations (2)
| Title |
|---|
| 宋少帅: "《心音信号分段定位与识别的研究》", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 * |
| 梅开煌: "《基于PVDF压电薄膜传感器的心音检测系统研究》", 《中国优秀博硕士学位论文全文数据库(硕士)医药卫生科技辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112971802A (en) * | 2021-02-08 | 2021-06-18 | 中北大学 | Heart sound signal detection method and system based on deep learning model |
| CN113679410A (en) * | 2021-07-08 | 2021-11-23 | 北京中医药大学东直门医院 | Pulmonary nodule patient uses sharing stethoscope diagnostic system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Phonocardiographic signal analysis method using a modified hidden Markov model | |
| Leung et al. | Classification of heart sounds using time-frequency method and artificial neural networks | |
| Javed et al. | A signal processing module for the analysis of heart sounds and heart murmurs | |
| WO1998020792A1 (en) | Apparatus and methods for analyzing body sounds | |
| CN112274174A (en) | Intelligent electronic auscultation control system, method, storage medium and electronic stethoscope | |
| CN113436726B (en) | Automatic lung pathological sound analysis method based on multi-task classification | |
| Gavrovska et al. | Automatic heart sound detection in pediatric patients without electrocardiogram reference via pseudo-affine Wigner–Ville distribution and Haar wavelet lifting | |
| WO2020151169A1 (en) | Method for automatic removal of frictional sound interference of electronic stethoscope | |
| CN112971802A (en) | Heart sound signal detection method and system based on deep learning model | |
| Omarov et al. | Artificial Intelligence in Medicine: Real Time Electronic Stethoscope for Heart Diseases Detection. | |
| Brusco et al. | Development of an intelligent PDA-based wearable digital phonocardiograph | |
| CN113796889A (en) | Auxiliary electronic stethoscope signal discrimination method based on deep learning | |
| Ghosh et al. | Deep layer kernel sparse representation network for the detection of heart valve ailments from the time-frequency representation of PCG recordings | |
| Elgendi et al. | Detection of Heart Sounds in Children with and without Pulmonary Arterial Hypertension―Daubechies Wavelets Approach | |
| CN111938691B (en) | Basic heart sound identification method and equipment | |
| CN201683910U (en) | Intelligent cardiopulmonary analyzing instrument | |
| CN215349053U (en) | Congenital heart disease intelligent screening robot | |
| CN111543978A (en) | Method and system for synchronously diagnosing based on electrocardio, heart sound and pulse | |
| Zannat et al. | Automated system for features extraction from pcg signal | |
| Zhong et al. | Heart murmur recognition based on hidden Markov model | |
| Botha et al. | Autonomous auscultation of the human heart employing a precordial electro-phonocardiogram and ensemble empirical mode decomposition | |
| Skopin et al. | Heartbeat feature extraction from vowel speech signal using 2D spectrum representation | |
| Ghaffari et al. | Phonocardiography signal processing for automatic diagnosis of ventricular septal defect in newborns and children | |
| CN110811673A (en) | Heart sound analysis system based on probabilistic neural network model | |
| Golpaygani et al. | Detection and identification of S1 and S2 heart sounds using wavelet decomposition method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210129 |
|
| RJ01 | Rejection of invention patent application after publication |