Intelligent analysis method and system for breast disease medical images

By combining management servers and generating adversarial models, the optimal shooting perspective for patients with breast disease is determined, which solves the problem of insufficient or many breast disease images in the prior art, and improves the efficiency and accuracy of breast disease diagnosis.

CN118675715BActive Publication Date: 2024-11-22NANJING GNOS BIOTECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411159401.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-22
Estimated Expiration
2044-08-22

AI Technical Summary

Technical Problem

The existing breast disease medical imaging methods cannot choose the best shooting angle based on the patient's specific condition, resulting in insufficient images or excessive irrelevant images, affecting the efficiency of doctors' diagnosis.

Method used

Through the management server, the medical image shooting request data of breast disease patients is received, the semantic analysis and feature extraction is performed using a large language model, and the generation of adversarial model prediction and matching analysis is combined to determine the optimal shooting perspective, and the target images used for diagnosis are screened out.

Benefits of technology

It improves the doctor's diagnosis efficiency for patients with breast disease, ensures that the image quality and quantity meet the diagnostic needs, and improves the accuracy of in-depth diagnosis of breast disease.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118675715B_ABST
    Figure CN118675715B_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of smart medical technology. An intelligent analysis method and system for breast disease medical images are provided. The method includes: a management server receives medical image shooting request data of a breast disease patient, uses a large language model to perform semantic analysis and feature extraction on the medical image shooting request data, and obtains breast disease features; inputs the breast disease features into a generative adversarial model, and the generative adversarial model outputs a number of first shooting angles; reversely infers the shooting angles of a number of breast disease medical images taken to obtain a number of second shooting angles, matches and analyzes the second shooting angles with the first shooting angles, and determines the breast disease medical image corresponding to the successfully matched second shooting angle as the target breast disease medical image. The solution of the present invention can automatically remove irrelevant breast disease medical images, thereby improving the efficiency of doctors in conducting in-depth diagnosis of patients' breast diseases.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical technology, and in particular to an intelligent analysis method and system for breast disease medical images. Background Art

[0002] Qualitative analysis of breast disease based on breast disease medical images is an important part of breast disease diagnosis and treatment. Breast disease medical images include breast X-ray images, breast MRI images, breast ultrasound images, etc.

[0003] Mammography: It is suitable for X-ray examinations of patients with breast diseases and socialized subjects for breast disease screening, especially for women over 40 years old. It can better show breast masses and small calcifications, and is an important means of screening and diagnosing breast cancer. In diagnosis, attention will be paid to the shape, edge, density and other characteristics of the mass, as well as whether there are other signs such as calcification. In addition, there are special techniques such as breast point compression magnification photography, breast duct angiography and mammography positioning puncture photography for further evaluation and location of lesions.

[0004] Breast MRI imaging: It is of greater value in diagnosing lesions that are difficult or hard to identify with mammary X-ray or ultrasound, especially in showing the shape, size, number and location of breast lesions. MRI examination is suitable for preoperative evaluation, treatment evaluation and follow-up of breast cancer, screening of high-risk populations, and MRI-guided puncture positioning or biopsy. MRI examination can provide multi-sequence, multi-parameter, and dynamic enhanced scanning information, which helps to more accurately evaluate breast lesions.

[0005] Breast ultrasound imaging: Although not described in detail in the search results, ultrasound examination is also a commonly used non-invasive examination method in the diagnosis of breast diseases. It is especially suitable for young women and pregnant women. It can show the internal characteristics of breast tissue structure and masses.

[0006] In the existing medical system, breast disease images are all shot according to a preset shooting method, and the appropriate shooting angle cannot be selected according to the specific condition of the patient, resulting in a lack of images with the best angle for assisting doctors in diagnosis, or too many irrelevant images in the multi-angle images shot, which is not conducive to doctors' efficient diagnosis and treatment of breast disease patients. The present invention aims to solve or improve this technical problem. Summary of the invention

[0007] In this regard, the present invention provides an intelligent analysis method, system, electronic device, computer storage medium and computer program product for breast disease medical images to solve the above technical problems.

[0008] The present invention discloses an intelligent analysis method for breast disease medical images, which comprises the following steps: a management server receives medical image shooting request data of a breast disease patient, uses a large language model to perform semantic analysis and feature extraction on the medical image shooting request data, and obtains breast disease features; the medical image shooting request data is initiated by an associated doctor of the breast disease patient, and at least includes the breast disease type, preliminary lesion area, and breast disease degree of the breast disease patient; the breast disease features are input into a generative adversarial model, and the generative adversarial model outputs a plurality of first shooting angles; the shooting angles of the plurality of breast disease medical images taken are reversed to obtain a plurality of second shooting angles, the second shooting angles are matched and analyzed with the first shooting angles, and the breast disease medical images corresponding to the second shooting angles that are successfully matched are determined as target breast disease medical images.

[0009] In some embodiments, the use of a large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain breast disease features includes: the management server calls the large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain preliminary breast disease features; the management server retrieves historical data of the generative adversarial model whose prediction accuracy for the first shooting angle is above a specified threshold, and obtains an average feature data volume corresponding to the breast disease features based on the historical data; if the feature data volume of the preliminary breast disease features is lower than the average feature data volume, the management server sends a data supplement request to the doctor associated with the breast disease patient, extracts supplementary breast disease features from the supplementary data, integrates the preliminary breast disease features with the supplementary breast disease features, and obtains the breast disease features; if the feature data volume of the preliminary breast disease features is higher than the average feature data volume, the preliminary breast disease features are used as the breast disease features.

[0010] In some embodiments, the specified threshold is determined by: ; In the formula, For the specified threshold, is the experience value of the specified threshold; The number of the first shooting angles for which the predicted label of the generative adversarial model is “accurate”, The number of the first shooting perspectives for which the predicted label of the generative adversarial model is “wrong”; is the average value of the confidence probability of the first shooting angle labeled as “accurate”, is the average value of the confidence probabilities of the first shooting angles labeled as “error”.

[0011] In some embodiments, the generative adversarial model is trained in the following manner: medical data of breast disease is obtained, and shooting angle data of each breast disease image data in the medical data is associated with at least one breast disease composite label to obtain a training data set; wherein the breast disease composite label includes a breast disease type sub-label, a preliminary lesion area sub-label, and a breast disease degree sub-label; and the generative adversarial model is gradually trained using the training data in the training data set until a convergence condition is met.

[0012] In some embodiments, the shooting angle of view of several breast disease medical images is reversed to obtain several second shooting angles, including: determining the image shooting posture of the breast disease patient based on the breast disease type, and determining the first distribution characteristics of multiple reference points in the breast disease medical images according to the image shooting posture; identifying the second distribution characteristics of the multiple reference points in each of the breast disease medical images; calculating the deviation value between the second distribution position and the first distribution position, and reversely inferring the second shooting angle of view of each breast disease medical image according to the deviation.

[0013] In some embodiments, the method also includes: the management server retrieves the first number of predictions made by the generative adversarial model for the first shooting angle, and the second number of predictions for the first shooting angle whose accuracy is above a specified threshold, and calculates a prediction success rate based on the first number and the second number; determines a third number based on the prediction success rate, and screens the breast disease medical images corresponding to the second shooting angles whose matching fails in the third number, and determines them as target breast disease medical images.

[0014] The present invention also discloses an intelligent analysis system for breast disease medical images, which is applied to a management server and includes a receiving module, a prediction module, and a matching module; the receiving module is used to receive medical image shooting request data of a breast disease patient, and use a large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain breast disease features; the medical image shooting request data is initiated by an associated doctor of the breast disease patient, and at least includes the breast disease type, preliminary lesion area, and degree of breast disease of the breast disease patient; the prediction module is used to input the breast disease features into a generative adversarial model, and the generative adversarial model outputs a plurality of first shooting angles; the matching module is used to reversely infer the shooting angles of a plurality of breast disease medical images taken to obtain a plurality of second shooting angles, match and analyze the second shooting angles with the first shooting angles, and determine the breast disease medical image corresponding to the successfully matched second shooting angle as the target breast disease medical image.

[0015] The present invention also discloses an electronic device, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement any of the above methods.

[0016] The present invention also discloses a computer storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement any of the above methods.

[0017] The present invention also discloses a computer program product, which implements any of the above methods when the computer program product is run on a terminal.

[0018] The beneficial technical effect of the present invention is that the present invention determines the optimal image shooting angle based on the specific condition of the breast disease patient, screens the breast disease medical images taken by the imaging equipment according to the image shooting angle, removes those breast disease medical images that do not meet the optimal image shooting angle, and the remaining breast disease medical images are all target breast disease medical images that are helpful for the doctor to conduct an in-depth diagnosis of the breast disease of the breast disease patient, thereby improving the doctor's diagnostic efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is a flow chart of an intelligent analysis method of breast disease medical images disclosed in an embodiment of the present invention.

[0021] Figure 2 It is a structural schematic diagram of an intelligent analysis system for breast disease medical images disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following is a description of the implementation of the present application by specific specific embodiments. People familiar with the technology can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0023] In addition, the technical features involved in the different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0024] like Figure 1 As shown, an embodiment of the present invention discloses an intelligent analysis method for breast disease medical images, and the method includes the following steps: a management server receives medical image shooting request data of a breast disease patient, uses a large language model to perform semantic analysis and feature extraction on the medical image shooting request data, and obtains breast disease features; the medical image shooting request data is initiated by an associated doctor of the breast disease patient, and at least includes the breast disease type, preliminary lesion area, and breast disease degree of the breast disease patient; the breast disease features are input into a generative adversarial model, and the generative adversarial model outputs a plurality of first shooting angles; the shooting angles of the plurality of breast disease medical images taken are reversed to obtain a plurality of second shooting angles, the second shooting angles are matched and analyzed with the first shooting angles, and the breast disease medical image corresponding to the successfully matched second shooting angle is determined as the target breast disease medical image.

[0025] In the present invention, doctors can make a preliminary diagnosis of the breast condition of breast disease patients based on conventional examination methods, and thereby obtain the breast disease type, preliminary lesion area, and degree of breast disease of the breast disease patient, and fill these data into the medical image shooting request data and send them to the management server. The management server calls the large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain the breast disease characteristics of the patient; and the breast disease characteristics are input into the generative adversarial model, which can predict several first shooting angles that are consistent with the actual situation of the patient's breast disease based on the breast disease characteristics, that is, the breast disease medical image obtained by the imaging device according to the first shooting angle is most suitable for the doctor to make an in-depth diagnosis of the patient's breast disease. The medical imaging device can capture a number of breast disease medical images according to preset rules, and the medical management server can infer the second shooting angle of each breast disease medical image. The second shooting angle is the shooting angle taken by the imaging device when capturing the breast disease medical image. The management server matches the second shooting angle of the captured breast disease medical image with the first shooting angle determined above one by one, and the target breast disease medical image that is successfully matched is the target breast disease medical image that has the best shooting angle (i.e., the first shooting angle).

[0026] Therefore, the present invention determines the optimal image shooting angle based on the specific condition of the breast disease patient, screens the breast disease medical images taken by the imaging equipment according to the image shooting angle, removes those breast disease medical images that do not meet the optimal image shooting angle, and the remaining breast disease medical images are all target breast disease medical images that are helpful for the doctor to conduct an in-depth diagnosis of the breast disease of the breast disease patient, thereby improving the doctor's diagnostic efficiency.

[0027] In some embodiments, the use of a large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain breast disease features includes: the management server calls the large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain preliminary breast disease features; the management server retrieves historical data of the generative adversarial model whose prediction accuracy for the first shooting angle is above a specified threshold, and obtains an average feature data volume corresponding to the breast disease features based on the historical data; if the feature data volume of the preliminary breast disease features is lower than the average feature data volume, the management server sends a data supplement request to the doctor associated with the breast disease patient, extracts supplementary breast disease features from the supplementary data, integrates the preliminary breast disease features with the supplementary breast disease features, and obtains the breast disease features; if the feature data volume of the preliminary breast disease features is higher than the average feature data volume, the preliminary breast disease features are used as the breast disease features.

[0028] In this embodiment, the management server can schedule the corresponding large language model and generative adversarial model, and store the historical data of the generative adversarial model, which is the first shooting angles obtained by the generative adversarial model for predictive analysis of the input breast disease features. The management server can perform statistical calculations on the historical data of the prediction accuracy of the generative adversarial model for the first shooting angle above the specified threshold, and obtain the average feature data amount, that is, the average amount of breast disease features input into the generative adversarial model. Thereby determining the minimum value of the feature data amount of the breast disease feature required for the generative adversarial model to output an accurate prediction result.

[0029] Therefore, when the amount of feature data of the preliminary breast disease feature obtained is lower than the aforementioned average feature data amount, it means that the amount of feature data of the preliminary breast disease feature contained in the medical imaging request data filled in by the doctor is insufficient, and it cannot guarantee the accuracy of the prediction result output by the generative adversarial model. At this time, the management server needs to send a data supplement request to the doctor associated with the breast disease patient, extract the supplementary breast disease feature from the supplementary data (also extracted by the large language model), and the preliminary breast disease feature and the supplementary breast disease feature are used as the breast disease feature as a whole. When the amount of feature data of the preliminary breast disease feature obtained is higher than the aforementioned average feature data amount, it means that the amount of feature data of the preliminary breast disease feature contained in the medical imaging request data filled in by the doctor is sufficient, which can guarantee the accuracy of the prediction result output by the generative adversarial model. At this time, the preliminary breast disease feature obtained above is directly determined as the breast disease feature.

[0030] BERT (Bidirectional Encoder Representations from Transformers) is a pre-trained language model designed based on the Transformer architecture. It learns text context information through large-scale unsupervised training and implements bidirectional encoder representation, which can better capture the semantic relationship between texts. The large language model in the present invention can be constructed using the BERT model. Before the large language model is used, it needs to be fine-tuned based on a variety of large amounts of medical imaging request data, which will not be described in detail.

[0031] In some embodiments, the specified threshold is determined by: ; In the formula, For the specified threshold, is the experience value of the specified threshold; The number of the first shooting angles for which the predicted label of the generative adversarial model is “accurate”, The number of the first shooting perspectives for which the predicted label of the generative adversarial model is “wrong”; is the average value of the confidence probability of the first shooting angle labeled as “accurate”, is the average value of the confidence probabilities of the first shooting angles labeled as “error”.

[0032] In this embodiment, after the generative adversarial model completes the prediction of the first shooting angle for multiple times, the management server can analyze the labels and confidence probabilities of each first shooting angle predicted each time, wherein when the confidence probability is higher than a preset value, the corresponding first shooting angle can be marked as "accurate", otherwise it can be marked as "wrong", and the marking can be automatically implemented. The management server counts the above prediction result data of all times of the generative adversarial model, and calculates the number of first shooting angles predicted by the generative adversarial model and labeled as "accurate" , the number of first shooting angles predicted by the generative adversarial model as “wrong” , the average confidence probability of the first shooting angle labeled as “accurate” , the average confidence probability of the first shooting angle labeled as "wrong" Substituting the above parameters into the above formula, we can get the specified threshold .

[0033] In some embodiments, the generative adversarial model is trained in the following manner: medical data of breast disease is obtained, and shooting angle data of each breast disease image data in the medical data is associated with at least one breast disease composite label to obtain a training data set; wherein the breast disease composite label includes a breast disease type sub-label, a preliminary lesion area sub-label, and a breast disease degree sub-label; and the generative adversarial model is gradually trained using the training data in the training data set until a convergence condition is met.

[0034] In this embodiment, medical image shooting request data initiated by a doctor are collected. The medical image shooting request data can be data filled in by the doctor for the condition of an actual patient, or data filled in by the doctor for the condition of a virtual patient. The collected medical image shooting request data also include breast disease types, preliminary lesion areas, and breast disease degrees. According to the preset label classification rules, corresponding sub-labels are set for different breast disease types, preliminary lesion areas, and breast disease degrees, and the three sub-labels constitute a group of breast disease composite labels.

[0035] At the same time, the collected medical imaging request data are also associated with the shooting angle data corresponding to the corresponding breast disease imaging data. The breast disease imaging data and shooting angle data can be images and their shooting angles corresponding to the actual patient's condition captured by the imaging equipment, or they can be images and their shooting angles predicted and generated based on the virtual patient's breast disease type, preliminary lesion area, and degree of breast disease.

[0036] The collected shooting angle data of each breast disease image data is associated with at least one breast disease composite label to obtain a training data set. The training data in the training data set is used to gradually train the generative adversarial model until the deviation between the prediction result of the generative adversarial network and the above composite label meets the convergence condition.

[0037] In some embodiments, the shooting angle of view of several breast disease medical images is reversed to obtain several second shooting angles, including: determining the image shooting posture of the breast disease patient based on the breast disease type, and determining the first distribution characteristics of multiple reference points in the breast disease medical images according to the image shooting posture; identifying the second distribution characteristics of the multiple reference points in each of the breast disease medical images; calculating the deviation value between the second distribution position and the first distribution position, and reversely inferring the second shooting angle of view of each breast disease medical image according to the deviation.

[0038] In this embodiment, multiple (for example, three) reference points are selected in the breast image of the breast disease patient. These reference points can be known lesion points and can be determined through historical image data. According to the type of breast disease (mainly including the location of the lesion on the breast), the most suitable image shooting posture of the breast disease patient can be determined, such as lying posture, side-lying posture, etc. After determining the patient's shooting posture, the first distribution characteristics of multiple reference points in the breast disease medical image (that is, the position of each reference point) can be calculated. Then, the second distribution characteristics of the above-mentioned reference points in the actual breast disease medical image are detected, and the deviation value of the second distribution characteristic and the first distribution characteristic obtained by the above-mentioned prediction is calculated. According to the deviation value, the second shooting angle of each breast disease medical image can be inferred.

[0039] In some embodiments, the method also includes: the management server retrieves the first number of predictions made by the generative adversarial model for the first shooting angle, and the second number of predictions for the first shooting angle whose accuracy is above a specified threshold, and calculates a prediction success rate based on the first number and the second number; determines a third number based on the prediction success rate, and screens the breast disease medical images corresponding to the second shooting angles whose matching fails in the third number, and determines them as target breast disease medical images.

[0040] In this embodiment, since the prediction success rate of the generative adversarial model is not 100%, the first shooting angle determined above may be wrong, resulting in the subsequent target breast disease medical image lacking the most suitable breast disease medical image for the doctor. In this regard, the present invention is further configured to determine a third number based on the prediction success rate of the generative adversarial model, and further randomly select a third number of breast disease medical images from the breast disease medical images corresponding to the second shooting angle that failed to match, and add them to the target breast disease medical images determined above. Therefore, by reasonably supplementing the breast disease medical images corresponding to the second shooting angle that failed to match, the present invention can overcome the defect that the generative adversarial model cannot predict 100% success.

[0041] like Figure 2As shown, an embodiment of the present invention also discloses an intelligent analysis system for breast disease medical images, which is applied to a management server and includes a receiving module, a prediction module, and a matching module; the receiving module is used to receive medical image shooting request data of a breast disease patient, and use a large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain breast disease features; the medical image shooting request data is initiated by an associated doctor of the breast disease patient, and at least includes the breast disease type, preliminary lesion area, and degree of breast disease of the breast disease patient; the prediction module is used to input the breast disease features into a generative adversarial model, and the generative adversarial model outputs a plurality of first shooting angles; the matching module is used to reverse the shooting angle of a plurality of breast disease medical images taken to obtain a plurality of second shooting angles, match and analyze the second shooting angle with the first shooting angle, and determine the breast disease medical image corresponding to the successfully matched second shooting angle as the target breast disease medical image.

[0042] An embodiment of the present invention further discloses an electronic device, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method described in the above embodiment.

[0043] An embodiment of the present invention further discloses a computer storage medium, wherein the computer storage medium stores a computer program, and the computer program is executed by a processor to implement the method described in the above embodiment.

[0044] The embodiment of the present invention further discloses a computer program product. When the computer program product is run on a terminal, the method described in the above embodiment is implemented.

[0045] The various component embodiments of the present invention can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It should be understood by those skilled in the art that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all functions of some or all components in the device according to an embodiment of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., computer program and computer program product) for executing part or all of the methods described herein. Such a program implementing the present invention can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0046] References herein to "one embodiment," "embodiment," or "one or more embodiments" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment is included in at least one embodiment of the present invention. In addition, please note that examples of the term "in one embodiment" do not necessarily all refer to the same embodiment.

[0047] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0048] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc., does not indicate any order. These words may be interpreted as names.

[0049] In addition, it should be noted that the language used in this specification is selected primarily for readability and teaching purposes, rather than for explaining or limiting the subject matter of the present invention. Therefore, many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the appended claims. The disclosure of the present invention is illustrative rather than restrictive with respect to the scope of the present invention, and the scope of the present invention is defined by the appended claims.

Claims

1. An intelligent analysis method for breast disease medical images, characterized in that: The method comprises the following steps: The management server receives medical imaging request data from a breast disease patient, and uses a large language model to perform semantic analysis and feature extraction on the medical imaging request data to obtain breast disease features; the medical imaging request data is initiated by an associated doctor of the breast disease patient, and includes at least the type of breast disease, preliminary lesion area, and degree of breast disease of the breast disease patient; Inputting the breast disease feature into a generative adversarial model, and the generative adversarial model outputs a plurality of first shooting angles; Reversely inferring the shooting angles of the plurality of breast disease medical images shot to obtain a plurality of second shooting angles, matching and analyzing the second shooting angles with the first shooting angles, and determining the breast disease medical image corresponding to the successfully matched second shooting angle as the target breast disease medical image; The generative adversarial model is trained as follows: Obtaining medical data of breast disease, associating shooting angle data of each breast disease image data in the medical data with at least one breast disease composite label, to obtain a training data set; wherein the breast disease composite label includes a breast disease type sub-label, a preliminary lesion area sub-label, and a breast disease degree sub-label; Using the training data in the training data set to gradually train the generative adversarial model until a convergence condition is met; The method of inversely calculating the shooting angles of the plurality of breast disease medical images to obtain a plurality of second shooting angles includes: Determine an image shooting posture of a breast disease patient based on the breast disease type, and determine first distribution positions of a plurality of reference points in the breast disease medical image according to the image shooting posture; Identifying second distribution positions of a plurality of reference points in each of the breast disease medical images; The deviation value between the second distribution position and the first distribution position is calculated, and the second shooting angle of view of each breast disease medical image is inferred based on the deviation.

2. The intelligent analysis method of breast disease medical images according to claim 1, characterized in that: The large language model is used to perform semantic analysis and feature extraction on the medical imaging request data to obtain breast disease features, including: The management server calls the large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain preliminary breast disease features; The management server retrieves historical data of the prediction accuracy of the generative adversarial model for the first shooting angle being above a specified threshold, and obtains an average feature data amount corresponding to the breast disease feature based on the historical data; If the feature data amount of the preliminary breast disease feature is lower than the average feature data amount, the management server sends a data supplement request to the doctor associated with the breast disease patient, extracts a supplementary breast disease feature from the supplementary data, integrates the preliminary breast disease feature with the supplementary breast disease feature, and obtains the breast disease feature; If the feature data amount of the preliminary breast disease feature is higher than the average feature data amount, the preliminary breast disease feature is used as the breast disease feature.

3. The intelligent analysis method of breast disease medical images according to claim 2, characterized in that: The specified threshold is determined as follows: ; In the formula, For the specified threshold, is the experience value of the specified threshold; The number of the first shooting angles for which the predicted label of the generative adversarial model is “accurate”, The number of the first shooting angles for which the predicted label of the generated adversarial model is “error”; is the average value of the confidence probabilities of the first shooting angles labeled as “accurate”, is the average value of the confidence probabilities of the first shooting angles labeled as “error”.

4. The intelligent analysis method for breast disease medical images according to any one of claims 1 to 3, characterized in that: The method further comprises: The management server retrieves a first number of predictions of the first shooting angle by the generative adversarial model and a second number of predictions of the first shooting angle with accuracy above a specified threshold, and calculates a prediction success rate according to the first number and the second number; A third number is determined according to the prediction success rate, and the breast disease medical images corresponding to the second shooting angles with failed matching in the third number are screened and determined as target breast disease medical images.

5. An intelligent analysis system for breast disease medical images, the system is based on the method described in any one of claims 1 to 4, applied to a management server, and comprises a receiving module, a prediction module, and a matching module; characterized in that: The receiving module is used to receive medical image shooting request data of a breast disease patient, and use a large language model to perform semantic analysis and feature extraction on the medical image shooting request data to obtain breast disease features; the medical image shooting request data is initiated by an associated doctor of the breast disease patient, and at least includes the breast disease type, preliminary lesion area, and degree of breast disease of the breast disease patient; The prediction module is used to input the breast disease feature into a generative adversarial model, and the generative adversarial model outputs a plurality of first shooting angles; The matching module is used to reversely infer the shooting angles of several breast disease medical images taken to obtain several second shooting angles, match and analyze the second shooting angles with the first shooting angles, and determine the breast disease medical image corresponding to the successfully matched second shooting angle as the target breast disease medical image.

6. An electronic device comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method according to any one of claims 1 to 4.

7. A computer storage medium storing a computer program, characterized in that: The computer program is executed by a processor to implement the method according to any one of claims 1 to 4.

8. A computer program product, characterized in that: When the computer program product runs on a terminal, the method according to any one of claims 1 to 4 is implemented.