Method, device, electronic device and storage medium for obtaining medical diagnosis information
By inputting the images collected by medical equipment into the image synthesis model, the attenuation degree of the object tissue to the light beam is obtained after processing, the diagnostic error caused by differences in optical hardware parameters is solved and the accuracy of medical diagnosis is improved.
Patent Information
- Application Number
- CN202110791497.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-07-13
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2041-07-13
AI Technical Summary
The system error caused by slight differences in optical hardware parameters during manufacturing of existing medical image imaging equipment affects the accuracy of disease diagnosis.
By obtaining the first medical image collected by the medical device and inputting it into the target image synthesis model for processing, a second medical image describing the degree of beam attenuation of the object tissue is obtained, thereby obtaining more accurate medical diagnostic information and avoiding systematic errors.
It effectively avoids systematic errors in image imaging equipment, improves the accuracy of medical diagnostic information, and helps to more accurately understand the condition of object tissue, thereby facilitating the treatment of diseases.
Smart Images

Figure CN113380381B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of image processing technology, and in particular, relates to a method, device, electronic device and storage medium for obtaining medical diagnosis information. Background Art
[0002] In the medical field, in order to detect whether a certain object tissue has lesions or other problems, the basic principle of weak coherent light interferometer is usually used to emit a light beam to the object tissue through an imaging device (or light source) so that the imaging device can generate a medical image including the object tissue. Then, based on the medical image and the optical hardware parameters of the imaging device, the medical diagnostic information is determined.
[0003] However, since there may be slight differences in optical hardware parameters when imaging devices are manufactured, medical diagnostic information determined based on medical images and optical hardware parameters may also contain errors, which is not conducive to the diagnosis of diseases. Summary of the invention
[0004] The embodiments of the present application provide a method, device, electronic device and storage medium for obtaining medical identification information, which can obtain relatively accurate medical diagnosis information.
[0005] In a first aspect, an embodiment of the present application provides a method for processing an image, comprising:
[0006] Acquire a first medical image acquired by a medical device, wherein the first medical image includes image information of a subject's tissue;
[0007] Inputting the first medical image into a target image synthesis model for processing to obtain a second medical image, wherein the second medical image is used to describe the attenuation degree of the object tissue to the light beam, the light beam being emitted by a light source used to scan the object tissue;
[0008] Medical diagnosis information of the object tissue is acquired according to the second medical image.
[0009] The method for obtaining medical diagnostic information provided in the embodiment of the present application obtains a first medical image collected by a medical device for describing an object tissue, and inputs the first medical image into a target image synthesis model for processing to obtain a second medical image that can be used to describe the attenuation degree of the object tissue with respect to a light beam, thereby effectively avoiding the situation in which a medical image that can more accurately understand the condition of the object tissue cannot be obtained due to system errors of the imaging device, and obtains medical diagnostic information of the object tissue based on the attenuation degree of the object tissue with respect to the light beam recorded in the second medical image, so as to more accurately understand the condition of the object tissue and facilitate the treatment of the disease.
[0010] In a second aspect, an embodiment of the present application provides a device for obtaining medical diagnosis information, including:
[0011] A first acquisition module, configured to acquire a first medical image acquired by a medical device, wherein the first medical image includes image information of a subject's tissue;
[0012] a processing module, configured to input the first medical image into a target image synthesis model for processing to obtain a second medical image, wherein the second medical image is used to describe the attenuation degree of the object tissue to the light beam, wherein the light beam is emitted by a light source for scanning the object tissue;
[0013] The second acquisition module is used to acquire the medical diagnosis information of the object tissue according to the second medical image.
[0014] In a third aspect, an embodiment of the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for obtaining medical diagnostic information described in any one of the first aspects above is implemented.
[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for obtaining medical diagnostic information described in any one of the first aspects above is implemented.
[0016] In a fifth aspect, an embodiment of the present application provides a computer program product. When the computer program product is executed on an electronic device, the electronic device executes the method for obtaining medical diagnosis information described in any one of the above-mentioned first aspects.
[0017] It can be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0019] Figure 1 It is a flowchart of a method for obtaining medical diagnosis information provided in one embodiment of the present application.
[0020] Figure 2This is an example diagram of a first medical image and a second medical image provided by an embodiment of the present application.
[0021] Figure 3 It is a schematic diagram of a specific implementation flow of step S13 of the method for obtaining medical diagnosis information provided in one embodiment of the present application.
[0022] Figure 4 It is a schematic diagram of a specific implementation flow of step S23 of the method for obtaining medical diagnosis information provided in one embodiment of the present application.
[0023] Figure 5 It is a schematic diagram of a specific implementation flow of step S33 of the method for obtaining medical diagnosis information provided in one embodiment of the present application.
[0024] Figure 6 It is a schematic diagram of a generated carpet display diagram provided in an embodiment of the present application.
[0025] Figure 7 It is a flowchart of a method for training a target image synthesis model provided in another embodiment of the present application.
[0026] Figure 8 This is an example diagram of performing forward cycle training provided by another embodiment of the present application.
[0027] Fig. 9 This is an example diagram of reverse cycle training provided by another embodiment of the present application.
[0028] Fig.10 It is a structural diagram of an apparatus for obtaining medical diagnosis information provided by another embodiment of the present application.
[0029] Fig.11 It is a structural schematic diagram of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION
[0030] In order to illustrate the technical solution described in this application, a specific embodiment is provided below for illustration.
[0031] See also Figure 1 , Figure 1 It is a flowchart of an implementation of a method for obtaining medical diagnostic information provided by an embodiment of the present application. In this embodiment, the execution subject of the method for obtaining medical diagnostic information is an electronic device. The electronic device can be a medical device itself for acquiring a first medical image, or it can be other devices other than medical devices. When the electronic device is other devices other than medical devices, data communication can be performed between the electronic device and the medical device to realize data interaction between the two, as well as operations such as control of the medical device.
[0032] The following is an example in which the electronic device is a medical device itself.
[0033] like Figure 1 As shown, a method for obtaining medical diagnosis information provided by an embodiment of the present application includes the following steps:
[0034] S11: Acquire a first medical image acquired by a medical device, where the first medical image includes image information of a subject's tissue.
[0035] In step S11, the medical device is a device capable of performing image acquisition, for example, an optical coherence tomography (OCT) device.
[0036] The first medical image is an image including the object tissue acquired by a medical device after the location of the object tissue is determined. Figure 2 In Figure 2 A is the OCT image of blood vessels in the human body collected by the OCT device.
[0037] The first medical image is used to display the surface condition of the object tissue. For example, the surface condition of the blood vessel is displayed through the OCT image of the human blood vessel, such as the plaque formed on the inner wall surface of the blood vessel due to lipid accumulation is displayed through the first medical image.
[0038] The target tissue may be a biological tissue in a human or animal, for example, a blood vessel or a heart in a human body.
[0039] In this embodiment, a light beam is emitted to the object tissue through the light source of the medical device, so that a first medical image is obtained based on the situation when the light beam irradiates the object tissue. Therefore, the first medical image includes image information of the object tissue, so that the object tissue can be described through the image information of the object tissue. The purpose is to determine the object tissue that needs to be medically diagnosed through the first medical image and provide a basis for data processing based on the first medical image.
[0040] The timing of acquiring the first medical image acquired by the medical device may include but is not limited to the following two scenarios.
[0041] Scenario 1: When it is detected that a medical device has acquired a first medical image, the first medical image is acquired.
[0042] Scenario 2: When the number of first medical images acquired by the medical device reaches a preset image number threshold, the first medical images acquired by the medical device are acquired in the time sequence in which the first medical images are acquired.
[0043] The preset image quantity threshold can be set according to actual needs.
[0044] In some embodiments, in order to improve the efficiency of data processing, when the acquired first medical image is a surface image, a dimensionality reduction process is performed on the first medical image to obtain a plane image of the first medical image.
[0045] Preferably, the surface graph is a three-dimensional graph, and the plane graph is a two-dimensional graph.
[0046] S12: Input the first medical image into the target image synthesis model for processing to obtain a second medical image, where the second medical image is used to describe the attenuation degree of the object tissue to the light beam, which is emitted by a light source used to scan the object tissue.
[0047] In step S12, the attenuation degree is used to describe the situation that when the light beam irradiates the object tissue, the object tissue absorbs part of the energy of the light beam, resulting in the attenuation of the light beam.
[0048] The light source is a light source provided on a medical device, for example, a light source provided in a medical device for emitting X-rays.
[0049] In this embodiment, when the light beam irradiates the object tissue, the object tissue absorbs part of the energy of the light beam, causing the light beam to attenuate. Therefore, based on the attenuation of the light beam at various positions in the object tissue, the condition of the object tissue can be understood. Therefore, in order to understand the attenuation degree of the light beam by the object tissue, after acquiring the first medical image, the first medical image is input into the target image synthesis model for image synthesis processing to obtain a second medical image corresponding to the first medical image and capable of describing the attenuation degree of the light beam by the object tissue, so as to more clearly understand the condition of various positions in the object tissue. Figure 2 The first medical image 2A in is input into the target image synthesis model for processing to obtain the second medical image 2B.
[0050] For example, when there is diseased tissue in the object tissue, the energy of the light beam absorbed by the diseased tissue will be different from that of other parts of the object tissue, thereby obtaining the location of the diseased tissue and obtaining medical diagnostic information of the object tissue.
[0051] It is understandable that, in the application, a pre-trained target image synthesis model is pre-stored in the medical device. The target image synthesis model can be pre-trained by the medical device, or it can be pre-trained by other devices and then the file corresponding to the target image synthesis model is transplanted to the medical device. In other words, the execution subject of training the target image synthesis model and the execution subject of using the target image synthesis model for image synthesis can be the same or different. For example, when other devices are used to train an image synthesis model that has not been trained, after the other devices finish training the image synthesis model that has not been trained, the model parameters of the image synthesis model that has not been trained are fixed, and the file corresponding to the target image synthesis model is obtained, and then the file is transplanted to the medical device.
[0052] S13: Acquire medical diagnosis information of the object tissue according to the second medical image.
[0053] In step S13, the medical diagnosis information is used to describe whether the object tissue has a lesion, and one or more of the information such as the location of the lesion, the size of the lesion area, and the type of the lesion.
[0054] In this embodiment, since the second medical image is used to describe the attenuation degree of the object tissue to the light beam, the attenuation degree of the light beam at different positions in the object tissue is used to understand whether the object tissue has a lesion, thereby obtaining medical diagnostic information of the object tissue.
[0055] The method for obtaining medical diagnostic information provided in the embodiment of the present application obtains a first medical image collected by a medical device for describing an object tissue, and inputs the first medical image into a target image synthesis model for processing to obtain a second medical image that can be used to describe the attenuation degree of the object tissue with respect to a light beam, thereby effectively avoiding the situation in which a medical image that can more accurately understand the condition of the object tissue cannot be obtained due to system errors of the imaging device, and obtains medical diagnostic information of the object tissue based on the attenuation degree of the object tissue with respect to the light beam recorded in the second medical image, thereby facilitating a more accurate understanding of the condition of the object tissue, which is beneficial to the treatment of the disease.
[0056] In some embodiments, after the first medical image is input into the target image synthesis model for processing and the second medical image is obtained, image staining is performed on the second medical image in order to better understand the condition of the object tissue.
[0057] The second medical image is dyed according to the attenuation degree of the light beam by the object tissue described in the second medical image. Specifically, the second medical image is dyed with preset colors according to different light attenuation degrees described.
[0058] For example, a coloring strategy of gradient color from blue to red to yellow is used to color the second medical image describing the blood vessels, wherein the closer to blue, the smaller the attenuation degree and the smaller the lipid plaque content of the blood vessels, and the closer to yellow, the greater the attenuation degree and the greater the lipid plaque content of the blood vessels.
[0059] In some embodiments, in order to facilitate observation of the condition of the object tissue, the first medical image and the second medical image are subjected to image fusion processing to obtain a fused second medical image.
[0060] Specifically, according to corresponding positions of pixels of the first medical image and the second medical image, pixel values in the first medical image are transformed into attenuation degree values, thereby obtaining a fused image.
[0061] Combination Figure 3 In one embodiment of the present application, obtaining medical diagnosis information of the object tissue according to the second medical image includes:
[0062] S21: Performing linear interpolation on coordinate information of pixel points in the second medical image to obtain an interpolated second medical image.
[0063] S22: Projecting the interpolated second medical image to polar coordinates to obtain a third medical image, wherein the third medical image includes multiple columns of attenuation values, each column of attenuation values corresponds to a light beam, and each column of attenuation values is used to indicate the attenuation degree of the object tissue to the light beam corresponding to each column of attenuation values.
[0064] S23: Acquire medical diagnosis information of the object tissue according to the third medical image.
[0065] In this embodiment, the attenuation value is used to describe the attenuation of the light beam at a tissue point of the object tissue in the irradiation direction of the light beam when the light beam irradiates the object tissue. Therefore, it can be understood that the attenuation of the light beam corresponding to each column of the attenuation value of the object tissue can be described by each column of the attenuation value.
[0066] In this embodiment, in order to better process the second medical image, the coordinate information of each pixel in the second medical image is linearly interpolated to obtain the coordinate information of each pixel in the second medical image in polar coordinates, and the interpolated second medical image is obtained based on the coordinate information of each pixel in polar coordinates. Further, the interpolated second medical image is projected to polar coordinates to obtain a third medical image, so as to facilitate better use of the distribution of multiple columns of attenuation values included in the third medical image in polar coordinates, quickly understand whether the attenuation values describing the object tissue are abnormal, or the position of the abnormal attenuation values distributed in the third medical image, or the area occupied by the abnormal attenuation values, and then obtain the medical diagnosis information of the object tissue.
[0067] It can be understood that in order to determine the coordinate information of each pixel point of the second medical image in polar coordinates, for each pixel point, the relationship between the pixel point and the adjacent pixel points is utilized, and the coordinates of the pixel point are interpolated based on a preset difference algorithm to obtain the coordinate information of the interpolated pixel point, and then based on the interpolated coordinate information corresponding to each pixel point, the second medical image is projected to polar coordinates to obtain the third medical image in polar coordinates.
[0068] Among them, since the second medical image is used to describe the attenuation degree of the object tissue to multiple light beams, and each light beam corresponds to multiple tissue points in the object tissue, and each tissue point absorbs the energy of the light beam, the attenuation degree of each tissue point to the light beam will be recorded in the second medical image, and the attenuation degree of the tissue point to the light beam is recorded through the pixel point and the attenuation degree value corresponding to the pixel point, the attenuation degree of the tissue point to the light beam of the object tissue will also be recorded in the interpolated second medical image corresponding to the second medical image. Then, after the third medical image is obtained by projecting the interpolated second medical image, the attenuation degree of the tissue point to the light beam of the object tissue will also be recorded in the third medical image, and since there are multiple light beams, the third medical image will also include multiple columns of attenuation degree values.
[0069] Among them, the preset difference algorithm can be, but is not limited to, one of the nearest neighbor interpolation algorithm, bilinear interpolation algorithm and bicubic interpolation algorithm. The specific implementation process can refer to the implementation process of the relevant interpolation algorithm in the prior art, which will not be repeated here.
[0070] Combination Figure 4 In one embodiment of the present application, obtaining medical diagnosis information of the object tissue according to the third medical image includes:
[0071] S31: Determine a target vector corresponding to the first medical image, where the target vector includes a maximum attenuation value in each column of attenuation values in the third medical image.
[0072] S32: Determine a target ratio corresponding to the first medical image based on the target vector, where the target ratio refers to the ratio of attenuation values in the target vector that are greater than a preset attenuation value.
[0073] S33: Determine the medical diagnosis information of the object tissue according to the target ratio.
[0074] In this embodiment, when the attenuation value is greater than the preset attenuation value, it indicates that there is abnormal tissue in the subject tissue and the attenuation degree of light is abnormal. For example, a diseased tissue point in the subject tissue absorbs more light energy, so that the attenuation degree of the light beam at this tissue point is greater than that of other normal tissue points. The preset attenuation value can be set according to actual needs.
[0075] Among them, the target ratio corresponding to the first medical image is used to describe the ratio of abnormal attenuation degree values. Its purpose is to determine whether the object tissue described by the first medical image is abnormal through the target ratio. For example, whether there is diseased tissue in the object tissue can be understood through the target ratio.
[0076] In this embodiment, in order to improve the efficiency of data processing and determine whether there are abnormal tissue points in the direction irradiated by each light beam irradiating the object tissue, the maximum attenuation value in each column of attenuation values in the third medical image is determined, so as to describe whether each light beam irradiates the abnormal tissue point through each maximum attenuation value, and then obtain a target vector from each maximum attenuation value, and describe the situation of the object tissue corresponding to the first medical image through the target vector. At the same time, in order to determine whether the object tissue described by the first medical image is abnormal, the proportion of attenuation values greater than the preset attenuation value in the target vector is determined, so as to determine the overall situation of the object tissue through the target proportion, that is, whether the object tissue described by the first medical image has a lesion is determined by whether the target proportion is greater than the preset proportion, thereby determining the medical diagnosis information of the object tissue.
[0077] In some embodiments, the target ratio corresponding to the first medical image is calculated by the following formula (1):
[0078]
[0079] Among them, IPA x represents the target ratio corresponding to the first medical image, μ t That is, the value in the target vector, N(μ t >x), indicating that μ is greater than the preset attenuation value x t The total number of N total represents the total number of light beams corresponding to the third medical image, that is, the number of columns included in the third medical image, and M is a preset parameter for amplifying the calculation result.
[0080] Combination Figure 5 , in one embodiment of the present application, the number of the first medical images is multiple;
[0081] Based on the target ratio, determine the medical diagnostic information of the subject tissue, including:
[0082] S41: determining a target value according to a target ratio corresponding to each first medical image in a plurality of first medical images, wherein the target value is used to indicate the number of first medical images in the plurality of first medical images whose corresponding target ratio is greater than a preset ratio.
[0083] S42: When the target value is greater than the preset value, the pathological information of the object tissue is determined, and the medical diagnosis information of the object tissue is obtained.
[0084] In this embodiment, the preset ratio refers to the minimum ratio value for determining whether the object tissue included in the first medical image has an abnormality. It can be understood that when the target ratio is greater than the preset ratio, the attenuation degree of the light beam by the object tissue is also greater than a certain attenuation degree, which means that the object tissue has an abnormality.
[0085] The preset value describes the minimum number of first medical images with a target ratio greater than the preset ratio when the object tissue is diagnosed to have a lesion. For example, the object tissue is determined to have a lesion only when the lesion length is greater than or equal to 4 mm, and since the length of the object tissue corresponding to a first medical image is 0.2 mm, 20 first medical images with a target ratio greater than the preset ratio are required to be combined to determine the existence of a lesion in the object tissue.
[0086] In order to be able to completely and accurately diagnose the location of the lesion in the human body or animal, the number of first medical images acquired by the medical equipment is multiple, and the target value is determined according to the target ratio corresponding to each first medical image in the multiple first medical images. The purpose is to determine the first medical image whose corresponding target ratio is greater than the preset ratio from the multiple first medical images, so as to obtain the lesion information of the object tissue by combining the one or more first medical images whose corresponding target ratio is greater than the preset ratio, thereby obtaining the medical diagnosis information corresponding to the object tissue.
[0087] In some embodiments, in order to better determine the lesion information of the object tissue and obtain the medical diagnosis information of the object tissue, a preset ratio is obtained in advance through a test experiment. For example, multiple test medical image samples are obtained in advance, and the lesion information in the test medical image samples is marked by a medical expert, and then the target ratio corresponding to each test medical image sample is confirmed by changing the reference attenuation value, and each test medical image sample is distinguished by the reference ratio corresponding to the reference attenuation value set each time, so as to understand the distinguishing effect corresponding to the set reference ratio, that is, each test medical sample with recorded lesion information can be distinguished and processed more accurately by the set reference ratio. By analogy, when it is determined that a reference ratio meets the actual needs, the reference ratio is fixed as the preset ratio, and the corresponding reference attenuation value is fixed as the preset attenuation value.
[0088] In some embodiments, the target value is used to indicate the number of consecutive first medical images whose corresponding target ratio is greater than a preset ratio among the plurality of first medical images.
[0089] In an embodiment of the present application, in order to more intuitively display the object organization, this embodiment provides a method for obtaining medical diagnosis information, and the specific implementation method includes:
[0090] A blanket image is generated according to the plurality of the first medical images, where the blanket image is used to display the object tissue, and the blanket image is one of a flat blanket image and a curved blanket image.
[0091] In this embodiment, in order to better display the condition of the object tissue, an attenuation degree value matrix is constructed based on the target vector corresponding to each first medical image, and the attenuation degree value matrix is used as a blanket image to display the medical condition of the object tissue through the blanket image.
[0092] In this embodiment, since the blanket image can be used for the medical condition of the object tissue, and each first medical image in each blanket image records a part of the object tissue, and the first medical image with a target ratio greater than the preset ratio describes the area where the lesion occurs in the object tissue, when the object tissue is diseased, the diseased area in the object tissue can be located based on the position of the first medical image with a target ratio greater than the preset ratio in the blanket image, thereby obtaining the lesion information of the object tissue and the medical diagnosis information of the object tissue.
[0093] It can be understood that the plane carpet display is a two-dimensional carpet display, and the curved carpet display is a three-dimensional curved carpet display. In order to observe the entire object organization more intuitively, the plane carpet display is converted into a three-dimensional curved carpet display.
[0094] Example, combined Figure 6 The target tissue is a blood vessel, and a planar blanket image is generated based on each first medical image describing the blood vessel. Figure 8 A, or three-dimensional curved carpet exhibition Figure 8 B.
[0095] In one example, when it is determined that a lesion exists according to the lesion information, a blanket map is generated.
[0096] In one embodiment, a specific implementation method of generating a blanket image according to the plurality of first medical images includes:
[0097] In a preset order, each of the first medical images is input into a target image synthesis model for processing to obtain a second medical image corresponding to each first medical image.
[0098] Linear interpolation is performed on the coordinate information of the pixel points in the second medical image corresponding to each first medical image to obtain the interpolated second medical image corresponding to each first medical image.
[0099] The interpolated second medical image corresponding to each first medical image is projected to polar coordinates to obtain a third medical image.
[0100] Determine the maximum attenuation value in each column of attenuation values in the third medical image corresponding to each first medical image, and obtain the target vector corresponding to each first medical image based on the maximum attenuation value in each column of attenuation values in the third medical image corresponding to each first medical image.
[0101] According to the target vector corresponding to each first medical image, an attenuation degree value matrix is constructed, and the attenuation degree value matrix is used as a blanket image.
[0102] Combination Figure 7 In one embodiment of the present application, the process of training and obtaining the target image synthesis model includes:
[0103] S51: Acquire multiple groups of training samples, each of the multiple groups of training samples includes a first medical image sample and a second medical image sample, the first medical image sample is obtained by collecting an object tissue sample by a medical device, the second medical image sample is determined based on the first medical image sample and an optical hardware test parameter of the medical device, and the second medical image sample is used to describe the attenuation degree of the object tissue sample to the light beam.
[0104] S52: Based on the multiple groups of training samples, the image synthesis model that has not been trained is trained to obtain a target image synthesis model.
[0105] In this embodiment, the first medical image sample includes image information of a tissue sample of the object.
[0106] The optical hardware test parameters of the medical device are obtained by testing and calculating the medical device that acquires the first medical image sample.
[0107] It is understandable that in order to understand whether there is an abnormality in the object tissue, after the medical image including the image information of the object tissue is acquired by the medical device, it is necessary to manually measure the optical hardware test parameters of the medical device, so as to determine the degree of attenuation of the object tissue to the light beam based on the acquired medical image and the optical hardware test parameters of the medical device, thereby determining the medical diagnostic information of the object tissue. This process is relatively time-consuming and labor-intensive, so multiple groups of training samples are obtained, and the image synthesis model that has not been fully trained is trained based on the multiple groups of training samples to obtain a target image synthesis model. The purpose is to directly synthesize the acquired medical image through the target image synthesis model to obtain a medical image that can describe the degree of attenuation of the object tissue to the light beam, and then obtain the medical diagnostic information of the object tissue, thereby improving the efficiency of data processing, and also preventing the medical diagnostic information determined based on the medical image and the optical hardware parameters from having errors due to the manufacturing process when the medical device is manufactured, which is not conducive to the diagnosis of the disease.
[0108] It is understandable that the first medical image sample in each group of training samples may be acquired based on the same medical device, or may be acquired based on different medical devices.
[0109] In some embodiments, since the hardware parameters of the medical device used to acquire the first medical image may be different, for each group of training samples, the optical hardware test parameters of the medical device used to acquire the first medical image sample in the training samples are calibrated so that each training sample can be distinguished, to facilitate the training operation of the image synthesis model that has not been completed.
[0110] In some embodiments, in practical applications, since the number of first medical image samples with lesion information annotated by experts is relatively small, the amount of training samples can be expanded by rotating, flipping, adjusting contrast, etc. to obtain multiple groups of training samples.
[0111] In one embodiment of the present application, based on multiple groups of training samples, an image synthesis model that has not been trained is trained to obtain a target image synthesis model, including:
[0112] A group of training samples is extracted from multiple groups of training samples according to a specified extraction method to obtain a training sample group.
[0113] Based on the training sample group, a forward loop training is performed on the image synthesis model that has not completed the training to obtain a first training result and a first image loss information. Based on the training sample group, a reverse loop training is performed on the image synthesis model that has not completed the training to obtain a second training result and a second image loss information. The first image loss information is used to indicate the loss information of the image synthesized by the image synthesis model that has not completed the training in the forward loop training, and the second image loss information is used to indicate the loss information of the image synthesized by the image synthesis model that has not completed the training in the reverse loop training.
[0114] If it is determined according to the first training result and the second training result that the training end condition has not been met, the model parameters of the image synthesis model that has not completed the training are adjusted based on the first image loss information and the second image loss information, and the process returns to the step of extracting a group of training samples from multiple groups of training samples according to the specified extraction method to obtain a training sample group.
[0115] Until it is determined according to the first training result and the second training result that the training end condition is met, the image synthesis model obtained through training is determined as the target image synthesis model.
[0116] In this embodiment, the specified extraction method describes a method of using multiple sets of training samples to train an image synthesis model that has not been trained. For example, a set of training samples is extracted from multiple sets of training samples according to the storage order of the training samples or the time order of constructing the training samples to obtain a training sample set.
[0117] The first training result is used to describe whether the training goal of the positive cycle training is achieved.
[0118] The second training result is used to describe whether the training goal of the reverse cycle training is achieved.
[0119] In this embodiment, in the process of performing forward loop training on the image synthesis model that has not been trained based on the training sample group, due to the inaccuracy of the model parameters, a part of the image information will be lost when performing forward loop training based on the medical image samples in the training sample group to obtain the synthesized image, and a part of the image information will be lost when performing reverse loop training to obtain the synthesized image. Therefore, in order to enable the image synthesis model to synthesize a better image, when it is determined according to the first training result and the second training result that the training end condition has not been met, the model parameters of the image synthesis model that has not been trained are continuously adjusted based on the first image loss information and the second image loss information, and a group of training samples are continuously extracted from multiple groups of training samples according to the specified extraction method to obtain a training sample group, so as to continuously adjust the model parameters of the image synthesis model that has not been trained based on the first image loss information and the second image loss information, so that the image synthesis model that has not been trained can better synthesize the input medical image based on the adjusted model parameters, so that the lost image information is reduced, and the synthesized image is more accurate.
[0120] In some embodiments, a group of training samples is extracted from multiple groups of training samples according to a specified extraction method to obtain a training sample group, and based on the training sample group, forward loop training and reverse loop training are performed on the image synthesis model that has not been completed. It is considered that after the forward loop training and reverse loop training of the image synthesis model that has not been completed are completed based on a group of training sample groups, another group of training samples is then extracted, and then the forward loop training and reverse loop training of the image synthesis model that has not been completed are performed until the training end conditions are met.
[0121] It can be understood that when each group of training samples from multiple groups of training samples is extracted according to the specified extraction method and the forward cycle training and reverse cycle training are performed on the incomplete trained image synthesis model respectively, and the training end condition is still not met, the step of extracting one group of training samples from multiple groups of training samples according to the specified extraction method to obtain a training sample group is continued.
[0122] In an embodiment of the present application, based on the training sample group, forward loop training is performed on the image synthesis model that has not been trained yet, and a specific implementation of obtaining the first training result and the first image loss information may include:
[0123] Based on the first medical image sample in the training sample group, a first synthesizer in the image synthesis model that has not completed training is used to determine a first synthesized medical image, and a first discriminator in the image synthesis model that has not completed training is used to determine the similarity between the first synthesized medical image and the second medical image sample in the training sample group to obtain a first training result.
[0124] Based on the first synthesized medical image, a second synthesizer in an image synthesis model that has not completed training is used to determine a second synthesized medical image, and based on the second synthesized medical image and the first medical image sample in the training sample group, the first image loss information is determined.
[0125] In this embodiment, based on the first medical image sample in the training sample group, after the first synthesized medical image is determined by the first synthesizer in the image synthesis model that has not completed training, in order to verify the synthesis accuracy of the first synthesizer, the similarity between the first synthesized medical image and the second medical image sample in the training sample group is determined by the first discriminator in the image synthesis model that has not completed training, and a first training result is obtained, so that the accuracy of the synthesis processing operation of the first synthesizer can be understood through the first training result.
[0126] At the same time, in order to determine the processing defects when the first synthesizer processes the medical image, based on the first synthesized medical image, the second synthesized medical image is determined by the second synthesizer in the image synthesis model that has not completed training, so as to determine the first image loss information based on the second synthesized medical image and the first medical image sample in the training sample group, and describe the image processing defects of the first synthesizer through the first image loss information, so as to adjust the model parameters of the first synthesizer based on the first image loss information, so that the subsequent first synthesizer can better process the medical image input to the first synthesizer based on the adjusted model parameters, that is, the first synthesizer can focus on processing the pixels in the image described by the first image loss information to avoid the loss of this part of the image when obtaining the synthesized medical image.
[0127] For example, see Figure 8 , the first medical image sample (Img OCT ) is input to the first synthesizer (Syn IPA ) is processed to obtain a first synthetic medical image (Syn IPA (Img OCT )), then, through the first discriminator (Dis IPA ) determine the first synthetic medical image (Syn IPA (Img OCT )) and the second medical image sample (Real IPA Img) to obtain the first training result.
[0128] In addition, the second synthesizer (Syn OCT ) for the first synthetic medical image (Syn IPA (Img OCT)) is synthesized to obtain a second synthesized medical image (Syn OCT (Syn IPA (Img OCT ))), and based on the second synthetic medical image (Syn OCT (Syn IPA (Img OCT ))) and the first medical image sample (Img OCT ), determine the first image loss information.
[0129] In some embodiments, the similarity between the first synthetic medical image and the second medical image sample in the training sample group is determined by the following formula (2):
[0130] Loss IPA =(1-Dis IPA (Img IPA )) 2 +Dis IPA (Syn IPA (Img OCT )) 2 (2)
[0131] Among them, Dis IPA () represents the discriminator used to determine whether the input image is real. IPA (Img IPA ) indicates that the first discriminator judges the second medical image sample to determine whether the second medical image sample is true. IPA (Img OCT ) indicates that the first synthesizer in the untrained image synthesis model synthesizes the first medical image sample to obtain a first synthesized medical image, Dis IPA (Syn IPA (Img OCT )) 2 It indicates that the first discriminator judges the first synthetic medical image to determine whether the first synthetic medical image is real. OCT Represents the first medical image sample, Img IPA represents a second medical image sample.
[0132] Example, Dis IPA (Img IPA ) indicates that when the first discriminator judges the second medical image sample, since the image input to the discriminator is the real second medical image sample, Dis IPA The output result of () is 1, which means that the input image is a real image. IPA (Syn IPA (ImgOCT )) 2 Indicates that the first discriminator judges the first synthetic medical image. IPA () of the composite image, if Dis IPA () The output result is 0, indicating that the synthesized image is completely different from the second medical image sample. If the first synthesized medical image obtained by the first synthesizer synthesizing the first medical image sample is not much different from the second medical image sample, then Dis IPA () The output result will be closer to 1, for example, when the synthesis performance of the first synthesizer is better, Dis IPA (Syn IPA (Img OCT )) 2 The corresponding output result is 0.5, so the synthesis performance of the first synthesizer is measured based on the output result.
[0133] It can be understood that, since the first discriminator is used to judge the authenticity of the input image, when the first discriminator judges the real second medical image sample, its output result is 1, and when the first discriminator judges the first synthetic medical image, since the first synthetic medical image is synthesized, and part of the image information may be lost when synthesizing the first synthetic medical image, there may be a certain difference between the first synthetic medical image and the second medical image sample, so the output result of the first discriminator is between 0 and 1. Therefore, Loss IPA It can be used to describe the similarity between the first synthesized medical image and the second medical image sample in the training sample group, and use it as the first training result to characterize the synthesis accuracy of the first synthesizer.
[0134] In an embodiment of the present application, based on the training sample group, reverse loop training is performed on the image synthesis model that has not been trained, and a specific implementation of obtaining the second training result and the second image loss information may include:
[0135] Based on the second medical image sample in the training sample group, determining a third synthesized medical image through a second synthesizer in the image synthesis model that has not completed training, and determining the similarity between the third synthesized medical image and the first medical image sample in the training sample group through a second discriminator in the image synthesis model that has not completed training, to obtain a second training result;
[0136] Based on the third synthesized medical image, a fourth synthesized medical image is determined by the first synthesizer in the image synthesis model that has not completed training, and based on the fourth synthesized medical image and the second medical image sample in the training sample group, the second image loss information is determined.
[0137] In this embodiment, based on the second medical image sample in the training sample group, after the third synthesized medical image is determined by the second synthesizer in the image synthesis model that has not completed training, in order to verify the synthesis accuracy of the second synthesizer, the similarity between the third synthesized medical image and the first medical image sample in the training sample group is determined by the second discriminator in the image synthesis model that has not completed training, and a second training result is obtained, so that the accuracy of the synthesis processing operation of the second synthesizer can be understood through the second training result.
[0138] At the same time, in order to determine the processing defects when the second synthesizer processes the medical image, based on the third synthesized medical image, the fourth synthesized medical image is determined by the first synthesizer in the image synthesis model that has not completed training, so as to determine the second image loss information based on the fourth synthesized medical image and the second medical image sample in the training sample group, and describe the image processing defects of the second synthesizer through the second image loss information, so as to adjust the model parameters of the second synthesizer based on the second image loss information, so that the subsequent second synthesizer can better process the medical image input to the second synthesizer based on the adjusted model parameters, that is, the second synthesizer can focus on processing the pixels in the image described by the second image loss information to avoid the loss of this part of the image when obtaining the synthesized medical image.
[0139] For example, see Fig. 9 , the second medical image sample (Img IPA ) is input to the second synthesizer (Syn OCT ) is processed to obtain a third synthetic medical image (Syn OCT (Img IPA )), then, through the second discriminator (Dis OCT ) determines the third synthetic medical image (Syn OCT (Img IPA )) and the first medical image sample in the training sample group (Real OCT Img) to obtain the second training result.
[0140] In addition, the first synthesizer (Syn IPA ) for the third synthetic medical image (Syn OCT (Img IPA )) is synthesized to obtain a fourth synthesized medical image (Syn IPA (Syn OCT (Img IPA ))), and based on the fourth synthetic medical image (Syn IPA (Syn OCT (ImgIPA ))) and the second medical image sample (Img IPA ), determine the second image loss information.
[0141] In some embodiments, the similarity between the third synthetic medical image and the first medical image sample in the training sample group is determined by the following formula (3):
[0142] Loss OCT =(1-Dis OCT (Img OCT )) 2 +Dis OCT (Syn OCT (Img OCT )) 2 (3)
[0143] Among them, Dis OCT () represents the discriminator used to determine whether the input image is real. OCT (Img OCT ) indicates that the second discriminator judges the first medical image sample to determine whether the first medical image sample is true. OCT (Img OCT ) represents that the second medical image sample is processed by the second synthesizer to obtain a third synthesized medical image, Dis OCT (Syn OCT (Img OCT )) 2 It indicates that the second discriminator judges the third synthesized image to determine whether the third synthesized medical image is real.
[0144] Example, Dis OCT (Img OCT ) represents when the second discriminator judges the first medical image sample, and since the image input to the second discriminator is the real first medical image sample, Dis OCT The output result of () is 1, which means that the input image is a real image. OCT (Syn OCT (Img OCT )) 2 Indicates that the second discriminator judges the third synthetic medical image. However, due to the input of the second discriminator Dis OCT () of the composite image, if Dis OCT () The output result is 0, indicating that the synthesized image is completely different from the first medical image sample. If the third synthesized medical image obtained by the second synthesizer synthesizing the second medical image sample is not much different from the first medical image sample, then Dis OCT() The output result will be closer to 1. For example, when the synthesis performance of the second synthesizer is good, less image information is lost in the process of synthesizing the third synthesized medical image, so that the third synthesized medical image is not much different from the first medical image sample, and at this time Dis OCT (Syn OCT (Img OCT )) 2 The corresponding output result is 0.8, and the synthesis performance of the second synthesizer can be measured based on the output result.
[0145] It can be understood that, since the second discriminator can be used to judge the authenticity of the input image, when the second discriminator judges the real first medical image sample, its output result is 1, and when the second discriminator judges the third synthetic medical image, since the third synthetic medical image is synthesized, and part of the image information may be lost when synthesizing the third synthetic medical image, there may be a certain difference between the third synthetic medical image and the first medical image sample, so the output result of the second discriminator is between 0 and 1. Therefore, Loss OCT It can be used to describe the similarity between the third synthesized medical image and the first medical image sample in the training sample group, and use it as the second training result to characterize the synthesis accuracy of the second synthesizer.
[0146] In some embodiments, in order to determine the training effect when the unfinished image synthesis model is trained based on the training samples, a third training result is further calculated to describe whether the training end condition is met through the third training result.
[0147] Specifically, the third training result is calculated by the following formula (4):
[0148]
[0149] Among them, Syn OCT (Syn IPA (Img OCT ) indicates that the first synthetic medical image is synthesized by the second synthesizer in the untrained image synthesis model to obtain a second synthetic medical image, |Syn OCT (Syn IPA (Img OCT ))-Img OCT | represents determining the first image loss information based on the second synthetic medical image and the first medical image sample in the training sample group, Syn IPA (Syn OCT (Img IPA) indicates that the third synthetic medical image is synthesized by the first synthesizer in the image synthesis model that has not completed training, so as to obtain a fourth synthetic medical image; |Syn IPA (Syn OCT (Img IPA ))-Img IPA | represents determining the second image loss information based on the fourth synthetic medical image and the second medical image sample in the training sample group. represents a preset weight coefficient for the first medical image sample, represents a preset weight coefficient for the second medical image sample.
[0150] It is understandable that when the third training result Loss Cycle When the indicated image loss reaches a preset image loss value, it indicates that the training end condition is reached. Therefore, according to the first training result, the second training result and the third training result, it is determined that the training end condition is reached, and the image synthesis model obtained by training is determined as the target image synthesis model. Among them, the preset image loss value can be preset according to actual needs.
[0151] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0152] Corresponding to the method for obtaining medical diagnosis information in the above embodiment, Fig.10 A structural block diagram of an apparatus for acquiring medical diagnosis information provided in an embodiment of the present application is shown. For ease of explanation, only the portion related to the embodiment of the present application is shown.
[0153] Reference Fig.10 , the device 100 comprises:
[0154] A first acquisition module 101 is used to acquire a first medical image acquired by a medical device, where the first medical image includes image information of a subject's tissue;
[0155] A processing module 102 is used to input the first medical image into the target image synthesis model for processing to obtain a second medical image, where the second medical image is used to describe the attenuation degree of the object tissue to the light beam, where the light beam is emitted by a light source used to scan the object tissue;
[0156] The second acquisition module 103 is used to acquire medical diagnosis information of the object tissue according to the second medical image.
[0157] In one embodiment, the second acquisition module 103 is further used to perform linear interpolation on the coordinate information of the pixel points in the second medical image to obtain an interpolated second medical image; project the interpolated second medical image to polar coordinates to obtain a third medical image, wherein the third medical image includes multiple columns of attenuation values, each column of attenuation values corresponds to a light beam, and each column of attenuation values is used to indicate the attenuation degree of the object tissue to the light beam corresponding to each column of attenuation values; and obtain medical diagnostic information of the object tissue based on the third medical image.
[0158] In one embodiment, the second acquisition module 103 is also used to determine a target vector corresponding to the first medical image, the target vector including the maximum attenuation value in each column of attenuation values in the third medical image; based on the target vector, determine a target ratio corresponding to the first medical image, the target ratio refers to the ratio of attenuation values in the target vector that are greater than a preset attenuation value; and determine the medical diagnostic information of the object tissue based on the target ratio.
[0159] In one embodiment, the number of the first medical images is multiple.
[0160] The second acquisition module 103 is further used to determine a target value based on a target ratio corresponding to each first medical image in a plurality of first medical images, where the target value is used to indicate the number of first medical images in the plurality of first medical images whose corresponding target ratio is greater than a preset ratio; when the target value is greater than the preset value, the lesion information of the object tissue is determined, and the medical diagnosis information of the object tissue is obtained.
[0161] In one embodiment, the device further includes a generating module.
[0162] A generating module is used to generate a blanket image based on the multiple first medical images, wherein the blanket image is used to display the object tissue, and the blanket image includes any one of a plane blanket image and a three-dimensional curved blanket image.
[0163] In one embodiment, the apparatus 100 further includes a training module.
[0164] A training module is used to obtain multiple groups of training samples, each of the multiple groups of training samples includes a first medical image sample and a second medical image sample, the first medical image sample is obtained by collecting an object tissue sample by a medical device, the second medical image sample is determined based on the first medical image sample and the optical hardware test parameters of the medical device, and the second medical image sample is used to describe the attenuation degree of the object tissue sample to the light beam; based on the multiple groups of training samples, an image synthesis model that has not been completed is trained to obtain a target image synthesis model.
[0165] In one embodiment, the training module is further used to extract a group of training samples from multiple groups of training samples in accordance with a specified extraction method to obtain a training sample group; based on the training sample group, forward loop training is performed on the image synthesis model that has not completed training to obtain a first training result and a first image loss information; and based on the training sample group, reverse loop training is performed on the image synthesis model that has not completed training to obtain a second training result and a second image loss information, wherein the first image loss information is used to indicate the loss information of the image synthesized by the image synthesis model that has not completed training in the forward loop training, and the second image loss information is used to indicate the loss information of the image synthesized by the image synthesis model that has not completed training in the reverse loop training; if it is determined according to the first training result and the second training result that the training end condition has not been met, the model parameters of the image synthesis model that has not completed training are adjusted based on the first image loss information and the second image loss information, and the step of extracting a group of training samples from multiple groups of training samples in accordance with the specified extraction method to obtain the training sample group is returned; until it is determined according to the first training result and the second training result that the training end condition has been met, the trained image synthesis model is determined as the target image synthesis model.
[0166] In one embodiment, the training module is also used to determine a first synthetic medical image based on a first medical image sample in the training sample group through a first synthesizer in an image synthesis model that has not completed training, and determine the similarity between the first synthetic medical image and a second medical image sample in the training sample group through a first discriminator in the image synthesis model that has not completed training, so as to obtain a first training result; based on the first synthetic medical image, determine a second synthetic medical image through a second synthesizer in the image synthesis model that has not completed training, and determine first image loss information based on the second synthetic medical image and the first medical image sample in the training sample group.
[0167] In one embodiment, the training module is also used to determine a third synthetic medical image based on the second medical image sample in the training sample group through the second synthesizer in the image synthesis model that has not completed training, and determine the similarity between the third synthetic medical image and the first medical image sample in the training sample group through the second discriminator in the image synthesis model that has not completed training, so as to obtain a second training result; based on the third synthetic medical image, determine a fourth synthetic medical image through the first synthesizer in the image synthesis model that has not completed training, and determine the second image loss information based on the fourth synthetic medical image and the second medical image sample in the training sample group.
[0168] This embodiment provides a device for obtaining medical diagnosis information, which is used to implement a method for obtaining medical diagnosis information in a method embodiment. The functions of each module can refer to the corresponding description in the method embodiment. The implementation principles and technical effects are similar and will not be repeated here.
[0169] Fig.11 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Fig.11 As shown, the electronic device 7 of this embodiment includes: at least one processor 70 ( Fig.11 Only one processor is shown in the figure), a memory 71, and a computer program 72 stored in the memory 71 and executable on at least one processor 70. When the processor 70 executes the computer program 72, the steps in any of the above-mentioned method embodiments for obtaining medical diagnostic information are implemented.
[0170] The electronic device 7 may be a computing device such as a medical device, a desktop computer, a notebook, a PDA, or a cloud server. The electronic device may include, but is not limited to, a processor 70 and a memory 71. Those skilled in the art will appreciate that Fig.11 It is only an example of the electronic device 7 and does not constitute a limitation on the electronic device 7. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.
[0171] The processor 70 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.
[0172] In some embodiments, the memory 71 may be an internal storage unit of the electronic device 7, such as a hard disk or memory of the electronic device 7. In other embodiments, the memory 71 may also be an external storage device of the electronic device 7, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 7. Further, the memory 71 may also include both an internal storage unit and an external storage device of the electronic device 7. The memory 71 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as program codes of a computer program. The memory 71 may also be used to temporarily store data that has been output or is to be output.
[0173] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.
[0174] The embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.
[0175] An embodiment of the present application provides a computer program product. When the computer program product runs on an electronic device, the electronic device can implement the steps in the above-mentioned method embodiments when executing the computer program product.
[0176] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.
Claims
1. A method for obtaining medical diagnosis information, characterized in that: The method comprises: Acquire a first medical image acquired by a medical device, wherein the first medical image includes image information of a subject's tissue, and the medical device is an OCT device; Inputting the first medical image into a target image synthesis model for processing to obtain a second medical image, wherein the second medical image is used to describe the attenuation degree of the object tissue to the light beam, the light beam being emitted by a light source used to scan the object tissue; acquiring medical diagnosis information of the object tissue according to the second medical image; The acquiring medical diagnosis information of the object tissue according to the second medical image includes: Performing linear interpolation on the coordinate information of the pixel points in the second medical image to obtain an interpolated second medical image; Projecting the interpolated second medical image to polar coordinates to obtain a third medical image, wherein the third medical image includes a plurality of columns of attenuation values, each column of attenuation values corresponds to a light beam, and each column of attenuation values is used to indicate the attenuation degree of the object tissue to the light beam corresponding to each column of attenuation values; acquiring medical diagnosis information of the object tissue according to the third medical image; The step of acquiring medical diagnosis information of the object tissue according to the third medical image includes: Determine a target vector corresponding to the first medical image, wherein the target vector includes a maximum attenuation value in each column of attenuation values in the third medical image; Based on the target vector, determining a target ratio corresponding to the first medical image, the target ratio refers to a ratio of attenuation values in the target vector that are greater than a preset attenuation value; determining medical diagnosis information of the subject tissue according to the target ratio; The number of the first medical images is multiple; and determining the medical diagnosis information of the object tissue according to the target ratio includes: Determining a target value according to a target ratio corresponding to each first medical image in the plurality of first medical images, wherein the target value is used to indicate the number of first medical images in the plurality of first medical images whose target ratio is greater than a preset ratio; When the target value is greater than a preset value, the pathological information of the object tissue is determined, and medical diagnosis information of the object tissue is obtained.
2. The method according to claim 1, characterized in that The method further comprises: A blanket image is generated according to the multiple first medical images, where the blanket image is used to display the object tissue, and the blanket image includes any one of a plane blanket image and a curved blanket image.
3. The method according to any one of claims 1 to 2, characterized in that: The method further comprises: Acquire multiple groups of training samples, each of the multiple groups of training samples includes a first medical image sample and a second medical image sample, the first medical image sample is acquired by a medical device from an object tissue sample, the second medical image sample is determined based on the first medical image sample and an optical hardware test parameter of the medical device, and the second medical image sample is used to describe the attenuation degree of the light beam by the object tissue sample; Based on the multiple groups of training samples, the image synthesis model that has not been trained is trained to obtain the target image synthesis model.
4. The method according to claim 3, characterized in that The step of training the image synthesis model that has not been trained based on the multiple groups of training samples to obtain the target image synthesis model includes: Extracting a group of training samples from the multiple groups of training samples according to a specified extraction method to obtain a training sample group; Based on the training sample group, forward loop training is performed on the image synthesis model that has not been trained yet, so as to obtain a first training result and first image loss information; and based on the training sample group, reverse loop training is performed on the image synthesis model that has not been trained yet, so as to obtain a second training result and second image loss information, wherein the first image loss information is used to indicate loss information of images synthesized by the image synthesis model that has not been trained yet in the forward loop training, and the second image loss information is used to indicate loss information of images synthesized by the image synthesis model that has not been trained yet in the reverse loop training; If it is determined according to the first training result and the second training result that the training end condition is not met, adjusting the model parameters of the image synthesis model that has not completed the training based on the first image loss information and the second image loss information, and returning to the step of extracting a group of training samples from the multiple groups of training samples in accordance with the specified extraction method to obtain a training sample group; Until it is determined according to the first training result and the second training result that the training end condition is met, the image synthesis model obtained through training is determined as the target image synthesis model.
5. The method according to claim 4, characterized in that The step of performing forward loop training on the untrained image synthesis model based on the training sample group to obtain a first training result and first image loss information includes: Based on the first medical image sample in the training sample group, determining a first synthesized medical image through a first synthesizer in the image synthesis model that has not been trained, and determining the similarity between the first synthesized medical image and a second medical image sample in the training sample group through a first discriminator in the image synthesis model that has not been trained, to obtain the first training result; Based on the first synthetic medical image, a second synthetic medical image is determined by a second synthesizer in the incompletely trained image synthesis model, and based on the second synthetic medical image and the first medical image sample in the training sample group, the first image loss information is determined.
6. The method according to claim 4, characterized in that The performing reverse loop training on the unfinished image synthesis model based on the training sample group to obtain a second training result and second image loss information includes: Based on the second medical image sample in the training sample group, determine a third synthesized medical image through a second synthesizer in the image synthesis model that has not been trained, and determine the similarity between the third synthesized medical image and the first medical image sample in the training sample group through a second discriminator in the image synthesis model that has not been trained, to obtain the second training result; Based on the third synthesized medical image, a fourth synthesized medical image is determined by the first synthesizer in the incompletely trained image synthesis model, and based on the fourth synthesized medical image and the second medical image sample in the training sample group, the second image loss information is determined.
7. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 6 when executing the computer program.
