Disclosure of Invention
The embodiment of the invention aims to provide a remote disease intelligent diagnosis system and a remote disease intelligent diagnosis method based on medical images and block chains, and aims to automatically realize noise reduction of the medical images and disease diagnosis. The specific technical scheme is as follows:
in a first aspect of embodiments of the present invention, there is provided a remote disease intelligent diagnosis system based on medical images and blockchains, the system including: the system comprises a user terminal, medical image acquisition equipment, a diagnosis server and a block chain server;
the medical image acquisition equipment is used for continuously acquiring a plurality of medical images aiming at a focus part of a user under the same acquisition environment and sending the plurality of medical images to the user terminal;
the user terminal is used for receiving the medical images, generating a disease diagnosis request and sending the disease diagnosis request to the diagnosis server; wherein the disease diagnosis request includes the plurality of medical images and user information of the user;
the diagnosis server is used for receiving the disease diagnosis request and calling an image denoising unit to denoise a plurality of medical images in the disease diagnosis request to obtain a denoised medical image, and then performing disease prediction based on the denoised medical image to obtain a disease prediction result; the image denoising unit performs denoising processing in the following manner: detecting whether all the multiple medical images are highlight pixels at the pixel point or not aiming at each pixel point, if so, determining the pixel point as the highlight pixel point, and if not, determining the pixel point as the pixel point with the brightness of 0, and finally obtaining a denoised medical image;
the diagnosis server is also used for extracting the user information from the disease diagnosis request, generating a public key query transaction, signing the public key query transaction, and then sending the signed public key query transaction to the block chain server; the public key inquiry transaction comprises the user information;
the block chain server is used for receiving the public key query transaction, performing signature verification on the public key query transaction by using the public key of the diagnosis server, calling a user public key corresponding to the user based on the user information in the public key query request under the condition that the signature verification is passed, and returning the user public key to the diagnosis server;
the diagnosis server is also used for receiving the user public key, encrypting the disease prediction result by using the user public key and sending the encrypted disease prediction result to the user terminal.
In a second aspect of the embodiments of the present invention, there is provided a remote intelligent disease diagnosis method based on medical images and blockchains, applied to a diagnosis server, the method including:
receiving a disease diagnosis request sent by a user terminal; the disease diagnosis request comprises a plurality of medical images and user information, wherein the plurality of medical images are continuously acquired aiming at a focus part of a user in the same acquisition environment;
calling an image denoising unit to denoise a plurality of medical images in the disease diagnosis request to obtain a denoised medical image, and then predicting a disease based on the denoised medical image to obtain a disease prediction result; the image denoising unit performs denoising processing in the following manner: detecting whether all the multiple medical images are highlight pixels at the pixel point or not aiming at each pixel point, if so, determining the pixel point as the highlight pixel point, and if not, determining the pixel point as the pixel point with the brightness of 0, and finally obtaining a denoised medical image;
extracting the user information from the disease diagnosis request, generating a public key query transaction, and signing the public key query transaction, wherein the public key query transaction comprises the user information;
sending the signed public key query transaction to a blockchain server so that the blockchain server performs signature verification by using the public key of the diagnosis server, calling a user public key corresponding to the user based on user information in the public key query request under the condition that the signature verification is passed, and returning the user public key to the diagnosis server;
and receiving a user public key returned by the block chain server, encrypting the disease prediction result by using the user public key, and sending the encrypted disease prediction result to the user terminal.
In the invention, the medical image acquisition equipment continuously acquires a plurality of medical images aiming at the focus part of a user under the same acquisition environment. Because the acquisition environment is the same, a plurality of medical images which are continuously acquired have the same human tissue and lesion tissue. And since the image noise is randomly distributed in each medical image, different image noise exists in each medical image. Therefore, when the image denoising unit denoises, for each pixel point, if all the plurality of medical images are highlight pixel points at the pixel point, the pixel point is probably a human body tissue or a focus tissue, and therefore the pixel point is determined to be the highlight pixel point. And if the plurality of medical images are not all highlight pixel points at the pixel point, the pixel point is probably image noise, and therefore the pixel point is determined as the pixel point with the brightness of 0. Finally, a denoised medical image is obtained. And then, disease prediction is carried out based on the denoised medical image, which is beneficial to improving the accuracy of disease prediction.
In addition, the diagnosis server acquires the user public key from the block chain server, encrypts the disease prediction result by using the user public key, and sends the encrypted disease prediction result to the user terminal. Therefore, the method can prevent the clear text of the disease prediction result from being intercepted by hackers to cause the leakage of the privacy of the user.
In addition, in the invention, the diagnosis server generates a public key query transaction in order to query the user public key from the blockchain server, and signs the public key query transaction by using the private key of the diagnosis server. And after receiving the public key inquiry transaction, the blockchain server performs signature verification on the public key inquiry transaction by using the public key of the diagnosis server. The blockchain server sends the user public key to the diagnostic server only if the signature verification passes. Thus, a hacker can be prevented from impersonating the diagnostic server and illegally obtaining the user public key from the blockchain server.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
With the application and popularization of the mobile internet, more and more users begin to conduct business online by means of mobile terminals. Taking the diagnosis of diseases by a remote medical system as an example, a patient user generally needs to upload medical data (such as medical images) of the patient user to a disease prediction server by using a user terminal, and the disease prediction server makes a corresponding disease prediction according to the medical data uploaded by the user terminal. However, there are many disordered and irregularly distributed image noises in the medical image data, which appear as white spots with fine particles. These image noises greatly interfere with the diagnosis of diseases, and image the accuracy of disease diagnosis, especially when the artificial intelligence technology is used for automatic diagnosis.
In view of this, the present invention provides a remote disease intelligent diagnosis system and a diagnosis method based on medical images and block chains through the following embodiments, aiming to automatically implement noise reduction on medical images and implement disease diagnosis.
Referring to fig. 1, fig. 1 is a system for remote intelligent diagnosis of diseases based on medical images and blockchains according to an embodiment of the present invention. As shown in fig. 1, the diagnostic system includes: the system comprises a user terminal, medical image acquisition equipment, a diagnosis server and a block chain server.
As shown in fig. 1, the medical image capturing device is configured to continuously capture a plurality of medical images for a lesion site of a user in the same capturing environment, and send the plurality of medical images to the user terminal.
It should be noted that, since the medical image capturing apparatus continuously captures a plurality of medical images for the focal region of the user in the same capturing environment, for example, 5 medical images are continuously captured for the focal region of the user within 1 second, each medical image should have the same human body composition and focal tissue. In other words, human tissue is located at the same pixel point in each medical image, and likewise, lesion tissue is also located at the same pixel point in each medical image. Due to the irregularity of the image noise, the image noise does not always exist at the same pixel point of each medical image.
As shown in fig. 1, the user terminal is configured to receive the plurality of medical images, generate a disease diagnosis request, and send the disease diagnosis request to the diagnosis server; wherein the disease diagnosis request includes the plurality of medical images and user information of the user.
As shown in fig. 1, the diagnosis server is configured to receive the disease diagnosis request, and is configured to invoke an image denoising unit to perform denoising processing on a plurality of medical images in the disease diagnosis request to obtain a denoised medical image, and then perform disease prediction based on the denoised medical image to obtain a disease prediction result.
The image denoising unit performs denoising processing in the following manner: and detecting whether all the multiple medical images are highlight pixels at the pixel point or not aiming at each pixel point, if so, determining the pixel point as the highlight pixel point, and if not, determining the pixel point as the pixel point with the brightness of 0, thereby finally obtaining a denoised medical image.
For ease of understanding, a blank image of the same size as the medical image is illustratively first provided. Then, the following steps are executed for each pixel point of the blank image in the sequence from left to right and from top to bottom: firstly, aiming at the first pixel point at the upper left corner, judging whether the brightness of the first pixel point at the upper left corner of all medical images is greater than 50, if so, determining that the pixel point is a human body tissue or a focus tissue, and displaying the pixel point on a blank image in a high-brightness mode. If not, determining that the pixel point is image noise, and then setting the brightness of the pixel point on the blank image to be 0. By analogy, after each pixel point on the blank image is traversed, the image is a denoised medical image.
Optionally, in some embodiments, when performing disease prediction based on the denoised medical image, the diagnosis server may specifically input the denoised medical image into a pre-trained disease prediction model, so as to obtain a disease prediction result output by the disease prediction model.
Wherein the disease prediction model is obtained by training in the following way: collecting a plurality of positive sample medical images with the target disease and a plurality of negative sample medical images without the target disease, wherein each positive sample medical image carries a positive mark and each negative sample medical image carries a negative mark; inputting each positive sample medical image and each negative sample medical image into a preset neural network model, and thus training the neural network model; and using the trained neural network model as a disease prediction model for detecting the target disease.
The disease prediction based on the denoised medical image comprises the following steps: inputting the denoised medical image into a pre-trained disease prediction model so as to obtain a disease prediction result output by the disease prediction model; the disease prediction model can be continuously and dynamically updated to achieve the effect of more accurate diagnosis.
In the training process of the neural network model, the initial weight and the threshold of the neural network are optimized by adopting a particle swarm algorithm, the ITAE performance index is adopted by the particle swarm algorithm as the fitness function of the particle swarm algorithm, the smaller the fitness function value of the particle is, the better the optimization result of the representative particle is, and the particle swarm algorithm adopted in the intelligent control unit is arranged to be updated according to the following mode:
Bi(r+1)=ωi(r)Bi(r)+c1r1(Pi(r)-Xi(r))+c2r2(G(r)-Xi(r))+θ(r)r3(Ki(r)-Xi(r))
Xi(r+1)=Xi(r)+Bi(r+1)
in the formula, Bi(r +1) denotes the step size of particle i at the (r +1) th iteration, Xi(r +1) represents the solution of particle i at the (r +1) th iteration, Bi(r) represents the step size, X, of particle i at the r-th iterationi(r) represents the solution, ω, of particle i at the r-th iterationi(r) represents the inertial weight factor of particle i at the r-th iteration, c1And c2A learning factor, r, representing a particle1、r2And r3Denotes a random number, P, between randomly generated (0,1)i(r) represents the historical optimal solution of the particle i at the r-th iteration, G (r) represents the global optimal solution of the particle swarm at the r-th iteration, theta (r) represents the optimizing adjustment value of the particle swarm at the r-th iteration, Ki(r) represents the solution to the optimization of particle i at the r-th iteration, and θ (r) and KiThe value of (r) is determined in the following manner:
marking particles in a population of particles, given a first optimal distribution detection threshold Q
1(r) and a second optimal distribution detection threshold Q
2(r) and
wherein,
representing the mean value, f, of the fitness function values of the population at the r-th iteration
min(r) a minimum value of the fitness function values for the population at the r-th iteration; let f
i(r) represents the fitness function value of particle i at the r-th iteration, when particle i satisfies f
i(r)≤Q
1(r), label particle i as 1, when particle i satisfies Q
1(r)<f
i(r)≤Q
2(r), label particle i as 2, when particle i satisfies f
i(r)>Q
2(r), then label particle i as 3; defining F (r) to represent the detection coefficient of the optimizing distribution of the particle swarm in the r iteration, wherein the expression of F (r) is as follows:
in the formula, n1(r) denotes at the r-th iterationNumber of particles, n, marked 1 in the population2(r) denotes the number of particles labeled 2 in the population at the r-th iteration, n3(r) represents the number of particles in the population labeled 3 at the r-th iteration, N represents the number of particles in the population;
optimizing distribution detection reference value T of given particle swarm in the r-th iterationF(r), and TFThe expression of (r) is:
wherein N (0) represents an initial value of a given detection reference value of the optimum profile, N (0) is a positive integer, and N (0) < N, γ (r) represents an adjustment coefficient of the corresponding detection reference value of the optimum profile at the r-th iteration of the particle swarm, and
r represents the current number of iterations of the particle swarm, R
maxRepresenting a maximum number of iterations of the particle swarm;
satisfying F (r) ≦ T when the particle population is iterated at the r-th timeF(r), when θ (r) is 0;
satisfying F (r) > T when the particle population is iterated at the r time
FWhen (r), let θ (r) be 1, and when (r) is greater than 1
When it is used, order
When in use
When it is used, order
Wherein, X
i,2(r) represents the solution at the r-th iteration for the particle in the population nearest to particle i and labeled 2, X
i,1(r) represents the solution at the r-th iteration of the particle, G, nearest to particle i and labeled 1, in the population of particles
2(r) represents a solution corresponding to the particle having the smallest fitness function value among the particles labeled 2 in the particle group, G
1(r) represents a solution corresponding to a particle having the smallest fitness function value among the particles marked as 1 in the particle group, ρ
i(r) represents the labeled value of particle i at the r-th iteration, when particle i is labeled as 1 at the r-th iteration, then ρ
i(r) 1, when particle i is labeled 2 at the r-th iteration, then ρ
i(r) 2, when particle i is labeled 3 at the r-th iteration, then ρ
i(r)=3。
The preferred embodiment provides a new updating method of particles in a particle swarm algorithm, additionally introduces a learning item of optimization solution in an updating formula of the traditional particle swarm algorithm, a first optimization distribution detection threshold and a second optimization distribution detection threshold are given to mark the particles in the particle swarm into three types, the first type is the particles with better optimization results, the second type is the particles with better optimization results, the third type is the particles with general optimization results, the particles with better optimization results and the better particles in the particle swarm have higher probability of finding the optimal solution as the particles with better optimization performance in the particle swarm, and also have higher probability of easily falling into a local extreme value in the optimization process The optimization method of the particles can ensure the convergence rate of the particle swarm algorithm and avoid the particle swarm algorithm from falling into a local extreme value, and specifically comprises the following steps: the method comprises the steps that the optimal distribution detection coefficient measures the distribution situation between particles with excellent optimization results in a particle swarm and particles with good optimization results in the particle swarm by counting the difference between the first type of particle number and the second type of particle number, when the value of the optimal distribution detection coefficient is in the range of a given optimal distribution detection reference value, the difference between the first type of particle number and the second type of particle number is shown to be in a given allowable range, namely the particles with good optimization performance in the particle swarm are distributed uniformly and have a smaller probability of falling into a local extreme value, and at the moment, the optimal adjustment value of the particle swarm is made to be, namely the particles in the particle swarm are updated in a traditional mode; when the value of the detection coefficient of the optimizing distribution exceeds a given reference value of the optimizing distribution detection, it indicates that the difference between the first type of particle number and the second type of particle number is large, that is, it indicates that the particles with better optimizing performance in the particle swarm have a phenomenon that part of the particles are distributed too intensively, at this time, the optimizing adjustment value of the particle swarm is made to be, that is, the particles with better optimizing performance in the particle swarm are adjusted, specifically: when the number of the first type particles exceeds the number of the second type particles and is outside the allowable range, when the particles are the second type particles, the optimization adjustment solution of the particles is made to be the solution corresponding to the particles with the minimum fitness function value in the particles marked in the particle swarm, when the particles are not the second type particles, the optimization adjustment solution of the particles is made to be the solution corresponding to the particles which are closest to the optimization adjustment solution and marked in the current iteration, namely, the distribution of the number of the particles between the first type particles and the second type particles is adjusted by additionally adding the solution to the second type particles in the update formula of the particles for learning, when the number of the second type particles exceeds the number of the first type particles and is outside the allowable range, the distribution of the number of the particles between the second type particles and the first type particles is also adjusted by additionally adding the solution to the first type particles in the update formula of the particles according to the logic, the particles with better optimization performance in the particle swarm are ensured to be distributed more uniformly, and the phenomenon that the particles with better optimization performance in the particle swarm are distributed too intensively is avoided, so that the phenomenon that the particle swarm algorithm falls into a local extreme value is avoided, and the optimization performance of the particle swarm algorithm can be effectively improved; in addition, the reference value adjusting coefficient introduced into the reference value of the optimizing distribution detection coefficient is used for increasing the numerical value of the optimizing distribution detection reference value at the later iteration stage of the particle swarm optimization, so that the convergence speed of the particle swarm optimization is guaranteed.
In particular, for the pulmonary tuberculosis disease, the lung images of a plurality of patients with the pulmonary tuberculosis disease can be collected as positive sample medical images and carry positive marks (the positive marks are equal to 1). Images of the lungs of a plurality of persons not suffering from tuberculosis disease are then collected as negative sample medical images and carry a negative marker (negative marker equals 0). Then, for each positive sample medical image, the positive sample medical image is input into a preset BP neural network model, and the BP neural network outputs a decimal between 0 and 1. The decimal is then subtracted from the positive marker (i.e., 1 is used to subtract the decimal) to obtain a loss value, and the loss value is propagated back in the BP neural network model to update the BP neural network model.
Similarly, for each negative sample medical image, it is input into the preset BP neural network model, and the BP neural network outputs a decimal between 0 and 1. And then subtracting the negative mark from the decimal (namely subtracting 0 from the decimal) to obtain a loss value, and then reversely propagating the loss value in the BP neural network model to update the BP neural network model.
And finally, obtaining a trained BP neural network model, and then taking the trained BP neural network model as a disease prediction model for detecting the tuberculosis disease. When the lung disease prediction model is used, the denoised lung medical image is output to the disease prediction model, and if the output of the disease prediction model is greater than 0.5, the user is indicated to have the pulmonary tuberculosis disease. If the disease prediction model outputs a value less than or equal to 0.5, it indicates that the user does not have tuberculosis disease.
As shown in fig. 1, the diagnosis server is further configured to extract the user information from the disease diagnosis request, generate a public key query transaction, sign the public key query transaction, and then send the signed public key query transaction to the blockchain server; and the public key inquiry transaction comprises the user information. When the diagnosis server signs the public key inquiry transaction, the public key of the diagnosis server is used.
As shown in fig. 1, the blockchain server is configured to receive the public key query transaction, perform signature verification on the public key query transaction by using the public key of the diagnostic server, invoke a user public key corresponding to the user based on user information in the public key query request when the signature verification passes, and return the user public key to the diagnostic server.
Optionally, in some embodiments, to further protect the user privacy, the user information included in the disease diagnosis request is a hash value of the user identity. Correspondingly, the account book database of the block chain server stores the hash value of the user identity and the corresponding user public key.
After the block chain server receives the public key query transaction, a hash value of the user identity is extracted from the public key query transaction, and then the user public key corresponding to the hash value is queried by taking the hash value as an index. Therefore, the plaintext of the user identity is not transmitted in the network in the whole disease diagnosis process, so that the privacy of the user is well protected.
As shown in fig. 1, the diagnosis server is further configured to receive the user public key, encrypt the disease prediction result by using the user public key, and send the encrypted disease prediction result to the user terminal.
In the invention, the medical image acquisition equipment continuously acquires a plurality of medical images aiming at the focus part of a user under the same acquisition environment. Because the acquisition environment is the same, a plurality of medical images which are continuously acquired have the same human tissue and lesion tissue. And since the image noise is randomly distributed in each medical image, different image noise exists in each medical image. Therefore, when the image denoising unit denoises, for each pixel point, if all the plurality of medical images are highlight pixel points at the pixel point, the pixel point is probably a human body tissue or a focus tissue, and therefore the pixel point is determined to be the highlight pixel point. And if the plurality of medical images are not all highlight pixel points at the pixel point, the pixel point is probably image noise, and therefore the pixel point is determined as the pixel point with the brightness of 0. Finally, a denoised medical image is obtained. And then, disease prediction is carried out based on the denoised medical image, which is beneficial to improving the accuracy of disease prediction.
In addition, the diagnosis server acquires the user public key from the block chain server, encrypts the disease prediction result by using the user public key, and sends the encrypted disease prediction result to the user terminal. Therefore, the method can prevent the clear text of the disease prediction result from being intercepted by hackers to cause the leakage of the privacy of the user.
Optionally, in some embodiments, in order to further improve the security of disease detection, as shown in fig. 1, the blockchain server is further configured to generate a unique character string after receiving the public key query transaction, and send the character string to the user terminal and the diagnosis server. It should be noted that, in order to further ensure confidentiality, the blockchain server only sends a character string to the diagnostic server and the user terminal, and the character string can be used as signaling for mutual identity confirmation between the user terminal and the diagnostic server.
When the diagnosis server encrypts the disease prediction result, specifically, the diagnosis server firstly splices the character string and the disease prediction result, and then encrypts the spliced character string and the disease prediction result by using the user public key.
And after receiving the encrypted character string and the disease prediction result, the user terminal decrypts by using a user private key to obtain a decryption result. The user terminal also reads the character string from the decryption result, compares the read character string with the character string sent by the block chain server, and if the character string and the character string are consistent, the disease prediction result is from the diagnosis server, so that the user terminal receives the disease prediction result. On the contrary, if the two are not identical, it is indicated that the disease prediction result is not from the diagnosis server, or that the disease prediction result is falsified by a hacker, and thus the user terminal does not accept the disease prediction result.
Preferably, when the blockchain server generates the character string, the time stamp when the public key query transaction is received and the user information included in the public key query request may be used as input data, a hash algorithm model is input to obtain a hash value, and then the hash value is used as the unique character string.
Or preferably, the blockchain server may randomly generate a character string when generating the character string.
Based on the same inventive concept, the invention provides a remote disease intelligent diagnosis method based on medical images and block chains. Referring to fig. 2, fig. 2 is a flowchart of a remote disease intelligent diagnosis method based on medical images and blockchains, which is applied to a diagnosis server according to an embodiment of the present invention. It should be noted that the diagnostic method shown in fig. 2 can be cross-referenced with the diagnostic system shown in fig. 1.
As shown in fig. 2, the diagnosis method includes the steps of:
step S21: receiving a disease diagnosis request sent by a user terminal; the disease diagnosis request comprises a plurality of medical images and user information, wherein the plurality of medical images are continuously acquired aiming at the focus part of the user in the same acquisition environment.
Step S22: calling an image denoising unit to denoise a plurality of medical images in the disease diagnosis request to obtain a denoised medical image, and then predicting a disease based on the denoised medical image to obtain a disease prediction result; the image denoising unit performs denoising processing in the following manner: and detecting whether all the multiple medical images are highlight pixels at the pixel point or not aiming at each pixel point, if so, determining the pixel point as the highlight pixel point, and if not, determining the pixel point as the pixel point with the brightness of 0, thereby finally obtaining a denoised medical image.
Step S23: and extracting the user information from the disease diagnosis request, generating a public key query transaction, and signing the public key query transaction, wherein the public key query transaction comprises the user information.
Step S24: and sending the signed public key inquiry transaction to a blockchain server so that the blockchain server performs signature verification by using the public key of the diagnosis server, calling a user public key corresponding to the user based on the user information in the public key inquiry request under the condition that the signature verification is passed, and returning the user public key to the diagnosis server.
Step S25: and receiving a user public key returned by the block chain server, encrypting the disease prediction result by using the user public key, and sending the encrypted disease prediction result to the user terminal.
Optionally, in some embodiments, the method further comprises the steps of: and receiving a character string with uniqueness, which is sent by the blockchain server.
When the diagnosis server executes the step S25, specifically, the diagnosis server concatenates the character string and the disease prediction result, and encrypts the concatenated character string and the disease prediction result by using the user public key.
Optionally, in some embodiments, the character string is generated by: and the block chain server inputs a hash algorithm model by taking the timestamp when the public key query transaction is received and the user information contained in the public key query request as input data to obtain a hash value, and then takes the hash value as the unique character string.
Optionally, in some specific embodiments, the user information included in the disease diagnosis request is a hash value of a user identity, and the ledger database of the blockchain server stores the hash value of the user identity and a corresponding user public key.
Optionally, in some specific embodiments, when executing the step S22, the diagnosis server specifically inputs the denoised medical image into a pre-trained disease prediction model, so as to obtain a disease prediction result output by the disease prediction model.
The diagnostic server also trains the disease prediction model by performing the following steps: collecting a plurality of positive sample medical images with target diseases and a plurality of negative sample medical images without the target diseases in advance, wherein each positive sample medical image carries a positive mark, and each negative sample medical image carries a negative mark; inputting each positive sample medical image and each negative sample medical image into a preset neural network model, and thus training the neural network model; and using the trained neural network model as a disease prediction model for detecting the target disease.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.