CN113768532A - Health detection method and system based on five-path heart sound signal classification algorithm - Google Patents
Health detection method and system based on five-path heart sound signal classification algorithm Download PDFInfo
- Publication number
- CN113768532A CN113768532A CN202110962279.1A CN202110962279A CN113768532A CN 113768532 A CN113768532 A CN 113768532A CN 202110962279 A CN202110962279 A CN 202110962279A CN 113768532 A CN113768532 A CN 113768532A
- Authority
- CN
- China
- Prior art keywords
- heart sound
- neural network
- network model
- preset
- sound signals
- 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
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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Pathology (AREA)
- Biophysics (AREA)
- Psychiatry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Acoustics & Sound (AREA)
- Power Engineering (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)
Abstract
The application relates to a health detection method and system based on a five-way heart sound signal classification algorithm, in particular to the field of health equipment. The health detection method based on the five-channel heart sound signal classification algorithm comprises the following steps: acquiring five paths of heart sound signals of a sample to be detected; extracting feature information in the five-path heart sound signals by using a first preset function; carrying out calculation by bringing an array corresponding to the characteristic information of the five-path heart sound signal into a preset neural network model; the state information represented by the calculation result of the preset neural network model is output as the detection result, the method can directly obtain the state information of the sample to be detected by extracting and analyzing the characteristic information of the five-path heart sound signal, calculating by using the preset neural network model and comparing the calculation result with the preset threshold value, thereby simplifying the detection process of the body state of the sample to be detected.
Description
Technical Field
The application relates to the field of health equipment, in particular to a health detection method and system based on a five-channel heart sound signal classification algorithm.
Background
In modern chinese and western medicine theory, heart sounds can represent a person's physical state, and heart sounds (heart sound) refer to sounds generated by mechanical wave phenomena caused by contraction of cardiac muscle, closure of heart valves, and impact of blood against the wall of the heart chamber, the wall of the aorta, etc. It can be heard by stethoscope at a certain position of chest wall, or can be used to record the mechanical wave of heart sound by transducer, etc. the graph of its mechanical wave changing with time is called as heart sound graph.
In the prior art, the diagnosis of cardiovascular and cerebrovascular diseases is generally to diagnose the state of a sample to be detected by a series of methods such as skull CT, magnetic resonance imaging, cerebrovascular angiography, transcranial color Doppler ultrasound and the like.
However, the above-mentioned detection of the sample to be detected requires a lot of instruments, the time required for the detection is long, and most of the instruments involved are equipped in hospitals, so that the monitoring may not be in time in some emergency situations, and the diagnosis time of the sample to be detected may be delayed when performing various detections.
Disclosure of Invention
The invention aims to provide a health detection method and system based on a five-channel heart sound signal classification algorithm aiming at the defects in the prior art, so as to solve the problems that in the prior art, because more instruments are needed for detecting a sample to be detected, the time needed for detection is longer, most of the involved instruments are equipped in hospitals, the monitoring is possibly not in time in some emergency situations, and in addition, the diagnosis time of the sample to be detected can be delayed in the process of carrying out various detections.
In order to achieve the above purpose, the embodiment of the present invention adopts the following technical solutions:
in a first aspect, the present application provides a health detection method based on a five-way heart sound signal classification algorithm, including:
acquiring five paths of heart sound signals of a sample to be detected, wherein the five paths of heart sound signals are acquired through a heart sound probe;
extracting feature information in the five-path heart sound signals by using a first preset function; wherein the characteristic information includes: at least one of an average value, a period number, a waveform picture, a power spectral density map, a Mel cepstrum, a spectrogram and an energy spectrum;
carrying out calculation by bringing an array corresponding to the characteristic information of the five-path heart sound signal into a preset neural network model;
and outputting the state information corresponding to the calculation result obtained by the preset neural network model as a detection result.
Optionally, the preset neural network model is:
wherein, h (x) is a calculation result of the preset neural network model, the value range of h (x) is (-1,1), x1, x2 and xn are all the arrays corresponding to the input characteristic information, and w1, w2, wn and b are all parameters.
Optionally, before the step of bringing the array corresponding to the feature information of the five-way heart sound signal into the preset neural network model for calculation, the method further includes:
acquiring state information of a plurality of groups of known samples and characteristic information corresponding to five-path heart sound signals of the plurality of groups of known samples;
and bringing the array corresponding to the state information of the multiple groups of known samples and the characteristic information corresponding to the five-way heart sound signals of the multiple groups of known samples into a preset neural network model, and calculating parameters in the preset neural network model.
Optionally, the step of obtaining five heart sound signals of the sample to be detected further includes:
carrying out digital filtering on the five paths of heart sound signals;
carrying out down-sampling operation according to a preset frequency;
carrying out maximum value normalization processing on the five paths of heart sound signals according to a preset interval;
and cutting the five paths of heart sound signals according to the preset length and the preset coincidence degree.
Optionally, the step of bringing the array corresponding to the feature information of the five-way heart sound signal into a preset neural network model for calculation specifically includes:
converting the characteristic information of the five-path heart sound signals into array information;
and bringing the array information into a preset neural network model for calculation.
In a second aspect, the present application provides a health detection system based on a five-way heart sound signal classification algorithm, the system including: the device comprises an acquisition module, an extraction module, a calculation module and an output module; the acquisition module is used for acquiring five paths of heart sound signals of a sample to be detected, wherein the five paths of heart sound signals are acquired through the heart sound probe; the extraction module is used for extracting feature information in the five-path heart sound signals by using a first preset function; wherein the characteristic information includes: the calculation module is used for bringing an array corresponding to the characteristic information of the five-path heart sound signals into a preset neural network model for calculation; the output module is used for outputting the state information corresponding to the calculation result obtained by the preset neural network model as the detection result.
Optionally, the preset neural network model is:
wherein h (x) is the calculation result of the preset neural network model, and the value range of h (x) is (-1,1), x1，x2、xnAre arrays of input feature information, w1、w2、wnAnd b are parameters.
Optionally, the calculation module is further configured to obtain state information of a plurality of groups of known samples and feature information corresponding to five-way heart sound signals of the plurality of groups of known samples; and bringing the array corresponding to the state information of the multiple groups of known samples and the characteristic information corresponding to the five-way heart sound signals of the multiple groups of known samples into a preset neural network model, and calculating parameters in the preset neural network model.
Optionally, the step of obtaining five heart sound signals of the sample to be detected further includes: the preprocessing module is used for carrying out digital filtering on the five paths of heart sound signals; carrying out down-sampling operation according to a preset frequency; carrying out maximum value normalization processing on the five paths of heart sound signals according to a preset interval; and cutting the five paths of heart sound signals according to the preset length and the preset coincidence degree.
Optionally, the calculation module is specifically configured to convert feature information of the five-way heart sound signal into array information; and bringing the array information into a preset neural network model for calculation.
In a third aspect, the present application provides an electronic device, comprising: the health detection method based on the five-way heart sound signal classification algorithm comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor executes the program to realize the health detection method based on the five-way heart sound signal classification algorithm.
In a fourth aspect, the present application provides a computer-readable storage medium, where the computer-readable storage medium includes a computer program, and the computer program controls, when running, an electronic device where the computer-readable storage medium is located to execute the health detection method based on the five-way heart sound signal classification algorithm.
The invention has the beneficial effects that:
the health detection method based on the five-channel heart sound signal classification algorithm comprises the following steps: acquiring five paths of heart sound signals of a sample to be detected, wherein the five paths of heart sound signals are acquired through a heart sound probe; extracting feature information in the five-path heart sound signals by using a first preset function; wherein the characteristic information includes: at least one of an average value, a period number, a waveform picture, a power spectral density map, a Mel cepstrum, a spectrogram and an energy spectrum; carrying out calculation by bringing an array corresponding to the characteristic information of the five-path heart sound signal into a preset neural network model; carrying out calculation by bringing an array corresponding to the characteristic information of the five-path heart sound signal into a preset neural network model; the method comprises the steps of extracting and analyzing characteristic information of five-way heart sound signals, calculating by using the preset neural network model, and comparing the calculation result with a preset threshold value according to the calculation result, so that the state information of the sample to be detected can be directly obtained, the detection flow of the body state of the sample to be detected is simplified, and the detection accuracy of the sample to be detected is higher due to the fact that the preset database is used for comparing to obtain the state information of the sample to be detected, namely subjective errors caused by artificial diagnosis are avoided.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
Fig. 1 is a schematic view of an application scenario provided in the present application;
fig. 2 is a schematic flowchart of a health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application;
fig. 3 is a schematic diagram of a two-dimensional characteristic log mel spectrum of a health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application;
fig. 4 is a schematic diagram of a two-dimensional characteristic log power spectrum of a health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application;
fig. 5 is a schematic flowchart of another health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application;
fig. 6 is a schematic block diagram of a health detection system based on a five-way heart sound signal classification algorithm according to an embodiment of the present application;
fig. 7 is a block diagram of another health detection system based on a five-way heart sound signal classification algorithm according to an embodiment of the present application.
Detailed Description
The technical solutions in the present application will be described clearly and completely with reference to the drawings in the present application, and it should be understood that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the present application, as generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
Fig. 1 is a schematic view of an application scenario provided by the present application, and as shown in fig. 1, the method of the present application may be applied to the electronic device 10 and the heart sound acquiring probe 20 shown in fig. 1. As shown in fig. 1, the electronic device 10 may include: the device comprises a memory 11, a processor 12 and a network module 13, wherein the heart sound acquiring probes 20 are generally provided with five or multiples of five, the five heart sound acquiring probes 20 are used for acquiring heart sound signals of a sample to be detected in five parts, all the five heart sound acquiring probes 20 are connected with the processor 12 through the network module 13, and the processor 12 is connected with the memory 11.
The memory 11, the processor 12 and the network module 13 are electrically connected to each other directly or indirectly to realize data transmission or interaction. For example, the components may be electrically connected to each other via one or more communication buses or signal lines. The memory 11 stores at least one functional module which can be stored in the memory 11 in the form of software or firmware (firmware), and the processor 12 executes various functional applications and data processing by running the functional module stored in the memory 11 in the form of software or hardware, that is, implements the method executed by the electronic device 10 in the present application.
The Memory 11 may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Read Only Memory (EPROM), an electrically Erasable Read Only Memory (EEPROM), a magnetic disk, a solid state disk, or the like. The memory 11 is used for storing a program, and the processor 12 executes the program after receiving an execution instruction.
The processor 12 may be an integrated circuit chip having data processing capabilities. The Processor 12 may be a general-purpose Processor including a Central Processing Unit (CPU), a Network Processor (NP), and the like. The methods, steps, and logic blocks of the present application may be implemented or performed. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, and typically, the processor 12 may be a chip of PCB603C01 with a sensitivity of 100mV/g, a span of ± 50g, a frequency band of 5-10kHz, and a suitable temperature of-54 ℃ to +121 ℃.
The network module 13 is used for establishing a communication connection between the electronic device 10 and an external communication terminal through a network, and implementing transceiving operations of network signals and data. The network signal may include a wireless signal or a wired signal.
It will be appreciated that the configuration shown in FIG. 1 is merely illustrative and that electronic device 10 may include more or fewer components than shown in FIG. 1 or may have a different configuration than shown in FIG. 1. The components shown in fig. 1 may be implemented in hardware, software, or a combination thereof.
On the basis of the above, the present application further provides a computer-readable storage medium, which includes a computer program, and the computer program controls the electronic device 10 to execute the following method when running.
In order to make the implementation of the present invention clearer, the following detailed description is made with reference to the accompanying drawings.
Fig. 2 is a schematic flowchart of a health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application; as shown in fig. 2, the method for health detection based on a five-way heart sound signal classification algorithm provided by the present application includes:
s101, acquiring five paths of heart sound signals of a sample to be detected.
In the clinical diagnosis process, five most frequently used heart sound positions are named as an aortic region, an Erb region, a pulmonary artery region, a tricuspid valve region and a mitral valve region according to the positions which are most easily heard by valves, five heart sound acquisition probes are respectively arranged at the five positions of a sample to be detected to respectively acquire heart sound signals at the five positions, and researches show that the heart sound signals of different auscultation regions contain different physiological information, and the result of multi-channel heart sound signal combined analysis is superior to the result of single-channel heart sound signal combined analysis. In order to more accurately obtain the heart sound signals, the five-path heart sound obtaining probe is arranged near the heart of the sample to be detected for a period of time, and the heart sound signals of the sample to be detected are obtained after the state of the sample to be detected is stable, so that the great errors of the heart sound signals of the sample to be detected caused by activities or other reasons are avoided, the accuracy of detecting the state of the sample to be detected is higher, and the device for obtaining the heart sound signals can be one five-path heart sound obtaining probe or can be used for respectively measuring five positions by using one heart sound obtaining probe.
S102, extracting feature information in the five-channel heart sound signals by using a first preset function.
The characteristic signal of the heart sound signal comprises: when the obtained heart sound signals of the sample to be detected are one-dimensional information, namely linear information, the linear information is analyzed, and the average value and the period number of the heart sound signals of the linear information are extracted, wherein a first preset function of the heart sound signals is an average value extraction function, namely the sum of the amplitudes of the heart sound signals in one period is compared with the time of the last period to obtain the average value of the average amplitudes in the period, the first preset function of the period number is an extraction function, namely the amplitude number in a detection time period, namely the number of the maximum values or the minimum values in the detection time is calculated, and the number of the maximum values or the minimum values in the detection time is the period number. When the acquired heart sound signal of the sample to be detected is two-dimensional information, a waveform picture, a power spectral density map, a mel cepstrum, a spectrogram and an energy spectrum of the heart sound signal are extracted through a first preset function, and the first preset function corresponding to the waveform picture, the power spectral density map, the mel cepstrum, the spectrogram and the energy spectrum is the prior art and is not repeated herein.
For convenience of explanation, a mel-frequency cepstrum obtained by extracting two-dimensional information is taken as an example for simple explanation, and if an original heart sound signal is a sound signal, a two-dimensional signal obtained by STFT expansion is a so-called spectrogram. The spectrogram is often a large image, and in order to obtain a sound feature of a suitable size, it is often transformed into a mel-frequency spectrum by a mel-scale filter bank (mel-scale filters).
The first preset function is a feature extraction algorithm and is used for extracting feature information in a heart sound signal, namely, the input of the first preset function is an original heart sound signal, the output is extracted features, if the first preset function is a two-dimensional feature, the first preset function is a picture type, if the first preset function is a one-dimensional feature, the first preset function is an array type, the algorithm is used for extracting different features determining the state of the heart sound signal, the operation amount is reduced, and in general application, the heart sound signal can be one-dimensional information or two-dimensional information, and the one-dimensional information and the two-dimensional information can be mutually converted, wherein the method for converting the one-dimensional signal into the two-dimensional signal comprises the following steps: framing and windowing the acquired one-dimensional signals, performing Fourier transform (FFT) on each frame, and finally stacking the result of each frame along another dimension to obtain a two-dimensional signal form similar to a graph, so that the conversion from one-dimensional information to two-dimensional information is completed; otherwise, the two-dimensional information is converted into one-dimensional information, and the specific steps are not described herein.
S103, bringing an array corresponding to the characteristic information of the five-path heart sound signal into a preset neural network model for calculation.
Optionally, the step S103 specifically includes:
and converting the characteristic information of the five-path heart sound signals into array information.
The characteristic signals in the five paths of heart sound signals are all converted into array information, the characteristic information at least comprises an average value, a period number, a waveform picture, a power spectral density graph, a Mel cepstrum, a spectrogram and an energy spectrum, the characteristic information is provided with image information, the image information can be converted into the array information after being digitized, and the converted array can represent the characteristic information. Taking a waveform picture as an example, the picture is a multi-dimensional array essentially, a color picture is a three-dimensional array, a black and white picture is a two-dimensional array, and as the waveform picture does not need color information and can be converted into the black and white picture, the waveform picture can be represented by a [ n ] [ n ], and the waveform picture can be converted into the one-dimensional array in various ways, wherein the simplest form is demonstrated, namely, the 2 nd column to the nth column are directly stacked behind the first column to form a [ n ] n.
And bringing the array information into a preset neural network model for calculation.
Because the preset neural network model is:
wherein h (x) is the calculation result of the preset neural network model, and the value range of h (x) is (-1,1), x1，x2、xnAre arrays of input feature information, w1、w2、wnB are parameters, and x is1，x2、xnEach array is one feature information, that is, if the five-way heart sound signal of the sample to be tested is five feature information, the number of the arrays corresponding to the feature information to be input in the preset neural network model should be five, that is, the number of the arrays to be input in the preset neural network model is related to the number of the features to be tested, and the input features to be testedThe more the number of features is, the more accurate the detection result finally using the preset neural network model is; in addition, w1、w2、wnB are parameters in the preset neural network model, before the preset neural network model is used, the value of the parameters in the preset neural network model needs to be obtained, and the parameter w1、w2、wnThe number of the preset neural network model is related to the number of the characteristic information according to the input sample to be tested, and in practical application, the preset neural network model is used for w1、w2、wnB, etc. have been calculated so that the unknowns in the pre-set neural network model are only input x1，x2、xnAnd h (x) which is output, namely the preset neural network model inputs an array corresponding to the characteristic information of the sample to be detected, so that the state information of the sample to be detected can be output, in practical application, a preset threshold value is set, the calculation result of the preset neural network model is compared with the preset threshold value, if the calculation result of the preset neural network model is greater than the preset threshold value, the state of the sample to be detected is represented as a normal state, if the calculation result of the preset neural network model is not greater than the preset threshold value, the state of the sample to be detected is represented as a state needing further detection, specific parameters of the preset threshold value are set according to actual needs, specific limitations are not made, and in addition, b in the preset neural network model parameters is an error parameter.
In practical application, before the step of bringing the array corresponding to the feature information of the five-way heart sound signal into the preset neural network model for calculation, the step of calculating the parameters in the preset neural network model should also be included, and the step of calculating the parameters in the preset neural network model specifically includes:
and acquiring the state information of the multiple groups of known samples and the characteristic information corresponding to the five-way heart sound signals of the multiple groups of known samples.
And bringing the array corresponding to the state information of the multiple groups of known samples and the characteristic information corresponding to the five-way heart sound signals of the multiple groups of known samples into a preset neural network model, and calculating parameters in the preset neural network model.
It should be noted that, in the step of calculating the parameters in the preset neural network model, the more the number of the input arrays corresponding to the state information of the known samples and the feature information corresponding to the five-way heart sound signals of the multiple groups of known samples is, the more accurate the parameters in the preset neural network model are obtained, and since b is an error parameter, the more times the parameters in the preset neural network model are calculated are, the smaller the error parameter is.
Fig. 3 is a schematic diagram of a two-dimensional characteristic log power spectrum of a health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application; as shown in fig. 3, the power spectral density map is also called power spectrum, and in practical application, it is defined as signal power in a unit frequency band. It shows the variation of signal power with frequency, i.e. the distribution of signal power in frequency domain. The power spectrum shows the relation of the signal power changing with the frequency, generally, in practical application, the log power spectrum is used for showing the power spectrum, the log power spectrum is 10 times lg on the basis of the power spectrum, and then the maximum value is subtracted, because the dynamic range of the numerical value obtained by Fourier transform of the signal is very large, but most of the numerical value is concentrated in a low-frequency area, if the numerical value is directly displayed, the display effect is not good, and the logarithm is used for compressing the dynamic range of the numerical value and improving the display effect.
Fig. 4 is a schematic diagram of a two-dimensional characteristic log mel spectrum of a health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application; as shown in fig. 4, the mel-frequency cepstrum is a relatively large feature used in the field of speech signal processing at present, and the sound signal of the heart sound signal is a one-dimensional signal, which can only visually see time domain information but cannot see frequency domain information. The transformation to the frequency domain by Fourier Transform (FT), short-time fourier transform, wavelets, etc. are all very common time-frequency analysis methods. Short-time fourier transform (STFT) is a fourier transform of a short-time signal. The principle is as follows: the method comprises the steps of framing and windowing a long speech signal, performing Fourier transform on each frame, and stacking results of each frame along the other dimension to obtain a graph (similar to a two-dimensional signal), wherein the graph is a spectrogram. Since the obtained spectrogram is large, in order to obtain a sound feature with a proper size, it is usually passed through a Mel-scale filter banks (Mel-scale filter banks) to become a Mel-scale spectrum. And finally, taking a log of the data, namely a log Mel spectrum, wherein the logarithm reason is the same as that of the log power spectrum.
And S104, outputting the state information corresponding to the calculation result obtained by the preset neural network model as a detection result.
Since the value range of h (x) in the preset neural network model is (-1,1), the calculation result of the preset neural network model can only be between-1 and 1, and the state information of the sample to be detected is divided into a normal state and a state to be further detected, i.e. the calculation result obtained by the preset neural network model corresponds to the normal state and the state to be further detected respectively, generally, the calculation result obtained by the preset neural network model is compared with a preset threshold value by setting the preset threshold value according to experimental data, the state of the sample to be detected is distinguished by comparison, in practical application, the threshold value is generally set to 0.5, the calculation result of the preset neural network model is compared with the preset threshold value, if the calculation result is greater than the preset threshold value, the state information corresponding to the calculation result is in the normal state, that is, the state information of the sample to be detected is a normal state, and if the calculation result is not greater than the preset threshold, the state information corresponding to the calculation result is a state requiring further detection, that is, the state information of the sample to be detected is the state requiring further detection.
After each sample to be detected obtains the state information of the sample to be detected by using the method, the state information of the sample to be detected and the characteristic information of the electrocardio information of the sample to be detected are added into the preset neural network model for calculation, and the obtained parameters in the preset neural network model are compared with the parameters used in calculation, so that the parameters in the preset neural network model tend to be accurate, the error parameters approach to 0, and the reliability and the accuracy of the preset database are further improved.
Fig. 5 is a schematic flowchart of another health detection method based on a five-way heart sound signal classification algorithm according to an embodiment of the present application; as shown in fig. 5, optionally, the step of obtaining five heart sound signals of the sample to be measured further includes:
s201, performing digital filtering on the five-path heart sound signals.
The five-path heart sound acquisition probe is used for acquiring a heart sound signal of a sample to be detected, wherein the heart sound signal inevitably has part of noise, so that digital filtering is required to be performed on the heart sound signal, namely impurity sound is filtered out through the digital filtering, a useful heart sound signal is kept as much as possible, common digital filtering comprises multiple types, for convenience of description, a second-order 25hz-600hz Butterworth median filter is taken as an example, data formats before and after algorithm processing are unchanged, the main function is to remove the noise in the signal, the useful signal is kept as much as possible, the processed heart sound signal is purer, interference information is less, and a better auxiliary effect is achieved on later-stage processing.
And S202, performing down-sampling operation according to a preset frequency.
The digital filtered heart sound signal is processed by down-sampling operation, because the audio is subjected to 25hz-600hz median filtering in the steps, the signal is down-sampled to 1000hz according to the nyquist sampling law, the data format before and after algorithm processing is unchanged, but the data volume is correspondingly reduced due to the reduction of the sampling rate so as to reduce the operation amount, and the function of the down-sampling operation is selected according to the actual requirement, which is not limited in detail here. The down-sampling can reduce the data volume on the basis of ensuring enough characteristic information, further reduce the resources required by the operation algorithm and enlarge the application range of the algorithm.
And S203, carrying out maximum value normalization processing on the five-path heart sound signals according to a preset interval.
Since the heart sound signals in different data sets have large differences, all the heart sound signals are normalized, and for convenience of description, the maximum absolute value normalization is taken as an example, so that the range of the normalization is in the range of [ -1,1 ]. The normalized data can reduce the calculation complexity and accelerate the convergence speed in the process of training the model; gradient supersaturation is avoided; the output of the activation function is centered at "zero" for subsequent iterative optimization.
And S204, cutting the five paths of heart sound signals according to the preset length and the preset coincidence degree.
Cutting the audio information of the processed heart sound signal, and in order to enable the processing of the heart sound signal to be more accurate, repeatedly cutting the heart sound signal, for convenience of description, a section of heart sound signal is 200s, the preset length of each section of heart sound signal is 50s, the preset overlap ratio is 50%, and the first section of cut heart sound signal is 0s-50 s; the second end heart sound signal is 25s-75 s; the third end heart sound signal is 50s-100 s; the fourth segment of heart sound signals are 75s-125 s; the fifth segment of heart sound signal is 100s-150 s; the sixth segment of heart sound signals are 125s-175 s; the seventh segment of heart sound signal is 150s-200 s. The data volume can be enlarged under the condition of less samples by cutting the heart sound signal, and the algorithm can obtain better effect when the initial sample volume is less.
Fig. 6 is a schematic block diagram of a health detection system based on a five-way heart sound signal classification algorithm according to an embodiment of the present application; as shown in fig. 6, the present application provides a health detection system based on a five-way heart sound signal classification algorithm, the system includes: the device comprises an acquisition module 30, an extraction module 40, a matching module 50 and an output module 60; the acquisition module 30 is configured to acquire five cardiac sound signals of a sample to be detected, where the five cardiac sound signals are acquired through five cardiac sound probes; the extraction module 40 is configured to extract feature information in the five-channel heart sound signals by using a first preset function; wherein the characteristic information includes: at least one of an average value, a period number, a waveform picture, a power spectral density map, a Mel cepstrum, a spectrogram and an energy spectrum; the calculation module 50 is configured to bring the array corresponding to the feature information of the five-way heart sound signal into a preset neural network model for calculation; the output module 60 is configured to output, as a detection result, state information represented by a calculation result of the preset neural network model.
Optionally, the preset neural network model is:
wherein h (x) is the calculation result of the preset neural network model, and the value range of h (x) is (-1,1), x1，x2、xnAre arrays of input feature information, w1、w2、wnAnd b are parameters.
Optionally, the calculating module 50 is further configured to obtain state information of a plurality of groups of known samples and feature information corresponding to five-way heart sound signals of the plurality of groups of known samples; and bringing the array corresponding to the state information of the multiple groups of known samples and the characteristic information corresponding to the five-way heart sound signals of the multiple groups of known samples into a preset neural network model, and calculating parameters in the preset neural network model.
FIG. 7 is a block diagram of another health detection system based on a five-way heart sound signal classification algorithm according to an embodiment of the present application; as shown in fig. 7, optionally, the step of obtaining five heart sound signals of the sample to be measured further includes: the preprocessing module 70, the preprocessing module 70 performs digital filtering on the five-path heart sound signals; carrying out down-sampling operation according to a preset frequency; carrying out maximum value normalization processing on the five paths of heart sound signals according to a preset interval; and cutting the five paths of heart sound signals according to the preset length and the preset coincidence degree.
Optionally, the calculating module 50 is specifically configured to convert the feature information of the five-way heart sound signal into array information; and bringing the array information into a preset neural network model for calculation.
The application provides an electronic device, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the grain identification method when executing the program.
The application provides a computer-readable storage medium, which includes a computer program, and the computer program controls an electronic device where the computer-readable storage medium is located to execute the grain identification method when running.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A health detection method based on a five-way heart sound signal classification algorithm is characterized by comprising the following steps:
acquiring five paths of heart sound signals of a sample to be detected, wherein the five paths of heart sound signals are acquired through a heart sound probe;
extracting feature information in the five-path heart sound signals by using a first preset function; wherein the feature information includes: at least one of an average value, a period number, a waveform picture, a power spectral density map, a Mel cepstrum, a spectrogram and an energy spectrum;
bringing an array corresponding to the characteristic information of the five-path heart sound signal into a preset neural network model for calculation;
and outputting the state information corresponding to the calculation result obtained by the preset neural network model as a detection result.
2. The health detection method based on five-way heart sound signal classification algorithm according to claim 1, wherein the preset neural network model is:
wherein h (x) is the calculation result of the preset neural network model, and the value range of h (x) is (-1,1), x1，x2、xnAn array, w, corresponding to the characteristic information which is input1、w2、wnAnd b are parameters.
3. The health detection method based on the five-way heart sound signal classification algorithm according to claim 2, wherein before the step of bringing the array corresponding to the feature information of the five-way heart sound signal into a preset neural network model for calculation, the method further comprises:
acquiring state information of a plurality of groups of known samples and characteristic information corresponding to five-path heart sound signals of the plurality of groups of known samples;
and bringing an array corresponding to the state information of a plurality of groups of known samples and the characteristic information corresponding to the five-way heart sound signals of the plurality of groups of known samples into the preset neural network model, and calculating parameters in the preset neural network model.
4. The health detection method based on the five-way heart sound signal classification algorithm according to claim 3, wherein the step of obtaining the five-way heart sound signals of the sample to be detected further comprises the following steps:
performing digital filtering on the five paths of heart sound signals;
carrying out down-sampling operation according to a preset frequency;
carrying out maximum value normalization processing on the five paths of heart sound signals according to a preset interval;
and cutting the five paths of heart sound signals according to a preset length and a preset coincidence degree.
5. The health detection method based on the five-way heart sound signal classification algorithm according to claim 4, wherein the step of bringing the array corresponding to the feature information of the five-way heart sound signal into a preset neural network model for calculation specifically comprises:
converting the characteristic information of the five-path heart sound signals into array information;
and bringing the array information into the preset neural network model for calculation.
6. A health detection system based on a five-way heart sound signal classification algorithm, the system comprising: the device comprises an acquisition module, an extraction module, a calculation module and an output module; the acquisition module is used for acquiring five paths of heart sound signals of a sample to be detected, wherein the five paths of heart sound signals are acquired through a heart sound probe; the extraction module is used for extracting feature information in the five paths of heart sound signals by using a first preset function; wherein the feature information includes: the calculation module is used for bringing an array corresponding to the characteristic information of the five-path heart sound signals into a preset neural network model for calculation; and the output module is used for outputting the state information corresponding to the calculation result obtained by the preset neural network model as a detection result.
7. The five-way heart sound signal classification algorithm-based health detection system according to claim 6, wherein the preset neural network model is:
wherein h (x) is the calculation result of the preset neural network model, and the value range of h (x) is (-1,1), x1，x2、xnAn array, w, corresponding to the characteristic information which is input1、w2、wnAnd b are parameters.
8. The system according to claim 7, wherein the computing module is further configured to obtain state information of a plurality of known samples and feature information corresponding to the five-way heart sound signals of the plurality of known samples; and bringing an array corresponding to the state information of a plurality of groups of known samples and the characteristic information corresponding to the five-way heart sound signals of the plurality of groups of known samples into the preset neural network model, and calculating parameters in the preset neural network model.
9. The five-way heart sound signal classification algorithm-based health detection system according to claim 8, wherein the step of obtaining five-way heart sound signals of the sample to be detected further comprises: the preprocessing module is used for carrying out digital filtering on the five paths of heart sound signals; carrying out down-sampling operation according to a preset frequency; carrying out maximum value normalization processing on the five paths of heart sound signals according to a preset interval; and cutting the five paths of heart sound signals according to a preset length and a preset coincidence degree.
10. The system for health detection based on a five-way heart sound signal classification algorithm according to claim 9, wherein the calculation module is specifically configured to convert the feature information of the five-way heart sound signal into array information; and bringing the array information into the preset neural network model for calculation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110962279.1A CN113768532A (en) | 2021-08-20 | 2021-08-20 | Health detection method and system based on five-path heart sound signal classification algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110962279.1A CN113768532A (en) | 2021-08-20 | 2021-08-20 | Health detection method and system based on five-path heart sound signal classification algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113768532A true CN113768532A (en) | 2021-12-10 |
Family
ID=78838392
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110962279.1A Pending CN113768532A (en) | 2021-08-20 | 2021-08-20 | Health detection method and system based on five-path heart sound signal classification algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113768532A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110755108A (en) * | 2019-11-04 | 2020-02-07 | 合肥望闻健康科技有限公司 | Heart sound classification method, system and device based on intelligent stethoscope and readable storage medium |
| CN111248938A (en) * | 2020-02-24 | 2020-06-09 | 国微集团(深圳)有限公司 | Real-time heart sound signal diagnosis system and diagnosis method thereof |
| CN111759345A (en) * | 2020-08-10 | 2020-10-13 | 北京中科信利技术有限公司 | Heart valve abnormality analysis method, system and device based on convolutional neural network |
| CN112137612A (en) * | 2020-10-27 | 2020-12-29 | 中北大学 | Electrocardio-heart sound synchronous acquisition device and method |
| CN112971802A (en) * | 2021-02-08 | 2021-06-18 | 中北大学 | Heart sound signal detection method and system based on deep learning model |
-
2021
- 2021-08-20 CN CN202110962279.1A patent/CN113768532A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110755108A (en) * | 2019-11-04 | 2020-02-07 | 合肥望闻健康科技有限公司 | Heart sound classification method, system and device based on intelligent stethoscope and readable storage medium |
| CN111248938A (en) * | 2020-02-24 | 2020-06-09 | 国微集团(深圳)有限公司 | Real-time heart sound signal diagnosis system and diagnosis method thereof |
| CN111759345A (en) * | 2020-08-10 | 2020-10-13 | 北京中科信利技术有限公司 | Heart valve abnormality analysis method, system and device based on convolutional neural network |
| CN112137612A (en) * | 2020-10-27 | 2020-12-29 | 中北大学 | Electrocardio-heart sound synchronous acquisition device and method |
| CN112971802A (en) * | 2021-02-08 | 2021-06-18 | 中北大学 | Heart sound signal detection method and system based on deep learning model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ismail et al. | Localization and classification of heart beats in phonocardiography signals—a comprehensive review | |
| Gupta et al. | A critical review of feature extraction techniques for ECG signal analysis | |
| CN109961017A (en) | A kind of cardiechema signals classification method based on convolution loop neural network | |
| CN110123367B (en) | Computer device, heart sound recognition method, model training device, and storage medium | |
| Wang et al. | Phonocardiographic signal analysis method using a modified hidden Markov model | |
| Samieinasab et al. | Fetal phonocardiogram extraction using single channel blind source separation | |
| US8992435B2 (en) | System and method for classifying a heart sound | |
| CN112971802B (en) | Method and system for detecting heart sound signals based on deep learning model | |
| Dokur et al. | Feature determination for heart sounds based on divergence analysis | |
| CN111370120B (en) | Heart diastole dysfunction detection method based on heart sound signals | |
| Kaddoura et al. | Acoustic diagnosis of pulmonary hypertension: automated speech-recognition-inspired classification algorithm outperforms physicians | |
| Kahankova et al. | A review of recent advances and future developments in fetal phonocardiography | |
| Mondal et al. | Boundary estimation of cardiac events S1 and S2 based on Hilbert transform and adaptive thresholding approach | |
| CN110772279A (en) | Lung sound signal acquisition device and analysis method | |
| CN112336369B (en) | Coronary heart disease risk index evaluation system of multichannel heart sound signals | |
| Touahria et al. | Discrete Wavelet based Features for PCG Signal Classification using Hidden Markov Models. | |
| Omarov et al. | Electronic stethoscope for heartbeat abnormality detection | |
| CN115553784B (en) | Coronary heart disease assessment method and system based on electrocardio and heart sound signal coupling analysis | |
| Morshed et al. | Automated heart valve disorder detection based on PDF modeling of formant variation pattern in PCG signal | |
| González–rodríguez et al. | Robust denoising of phonocardiogram signals using time-frequency analysis and U-Nets | |
| CN113768532A (en) | Health detection method and system based on five-path heart sound signal classification algorithm | |
| Balasubramaniam et al. | Efficient computation of phonocardiographic signal analysis in digital signal processor based system | |
| CN105489228A (en) | Rhonchus identification method based on frequency domain image processing | |
| KR101453644B1 (en) | Peak detection method, peak detection apparatus and fetal heart beat detection apparatus | |
| WO2014027962A1 (en) | Device, system and method for detection of fluid accumulation |
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 |