CN110251152A - A kind of osteoacusis formula number auscultation system being automatically separated heart and lung sounds - Google Patents
A kind of osteoacusis formula number auscultation system being automatically separated heart and lung sounds Download PDFInfo
- Publication number
- CN110251152A CN110251152A CN201810198656.7A CN201810198656A CN110251152A CN 110251152 A CN110251152 A CN 110251152A CN 201810198656 A CN201810198656 A CN 201810198656A CN 110251152 A CN110251152 A CN 110251152A
- Authority
- CN
- China
- Prior art keywords
- heart
- signal
- sound
- lung sounds
- amplitude spectrum
- 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
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B7/00—Instruments for auscultation
- A61B7/003—Detecting lung or respiration noise
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Mathematical Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Veterinary Medicine (AREA)
- Acoustics & Sound (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Public Health (AREA)
- Artificial Intelligence (AREA)
- Pulmonology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)
Abstract
The invention discloses a kind of auscultation systems that heart and lung sounds separation is carried out based on deep learning, and auscultated using osteoacusis, which includes auscultation head, signal acquisition module, heart and lung sounds signal separation module, memory, user interface, output driving unit and osteoacusis unit；Wherein, signal acquisition module acquisition heart and lung sounds signal simultaneously digitizes, heart and lung sounds signal separation module carries out heart and lung sounds signal separation using time-frequency masking deep-cycle neural network, user interface can carry out heart and lung sounds model selection and expert marks storage operation, output driving unit receives the heart sound that prime is isolated or lungs sound time-domain signal carries out power amplification, and driving osteoacusis unit generates vibration.This stethoscope can be automatically separated out clean heart sound and lungs sound from the heart and lung sounds signal of aliasing, facilitate data acquisition and post analysis, promote auscultation experience.
Description
Technical field
The invention belongs to digital medical equipment technical fields, propose a kind of osteoacusis formula number for being automatically separated heart and lung sounds
Auscultation system.
Background technique
Stethoscope is the tool be unableing to do without in doctor's routine work, once becomes the synonym of doctor.So far stethoscope has been
There is development in more than 200 years, shape and acoustic mode are continuously improved, but basic structure change is little, mainly there is pickup portion
Divide, acoustic part, audition part composition.Conventional stethoscope generally uses Y type fixture, exist wear it is uncomfortable, vulnerable to ambient sound
The defects of interference, and heart sound is not only contained in the collected voice signal of stethoscope, also contain lungs sound and external ring
Border noise, this voice signal for being mixed with heart sound, lungs sound and environmental noise are difficult to the illness analysis of fining.In recent years
Electronic auscultation device will only pick up merely mostly since unique advantage in terms of amplifying sound, storage is quickly grown
The voice signal of sound portion collection is digitized, and does not carry out further signal processing to improve heart sound or lungs sound
Auscultate effect.
In response to this problem, existing scheme mainly has following: such as Chinese utility model patent 201620968593.5, also mentioning
And there is heart and lung sounds separation function, but it is using linear bandpass filter, it can not be by the heart sound and lungs sound of time-frequency aliasing
Fine separation, in some instances it may even be possible to lead to heart sound or a part of band information missing of lungs sound, to make auscultation inaccuracy；Such as Chinese invention
Patent 201710470514.7 carries out blind channel separation to heart and lung sounds signal using the thought of difference enhancing, but this method needs
It wants multiple signals to acquire simultaneously, inevitably makes that stethoscope volume is big, structure is complicated；For another example Chinese invention patent
201710651856.9, heart and lung sounds separation is carried out using Non-negative Matrix Factorization, since this method is assumed between heart sound and lungs sound
Independence, and fail nonlinear coupling relationship complicated between Efficient Characterization heart sound and lungs sound, this is just virtually being limited
The performance of this method.
In addition, electric signal is finally reduced into sound by earphone by general electronic auscultation device, auscultated, multi-purpose is general
Logical air transmitted earphone, voice signal is propagated in air to be interfered vulnerable to noise, and auscultation effect is simultaneously not ideal enough.
Summary of the invention
It is improved, is disclosed a kind of automatic in view of the above problems for deficiency, the present invention in the prior art
Separate the osteoacusis formula number auscultation system of heart sound lungs sound.The present invention is converted heart and lung sounds signal to using miniature sound transducer
Electric signal is further converted to digital signal, and carries out heart and lung sounds signal separation using time-frequency masking deep-cycle neural network,
Signal after final separation can pass to human auditory system nerve by way of osteoacusis.Based on big data training, deep-cycle
Fine noise reduction filtering may be implemented in neural network DRNN；Meanwhile DRNN can sufficiently learn in heart sound and the mutual aliasing of lungs sound
Complicated nonlinear coupling relationship carries out non-linear decoupling to heart and lung sounds signal, realizes good separating effect.In addition, bone passes
The mode of leading has independent acoustic path, and can bypass external auditory canal stimulation Middle inner ear not by the interference of sound wave in air makes auscultation more
Accurately；Secondly as being not required to pleasant wearing, osteoacusis mode is also beneficial to the hygiene and health of ear canal.
To achieve the goals above, it is automatically separated heart sound lungs sound, facilitates data acquisition and post analysis, experience is listened in promotion,
The present invention is as follows using concrete scheme:
Signal acquisition unit (2) acquires human body cardiopulmonary physiology by the miniature sound transducer being built in auscultation head (1)
Activity generate sound, be converted into electric signal, and it is amplified, filter and A/D conversion, ultimately generate digital audio letter
Number, input the heart and lung sounds signal separative unit (3) of rear class；
User interface (5) is responsible for marking (52) two functions of storage to be configured heart and lung sounds model selection (51), expert,
Wherein, heart and lung sounds model selection (51) is made of heart sound mode, lungs sound mode, general mode Three models；Expert marks storage
(52) it is responsible for that storage is marked to four patient's name, heart and lung sounds mode, auscultation position, severity information, and settable
Recording；
Heart and lung sounds signal separative unit (3) is using time-frequency masking deep-cycle neural network separation signal acquisition unit (2)
The heart and lung sounds mixed signal transmitted, and the option information according to selected by heart and lung sounds model selection (51) function of user interface (5),
Selection output heart sound, lungs sound or unsegregated mixed signal；
Storage unit (4) can store the recording with expert's mark information using Micro SD card；
Output driving unit (6) carries out D/A digital-to-analogue conversion and function to the signal of heart and lung sounds signal separative unit (3) output
Rate amplification, to drive the osteoacusis unit (7) of rear class to generate vibration；
Osteoacusis unit (7) converts vibration signal for the electrical power signal that output driving unit (6) export, passes through c-type
Support ring, jawbone, skull before so that the vibration unit of its left and right two is affixed on human ear, around the air acoustic path of external ear, directly
Auditory nerve is stimulated by osteoacusis, makes one to experience sound；
In addition, system is also equipped with power supply, it is made of management of charging and discharging circuit and rechargeable battery, is provided for system each unit
Electric power.
Preferably, the amplifier that the signal acquisition unit (2) uses is trans-impedance amplifier, it can be effectively by miniature sound
The current signal of sound sensor output is converted to easy-to-handle voltage signal；The filter of use is 2 rank Sallen-Key types
Bessel filter can make former letter because Bessel filter has linear phase to avoid during inhibiting out-of-band noise
Number phase is distorted；The A/D of use is delta-sigma type ADC, and sampling precision is improved by the way of over-sampling.
Preferably, the heart and lung sounds signal separative unit (3), carries out the heart using time-frequency masking deep-cycle neural network
Lung Sounds separation, specific network frame is as shown in Fig. 2, input heart and lung sounds mixed signal generates width through Short Time Fourier Transform
Degree spectrum A and phase spectrum P, wherein amplitude spectrum A is inputted in time-frequency masking deep-cycle neural network as feature and is carried out prediction separation,
Heart sound amplitude spectrum after separationWith lungs sound amplitude spectrumInverse Fourier transform in short-term is carried out with phase spectrum, is reduced to respective
Time-domain signal is achieved in the separation of heart sound and lungs sound.The structure of specific deep-cycle neural network is as shown in figure 3, introduce
It is as follows:
1) network inputs: in moment t, the amplitude spectrum x of network inputs heart and lung sounds mixed signalt；
2) network hidden layer: 3 layer depth Recognition with Recurrent Neural Network are used, for every node layer quantity 128~1024, activation primitive is complete
Portion is line rectification function ReLU；
3) network exports: the output target of network is clean heart sound amplitude spectrum y1tWith lungs sound amplitude spectrum y2t, corresponding
Output is the heart sound amplitude spectrum of prediction respectivelyWith lungs sound amplitude spectrum
4) in order to further smooth separating resulting, the additional layer time-frequency masking layer after network output layer, constraint separation it
The sum of heart and lung sounds spectrum amplitude afterwards keeps it equal with the spectrum amplitude for separating preceding heart and lung sounds mixed signal, wherein time-frequency masking
It is defined as follows:
mtTime-frequency masking is represented, f is frequency, by mt(f) it is applied to the amplitude spectrum x of mixed signaltTo obtain the heart sound of separation
Amplitude spectrumWith lungs sound amplitude spectrum
⊙ indicates Hadamard product.
5) loss function: final separation output is comparedWithWith output target y1tAnd y2t, missed by least square
Difference constrains neural network parameter:
By reducing JMSEIncrease the similarity between prediction and target.
Preferably, entire heart and lung sounds separation process includes training process S1 and separation process S2:
Training process S1 are as follows:
1) construct training set: the lungs sound that right chest is arrived far from the station acquisition of heart is as clean lungs sound；It feels suffocated, it will be left
For the heart sound that the station acquisition of chest close to heart arrives as clean heart sound, sample rate is 16kHz.By clean heart sound and completely
Lungs sound carry out being mixed to get heart and lung sounds according to different power proportions ([- 40, -30, -20, -10,0,10,20,30,40] dB)
Mixed signal；Short Time Fourier Transform is carried out to the mixed signal and generates mixed signal amplitude spectrum xt, wherein frame length be 20ms~
60ms, frame shifting is 10ms~30ms；Clean heart sound and lungs sound are subjected to same Short Time Fourier Transform；Each pair of training data
By the amplitude spectrum x of the continuous heart and lung sounds mixed signal of any N frametAnd the amplitude spectrum y of corresponding clean heart sound1tWith clean lungs sound width
Degree spectrum y2tComposition, N are 5~10.
2) training: by xtTraining pattern, y are inputted as feature1tAnd y2tIt as corresponding label, is trained, until mould
Type convergence, obtains trained model；
Separation process S2 are as follows:
The heart and lung sounds mixed signal of signal acquisition unit acquisition carries out short time discrete Fourier transform and generates amplitude spectrum A and phase spectrum
P successively chooses N frame continuous signal from amplitude spectrum A and is input to trained model in S1, exports the heart sound amplitude spectrum of separationWith lungs sound amplitude spectrumWithInverse Fourier transform in short-term is carried out with phase spectrum P respectively, generates the heart after separation
Sound time-domain signal and lungs sound time-domain signal.
The invention has the following beneficial effects:
This auscultation system is improved and is had a clear superiority in auscultation quality in separation heart and lung sounds signal, using time-frequency masking depth
It spends Recognition with Recurrent Neural Network and carries out heart and lung sounds signal separation, non-linear decoupling effectively can be carried out to heart and lung sounds signal, heart and lung sounds exist
Mutual interference is not will receive when auscultation；During auscultation, expert can be carried out to heart and lung sounds signal and mark storage, facilitate number
It is searched according to management, later period；This auscultation system is different from general stethoscope, stimulates auditory nerve, Ke Yiyou eventually by osteoacusis
Effect avoids the air propagation path of external ear, and the sound that avoids that treated is improved auscultation quality by secondary interference.
Detailed description of the invention
Fig. 1: structural block diagram of the invention；
Fig. 2: the frame of heart and lung sounds signal separation is carried out using time-frequency masking deep-cycle neural network；
Fig. 3: deep-cycle neural network structure.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.
As shown in Figure 1, embodiment provides a kind of osteoacusis formula number auscultation system for being automatically separated heart and lung sounds, comprising:
1) auscultation head (1), one side are vibrating membranes, and the other side is bell-jar, in internal resonant cavity center face vibrating membrane
Position install a miniature electret microphone；
2) signal acquisition unit (2), the internal miniature electret microphone of driving auscultation head (1) will carry sound letter
The current signal of breath is converted to voltage signal and is further converted into digital signal, the specific steps are as follows:
Step 1: driving mini microphone using reverse phase trans-impedance amplifier, current signal is converted into easy-to-handle voltage
The input of signal, amplifier uses AC coupled；
Step 2: filtering out out-of-band noise, pressed down using the second order Sallen-Key type Bessel filter that cutoff frequency is 3KHz
High-frequency noise processed；
Step 3: being counted the voltage signal that filter exports with the sample frequency of 16KHz using 16 ADC of delta-sigma type
Word.
3) heart and lung sounds signal separative unit (3) receives the heart and lung sounds mixed signal for coming from signal acquisition unit (2), when use
Frequency masking deep-cycle neural network carries out heart and lung sounds signal separation,
Network training process is as follows:
Acquire 100 subjects, 3 groups of everyone repeated acquisition clean cardiechema signals and Lung Sounds, every group of lasting 10s, by 9
The power ratio ([- 40, -30, -20, -10,0,10,20,30,40] dB) of a grade is mixed, and is selected from every group of mixed signal
Take 33 sections of continuous 5 frames, totally 445500 frame heart and lung sounds mixed signal amplitude spectrums and corresponding clean cardiechema signals amplitude spectrum and
Lung Sounds amplitude spectrum is trained network as training set；
Specific separating step refers to Fig. 2, is described below:
Step 1: the heart and lung sounds mixed signal that sample rate is 16kHz generates amplitude spectrum A and phase by Short Time Fourier Transform
Position spectrum P, wherein the frame length of Short Time Fourier Transform is 60ms, and it is 30ms that frame, which moves,；The continuous amplitude spectrum A of 5 frames is inputted into time-frequency masking
Depth neural circuitry network；
Step 2: by 3 layer depth Recognition with Recurrent Neural Network, every node layer quantity is 512, is activated using ReLU function, by
Time-frequency masking layer generates heart sound amplitude spectrumWith lungs sound amplitude spectrum
Step 3: by heart sound amplitude spectrumWith lungs sound amplitude spectrumInverse Fourier transform in short-term is carried out with phase spectrum P respectively,
Heart sound time-domain signal and lungs sound time-domain signal after generating separation are provided according to heart and lung sounds model selection (51) in user interface
Heart sound time-domain signal, is passed to the output driving unit of rear class by the instruction of " heart sound mode " option.
4) the heart sound time domain isolated is believed using two-channel 16 DAC, sample rate 16kHz output driving unit (6)
Number analog signal is converted to, and is transferred to dual-channel audio power amplifier, the osteoacusis unit of rear class is driven to generate vibration.
5) electrical power signal is reduced to voice signal by vibration unit by osteoacusis unit (7), and it is pleasant to be different from tradition
Formula stethoscope, which transmits without external ear ear canal, but is directly applied to the jawbone on front side of human ear, passes through osteoacusis
Auditory nerve is directly stimulated, to be auscultated.
It is noted that in the absence of conflict, the feature in embodiment and embodiment in the present invention can be mutual
Combination.Although present invention has been a degree of descriptions, it will be apparent that, in the condition for not departing from the spirit and scope of the present invention
Under, the appropriate variation of each condition can be carried out.It is appreciated that the present invention is not limited to the embodiments, and it is attributed to claim
Range comprising the equivalent replacement of each factor.
Claims (5)
1. a kind of osteoacusis auscultation system, which is characterized in that by auscultation head, signal acquisition unit, heart and lung sounds signal separative unit,
Storage unit, user interface, output driving unit, osteoacusis unit and power supply composition；
Wherein, signal acquisition unit is converted heart and lung sounds signal using the miniature sound transducer being built in auscultation head resonant cavity
For electric signal, and the electric signal is amplified, is filtered, A/D is converted to digital audio and video signals；
User interface is responsible for marking two functions of storage to be configured heart and lung sounds model selection, expert, wherein heart and lung sounds mode
Selection is made of heart sound mode, lungs sound mode, general mode Three models；Expert marks storage to be responsible for patient's name, cardiopulmonary
Storage is marked in four sound mode, auscultation position, severity information；
Heart and lung sounds signal separative unit be responsible for by signal acquisition unit export heart and lung sounds mixed signal separate, and according to
Option information provided by heart and lung sounds model selection function in the interface of family, selection output heart sound, lungs sound or unsegregated mixing
Signal；
Output driving unit is made of D/A converter and power amplifier, is responsible for the letter exported to heart and lung sounds signal separative unit
Number carry out power amplification, drive osteoacusis unit；
Osteoacusis unit is responsible for converting vibration signal for the electric signal of output driving unit.
2. signal acquisition unit as described in claim 1, which is characterized in that amplifier is reverse phase trans-impedance amplifier, filter
For second order Sallen-Key type Bessel filter, A/D converter is delta-sigma type ADC.
3. heart and lung sounds signal separative unit as described in claim 1, which is characterized in that using time-frequency masking deep-cycle nerve
Network carries out heart and lung sounds signal separation, specifically includes training process S1 and separation process S2,
Wherein, training process S1 are as follows:
Step 1: construction training set, using in the collected lungs sound of right chest as clean lungs sound；It will be collected when left chest is felt suffocated
Heart sound as clean heart sound；By clean heart sound and clean lungs sound according to different power proportions ([- 40, -30, -
20, -10,0,10,20,30,40] dB) it carries out being mixed to get heart and lung sounds mixed signal；Fourier in short-term is carried out to the mixed signal
Transformation generates heart and lung sounds mixed signal amplitude spectrum xt, wherein Short Time Fourier Transform frame length is 20ms~60ms, and frame shifting is 10ms
~30ms；Clean heart sound and lungs sound are subjected to same Short Time Fourier Transform；Each pair of training data is by any continuous heart of N frame
The amplitude spectrum x of lungs sound mixed signaltAnd the amplitude spectrum y of corresponding clean heart sound1tWith the amplitude spectrum y of clean lungs sound2tComposition, N 5
~10；
2) training: by xtTraining pattern, y are inputted as feature1tAnd y2tIt as corresponding label, is trained, until model is received
It holds back, obtains trained model；
Separation process S2 are as follows:
Short time discrete Fourier transform is carried out to the heart and lung sounds mixed signal of signal acquisition unit acquisition and generates amplitude spectrum A and phase spectrum P,
N frame continuous signal is successively chosen from amplitude spectrum A and is input to trained model in S1, exports the heart sound amplitude spectrum of separation
With lungs sound amplitude spectrumWithInverse Fourier transform in short-term is carried out with phase spectrum P respectively, generates the heart sound after separation
Time-domain signal and lungs sound time-domain signal.
4. time-frequency masking deep-cycle neural network as claimed in claim 3, which is characterized in that network structure is as follows:
1) network inputs: in moment t, the amplitude spectrum x of network inputs heart and lung sounds mixed signalt；
2) network hidden layer: 3~5 layer depth Recognition with Recurrent Neural Network are used, for every node layer quantity 128~1024, activation primitive is equal
For line rectification function ReLU；
3) network exports: the output target of network is clean heart sound amplitude spectrum y1tWith lungs sound amplitude spectrum y2t, corresponding
Output is the heart sound amplitude spectrum of prediction respectivelyWith lungs sound amplitude spectrum
4) time-frequency masking processing is carried out after network output layer, wherein time-frequency masking is defined as follows:
mtTime-frequency masking is represented, f is frequency, by mt(f) it is applied to the amplitude spectrum x of mixed signaltTo obtain the heart sound width of separation
Degree spectrumWith lungs sound amplitude spectrum
⊙ indicates Hadamard product；
5) loss function: final separation output is comparedWithWith output target y1tAnd y2t, by least squares error come
Constrain neural network parameter:
By reducing JMSEIncrease the similarity between prediction and target.
5. osteoacusis unit as described in claim 1, which is characterized in that include a C-shaped support ring for wearing, support
Respectively there are a vibration unit in ring two sides, and sound vibration signal is applied on front side of human ear on the jawbone of left and right, around the air of external ear
Acoustic path stimulates auditory nerve to experience auscultatory sound by osteoacusis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810198656.7A CN110251152A (en) | 2018-03-12 | 2018-03-12 | A kind of osteoacusis formula number auscultation system being automatically separated heart and lung sounds |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810198656.7A CN110251152A (en) | 2018-03-12 | 2018-03-12 | A kind of osteoacusis formula number auscultation system being automatically separated heart and lung sounds |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110251152A true CN110251152A (en) | 2019-09-20 |
Family
ID=67911396
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810198656.7A Pending CN110251152A (en) | 2018-03-12 | 2018-03-12 | A kind of osteoacusis formula number auscultation system being automatically separated heart and lung sounds |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110251152A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110559012A (en) * | 2019-10-21 | 2019-12-13 | 江苏鹿得医疗电子股份有限公司 | Electronic stethoscope, control method thereof and control method of medical equipment |
| CN110705624A (en) * | 2019-09-26 | 2020-01-17 | 广东工业大学 | Cardiopulmonary sound separation method and system based on multi-signal-to-noise-ratio model |
| CN112669870A (en) * | 2020-12-24 | 2021-04-16 | 北京声智科技有限公司 | Training method and device of speech enhancement model and electronic equipment |
| CN113229842A (en) * | 2021-05-19 | 2021-08-10 | 苏州美糯爱医疗科技有限公司 | Heart and lung sound automatic separation method based on complex deep neural network |
| RU2766751C1 (en) * | 2021-03-05 | 2022-03-15 | Федеральное государственное бюджетное образовательное учреждение высшего образования «Юго-Западный государственный университет» (ЮЗГУ) (RU) | Method for diagnosing asthmatic bronchitis in process of lung auscultation in adults |
| CN115422976A (en) * | 2022-09-14 | 2022-12-02 | 湖南万脉医疗科技有限公司 | Artificial network-based cardiopulmonary coupling relationship analysis method and monitoring system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1488319A (en) * | 2002-10-08 | 2004-04-14 | 合世股份有限公司 | Multifunctional electronic stethoscope |
| CN106419953A (en) * | 2016-09-26 | 2017-02-22 | 珠海爱珂索移动医疗科技有限公司 | Wireless auscultation system based on double-frequency collection and vibration reduction |
| CN106510752A (en) * | 2016-01-20 | 2017-03-22 | 曹松林 | Domestic and medical two-purpose sound recording and fetal heart sound listening stethoscope |
| CN107174277A (en) * | 2017-06-20 | 2017-09-19 | 深圳市泽科科技有限公司 | A kind of heart and lung sounds separation multichannel stethoscope based on FPGA |
-
2018
- 2018-03-12 CN CN201810198656.7A patent/CN110251152A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1488319A (en) * | 2002-10-08 | 2004-04-14 | 合世股份有限公司 | Multifunctional electronic stethoscope |
| CN106510752A (en) * | 2016-01-20 | 2017-03-22 | 曹松林 | Domestic and medical two-purpose sound recording and fetal heart sound listening stethoscope |
| CN106419953A (en) * | 2016-09-26 | 2017-02-22 | 珠海爱珂索移动医疗科技有限公司 | Wireless auscultation system based on double-frequency collection and vibration reduction |
| CN107174277A (en) * | 2017-06-20 | 2017-09-19 | 深圳市泽科科技有限公司 | A kind of heart and lung sounds separation multichannel stethoscope based on FPGA |
Non-Patent Citations (1)
| Title |
|---|
| PO-SEN HUANG等: "Singing-Voice Separation from Monaural Recordings using Deep Recurrent Neural Networks", 《15TH INTERNATIONAL SOCIETY FOR MUSIC INFORMATION RETRIEVAL CONFERENCE》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705624A (en) * | 2019-09-26 | 2020-01-17 | 广东工业大学 | Cardiopulmonary sound separation method and system based on multi-signal-to-noise-ratio model |
| CN110559012A (en) * | 2019-10-21 | 2019-12-13 | 江苏鹿得医疗电子股份有限公司 | Electronic stethoscope, control method thereof and control method of medical equipment |
| CN110559012B (en) * | 2019-10-21 | 2022-09-09 | 江苏鹿得医疗电子股份有限公司 | Electronic stethoscope, control method thereof and control method of medical equipment |
| CN112669870A (en) * | 2020-12-24 | 2021-04-16 | 北京声智科技有限公司 | Training method and device of speech enhancement model and electronic equipment |
| CN112669870B (en) * | 2020-12-24 | 2024-05-03 | 北京声智科技有限公司 | Training method and device for voice enhancement model and electronic equipment |
| RU2766751C1 (en) * | 2021-03-05 | 2022-03-15 | Федеральное государственное бюджетное образовательное учреждение высшего образования «Юго-Западный государственный университет» (ЮЗГУ) (RU) | Method for diagnosing asthmatic bronchitis in process of lung auscultation in adults |
| CN113229842A (en) * | 2021-05-19 | 2021-08-10 | 苏州美糯爱医疗科技有限公司 | Heart and lung sound automatic separation method based on complex deep neural network |
| CN113229842B (en) * | 2021-05-19 | 2022-10-14 | 苏州美糯爱医疗科技有限公司 | Heart and lung sound automatic separation method based on complex deep neural network |
| CN115422976A (en) * | 2022-09-14 | 2022-12-02 | 湖南万脉医疗科技有限公司 | Artificial network-based cardiopulmonary coupling relationship analysis method and monitoring system |
| CN115422976B (en) * | 2022-09-14 | 2023-11-21 | 湖南万脉医疗科技有限公司 | Cardiopulmonary coupling relation analysis method and monitoring system based on artificial network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110251152A (en) | A kind of osteoacusis formula number auscultation system being automatically separated heart and lung sounds | |
| JP5577449B2 (en) | Hearing aid suitable for EEG detection and method for adapting such a hearing aid | |
| CN102973277B (en) | Frequency following response signal test system | |
| CN110353725A (en) | A kind of heart sound acquisition and analysis system and method based on cloud framework | |
| US20110096936A1 (en) | Electronic stethoscope | |
| CN111657991B (en) | Intelligent array sensor electronic auscultation system | |
| WO2019241645A1 (en) | Device for monitoring activities of daily living and physiological parameters to determine a condition and diagnosis of the human brain and body | |
| EP0778747A1 (en) | An electronic stethoscope | |
| CN201755228U (en) | Wireless electrocardio (ECG) and Phonocardiogram (PCG) all-in-one machine | |
| CN111631728B (en) | Method and device for measuring bone conduction transfer function and storage medium | |
| CN113855052B (en) | Nerve feedback intervention system and method based on positive idea meditation | |
| CN103120592A (en) | Heart sound electronic auscultation and intelligent health analysis system | |
| CN202198616U (en) | Electronic stethoscope | |
| RU2644546C1 (en) | Electronic medical stethoscope | |
| US11471088B1 (en) | Handheld or wearable device for recording or sonifying brain signals | |
| CN115024719B (en) | Tinnitus detection method and device based on hearing bionics | |
| CN110010145B (en) | Method for eliminating friction sound of electronic stethoscope | |
| Joshi et al. | Bluetooth-based wireless digital stethoscope | |
| CN104306015A (en) | Portable otoacoustic emission detection system | |
| Malwade et al. | Wireless stethoscope with Bluetooth technology | |
| CN106073817A (en) | A kind of based on the portable Intelligent stethoscope of wireless network | |
| CN209234005U (en) | It returns sound functional component and assists the sound insulation feedback earphone of Foreigh-language oral-speech training | |
| CN208540163U (en) | A kind of auditory prosthesis | |
| CN207898490U (en) | A kind of Multi-purpose stethoscope used for Cardiologists | |
| CN204765703U (en) | Stethophone |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190920 |