Iconography report automatic generation and quality inspection system based on open source large language model

Abstract

The invention provides an automatic iconography report generation and quality control system based on an open source large language model, and the system comprises an obtaining module which is configured to obtain task information related to iconography examination; a classification module configured to store a neural network model; the relevancy sorting module is configured to extract an image case set from a specified database containing iconography cases and corresponding reports according to the image sign description; the priority ranking module is configured to screen the relevancy ranking sequence from a specified database according to the image sign description and the report requirement description to obtain a priority ranking sequence, and generating a plurality of paradigm reports of the iconography report according to the relevancy sequence, matching the matching degrees in the paradigm reports according to the priority sequence, and deleting the reports of which the matching degrees are outside a preset interval in the paradigm reports.

Description

Technical Field

[0001] The present invention belongs to the technical field of large models, and particularly relates to a system for automatically generating and quality inspecting imaging reports based on an open-source large language model. Background Art

[0002] In the process of medical diagnosis, imaging examinations (such as X-ray, CT, MRI, ultrasound, etc.) play a crucial role. These examinations can provide rich information about the internal structure of the human body.

[0003] Currently, the interpretation of imaging data and the generation of reports mainly rely on professional imaging doctors. In large medical institutions, doctors in the imaging department need to process a large amount of imaging materials every day. This process requires doctors to carefully observe various features in the images, including but not limited to organ morphology, tissue density, lesion location, etc., and make judgments based on their professional knowledge and clinical experience.

[0004] In view of the above-mentioned status of the prior art, there is an urgent need in the field of medical imaging for a system that can generate imaging reports efficiently, accurately, and flexibly, and can effectively control the quality of the reports. Such a system needs to overcome the problems in the prior art such as large manual workload, excessive dependence on doctor experience, lack of flexibility in software systems, and insufficient generalization ability, so as to improve the efficiency and quality of medical imaging diagnosis. Summary of the Invention

[0005] To this end, the present invention provides a system for automatically generating and quality controlling imaging reports based on an open-source large language model.

[0006] In the first aspect of the present invention, there is provided a system for automatically generating and quality controlling imaging reports based on an open-source large language model, comprising:

[0007] An acquisition module configured to acquire task information related to imaging examinations;

[0008] A classification module configured to store a neural network model, classify text keywords in the task information into imaging sign descriptions related to imaging image features and report requirement descriptions for imaging report writing and quality control, and construct corresponding data sets for the imaging sign descriptions and the report requirement descriptions, wherein the neural network model is constructed and optimized based on an open-source large language model architecture;

[0009] A relevance ranking module configured to extract an imaging case set from a specified database containing imaging cases and corresponding reports according to the imaging sign descriptions, and sort the imaging case set to obtain a relevance ranking sequence for evaluating the degree of tight association between different cases and the current imaging task in terms of imaging features;

[0010] A priority sorting module, configured to obtain a priority sorting sequence by screening the relevance sorting sequence from a specified database according to the imaging sign description and the report requirement description.

[0011] Generate multiple template reports of the imaging report according to the relevance sorting.

[0012] Match the matching degrees in each of the template reports according to the priority sorting, and delete the reports in the template reports whose matching degrees are outside the preset interval.

[0013] Further, the system is configured to perform the following steps:

[0014] S1. Obtain an imaging examination task, extract the imaging report generation requirement description and the imaging content feature description in the examination task, and respectively establish corresponding first and second data sets according to the requirement description and the content description.

[0015] S2. Establish a first index between the first data set and the second data set.

[0016] S3. Obtain an imaging history database, which at least includes multiple imaging cases in historical imaging examinations marked with imaging report requirement descriptions, and screen a required imaging case set from the requirement descriptions of the first data set.

[0017] S4. Obtain an imaging feature database, which includes multiple imaging cases marked with imaging content feature descriptions, and screen out a feature imaging case set from the content descriptions of the second data set.

[0018] S5. Form an associated data set of the requirement description and the content description. The associated data set forms multiple imaging case pairs based on the first index, and each imaging case pair includes at least one feature imaging case and one required imaging case.

[0019] S6. Construct a target image set, which includes multiple imaging case pairs determined according to the first index, obtain the relevance between each imaging case in the target image set and the examination task, and obtain the relevance sorting sequence of the target image set.

[0020] S7. Delete the imaging case pairs in the relevance sorting sequence whose relevance is below the relevance threshold.

[0021] S8. Obtain the priority sorting sequence of the imaging case pairs in the relevance sorting sequence, and perform imaging report generation and quality control operations.

[0022] Further, in S2, through natural language processing, obtain the key text segments in the inspection task, and construct a text time series set according to the positions of the key text segments in the inspection task;

[0023] Determine the part-of-speech meanings of the key text segments from the text time series set, and classify the key text segments into image content feature descriptions and image report generation requirement descriptions according to the part-of-speech meanings to construct the first data set and the second data set.

[0024] Further, when constructing the text time series set, also construct an extended text set, including the following steps:

[0025] Using any specified key text segment as an extension point, extend to adjacent text positions in the inspection task by a byte amount at a preset interval, and each extension is by squaring or summing;

[0026] After obtaining the extended text set, retrieve the adjacent key text segments of the key text segments from the extended text set, and write an index to the corresponding index positions in the first data set and the second data set.

[0027] Further, the first index is set as the value written when any image content feature description or image report generation requirement description associates with a description within the set extended text set range.

[0028] Further, the relevance between the image content feature description and the image report generation requirement description is calculated through the following steps:

[0029] Let the image content feature description vector of the current imaging inspection task be V C , whose dimension is n, where each dimension represents a specific image content feature, and the vector element represents the weight of the i-th feature;

[0030] Let the image content feature description vector of the image case be V I , and the dimension is also n.

[0031] Let the image report generation requirement description vector of the current imaging inspection task be V R , with a dimension of m, and the vector element V R (j) represents the weight of the j-th requirement feature;

[0032] Let the image report generation requirement description vector of the image case be V J , with a dimension of m.

[0033] Calculate the cosine similarity of the content features

[0034]

[0035] Calculate the weighted matching degree of content features

[0036]

[0037] Calculate the relevance of content features

[0038]

[0039] Where α is a parameter that balances the cosine similarity and the weighted matching degree, and its value range is between 0 and 1;

[0040] Calculate the cosine similarity of report requirements

[0041]

[0042] Calculate the weighted matching degree of report requirements

[0043]

[0044] Calculate the relevance of internal report requirements

[0045]

[0046] Where β is a parameter that balances the cosine similarity and the weighted matching degree, and its value range is between 0 and 1;

[0047] Calculate the comprehensive relevance,

[0048] ∑(x + i) n (A x + B x )

[0049]

[0050] Where N is the number of expansion times, x is the current expansion times, A is the content overlap degree of the text under the current expansion in the whole text, and B is the requirement overlap degree of the text under the current expansion in the whole text.

[0051] In the second aspect of the present invention, there is provided a method for automatically generating and quality controlling imaging reports based on an open-source large language model. S1: Obtain an imaging examination task, extract the imaging report generation requirement description and the imaging content feature description in the examination task, and establish corresponding first and second data sets according to the requirement description and the content description respectively;

[0052] S2: Establish a first index between the first data set and the second data set;

[0053] S3. Obtain the imaging history database, which at least includes multiple imaging cases in historical imaging examinations marked with imaging report requirement descriptions, and screen the required imaging case set from the requirement descriptions of the first data set;

[0054] S4. Obtain the imaging feature database, which includes multiple imaging cases marked with imaging content feature descriptions, and screen the feature imaging case set from the content descriptions of the second data set;

[0055] S5. Form the associated data set of the requirement description and the content description. The associated data set forms multiple imaging case pairs based on the first index, and each imaging case pair includes at least one feature imaging case and one required imaging case;

[0056] S6. Construct the target image set, which includes multiple imaging case pairs determined according to the first index, obtain the relevance of each imaging case in the target image set to the examination task, and obtain the relevance sorting sequence of the target image set;

[0057] S7. Delete the imaging case pairs with relevance below the relevance threshold in the relevance sorting sequence;

[0058] S8. Obtain the priority sorting sequence of the imaging case pairs in the relevance sorting sequence, and perform imaging report generation and quality control operations.

[0059] The above technical solution of the present invention has the following advantages compared with the prior art:

[0060] The present invention realizes the automatic generation of imaging reports through a semantic large model, and at the same time performs quality inspection on the templates according to the order of multiple generated templates. Description of the Drawings

[0061] Figure 1 It is a schematic diagram of the module connection of the system provided in Embodiment 1 of the present invention.

[0062] Figure 2 It is a schematic diagram of the electronic device provided in Embodiment 2 of the present invention.

[0063] Among them, 30. Processor; 31. Memory; 32. Communication interface; 33. Bus. Detailed Embodiment

[0064] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0065] Embodiment 1

[0066] In the first aspect of the present invention, there is provided an automatic generation and quality control system for imaging reports based on an open-source large language model, including:

[0067] An acquisition module, configured to acquire task information related to imaging examinations;

[0068] A classification module, configured to store a neural network model, classify the text keywords in the task information into imaging sign descriptions related to imaging image features and report requirement descriptions for imaging report writing and quality control, and construct corresponding data sets for the imaging sign descriptions and the report requirement descriptions, wherein the neural network model is constructed and optimized based on the open-source large language model architecture;

[0069] A relevance ranking module, configured to extract an imaging case set from a specified database containing imaging cases and corresponding reports according to the imaging sign descriptions, and sort the imaging case set to obtain a relevance ranking sequence for evaluating the degree of tight association between different cases and the current imaging task in terms of imaging features;

[0070] A priority ranking module, configured to screen the relevance ranking sequence from the specified database according to the imaging sign descriptions and the report requirement descriptions to obtain a priority ranking sequence,

[0071] Generate multiple template reports for the imaging report according to the relevance ranking

[0072] Match the matching degrees in each of the template reports according to the priority ranking, and delete the reports in the template reports whose matching degrees are outside the preset interval.

[0073] Further, the system is configured to perform the following steps:

[0074] S1. Obtain an imaging examination task, extract the imaging report generation requirement description and the imaging content feature description in the examination task, and respectively establish corresponding first and second data sets according to the requirement description and the content description;

[0075] S2. Establish a first index between the first data set and the second data set;

[0076] S3. Obtain the imaging history database, which at least includes multiple imaging cases in historical imaging examinations marked with imaging report requirement descriptions, and screen the required imaging case set from the requirement descriptions of the first data set;

[0077] S4. Obtain the imaging feature database, which includes multiple imaging cases marked with imaging content feature descriptions, and screen out the feature imaging case set from the content descriptions of the second data set;

[0078] S5. Form the associated data set of the requirement description and the content description. The associated data set forms multiple imaging case pairs based on the first index, and each imaging case pair includes at least one feature imaging case and one required imaging case;

[0079] S6. Construct the target image set, which includes multiple imaging case pairs determined according to the first index. Obtain the relevance of each imaging case in the target image set to the examination task, and obtain the relevance ranking sequence of the target image set;

[0080] S7. Delete the imaging case pairs with relevance below the relevance threshold in the relevance ranking sequence;

[0081] S8. Obtain the priority ranking sequence of the imaging case pairs in the relevance ranking sequence, and perform imaging report generation and quality control operations.

[0082] Further, in S2, through natural language processing, obtain the text key word segments in the examination task, and construct a text time series set according to the positions of the text key word segments in the examination task;

[0083] Determine the part-of-speech meanings of the text key word segments from the text time series set, and classify the text key word segments into imaging content feature descriptions and imaging report generation requirement descriptions according to the part-of-speech meanings to construct the first data set and the second data set.

[0084] Further, when constructing the text time series set, also construct an extended text set, including the following steps:

[0085] Take any specified text key word segment as the extension point, extend to the adjacent text positions in the examination task by a byte amount at a preset interval, and each extension is by squaring or summing;

[0086] After obtaining the extended text set, retrieve the adjacent key word segments of the text key word segment from the extended text set, and write an index to the corresponding index positions in the first data set and the second data set.

[0087] Further, the first index is set to the value written when any description of image content features or description of image report generation requirements is associated with a description within the set extended text set.

[0088] Further, the relevance between the description of image content features and the description of image report generation requirements is calculated through the following steps:

[0089] Let the vector of the description of image content features of the current imaging examination task be V C , whose dimension is n, where each dimension represents a specific image content feature, and the vector element represents the weight of the i-th feature;

[0090] Let the vector of the description of image content features of the image case be V I , and the dimension is also n.

[0091] Let the vector of the description of image report generation requirements of the current imaging examination task be V R , with a dimension of m, and the vector element V R (j) represents the weight of the j-th requirement feature;

[0092] Let the vector of the description of image report generation requirements of the image case be V J , with a dimension of m.

[0093] Calculate the cosine similarity of content features

[0094]

[0095] Calculate the weighted matching degree of content features

[0096]

[0097] Calculate the relevance of content features

[0098]

[0099] Among them, α is a parameter that balances the cosine similarity and the weighted matching degree, and its value range is between 0 and 1;

[0100] Calculate the cosine similarity of report requirements

[0101]

[0102] Calculate the weighted matching degree of report requirements

[0103]

[0104] Calculate the relevance of internal report requirements

[0105]

[0106] Among them, β is a parameter that balances the cosine similarity and the weighted matching degree, and its value range is between 0 and 1;

[0107] Calculate the comprehensive relevance,

[0108] ∑(x + i) n (A x + B x )

[0109]

[0110] Among them, N is the number of expansions, x is the current expansion number, A is the content overlap degree of the text under the current expansion in all texts, and B is the requirement overlap degree of the text under the current expansion in all texts.

[0111] In the second aspect of the present invention, a method for automatically generating and quality controlling imaging reports based on an open-source large language model is provided. S1: Obtain an imaging examination task, extract the imaging report generation requirement description and the imaging content feature description in the examination task, and establish corresponding first and second data sets according to the requirement description and the content description respectively;

[0112] S2: Establish a first index between the first data set and the second data set;

[0113] S3: Obtain an imaging history database, where the database includes at least multiple imaging cases in historical imaging examinations marked with imaging report requirement descriptions, and screen a required imaging case set from the requirement descriptions of the first data set;

[0114] S4: Obtain an imaging feature database, where the database includes multiple imaging cases marked with imaging content feature descriptions, and screen a feature imaging case set from the content descriptions of the second data set;

[0115] S5: Form an associated data set of the requirement description and the content description. The associated data set forms multiple imaging case pairs based on the first index, and each imaging case pair includes at least one feature imaging case and one required imaging case;

[0116] S6: Construct a target image set, where the target image set includes multiple imaging case pairs determined according to the first index, obtain the relevance between each imaging case in the target image set and the examination task, and obtain the relevance ranking sequence of the target image set;

[0117] S7: Delete the imaging case pairs with relevance below the relevance threshold in the relevance ranking sequence;

[0118] S8. Obtain the priority sorting sequence of the image case pairs in the relevance sorting sequence, and perform image report generation and quality control operations.

[0119] Embodiment 2

[0120] Combined with Figure 2 As shown, an embodiment of the present disclosure provides an electronic device, including a processor 30 and a memory 31. Optionally, the electronic device may further include a communication interface 32 and a bus 33. Among them, the processor 30, the communication interface 32, and the memory 31 can communicate with each other through the bus 33. The communication interface 32 can be used for information transmission. The processor 30 can call the logical instructions in the memory 31 to execute the method of the first embodiment above.

[0121] An embodiment of the present disclosure provides a storage medium storing computer-executable instructions, and the computer-executable instructions are set to execute the method as in the first embodiment.

[0122] The above storage medium may be a transient computer-readable storage medium or a non-transient computer-readable storage medium. The non-transient storage medium includes: various media such as USB flash drives, external hard drives, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks or optical discs that can store program codes, and may also be a transient storage medium.

[0123] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure, enabling those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, process, and other changes. Embodiments merely represent possible variations. Unless explicitly required, individual components and functions are optional, and the order of operations may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the terms used in this application are only for describing embodiments and do not limit the claims. As used in the description of the embodiments and the claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to also include the plural forms. Similarly, as used in this application, the term "and / or" refers to any and all possible combinations of one or more of the associated listed items. Additionally, when used in this application, the term "comprise" and its variants "comprises" and / or "comprising" etc. mean the presence of the stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or groups of these. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, or apparatus comprising the element. Herein, each embodiment may focus on the differences from other embodiments, and the same or similar parts among the embodiments may be referred to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method parts disclosed in the embodiments, the relevant parts may refer to the description of the method parts.

[0124] Those skilled in the art will appreciate that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software may depend on the specific application and design constraints of the technical solution. The skilled person may use different methods for each specific application to achieve the described functions, but such implementation should not be considered to exceed the scope of the embodiments of the present disclosure. The skilled person can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described devices, apparatuses, and units can refer to the corresponding processes in the foregoing method embodiments and will not be repeated here.

[0125] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functions, and operations of possible implementations of apparatuses, methods, and computer program products according to embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the block may occur in a different order from that marked in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. Each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware device that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. An automatic generation and quality control system for imaging reports based on an open source large language model, characterized by: include: An acquisition module, configured to acquire imaging examination-related task information; a classification module configured to store a neural network model, classify text keywords in the task information into image sign descriptions related to image features and report requirement descriptions for image report writing and quality control, and construct corresponding data sets for the image sign descriptions and the report requirement descriptions, wherein the neural network model is constructed and optimized based on an open source large language model architecture; A relevance ranking module is configured to extract an imaging case set from a designated database containing imaging cases and corresponding reports according to the imaging sign description, and to sort the imaging case set to obtain a relevance ranking sequence for evaluating the degree of relevance between different cases and the current imaging task in terms of imaging features; A priority sorting module is configured to obtain a priority sorting sequence after filtering the relevance sorting sequence from a specified database according to the image sign description and the report requirement description, Generating a plurality of sample reports of imaging reports according to the correlation ranking; The matching degrees in each of the template reports are matched according to the priority ranking, and the reports in the template reports whose matching degrees are outside a preset range are deleted.

2. The automatic generation and quality control system for radiology reports based on an open source large language model according to claim 1, characterized in that: The system is configured to perform the following steps: S1. Obtain an imaging examination task, extract an imaging report generation requirement description and an imaging content feature description in the examination task, and establish a corresponding first data set and a second data set according to the requirement description and the content description; S2. Establishing a first index between the first data set and the second data set; S3, obtaining an image history database, wherein the database at least includes a plurality of image cases in historical imaging examinations marked with an image report requirement description, and selecting a required image case set from the requirement description of the first data set; S4, obtaining an image feature database, wherein the database includes a plurality of image cases marked with image content feature descriptions, and filtering out a feature image case set from the content description of the second data set; S5, forming a correlation data set of the demand description and the content description, wherein the correlation data set forms a plurality of image case pairs based on the first index, and each of the image case pairs includes at least one feature image case and one demand image case; S6, constructing a target image set, wherein the target image set includes a plurality of image case pairs determined according to the first index, obtaining the relevance between each image case in the target image set and the inspection task, and obtaining a relevance ranking sequence of the target image set; S7, deleting image case pairs whose correlation is below the correlation threshold in the correlation sorting sequence; S8. Obtain a priority ranking sequence of image case pairs in the relevance ranking sequence, and perform image report generation and quality control operations.

3. The automatic generation and quality control system of radiology reports based on an open source large language model according to claim 2 is characterized in that: In S2, natural semantic processing is used to obtain text key words in the inspection task, and a text time series set is constructed according to the positions of the text key words in the inspection task; The part-of-speech meanings of the text key words are determined from the text time series set, and the text key words are classified into image content feature descriptions and image report generation requirement descriptions according to the part-of-speech meanings to construct the first data set and the second data set.

4. The automatic generation and quality control system for radiology reports based on an open source large language model according to claim 2, characterized in that: When constructing the text time series set, an extended text set is also constructed, including the following steps: Taking any specified text keyword segmentation as the expansion point, expand to the adjacent text position in the inspection task with a preset interval of bytes, and each expansion is expanded by square or sum. After the extended text set is obtained, adjacent key words of the key word of the question text are retrieved from the extended text set, and an index is written into the corresponding index position in the first data set and the second data set.

5. The automatic generation and quality control system of radiology reports based on an open source large language model according to claim 2 is characterized in that: The first index is set to a value written when any image content feature description or image report generation requirement description is written to an associated description within the set extended text set range.

6. The automatic generation and quality control system of radiology reports based on an open source large language model according to claim 5 is characterized in that: The correlation between the image content feature description and the image report generation requirement description is calculated by the following steps: Suppose the image content feature description vector of the current imaging examination task is V C , whose dimension is n, where each dimension represents a specific image content feature, and the vector element represents the weight of the i-th feature; Suppose the image content feature description vector of the image case is V I , the dimension is also n. Suppose the image report generation requirement description vector of the current imaging examination task is V R , dimension is m, vector element V R (j) represents the weight of the jth demand feature; Suppose the image report generation requirement description vector of the image case is V J , with dimension m. Calculate the cosine similarity of content features Calculate the weighted matching degree of content features Calculate content feature relevance Among them, α is a parameter that balances cosine similarity and weighted matching, and its value range is between 0 and 1; Calculate the cosine similarity of the report requirements Calculate the weighted match of the reporting requirements Calculate the relevance of the report requirements Among them, β is a parameter that balances cosine similarity and weighted matching, and its value range is between 0 and 1; Calculate the comprehensive correlation. Wherein, N is the number of expansions, x is the current number of expansions, A is the content overlap of the text under the current expansion in the entire text, and B is the demand overlap of the text under the current expansion in the entire text.

7. A method for automatically generating and controlling radiology reports based on an open source large language model, characterized in that: The steps include: S1. Obtain an imaging examination task, extract an imaging report generation requirement description and an imaging content feature description in the examination task, and establish a corresponding first data set and a second data set according to the requirement description and the content description; S2. Establishing a first index between the first data set and the second data set; S3, obtaining an image history database, wherein the database at least includes a plurality of image cases in historical imaging examinations marked with an image report requirement description, and selecting a required image case set from the requirement description of the first data set; S4, obtaining an image feature database, wherein the database includes a plurality of image cases marked with image content feature descriptions, and filtering out a feature image case set from the content description of the second data set; S5, forming a correlation data set of the demand description and the content description, wherein the correlation data set forms a plurality of image case pairs based on the first index, and each of the image case pairs includes at least one feature image case and one demand image case; S6, constructing a target image set, wherein the target image set includes a plurality of image case pairs determined according to the first index, obtaining the relevance between each image case in the target image set and the inspection task, and obtaining a relevance ranking sequence of the target image set; S7, deleting image case pairs whose correlation is below the correlation threshold in the correlation sorting sequence; S8. Obtain a priority ranking sequence of image case pairs in the relevance ranking sequence, and perform image report generation and quality control operations.

8. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method as claimed in claim 7 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method as claimed in claim 7 is implemented.

Images