Image processing method and device based on medical image processing model

Abstract

The present application provides an image processing method and device based on a medical image processing model, electronic equipment and storage medium, the method comprising: obtaining a standard medical image, converting the standard medical image, determining a medical image training sample set matched with the use environment of the medical image processing model, training a correction network of the medical image processing model, and determining model parameters adapted to the correction network of the medical image processing model. Thus, the corresponding light field background can be predicted by the medical image processing model, and in the test stage, the trained medical image processing model only needs to obtain one medical image to complete the prediction of the light field background, and can complete the correction of the medical image light field based on the predicted light field background.

Description

TECHNICAL FIELD

[0001] The present application relates to medical image processing model training technology, and in particular to an image processing method and device based on a medical image processing model, an electronic device, and a storage medium. BACKGROUND

[0002] In various types of classification based on deep learning, the processing of medical information can be assisted by a neural network model. For example, in an AI+medical scenario, the learning of a large number of medical images by a deployed neural network model can classify and identify the medical images, so as to quickly and accurately obtain relevant medical image analysis results.

[0003] However, in the related art, due to uneven environmental light sources and shaking in the camera imaging process, the obtained medical images usually need to be corrected for light field. In order to obtain clear medical images, it is often necessary to repeatedly collect multiple images of the same target, determine the equations corresponding to the images, and then correct the light field. This not only increases the cost of image storage, but also increases the time for light field correction, which is not conducive to the timely use of medical personnel. SUMMARY

[0004] Therefore, the embodiments of the present application provide an image processing method and device based on a medical image processing model, an electronic device, and a storage medium, which reduce the number of medical image collections, train the medical image processing model only by a standard medical image, predict the corresponding light field background through the medical image processing model, and, in the test use stage of the medical image processing model, predict the light field background through the trained medical image processing model only by obtaining one medical image, correct the medical image light field based on the predicted light field background, and speed up the processing speed of the light field correction.

[0005] The technical solutions of the embodiments of the present application are as follows:

[0006] The embodiments of the present application provide an image processing method based on a medical image processing model, which comprises the following steps:

[0007] Obtaining a standard medical image matched with the medical image processing model;

[0008] Based on the standard medical image, determining a light field change parameter corresponding to the use environment of the medical image processing model;

[0009] Determining a medical image training sample set matched with the use environment of the medical image processing model through the light field change parameter, wherein the medical image training sample set comprises medical images of different light field backgrounds;

[0010] The correction network of the medical image processing model is trained by using a training sample set matched with the use environment of the medical image processing model, model parameters matched with the correction network of the medical image processing model are determined, and the light field corresponding to the medical image in the use environment is predicted by using the medical image processing model.

[0011] A single medical image is acquired, and the corresponding light field information is acquired by processing the medical image by using the medical image processing model.

[0012] The light field of the single medical image is corrected according to the light field information.

[0013] The embodiment of the application provides an image processing device based on a medical image processing model, which comprises:

[0014] An information transmission module is configured to acquire a standard medical image matched with the medical image processing model.

[0015] An information processing module is configured to determine a light field change parameter corresponding to the use environment of the medical image processing model based on the standard medical image.

[0016] The information processing module is configured to determine a medical image training sample set matched with the use environment of the medical image processing model by using the light field change parameter, wherein the medical image training sample set comprises medical images in different light field backgrounds.

[0017] The information processing module is configured to train the correction network of the medical image processing model by using a training sample set matched with the use environment of the medical image processing model, determine model parameters matched with the correction network of the medical image processing model, and predict the light field corresponding to the medical image in the use environment by using the medical image processing model.

[0018] The information processing module is configured to acquire a single medical image, process the medical image by using the medical image processing model, and acquire corresponding light field information.

[0019] The information processing module is configured to correct the light field of the single medical image according to the light field information.

[0020] In the above scheme,

[0021] The information processing module is configured to determine a standard medical image matched with the standard medical image.

[0022] The information processing module is configured to determine the proportional relationship between the pixel points of the standard medical image and the pixel points of the standard medical image based on the standard medical image and the standard medical image.

[0023] The information processing module is configured to determine a light field variation parameter corresponding to a use environment of the medical image processing model based on a proportional relationship between the pixel points of the standard medical image and the pixel points of the standard medical image.

[0024] In the above scheme,

[0025] The information processing module is configured to determine a light field brightness variation range matching the use environment of the medical image processing model.

[0026] Based on the light field brightness variation range, the standard medical image is processed by the light field variation parameter to form a simulation medical image and a light field image matching the simulation medical image.

[0027] The simulation medical image and the light field image matching the simulation medical image are combined to form a medical image training sample set matching the use environment of the medical image processing model.

[0028] In the above scheme,

[0029] The information processing module is configured to determine a target smoothness parameter according to the use environment of the medical image processing model.

[0030] The simulation medical image is adjusted by the target smoothness parameter, and the smoothness of the simulation medical image is adapted to the use environment of the medical image processing model.

[0031] In the above scheme,

[0032] The information processing module is configured to process the training sample set by the encoder and the decoder of the correction network in the medical image processing model to determine initial parameters of the encoder and the decoder of the correction network.

[0033] In response to the initial parameters of the encoder and the decoder of the correction network, the training sample set is processed by the encoder and the decoder of the correction network to determine update parameters of the encoder and the decoder of the correction network.

[0034] According to the update parameters of the encoder and the decoder of the correction network, the parameters of the encoder and the decoder of the correction network are iteratively updated by the training sample set.

[0035] In the above scheme,

[0036] The information processing module is configured to substitute different training samples in the training sample set into a loss function corresponding to the encoder and the decoder of the correction network.

[0037] The encoder and the decoder of the correction network correspond to the updated parameters when it is determined that the loss function satisfies a corresponding convergence condition.

[0038] In the above scheme,

[0039] The information processing module is configured to perform noise reduction processing on the standard medical image by using an image preprocessing network of the medical image processing model.

[0040] The noise-reduced medical image is processed by a convolution layer and a maximum value pooling layer of the image preprocessing network of the medical image processing model to obtain a down-sampling result of the medical image.

[0041] The down-sampling result of the medical image is normalized by a fully connected layer of the image preprocessing network of the medical image processing model.

[0042] The normalized down-sampling result of the medical image is processed by the image preprocessing network of the medical image processing model to determine a light field feature vector matched with the medical image.

[0043] In the above scheme,

[0044] The information processing module is configured to determine a dynamic noise threshold matched with a use environment of the medical image processing model according to a position of a target region corresponding to the medical image.

[0045] The medical image is processed by the image preprocessing network of the medical image processing model according to the dynamic noise threshold to form a medical image matched with the dynamic noise threshold.

[0046] In the above scheme,

[0047] The information processing module is configured to determine a dynamic noise threshold matched with a use environment of the medical image processing model according to an image type of the medical image.

[0048] The medical image is processed by the image preprocessing network of the medical image processing model according to the dynamic noise threshold to form a medical image matched with the dynamic noise threshold.

[0049] In the above scheme,

[0050] The information processing module is configured to display a user interface, and the user interface includes a view picture for observing different positions of a target object from a use environment perspective.

[0051] When a trigger operation of light field correction on a medical image in medical information of a target object is received, display an original medical image collected by a microscope on the user interface;

[0052] Perform light field correction processing on the original medical image collected by the microscope through a medical image processing model to obtain a medical image processed by light field correction processing;

[0053] Present the medical image processed by light field correction processing through the user interface, wherein the medical image processing model is trained based on the method as described above.

[0054] In the above scheme,

[0055] The information processing module is configured to send a model identifier of the medical image processing model, the original medical image, and the medical image processed by light field correction processing to a blockchain network, so that a node of the blockchain network fills the model identifier of the medical image processing model, the original medical image, and the medical image processed by light field correction processing into a new block, and when a consensus is reached on the new block, the new block is appended to the tail of the blockchain.

[0056] The embodiment of the application provides a kind of based on medical image processing model image processing device, the training device includes:

[0057] Memory, for storing executable instructions;

[0058] Processor, for running the executable instructions stored in the memory, realize the image processing method based on medical image processing model provided in the embodiment of the application.

[0059] The embodiment of the application provides a kind of based on medical image processing model image processing device, the based on medical image processing model image processing device includes:

[0060] Memory, for storing executable instructions;

[0061] Processor, for running the executable instructions stored in the memory, realize the image processing method based on medical image processing model of medical image processing model provided in the embodiment of the application.

[0062] The embodiment of the application provides an electronic device, the electronic device includes:

[0063] Memory, for storing executable instructions;

[0064] Processor, for running the executable instructions stored in the memory, realize the image processing method based on medical image processing model of medical image processing model provided in the embodiment of the application.

[0065] The embodiment of the present application provides a computer readable storage medium, which stores executable instructions, and the executable instructions are executed by a processor to implement the image processing method based on a medical image processing model or the image processing method based on a medical image processing model.

[0066] The embodiment of the present application has the following beneficial effects:

[0067] By acquiring a standard medical image matched with the medical image processing model, determining a light field change parameter corresponding to a use environment of the medical image processing model based on the standard medical image, determining a medical image training sample set matched with the use environment of the medical image processing model through the light field change parameter, wherein the medical image training sample set comprises medical images of different light field backgrounds, training a correction network of the medical image processing model through the training sample set matched with the use environment of the medical image processing model, determining a model parameter adapted to the correction network of the medical image processing model, predicting a light field corresponding to a medical image in a use environment through the medical image processing model, acquiring a single medical image, processing the single medical image through the medical image processing model to acquire corresponding light field information, and correcting the light field of the single medical image according to the light field information. Therefore, the medical image processing model can be used to predict the corresponding light field background, and in the test stage, the trained medical image processing model only needs to acquire one medical image to complete the prediction of the light field background, and can complete the correction of the medical image light field based on the predicted light field background, thereby reducing the waiting time of the light field correction. BRIEF DESCRIPTION OF DRAWINGS

[0068] Figure 1 A structure diagram of the medical image processing model provided by the embodiment of the present application is provided.

[0069] Figure 2 A structure diagram of the electronic device provided by the embodiment of the present application is provided.

[0070] Figure 3 An optional structure of a microscope system in the related art of the present application is provided.

[0071] Figure 4 An optional flowchart of the image processing method based on a medical image processing model provided by the embodiment of the present application is provided.

[0072] Figure 5 A diagram of a training sample in the embodiment of the present application is provided.

[0073] Figure 6 A training process diagram of the medical image processing model in the embodiment of the present application is provided.

[0074] Figure 7 A use scenario diagram of the image processing method based on the medical image processing model provided by the embodiment of the present application is shown in the figure.

[0075] Figure 8 An optional flow diagram of the image processing method based on the medical image processing model provided by the embodiment of the present application is shown in the figure.

[0076] Figure 9 A diagram of the training sample set of the cell slide microscope image of human epidermal growth factor receptor-2 in the embodiment of the present application is shown in the figure.

[0077] Figure 10 A correction diagram of the cell slide microscope image of human epidermal growth factor receptor-2 in the embodiment of the present application is shown in the figure.

[0078] Figure 11 A network structure diagram of the medical image processing model provided by the embodiment of the present application is shown in the figure.

[0079] Figure 12 A diagram of presenting the microscope image corrected by the light field in the embodiment of the present application through the display interface is shown in the figure. DETAILED DESCRIPTION

[0080] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings, and the described embodiments should not be regarded as limiting the present application. All other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0081] In the following description, "some embodiments" are described, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0082] Before the embodiments of the present application are further described in detail, the terms and terms involved in the embodiments of the present application are explained, and the terms and terms involved in the embodiments of the present application are applicable to the following explanations.

[0083] 1) based, used to indicate the condition or state on which the operation is performed, when the dependent condition or state is met, one or more operations performed can be real-time or have a set delay; in the absence of special instructions, there is no restriction on the execution order of the multiple operations performed.

[0084] 2) Client, the carrier of specific functions in the terminal, for example, mobile client (APP) is the carrier of specific functions in the mobile terminal, such as the function of performing online live broadcast (video push stream) or the function of playing online video.

[0085] 3) Convolutional Neural Networks (CNN) is a kind of feed forward neural network containing convolution calculation and having deep structure, which is one of the representative algorithms of deep learning. Convolutional Neural Networks has the ability of representation learning, and can perform shift-invariant classification on input information according to its hierarchical structure.

[0086] 4) Model training, multi-classification learning on image data set. The model can be constructed by using deep learning frameworks such as Tensor Flow and torch, and multi-layer combination of neural network layers such as CNN to form a multi-medical image processing model.

[0087] 5) Neural Network (Neural Network, NN): Artificial Neural Network (Artificial Neural Network, ANN), simply called neural network or neural network, in the field of machine learning and cognitive science, is a mathematical model or computational model that simulates the structure and function of biological neural network (animal central nervous system, especially brain), which is used to estimate or approximate function.

[0088] 6) Contrastive loss: contrastive loss function, which can learn a mapping relationship, so that in high-dimensional space, points of the same class but far apart will become closer after mapping to low-dimensional space, and points of different classes but close together will become farther apart in low-dimensional space. The result is that in low-dimensional space, points of the same class will produce clustering effect, and different classes will be separated. Similar to fisher dimension reduction, but fisher dimension reduction does not have the effect of out-of-sample extension, and cannot act on new samples.

[0089] 7) Chinese name computer aided diagnosis (AD Computer Aided Diagnosis), in which CAD is used to find lesions and improve the accuracy of diagnosis by combining imageology, medical image processing model training technology and other possible physiological and biochemical means with computer analysis and calculation.

[0090] 8) Endoscopic video stream: Pathological information of video state formed by image acquisition of body parts (different target organs of the human body or lesions in the body) by image acquisition devices (such as endoscopes).

[0091] 9) Lesion: Lesion generally refers to the part of the body where the lesion occurs. In another expression, a localized lesion with pathogenic microorganisms can be called a lesion.

[0092] The image processing method based on the medical image processing model provided by the present application will be described below by taking the observation of the corresponding lesion cell section by a microscope as an example, referring to Figure 1 , Figure 1 The use scenario diagram of the image processing method based on the medical image processing model provided by the present application is shown in Figure 1 , The terminal (including terminal 10-1 and terminal 10-2) is provided with a client capable of performing different functions, wherein the client is obtained by the terminal (including terminal 10-1 and terminal 10-2) from the corresponding server 200 to browse different slice images, the terminal connects the server 200 through the network 300, and the network 300 can be a wide area network or a local area network, or a combination of the two, and the data transmission is realized by using a wireless link, wherein the slice image type obtained by the terminal (including terminal 10-1 and terminal 10-2) from the corresponding server 200 can be the same or different, for example: the terminal (including terminal 10-1 and terminal 10-2) can obtain the pathological image or pathological video matched with the target object from the corresponding server 200 through the network 300, or can obtain the pathological slice matched with the current target from the corresponding server 200 through the network 300 for browsing. Microscope images in the corresponding light field environment can also be obtained from the microscope 400, and the server 200 can save the slice images corresponding to different target objects respectively, and can save the auxiliary analysis information matched with the slice images of the target object.

[0093] Among them, the neural network model in the field of artificial intelligence deployed by the server can use a camera to collect the image of the sample to be observed on a traditional optical microscope, and combine a machine learning algorithm to analyze the real-time image. Artificial intelligence (AI Artificial Intelligence) is to use digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. Theory, method, technology and application system.

[0094] Specifically, artificial intelligence is a comprehensive technology of computer science, which attempts to understand the essence of intelligence and produce a new intelligent machine that can react similarly to human intelligence. Artificial intelligence is the study of the design principles and implementation methods of various intelligent machines, enabling machines to have perception, reasoning, and decision-making functions. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0095] It should be noted that the patient lesion viewed under the microscope system (a medical device in contact with the pathological cell section of the target object) can include a variety of different application scenarios, such as lung cancer cell screening, early screening of cervical cancer, and different cell section screening. The image processing method based on the medical image processing model of the microscope system based on the embodiment can be deployed to a variety of application scenarios, thereby facilitating remote review and use by doctors.

[0096] The server 200 sends the pathological information of the same target object to the terminal (terminal 10-1 and / or terminal 10-2) through the network 300. The user of the terminal (terminal 10-1 and / or terminal 10-2) analyzes the pathological information of the target object. As an example, the server 200 deploys a corresponding neural network model for analyzing clear image information output by a microscope system. The processing of the medical image by the microscope system can be achieved by the following manner: obtaining a standard medical image matched with a medical image processing model; determining a light field change parameter corresponding to the use environment of the medical image processing model based on the standard medical image; determining a medical image training sample set matched with the use environment of the medical image processing model through the light field change parameter, wherein the medical image training sample set includes medical images in different light field backgrounds; training the correction network of the medical image processing model through the training sample set matched with the use environment of the medical image processing model, determining the model parameter adapted to the correction network of the medical image processing model, and predicting the light field corresponding to the medical image in the use environment through the medical image processing model. In some embodiments of the present application, the medical image can be any medical image affected by light conditions in the imaging process in the medical environment, for example, a set of multi-view pathological pictures obtained by repeatedly observing suspected lesion areas by moving the camera, switching the magnification, etc. during the use of the endoscope by the doctor, which integrates the information of a specific view under the endoscope. Since the endoscope video stream records all the information in the endoscope view during the doctor's observation of the patient's lesion, the information of the doctor's observation of a single patient's lesion in the endoscope view is utilized as a continuous video stream, which avoids the doctor's neglect of small lesion areas during the rapid movement of the endoscope, thereby providing more information than a single frame of picture to assist the doctor in diagnosis and finding small lesion areas.

[0097] In the present application, the cloud technology can be used to realize the present application. The cloud technology refers to a kind of hosting technology that unifies hardware, software and network series resources in wide area network or local area network to realize data calculation, storage, processing and sharing. It can also be understood as a general term of network technology, information technology, integration technology, management platform technology and application technology based on cloud computing business model application. The background service of technical network system needs a lot of calculation and storage resources, such as video website, picture website and more portal website, so the cloud technology needs to support cloud computing.

[0098] It should be noted that cloud computing is a computing mode, which distributes computing tasks on a resource pool formed by a large number of computers, so that various application systems can obtain computing power, storage space and information services according to needs. The network providing resources is called "cloud". The resources in the "cloud" are infinitely expandable to the user and can be obtained at any time, used on demand, expanded at any time, and paid according to use. As a basic capability provider of cloud computing, a cloud computing resource pool platform, referred to as a cloud platform, is generally called Infrastructure as a Service (IaaS), which deploys various types of virtual resources in the resource pool for external customers to select and use. The cloud computing resource pool mainly includes: computing devices (which can be virtual machines containing operating systems), storage devices and network devices.

[0099] The structure of the electronic device of the embodiment of the present application will be described in detail below. The electronic device can be implemented in various forms, such as a special terminal with an endoscopic medical image processing model training function, and can also be an electronic device or a cloud server with an endoscopic medical image processing model training function, such as the server 200 in the foregoing Figure 1 . Figure 2 The schematic diagram of the component structure of the electronic device provided by the embodiment of the present application can be understood as follows: Figure 2 only an exemplary structure of the electronic device is shown, not all structures, and part of the structure or all the structure shown can be implemented according to needs. Figure 2

[0100] The electronic device provided by the embodiment of the present application includes at least one processor 201, a memory 202, a user interface 203 and at least one network interface 204. The various components in the electronic device are coupled together through a bus system 205. It can be understood that the bus system 205 is used to realize the connection and communication between the components. In addition to including a data bus, the bus system 205 also includes a power bus, a control bus and a status signal bus. However, for the purpose of clear illustration, all kinds of buses are marked as the bus system 205 in Figure 2 .

[0101] The user interface 203 can include a display, a keyboard, a mouse, a trackball, a click wheel, a key, a button, a touchpad or a touch screen, etc.

[0102] ​It is to be understood that the memory 202 can be volatile or nonvolatile memory, or both. The memory 202 in the embodiments of the present application is capable of storing data to support the operation of the terminal (e.g., 10-1). Examples of such data include any computer programs for operation on the terminal (e.g., 10-1), such as an operating system and application programs. The operating system includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application programs can include various application programs.

[0103] In some embodiments, the image processing apparatus based on the medical image processing model provided by the embodiments of the present application can be implemented in a combination of software and hardware. For example, the image processing apparatus based on the medical image processing model provided by the embodiments of the present application can be a processor in the form of a hardware decoding processor programmed to perform the image processing method based on the medical image processing model provided by the embodiments of the present application. For example, the processor in the form of a hardware decoding processor can use one or more application specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), or other electronic components.

[0104] As an example of the image processing apparatus based on the medical image processing model provided by the embodiments of the present application implemented in a combination of software and hardware, the image processing apparatus based on the medical image processing model provided by the embodiments of the present application can be directly embodied as a combination of software modules executed by the processor 201. The software modules can be located in a storage medium, and the storage medium is located in the memory 202. The processor 201 reads executable instructions included in the software modules in the memory 202, and in combination with necessary hardware (e.g., including the processor 201 and other components connected to the bus 205), completes the image processing method based on the medical image processing model provided by the embodiments of the present application.

[0105] As an example, the processor 201 can be an integrated circuit chip with signal processing capability, such as a general purpose processor, a digital signal processor (DSP), or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. The general purpose processor can be a microprocessor or any conventional processor.

[0106] As an example of the hardware implementation of the image processing device based on the medical image processing model provided by the embodiments of the present application, the device provided by the embodiments of the present application can be directly implemented by using a hardware decoding processor in the form of a processor 201, for example, one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD), field programmable gate array (FPGA) or other electronic components to implement the image processing method based on the medical image processing model provided by the embodiments of the present application.

[0107] The memory 202 in the embodiments of the present application is used to store various types of data to support the operation of the electronic device. Examples of these data include any executable instructions for operating on the electronic device, such as executable instructions, programs implementing the image processing method based on the medical image processing model provided by the embodiments of the present application can be included in the executable instructions.

[0108] In other embodiments, the image processing device based on the medical image processing model provided by the embodiments of the present application can be implemented in software, Figure 2 The image processing device based on the medical image processing model 2020 stored in the memory 202 can be software in the form of programs and plug-ins, and includes a series of modules. As an example of the program stored in the memory 202, it can include the image processing device based on the medical image processing model 2020, and the image processing device based on the medical image processing model 2020 includes the following software modules:

[0109] The information transmission module 2081 is configured to obtain a standard medical image matched with the medical image processing model;

[0110] The information processing module 2082 is configured to determine a light field change parameter corresponding to the use environment of the medical image processing model based on the standard medical image;

[0111] The information processing module 2082 is configured to determine a medical image training sample set matched with the use environment of the medical image processing model through the light field change parameter, wherein the medical image training sample set includes medical images of different light field backgrounds;

[0112] The information processing module 2082 is configured to train the correction network of the medical image processing model by using a training sample set matched with a use environment of the medical image processing model, determine model parameters matched with the correction network of the medical image processing model, and predict the light field corresponding to the medical image in the use environment by using the medical image processing model.

[0113] The information processing module 2082 is configured to acquire a single medical image, process the single medical image by using the medical image processing model, and acquire light field information corresponding to the single medical image.

[0114] The information processing module 2082 is configured to correct the light field of the single medical image according to the light field information.

[0115] In combination with Figure 2 The image processing apparatus based on the medical image processing model shown in the figure illustrates the image processing method based on the medical image processing model provided in the embodiment of the application. Before introducing the image processing method based on the medical image processing model provided in the application, the process of collecting a medical image by a microscope in the related art is first described, which is described with reference to FIG. 1. Figure 3 , Figure 3 is an optional structure of the microscope system in the related art. In the related embodiments, a microscope 500 is provided, which has a microscope body 301, a microscope body stage focusing knob 302, a microscope body stage 303, a sample to be observed 304, a microscope body objective lens 305, a trinocular tube 306, a camera 307, and an eyepiece 308. The microscope body 301 is provided with the microscope body stage 303 above, the sample to be observed 304 is placed on the microscope body stage 303, the microscope body stage focusing knob 302 is arranged on both sides of the microscope body 301, the microscope body objective lens 305 is located above the microscope body stage 303, the trinocular tube 306 is further arranged above the microscope body objective lens 305, and the trinocular tube 306 is connected with the camera 307 and the eyepiece 308, respectively. Adjusting the microscope body stage focusing knob 302 can adjust the microscope body stage 303 to rise or fall in the vertical direction, so as to change the distance between the microscope body stage 303 and the microscope body objective lens 305 for focusing. Of course, the microscope body objective lens 305 can also be moved to change the distance between the microscope body stage 303 and the microscope body objective lens 305 to achieve focusing.

[0116] In the field of automatic analysis of microscope images, the camera can collect high-quality images, which is the guarantee of the accuracy of the algorithm of the augmented reality microscope. The medical images taken by the camera out of focus or affected by the light source will lose a lot of important optical information, affecting the observation of the doctor. In order to obtain clear medical images as training samples, it is necessary to repeatedly collect multiple images of the same target to correct the light field, which not only increases the shooting cost and image storage cost.

[0117] In order to solve the above problems, see Figure 4 , Figure 4 An optional flowchart of the image processing method based on the medical image processing model provided by the embodiment of the present application can be understood as Figure 4 The image processing method based on the medical image processing model shown in the above can be applied to the field of medical image processing model training, and the segmentation of the medical image is realized, wherein Figure 4 The steps shown in the above can be executed by various electronic devices running the image processing device based on the medical image processing model, such as a special terminal, a server or a server cluster with a medical image processing model training function. The steps shown in the above will be described below. Figure 4

[0118] Step 401: The image processing device based on the medical image processing model acquires a standard medical image matched with the use environment of the medical image processing model.

[0119] Among them, the use and environment in the embodiment can be expressed as the light environment, and the use environment that the light environment changes and affects the accuracy of image acquisition is all tried by the image processing method based on the medical image processing model provided by the present application. Specifically, due to the different use environments of the medical image processing model, the light conditions and the imaging camera are also different, and by collecting standard medical images, the difficulty of acquiring sample data can be effectively reduced.

[0120] Step 402: The image processing device based on the medical image processing model determines the light field change parameter corresponding to the use environment of the medical image processing model based on the standard medical image.

[0121] In some embodiments of the present application, based on the standard medical image, the light field change parameter corresponding to the use environment of the medical image processing model can be determined by the following way:

[0122] ​determine a standard medical image matched with the standard medical image; determine a proportional relationship between a pixel point of the standard medical image and a pixel point of the standard medical image based on the standard medical image and the standard medical image; and determine a light field change parameter corresponding to the use environment of the medical image processing model based on the proportional relationship between the pixel point of the standard medical image and the pixel point of the standard medical image. For example, for the microscope image, the change of the non-uniform brightness and the brightness of the microscope image can be linearly transformed by formula 1:

[0123] I 成像 (x)=I 真实 (x)×S(x)+D(x) Formula 1

[0124] where I 成像 is the non-uniform light field microscope image captured by the imaging mechanism, I 真实 is the ideal brightness uniform image, S is the multiplicative light intensity, D is the additive dark field, and x represents any pixel in the image.

[0125] For the use environment of the microscope image analysis, the change shown in formula 1 can be simplified, and a single light field M is controlled by formula 2:

[0126] I 成像 (x)=I 真实 (x)×M(x) Formula 2

[0127] where M is a uniform background light field, and the transformation between the microscope image and the ideal image can still be represented as a linear relationship. Since each pixel point of the microscope image has a specific linear transformation coefficient, a complete light field image can be formed by combining the pixel points.

[0128] Step 403: The image processing device of the medical image processing model determines a medical image training sample set matched with the use environment of the medical image processing model based on the light field change parameter.

[0129] Reference Figure 5 , Figure 5 is a schematic diagram of the training sample in the embodiment of the application. In some embodiments of the application, the medical image training sample set matched with the use environment of the medical image processing model is determined based on the light field change parameter, which can be realized by the following method:

[0130] determine a light field brightness variation range matching the use environment of the medical image processing model; based on the light field brightness variation range, process the standard medical image through the light field variation parameter to form a simulation medical image and a light field image matching the simulation medical image; and combine the simulation medical image and the light field image matching the simulation medical image to form a medical image training sample set matching the use environment of the medical image processing model. Since the obtained medical image is only a single image, a pair of brightness non-uniform images and their corresponding background light fields need to be constructed as training samples through simulation data, wherein the image simulates a Gaussian distribution of light field, and the brightness range varies from [a, b]. Wherein a and b are parameters of light field brightness variation, and are preferably set to a = 0.5 and b = 1.2 when training the microscope medical image processing model, so that the medical image training sample set includes medical images of different light field backgrounds. Of course, the medical images involved in the present application can be images of different data domains, that is, images of different modalities formed by scanning the human body or a part of the human body by different medical instruments. The medical images obtained from different medical application scenarios belong to different data domains, which can represent that the medical images belong to a certain medical device or a certain imaging modality. For example: the medical image can be a microscope image obtained by an optical microscope, or an electron microscope image obtained by an electron microscope.

[0131] In some embodiments of the present application, in the medical field, the inherent heterogeneity of different symptoms can be reflected in the medical image, such as the existence of different degrees of difference in the appearance (such as shape) of a part of the human body. Therefore, the medical image can be used as a kind of medical judgment means or reference factor for assisting clinical diagnosis. Wherein, the terminal running the medical image processing model can select the corresponding medical image from the image database according to the input image selection instruction; or the terminal and the medical instrument establish a communication connection, such as a wired communication connection or a wireless communication connection, and when the medical instrument forms a medical image by scanning, the medical image formed by the medical instrument is obtained.

[0132] In some embodiments of the present application, the target smoothness parameter can also be determined according to the use environment of the medical image processing model; and the simulation medical image is adjusted through the target smoothness parameter, and the smoothness of the simulation medical image is adapted to the use environment of the medical image processing model. Wherein, the smoothness constraint can be realized by controlling the minimum difference of adjacent pixels of the background light field. For example: if all the pixels correspond to consistent (constant) light field variation, the difference result is 0, and the result is the most smooth. If there is mutation in adjacent pixels, the difference result is too large, and it is not smooth. Therefore, the difference minimization constraint can control the smoothness of the output light field.

[0133] Step 404: The image processing device based on the medical image processing model trains the correction network of the medical image processing model through a training sample set matched with the use environment of the medical image processing model, determines model parameters matched with the correction network of the medical image processing model, and predicts the light field corresponding to the medical image in the use environment through the medical image processing model.

[0134] Step 405: The image processing device based on the medical image processing model acquires a single medical image, processes the single medical image through the medical image processing model, and acquires corresponding light field information.

[0135] Step 406: The image processing device based on the medical image processing model corrects the light field of the single medical image according to the light field information.

[0136] Reference Figure 6 , Figure 6 is a schematic diagram of a training process of a medical image processing model in an embodiment of the present application, and specifically includes the following steps:

[0137] Step 601: The initial parameters of the encoder and the decoder of the correction network are determined by processing the training sample set through the encoder and the decoder of the correction network in the medical image processing model.

[0138] Step 602: The update parameters of the encoder and the decoder of the correction network are determined by processing the training sample set through the encoder and the decoder of the correction network in response to the initial parameters of the encoder and the decoder of the correction network.

[0139] The different training samples in the training sample set can be substituted into the loss function corresponding to the encoder and the decoder of the correction network.

[0140] The update parameters corresponding to the encoder and the decoder of the correction network are determined when the loss function satisfies the corresponding convergence condition.

[0141] Step 603: The parameters of the encoder and the decoder of the correction network are iteratively updated through the training sample set according to the update parameters of the encoder and the decoder of the correction network.

[0142] The different training samples in the training sample set can be substituted into the loss function corresponding to the encoder and the decoder of the correction network when the data of the encoder and the decoder of the correction network is updated.

[0143] In some embodiments of the present application, in order to improve the processing accuracy of the correction network, noise adjustment processing can be performed before processing the medical image. Specifically, the standard medical image can be denoised by the image preprocessing network of the medical image processing model; the medical image after denoising is processed by the convolution layer and the maximum value pooling layer of the image preprocessing network of the medical image processing model to obtain the down-sampling result of the medical image; the down-sampling result of the medical image is normalized by the full connection layer of the image preprocessing network of the medical image processing model; and the normalized down-sampling result of the medical image is decomposed in depth by the image preprocessing network of the medical image processing model to determine the light field feature vector matched with the medical image. The medical image involved in the present application can be a microscope image collected in different light environments, i.e., different modal images formed by scanning human tissue or biological sample tissue by a microscope in different light environments. The medical images obtained in different medical application scenarios belong to different data domains, which can represent that the medical image belongs to a certain medical device or a certain imaging modality. For example, when collecting microscope images, both transmission images and reflection images can reflect the surface information of the target object. Taking human tissue as the target object, when collecting transmission images, due to the different light transmittances of different types of human tissue and different light environments, some types of tissue have poor light transmittance, while some types of tissue have good light transmittance. Therefore, when light signals pass through human tissue, the area where the tissue with poor light transmittance is located transmits less light signal, and the area where the tissue with good light transmittance is located transmits more light signal. Therefore, in the transmission image collected by the image collection device, the darker area is the area where the tissue with poor light transmittance is located in the human tissue, the brighter area is the area where the tissue with good light transmittance is located in the human tissue, and the boundary between the darker area and the brighter area is also obvious. Therefore, the transmission image can more reflect the division of the areas where different types of tissue are located. The terminal running the medical image processing model can select the corresponding medical image from the image database according to the input image selection instruction; or the terminal and the medical instrument establish a communication connection, such as a wired communication connection or a wireless communication connection, and when the medical instrument forms a medical image by scanning, the medical image formed by the medical instrument is obtained.

[0144] In some embodiments of the present application, a dynamic noise threshold value matched with the use environment of the medical image processing model can be determined according to the position of the target region corresponding to the medical image; and the medical image is subjected to noise reduction processing by an image preprocessing network of the medical image processing model according to the dynamic noise threshold value, so as to form a medical image matched with the dynamic noise threshold value. Wherein, the dynamic noise value in different medical images is not the same due to the different lesion positions, and the noise of the medical image will produce different speckle effects, which will affect the observation results under the same light condition, for example, the speckle of the microscope image. The speckle produced by the noise will superimpose the influence of the light, which is not conducive to the accuracy of auxiliary diagnosis. After obtaining the microscope scanning image with noise, speckle recognition needs to be performed on the obtained medical image to determine whether there is speckle in the medical image and the severity value of the speckle. Further, the medical image can be subjected to noise reduction processing by the image information processing network according to the determined dynamic noise threshold value, so as to eliminate the speckle produced by the dynamic noise in the medical image at the lesion position. Wherein, the main reason for the motion speckle is that the noise superimposes the light field influence, the coding and the signal acquisition position or form change during the scanning process of the microscope, so that the phase error occurs and the uneven speckle is produced. The medical image can be subjected to noise reduction processing by the image information processing network according to the determined dynamic noise threshold value, so as to eliminate the speckle produced by the dynamic noise in the microscope image acquisition.

[0145] In some embodiments of the present application, a fixed noise threshold value matched with the use environment of the medical image processing model can also be determined according to the image type of the medical image; and the medical image is subjected to noise reduction processing by an image preprocessing network of the medical image processing model according to the fixed noise threshold value, so as to form a medical image matched with the fixed noise threshold value. Wherein, when the immunohistochemical HER2 (human epidermal growth factor receptor-2) picture is processed by the image shooting of the microscope, 500 images can be acquired under the same light field environment. Since the noise is relatively single, the training speed of the medical image processing model can be effectively improved and the waiting time of the user can be reduced by using the fixed noise threshold value corresponding to the fixed medical image processing model.

[0146] Step 604: deploying the trained medical image processing model in the corresponding server.

[0147] Therefore, the light field corresponding to the medical image in the use environment can be predicted by the medical image processing model.

[0148] Step 605: When receiving a trigger operation of performing light field correction on a medical image in medical information of a target object, display an original medical image collected by a microscope on the user interface.

[0149] Step 606: Process the collected original medical image by a medical image processing model to determine the light field information corresponding to the collected original medical image.

[0150] Step 607: Based on the light field information corresponding to the original medical image, perform light field correction processing on the original medical image collected by the microscope to obtain a medical image after light field correction processing.

[0151] Perform light field correction processing on the original medical image collected by the microscope to obtain a medical image after light field correction processing.

[0152] Wherein, after obtaining the medical image after light field correction processing, the corresponding original medical image and the medical image after light field correction can also be stored by using cloud network technology. Specifically, the embodiment of the application can be implemented in combination with cloud technology. Cloud technology refers to a kind of hosting technology that unifies a series of resources such as hardware, software and network in a wide area network or local area network to realize data calculation, storage, processing and sharing. It can also be understood as a general term for network technology, information technology, integration technology, management platform technology and application technology based on cloud computing business model application. The background service of the technical network system needs a large amount of calculation and storage resources, such as video websites, picture websites and more portal websites, so cloud technology needs to be supported by cloud computing.

[0153] It should be noted that cloud computing is a computing mode that distributes computing tasks on a resource pool composed of a large number of computers, so that various application systems can obtain computing power, storage space and information services according to needs. The network that provides resources is called "cloud". The resources in the "cloud" can be infinitely expanded in the eyes of the user, and can be obtained at any time, used on demand, expanded at any time, and paid according to use. As a basic capability provider of cloud computing, a cloud computing resource pool platform, referred to as a cloud platform, is generally called Infrastructure as a Service (IaaS), which deploys various types of virtual resources in the resource pool for external customers to choose and use. The cloud computing resource pool mainly includes: computing devices (which can be virtualized machines, including operating systems), storage devices and network devices.

[0154] Combination of embodiments Figure 1As shown, the target object determination method provided by the embodiments of the present application can be implemented by a corresponding cloud device, for example: a terminal (including terminal 10-1 and terminal 10-2) connects a server 200 located in the cloud through a network 300, and the network 300 can be a wide area network or a local area network, or a combination of the two. It should be noted that the server 200 can be a physical device, or a virtualized device.

[0155] Specifically, in combination with the preceding embodiment Figure 1 As shown, the server 200 can be a standalone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, and basic cloud computing services such as big data and artificial intelligence platforms. The terminal can be a smartphone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited thereto. The terminal and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in the present application.

[0156] In combination with the preceding Figure 1 As shown, the target object determination method provided by the embodiments of the present application can be implemented by a corresponding cloud device, for example: a terminal (including terminal 10-1 and terminal 10-2) connects a server 200 located in the cloud through a network 300, and the network 300 can be a wide area network or a local area network, or a combination of the two. It should be noted that the server 200 can be a physical device, or a virtualized device.

[0157] In some embodiments of the present application, the target object determination method provided by the present application further comprises:

[0158] receiving a data synchronization request of another node in the blockchain network; in response to the data synchronization request, verifying the authority of the other node; when the authority of the other node is verified, controlling the data synchronization between the current node and the other node, and the other node obtains the model identifier of the medical image processing model, the original medical image, and the medical image processed by the light field correction.

[0159] Further, in some embodiments of the present application, in response to the query request, the query request is parsed to obtain a corresponding object identifier; according to the object identifier, the permission information in the target block in the blockchain network is obtained; the matching of the permission information and the object identifier is checked; when the permission information and the object identifier match, the model identifier of the corresponding medical image processing model, the original medical image and the medical image processed by the light field correction in the blockchain network are obtained; in response to the query instruction, the obtained model identifier of the corresponding medical image processing model, the original medical image and the medical image processed by the light field correction are pushed to the corresponding client, and the client obtains the model identifier of the corresponding medical image processing model, the original medical image and the medical image processed by the light field correction saved in the blockchain network.

[0160] The image light field of the cell slide of human epidermal growth factor receptor-2 is taken as an example for adjustment, and the image processing method based on the medical image processing model provided by the present application is described below,

[0161] Figure 7 The use scene diagram of the image processing method based on the medical image processing model provided by the embodiment of the present application is shown in Figure 7 The terminal 200 can be located in various institutions with medical properties (such as hospitals, medical research institutes), and can be used to collect various types of microscope images or endoscope images (for example, the image collection device of the terminal 200, or the cell slide (i.e. the to-be-processed blood vessel image) of human epidermal growth factor receptor-2 of a patient collected by other medical terminal 400 (such as an image collection device).

[0162] In some embodiments, the terminal 200 locally executes the image processing method based on the medical image processing model provided by the embodiment of the present application to complete the type identification of the cell slide of human epidermal growth factor receptor-2, and the result of the type identification is output in a graphical manner, so that doctors, researchers can diagnose diseases, recheck and research treatment methods, for example, the detection result of the cell slide of human epidermal growth factor receptor-2 can be used to realize the screening of breast cancer, ovarian cancer, endometrial cancer, fallopian tube cancer, gastric cancer and prostate cancer of a patient, determine the morphological performance of different types of human epidermal growth factor receptor-2, and then assist or directly diagnose whether the patient has a risk of cancer disease and lung tumor lesion.

[0163] The terminal 200 can also send the cell microscope image of human epidermal growth factor receptor-2 to the server 100 through the network 300, and call the function of the remote diagnosis service provided by the server 100. The server 100 deploys the trained medical image processing model based on the image processing method of the medical image processing model provided by the embodiment of the application, and obtains the cell microscope image of human epidermal growth factor receptor-2 corrected by light field through the medical image processing model, and performs type identification. The result of type identification is returned to the terminal 200 for doctors, researchers to diagnose diseases, recheck and research treatment methods.

[0164] The terminal 200 can display various intermediate results and final results of medical image processing model training in the graphical interface 210, such as cell slides of human epidermal growth factor receptor-2 and corresponding classification results, etc.

[0165] Reference Figure 8 , Figure 8 An optional flowchart of the image processing method based on the medical image processing model provided by the embodiment of the application is provided, wherein the user can be a doctor, and the cell microscope image of human epidermal growth factor receptor-2 to be corrected is processed, which specifically includes the following steps:

[0166] Step 801: Obtain a standard cell microscope image of human epidermal growth factor receptor-2.

[0167] Step 802: Determine the light field change parameter, and expand the cell microscope image of human epidermal growth factor receptor-2 through the light field change parameter to form a training sample set.

[0168] Wherein, Figure 9 An illustration of the training sample set of the cell microscope image of human epidermal growth factor receptor-2 in the embodiment of the application is provided. The light field brightness change range of the cell microscope image of human epidermal growth factor receptor-2 is determined; based on the light field brightness change range, the single cell microscope image of human epidermal growth factor receptor-2 is expanded through the light field change parameter to form simulation data as a training sample.

[0169] Step 803: Train the correction network of the medical image processing model, and deploy the trained medical image processing model.

[0170] Figure 10 An illustration of the correction of the cell microscope image of human epidermal growth factor receptor-2 in the embodiment of the application is provided. The correction network of the image processing model training model can use U-Net network or LinkNe network. Taking LinkNet as an example, Figure 11The network structure of the medical image processing model provided by the embodiment of the present application intends to adopt a mean square error as a loss function, adopt an Adam method as an optimization method, and set a learning rate to 0.001. In the LinkNet network structure, each encoder (Encoder Block) is connected with a decoder (Decoder Block), and the structure further includes a maximum pooling layer and a full convolution layer. As shown in Figure 11 , a microscope image to be subjected to light field correction is input into the network structure, and first, the input target detection object image is encoded and compressed by the encoder to obtain low-dimensional low-level semantic feature information such as color and brightness. The decoder is connected with the encoder, the low-level semantic feature information output by the encoder is input into the decoder, the decoder performs decoding operation on the low-level semantic feature information, and a feature object segmentation graph with the same size as the target detection object image is output. In the feature object segmentation graph, the white area is the area where the cells are located, and the black area is a light spot generated due to uneven light field distribution. In the LinkNet network structure, the input of the encoder is connected to the output of the corresponding decoder, and before the decoder outputs the feature object segmentation graph, the encoder can integrate the low-level semantic feature information into the decoder, so that the decoder fuses the low-level semantic feature information and high-level semantic feature information, can effectively reduce the loss of spatial information during the down-sampling operation, and the decoder shares the parameters learned from each layer of the encoder, which can effectively reduce the parameters of the decoder

[0171] Further, since the light field changes gradually, a smooth constraint can be added to the light field. The smooth constraint can be realized by controlling the minimum difference of adjacent pixels of the background light field. For example, if the light field changes of all pixels are consistent (constant), the difference result is 0, and the result is the most smooth. If there is a mutation in the adjacent pixels, the difference result is too large, and it is not smooth. Therefore, the difference minimization constraint can control the smoothness of the output light field. The smoothness can be controlled by a coefficient c, and c is set to an empirical value of 0.01.

[0172] Step 804: acquiring any one of the cell slide microscope images of human epidermal growth factor receptor-2 collected by the microscope, and determining the corresponding light field information by the medical image processing model.

[0173] Step 805: based on the acquired light field information, performing light field correction processing on the cell slide microscope image of human epidermal growth factor receptor-2.

[0174] Step 806: presenting the cell slide microscope image of human epidermal growth factor receptor-2 subjected to the light field correction processing on the user interface in the medical device.

[0175] Reference Figure 12 ,Figure 12 For the embodiment of the present application, the microscope image corrected by the light field is presented through the display interface, so that only one cell slide microscope image of human epidermal growth factor receptor-2 needs to be obtained to train and use the medical image processing model, and the cell slide microscope image of human epidermal growth factor receptor-2 in different light field environments can be processed, and the image with uneven brightness is well corrected, and the image that is too dark can also be restored to the brightness corresponding to the common scanning image.

[0176] In summary, the embodiments of the present application have the following beneficial effects:

[0177] By obtaining a standard medical image matched with the medical image processing model, determining a light field variation parameter corresponding to the use environment of the medical image processing model based on the standard medical image, determining a medical image training sample set matched with the use environment of the medical image processing model through the light field variation parameter, wherein the medical image training sample set includes medical images in different light field backgrounds, training the correction network of the medical image processing model through the training sample set matched with the use environment of the medical image processing model, determining model parameters adapted to the correction network of the medical image processing model, predicting the light field corresponding to the medical image in the use environment through the medical image processing model, obtaining a single medical image, processing the medical image through the medical image processing model to obtain corresponding light field information, and correcting the light field of the single medical image according to the light field information. Thus, the corresponding light field background can be predicted through the medical image processing model, and the trained medical image processing model only needs to obtain one medical image to complete the prediction of the light field background during the test use stage of the medical image processing model, and can complete the correction of the medical image light field based on the predicted light field background, thereby accelerating the processing speed of the light field correction.

[0178] The above is only an embodiment of the present application, and is not used to limit the protection scope of the present application, and any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An image processing method based on a medical image processing model, characterized by, The method comprises: acquiring a brightness uniform image matched with a medical image processing model; determining a light field non-uniform image matched with the brightness uniform image, wherein the light field non-uniform image is a product of the brightness uniform image and multiplicative light intensity and a sum of additive dark field; based on the brightness uniform image and the light field non-uniform image, determining a proportional relationship between a pixel point of the brightness uniform image and the light field non-uniform image; based on the proportional relationship, determining a light field change parameter corresponding to a use environment of the medical image processing model; determining a medical image training sample set matched with the use environment of the medical image processing model through the light field change parameter, wherein the medical image training sample set comprises medical images of different light field backgrounds; training a correction network of the medical image processing model through the medical image training sample set matched with the use environment of the medical image processing model, determining a model parameter adapted to the correction network of the medical image processing model, and predicting a light field corresponding to a medical image in a use environment through the medical image processing model; acquiring a single medical image, processing the single medical image through the medical image processing model, and acquiring corresponding light field information; correcting a light field of the single medical image according to the light field information.

2. The method of claim 1, wherein, The determination of the medical image training sample set matched with the use environment of the medical image processing model through the light field change parameter comprises: determining a light field brightness change range matched with the use environment of the medical image processing model; based on the light field brightness change range, processing the brightness uniform image through the light field change parameter to form a simulation medical image and a light field image matched with the simulation medical image; combining the simulation medical image and the light field image matched with the simulation medical image to form the medical image training sample set matched with the use environment of the medical image processing model.

3. The method of claim 2, wherein, The method further comprises: determining a target smoothness parameter according to the use environment of the medical image processing model; adjusting the simulation medical image through the target smoothness parameter, so that the smoothness of the simulation medical image is adapted to the use environment of the medical image processing model.

4. The method of claim 1, wherein, The training of the correction network of the medical image processing model through the medical image training sample set matched with the use environment of the medical image processing model, and the determination of the model parameter adapted to the correction network of the medical image processing model, comprise: processing the medical image training sample set through an encoder and a decoder of the correction network of the medical image processing model to determine initial parameters of the encoder and the decoder of the correction network; in response to the initial parameters of the encoder and the decoder of the correction network, processing the medical image training sample set through the encoder and the decoder of the correction network to determine updated parameters of the encoder and the decoder of the correction network; According to the update parameters of the encoder and the decoder of the correction network, the parameters of the encoder and the decoder of the correction network are iteratively updated through the medical image training sample set.

5. The method of claim 4, wherein, According to the initial parameters of the encoder and the decoder of the correction network, the training sample set is processed through the encoder and the decoder of the correction network to determine the update parameters of the encoder and the decoder of the correction network, comprising: Substitute different training samples in the medical image training sample set into the loss function corresponding to the encoder and the decoder of the correction network; When the loss function satisfies the corresponding convergence condition, the encoder and the decoder of the correction network correspond to the update parameters.

6. The method of claim 1, wherein, The method further comprises: The medical image is denoised by the image preprocessing network of the medical image processing model; The denoised medical image is processed by the convolution layer and the maximum value pooling layer of the image preprocessing network of the medical image processing model to obtain the down-sampling result of the medical image; The down-sampling result of the medical image is normalized by the full connection layer of the image preprocessing network of the medical image processing model; The normalized result of the down-sampling of the medical image is processed by the image preprocessing network of the medical image processing model to determine the light field feature vector matched with the medical image.

7. The method of claim 6, wherein, The medical image is denoised by the image preprocessing network of the medical image processing model, comprising: According to the position of the target area corresponding to the medical image, a dynamic noise threshold value matched with the use environment of the medical image processing model is determined; According to the dynamic noise threshold value, the medical image is denoised by the image preprocessing network of the medical image processing model to form a medical image matched with the dynamic noise threshold value.

8. The method of claim 6, wherein, The medical image is denoised by the image preprocessing network of the medical image processing model, comprising: According to the image type of the medical image, a fixed noise threshold value matched with the use environment of the medical image processing model is determined; According to the fixed noise threshold value, the medical image is denoised by the image preprocessing network of the medical image processing model to form a medical image matched with the fixed noise threshold value.

9. The method of claim 1, wherein, The single medical image is obtained, and the corresponding light field information is obtained by processing the medical image processing model, comprising: When the use environment of the medical image processing model is a microscope for collecting images, an original medical image collected by the microscope is obtained; The light field information corresponding to the original medical image collected by the microscope is determined by processing the original medical image collected by the medical image processing model; Based on the light field information corresponding to the original medical image, the original medical image collected by the microscope is processed by light field correction to obtain a medical image processed by light field correction.

10. The method according to any one of claims 1 to 9, characterized in that, The method further comprises: The model identifier of the medical image processing model, the original medical image, and the medical image processed by the light field correction are sent to the blockchain network, so that the nodes of the blockchain network fill the model identifier of the medical image processing model, the original medical image, and the medical image processed by the light field correction into a new block, and when the new block is consensus, the new block is appended to the tail of the blockchain.

11. An image processing apparatus based on a medical image processing model, characterized by, The device comprises: An information transmission module configured to obtain a brightness uniform image matched with the medical image processing model; An information processing module configured to determine a light field non-uniform image matched with the brightness uniform image, wherein the light field non-uniform image is a product of the brightness uniform image and multiplicative light intensity and a sum of additive dark field; determine a proportional relationship between a pixel point of the brightness uniform image and the light field non-uniform image based on the brightness uniform image and the light field non-uniform image; and determine a light field change parameter corresponding to a use environment of the medical image processing model based on the proportional relationship; The information processing module is configured to determine a medical image training sample set matched with the use environment of the medical image processing model by using the light field change parameter, wherein the medical image training sample set comprises medical images of different light field backgrounds; The information processing module is configured to train a correction network of the medical image processing model by using the medical image training sample set matched with the use environment of the medical image processing model, determine a model parameter adapted to the correction network of the medical image processing model, and predict a light field corresponding to a medical image in a use environment by using the medical image processing model; The information processing module is configured to obtain a single medical image, process the single medical image by using the medical image processing model, and obtain corresponding light field information; The information processing module is configured to correct a light field of the single medical image according to the light field information.

12. The device of claim 11, wherein The information processing module is further configured to determine a light field brightness change range matched with the use environment of the medical image processing model, process the brightness uniform image by using the light field change parameter based on the light field brightness change range, form a simulation medical image and a light field image matched with the simulation medical image, and combine the simulation medical image and the light field image matched with the simulation medical image to form a medical image training sample set matched with the use environment of the medical image processing model.

13. The device of claim 12, wherein The information processing module is further configured to determine a target smoothness parameter according to the use environment of the medical image processing model, and adjust the simulation medical image by using the target smoothness parameter, so that the smoothness of the simulation medical image is adapted to the use environment of the medical image processing model.

14. An electronic device, comprising: The electronic device comprises: A memory configured to store executable instructions; and A processor, when executing the executable instructions stored in the memory, implements the image processing method based on the medical image processing model according to any one of claims 1-10.

15. A computer-readable storage medium storing executable instructions, the method comprising: The executable instructions, when executed by the processor, implement the image processing method based on the medical image processing model according to any one of claims 1-10.

Images