PCR image analysis method and device and computer equipment

Abstract

The invention relates to a PCR image analysis method and device and computer equipment. The method comprises the steps of obtaining data to be discriminated, inputting the data to be discriminated into a discrimination model to obtain image analysis data, storing the image analysis data and outputting the image analysis data to a terminal. Wherein the discrimination model is established based on a multi-modal large model and PCR historical data; the discrimination model is used for performing feature extraction and feature analysis on the data to be discriminated and outputting image analysis data for representing a PCR anomaly discrimination result; under the condition that the PCR anomaly judgment result represents that the to-be-judged data is abnormal, the image analysis data further comprises an anomaly removing instruction for representing an anomaly reason and an anomaly solving measure. By using the image-text processing advantage of the multi-modal large model and combining the PCR historical data, the discrimination model can accurately analyze the to-be-discriminated data, especially deeply analyze the abnormal PCR image, so as to obtain and output accurate and reliable image analysis data, and the abnormality discrimination efficiency of the PCR image can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of data analysis, and in particular to a PCR image analysis method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] During the PCR (Polymerase Chain Reaction) experiment, the obtained PCR curve may have various problems, such as failure to show a normal S-shaped curve, or the curve has jagged edges, etc. These abnormal curves will affect the final experimental results. Therefore, when conducting a PCR experiment, it is necessary to identify abnormal PCR curves to facilitate the completion of accurate PCR experiments.

[0003] The identification of PCR curves is usually based on the industry experience of the experimenter combined with the literature of experts, or through machine learning and the establishment of rule models, which requires a large number of historical PCR image samples to construct.

[0004] However, for the same PCR curve, different experts may come to different conclusions, which requires a comprehensive assessment based on different literature and experience, which is time-consuming and laborious. At the same time, in data annotation using machine learning modeling, the accuracy of the assessment may decrease due to differences in the expert's interpretation of an image, which requires subsequent adjustments and improvements. Therefore, there are still problems with low efficiency and insufficient accuracy in the abnormal assessment of PCR curves. Summary of the invention

[0005] Based on this, it is necessary to provide a PCR image analysis method, device, computer equipment, computer-readable storage medium and computer program product that can improve the efficiency and accuracy of abnormality identification in order to address the above technical problems.

[0006] In a first aspect, the present application provides a PCR image analysis method, the method comprising:

[0007] Obtaining data to be judged;

[0008] The data to be judged is input into a discrimination model to obtain image analysis data; the discrimination model is established based on a multimodal large model and PCR historical data; the discrimination model is used to extract and analyze features of the data to be judged, and output the image analysis data used to characterize the PCR abnormality discrimination result; in the case where the PCR abnormality discrimination result characterizes that the data to be judged is abnormal, the image analysis data also includes an abnormality resolution indication characterizing the abnormality cause and abnormality resolution measures;

[0009] The image analysis data is stored and output to a terminal.

[0010] In one embodiment, the method further includes:

[0011] Establish a basic knowledge base based on the PCR historical data, and optimize the basic knowledge base by using the image-text conversion function of the multimodal large model to obtain a structured knowledge base;

[0012] Solidify the structure of the multimodal large model according to the structured knowledge base, and customize to obtain a discrimination model.

[0013] In one embodiment, the PCR historical data includes document data and historical image data. The establishment of the basic knowledge base based on the PCR historical data and the optimization of the basic knowledge base by using the image-text conversion function of the multimodal large model to obtain a structured knowledge base includes:

[0014] Input the document data into a large language model, and extract to obtain the basic knowledge base; the basic knowledge base includes an abnormal image feature library and an abnormal cause solution library;

[0015] Perform image-text conversion on the historical image data based on the multimodal large model, generate problem image knowledge, and incorporate the problem image knowledge into the abnormal image feature library to update the abnormal image feature library;

[0016] Determine the structured knowledge base according to the updated abnormal image feature library and the abnormal cause solution library.

[0017] In one embodiment, the solidifying the structure of the multimodal large model according to the structured knowledge base and customizing to obtain a discrimination model includes:

[0018] Convert the structured knowledge base into a Q&A document in a retrieval-augmented generation manner;

[0019] Vectorize and save the Q&A document, and input it into the multimodal large model to obtain the discrimination model.

[0020] In one embodiment, in the step of inputting the data to be discriminated into the discrimination model to obtain image analysis data, the discrimination model is used for:

[0021] Obtain the data to be discriminated, extract keywords of the data to be discriminated, and perform vectorization processing on the keywords;

[0022] Perform vector similarity matching based on the vectorized keywords and the vectorized and saved Q&A document, and output the image analysis data.

[0023] In one embodiment, the image analysis data includes an anomaly determination representing the result of PCR anomaly discrimination. Based on the vector similarity matching between the keyword after vectorization processing and the vectorized saved Q&A document, the output of the image analysis data includes:

[0024] Perform image feature matching on the keyword after vectorization processing and the vectorized saved Q&A document to obtain an anomaly determination.

[0025] In one embodiment, the image analysis data further includes an anomaly resolution instruction representing the anomaly cause and the anomaly resolution measure. After performing image feature matching on the keyword after vectorization processing and the vectorized saved Q&A document to obtain an anomaly determination, the discrimination model is further configured to:

[0026] When the anomaly determination represents that the data to be discriminated is abnormal, perform solution feature matching on the anomaly determination and the vectorized saved Q&A document to obtain an anomaly resolution instruction.

[0027] In one embodiment, the obtaining of the data to be discriminated includes:

[0028] Obtain the original output signal of the PCR device and convert the original output signal into image data in a fixed format and size; the image data is the data to be discriminated.

[0029] In a second aspect, the present application further provides a PCR image analysis device, which includes:

[0030] A data acquisition module, configured to acquire data to be discriminated;

[0031] An input module, configured to input the data to be discriminated into a discrimination model to obtain image analysis data; the discrimination model is established based on a multi-modal large model and PCR historical data; the discrimination model is configured to perform feature extraction and feature analysis on the data to be discriminated and output the image analysis data for representing the result of PCR anomaly discrimination; in the case where the PCR anomaly discrimination result represents that the data to be discriminated is abnormal, the image analysis data further includes an anomaly resolution instruction representing the anomaly cause and the anomaly resolution measure;

[0032] An output module, configured to store the image analysis data and output the image analysis data to a terminal.

[0033] In a third aspect, the present application further provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0034] Obtain data to be discriminated;

[0035] Input the data to be discriminated into the discrimination model to obtain image analysis data; the discrimination model is established based on a multi-modal large model and PCR historical data; the discrimination model is used to extract and analyze features of the data to be discriminated, and output the image analysis data for characterizing the PCR abnormal discrimination result; in the case that the PCR abnormal discrimination result indicates that the data to be discriminated is abnormal, the image analysis data further includes an abnormality resolution instruction characterizing the cause of the abnormality and the solution to the abnormality.

[0036] Store the image analysis data and output the image analysis data to the terminal.

[0037] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0038] Obtain the data to be discriminated;

[0039] Input the data to be discriminated into the discrimination model to obtain image analysis data; the discrimination model is established based on a multi-modal large model and PCR historical data; the discrimination model is used to extract and analyze features of the data to be discriminated, and output the image analysis data for characterizing the PCR abnormal discrimination result; in the case that the PCR abnormal discrimination result indicates that the data to be discriminated is abnormal, the image analysis data further includes an abnormality resolution instruction characterizing the cause of the abnormality and the solution to the abnormality.

[0040] Store the image analysis data and output the image analysis data to the terminal.

[0041] In a fifth aspect, the present application further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0042] Obtain the data to be discriminated;

[0043] Input the data to be discriminated into the discrimination model to obtain image analysis data; the discrimination model is established based on a multi-modal large model and PCR historical data; the discrimination model is used to extract and analyze features of the data to be discriminated, and output the image analysis data for characterizing the PCR abnormal discrimination result; in the case that the PCR abnormal discrimination result indicates that the data to be discriminated is abnormal, the image analysis data further includes an abnormality resolution instruction characterizing the cause of the abnormality and the solution to the abnormality.

[0044] Store the image analysis data and output the image analysis data to the terminal.

[0045] The above PCR image analysis method, device, computer device, computer-readable storage medium, and computer program product include obtaining data to be discriminated, inputting the data to be discriminated into a discrimination model to obtain image analysis data, storing the image analysis data, and outputting the image analysis data to a terminal. Among them, the discrimination model is established based on a multimodal large model and PCR historical data; the discrimination model is used to extract and analyze features of the data to be discriminated and output image analysis data for characterizing the PCR anomaly discrimination result; when the PCR anomaly discrimination result indicates that the data to be discriminated is abnormal, the image analysis data further includes an anomaly resolution instruction for characterizing the anomaly cause and the anomaly resolution measure. By establishing a discrimination model based on PCR historical data and a multimodal large model, leveraging the text and image processing advantages of the multimodal large model and combining with PCR historical data, the discrimination model can accurately analyze the data to be discriminated, especially deeply analyze abnormal PCR images, so as to obtain and output accurate and reliable image analysis data, which can improve the efficiency of PCR image anomaly discrimination. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] To more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0047] Figure 1 It is an application environment diagram of the PCR image analysis method in an embodiment;

[0048] Figure 2 It is a flowchart of the PCR image analysis method in an embodiment;

[0049] Figure 3 It is a schematic diagram of the discrimination model establishment process of the PCR image analysis method in an embodiment;

[0050] Figure 4 It is a flowchart of the steps of establishing a basic knowledge base based on PCR historical data and optimizing the basic knowledge base using the text-image conversion function of the multimodal large model to obtain a structured knowledge base in an embodiment;

[0051] Figure 5 It is a flowchart of the steps of customizing a discrimination model by structurally solidifying the multimodal large model according to the structured knowledge base in an embodiment;

[0052] Figure 6 It is a schematic diagram of the execution steps of the discrimination model in an embodiment;

[0053] Figure 7Schematic diagram of the execution steps of the discrimination model in another embodiment;

[0054] Figure 8 Schematic diagram of the process of establishing the discrimination model of the PCR image analysis method in another embodiment;

[0055] Figure 9 Schematic diagram of the process of the PCR image analysis method in another embodiment;

[0056] Figure 10 Structural block diagram of the PCR image analysis device in one embodiment;

[0057] Figure 11 Internal structure diagram of a computer device in one embodiment. Detailed implementation manners

[0058] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0059] It can be understood that the terms "first", "second", etc. used in the present application may be used herein to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from another element. For example, without departing from the scope of the present application, the first resistor may be referred to as the second resistor, and similarly, the second resistor may be referred to as the first resistor. Both the first resistor and the second resistor are resistors, but they are not the same resistor.

[0060] It can be understood that in the following embodiments, "connection", if there is an electrical signal or data transmission between the connected circuits, modules, units, etc., should be understood as "electrical connection", "communication connection", etc.

[0061] As used herein, the singular forms "a", "an" and "the" may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms "comprise / include" or "have" etc. specify the presence of the stated features, wholes, steps, operations, components, parts or combinations thereof, but do not preclude the presence or addition of one or more other features, wholes, steps, operations, components, parts or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the related listed items.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0063] During the PCR experiment, abnormal PCR data may cause the PCR curve to fail to show a normal S-shape, or have abnormal shapes such as sawtooth and the like. These abnormal curves will affect the final experimental results. Currently, the identification of PCR curve abnormalities is usually done by experimenters or researchers based on their actual operational experience. The accuracy of personnel's judgment varies from person to person. For the same PCR curve, each person may get different judgment results based on different experiences.

[0064] The PCR image analysis method provided in the present application example can be applied to Figure 1 In the application environment shown, the processor 102 establishes a communication connection with the PCR device 104 through a network or a data transmission line. Among them, the PCR device 104 is also called a PCR amplification device, also known as a PCR gene amplifier, a PCR nucleic acid amplifier or a polymerase chain reaction nucleic acid amplifier, and is an instrument specially used for specific DNA amplification in PCR (polymerase chain reaction) technology. The PCR amplification device controls the reaction temperature and performs multiple cycles of high temperature denaturation, low temperature annealing and medium temperature extension to achieve gene amplification, and obtains corresponding PCR curves, PCR amplification results and PCR experimental data. The processor 102 is connected to the PCR device 104 in communication, and can interact with the PCR device 104 to obtain the PCR data obtained by the PCR device 104, and perform abnormal analysis on these data to obtain whether there is an abnormality in the PCR data. Optionally, the processor 102 can be, but is not limited to, various personal computers, laptops, smart phones and tablet computers.

[0065] In an exemplary embodiment, Figure 2 As shown, a PCR image analysis method is provided, which is applied to Figure 1 The processor 102 in FIG. 1 is taken as an example to illustrate, and the following steps 202 to 206 are included. Among them:

[0066] Step 202, obtaining data to be determined.

[0067] Among them, the data to be discriminated refers to the PCR data obtained in the PCR experiment. For example, the PCR amplification curve, the fluorescence signal intensity of the PCR reaction product, etc. are the PCR data obtained from the actual experiment. Specifically, the processor obtains the data to be discriminated from the PCR device through a communication connection. Optionally, the data to be discriminated may not be obtained from the PCR device. The processor may be connected to other devices storing PCR data to obtain the PCR data as the data to be discriminated.

[0068] Further, the data to be discriminated may be PCR curve data. That is to say, the PCR data obtained in the PCR experiment has been preprocessed and converted into the form of a PCR curve. The PCR curve data is used as the data to be discriminated and is obtained by the processor. The process of preprocessing does not limit the implementation method and the implementation device, as long as it is determined that the data to be discriminated obtained by the processor is PCR curve data.

[0069] Among them, the PCR curve mainly includes an amplification curve, a melting curve, and a standard curve. In this application, the amplification curve is taken as an example for illustration. The amplification curve is a curve describing the dynamic process of PCR, with the cycle number as the abscissa and the real-time fluorescence signal intensity during the reaction as the ordinate. Theoretically, the reaction product in the PCR process grows exponentially. However, as the number of PCR cycles increases, the DNA polymerase becomes inactivated, the dNTPs and primers are depleted, and the reaction by-products hinder the reaction, resulting in a decrease in the amplification efficiency of PCR and a gradual slowdown in the product generation rate. Therefore, the amplification curve is an "S-shaped" curve. If the amplification curve cannot normally present an S shape or shows serrations, it indicates that the amplification curve is abnormal. If the abnormal amplification curve cannot be accurately discriminated, it will lead to deviations in the PCR experiment results.

[0070] When the data to be discriminated is PCR curve data, the processor may not obtain the data to be discriminated from the PCR device and needs to perform data processing or data format conversion to obtain the data to be discriminated. In one embodiment, step 202 may be: obtaining the original output signal of the PCR device and converting the original output signal into image data with a fixed format and size. Among them, the image data is the data to be discriminated.

[0071] Specifically, the original output signal output by the PCR device to the processor is PCR experiment data, that is, the original data obtained from the PCR experiment. The processor can obtain the original output signal and convert the original output signal into image data with a fixed size and format through format conversion, that is, display the original data in the form of a curve graph, also known as PCR curve data. This is used as the data to be discriminated to facilitate directly performing subsequent steps to complete the abnormal analysis of the PCR image.

[0072] In this embodiment, the normalization of the data to be discriminated can improve the analysis efficiency of the discrimination model for PCR curve data, thereby enhancing the efficiency of PCR image analysis.

[0073] Step 204: Input the data to be discriminated into the discrimination model to obtain image analysis data.

[0074] Among them, the discrimination model is established based on a multimodal large model and PCR historical data. The establishment process may include performing text-image conversion on PCR historical data based on the multimodal large model. The discrimination model is used to extract features and analyze features of the data to be discriminated, and output image analysis data for characterizing the PCR anomaly discrimination result. In the case where the PCR anomaly discrimination result indicates that the data to be discriminated is abnormal, the discrimination model can further extract features and analyze features of the abnormal data to be discriminated, so that the image analysis data can include an anomaly resolution indication characterizing the anomaly cause and the anomaly resolution measure, making the content of the image analysis data richer.

[0075] A multimodal large model (Multimodal Large Models, abbreviated as MLLM for short) is an artificial intelligence model that can simultaneously process and understand multiple types of data (such as text, images, audio, video, etc.). It can receive multiple types of data such as text, images, audio, and video at the same time, understand the relationships between different modal data, for example, associate an image with a text description to achieve cross-modal semantic understanding; at the same time, it can also generate various types of data outputs according to the input different modal data, such as generating an image according to text, or generating a text description according to an image. In this embodiment, the multimodal large model can be used to perform cross-modal understanding and analysis of PCR historical data, that is, text-image conversion. Among them, PCR historical data includes historical documents related to PCR experiments and historical analysis records, etc., and also includes PCR curve data that has been analyzed for anomalies.

[0076] Specifically, the discriminant model is stored in the processor. The discriminant model is established based on a multimodal large model and PCR historical data, and the multimodal large model can perform text-image conversion on PCR historical data and can fully understand and analyze PCR historical data. After the processor obtains the data to be discriminated, it inputs the data to be discriminated into the discriminant model, and the discriminant model extracts features from the data to be discriminated and analyzes based on the extracted features. Since the discriminant model can accurately perform anomaly analysis on the data to be discriminated when extracting and analyzing features, the processor can obtain the image analysis data output by the discriminant model for characterizing the PCR anomaly discrimination result.

[0077] Optionally, the discrimination model may not be stored in the processor. The discrimination model may be stored in a server communicatively connected to the processor. The processor may call the discrimination model through remote calls such as Rest, access the data to be discriminated to the discrimination model, and complete the anomaly analysis of the PCR curve data. For example, call the transmission of PCR curve data, prompt (instruction) text, and call the corresponding API (Application Programming Interface) of the LLM (Large Language Model), etc.

[0078] Exemplarily, the image analysis data may indicate that the PCR curve data is normal or that the PCR curve data is abnormal. When the image analysis data indicates that the PCR curve data is abnormal, the image analysis data may further include the reason for the abnormality of the PCR curve data and the solution to the abnormality of the PCR curve data.

[0079] Step 206, store the image analysis data and output the image analysis data to the terminal.

[0080] After the processor obtains the image analysis data, the processor will perform local storage on the image analysis data. Optionally, the processor may package the image analysis data and the data to be discriminated into a sample data packet for local storage. The processor may also be connected to the terminal and output the image analysis data to the terminal for the experimenter to view.

[0081] Among them, the terminal may be other personal computers, tablets, laptops, etc. After receiving the image analysis data, the terminal may either back up and store the image analysis data or display the image analysis data correspondingly. For example, it may display the anomaly analysis results of the PCR curve data to the experimenter through a web page or a small program.

[0082] Optionally, in the case where the processor includes a display screen and is not connected to the terminal, after obtaining the image analysis data, the processor may perform local storage and local display without outputting to the terminal. In the case where the processor is connected to the PCR device, after obtaining the image analysis data, the processor may perform local storage and then return the image analysis data to the PCR device, and the display screen of the PCR device will display the anomaly analysis results.

[0083] The above-mentioned PCR image analysis method includes obtaining the data to be discriminated, inputting the data to be discriminated into the discriminant model, obtaining image analysis data, storing the image analysis data, and outputting the image analysis data to the terminal. Among them, the discriminant model is established based on the multimodal large model and PCR historical data; the discriminant model is used to extract and analyze the features of the data to be discriminated, and output image analysis data used to characterize the PCR abnormality discrimination result; when the PCR abnormality discrimination result characterizes the abnormality of the data to be discriminated, the image analysis data also includes an abnormality release indication characterizing the abnormal cause and abnormal solution measures. By establishing a discriminant model based on PCR historical data and a multimodal large model, utilizing the advantages of the multimodal large model in image and text processing, and combining PCR historical data, the discriminant model can accurately analyze the data to be discriminated, especially deeply analyze the abnormal PCR images, so as to obtain and output accurate and reliable image analysis data, which can improve the abnormality discrimination efficiency of PCR images.

[0084] In an exemplary embodiment, Figure 3 As shown, the processor can be used not only to use the discriminant model, but also to establish the discriminant model. That is, the PCR image analysis method can also include a process of establishing the discriminant model, that is, how to obtain the discriminant model, including steps 302 to 304.

[0085] Step 302, a basic knowledge base is established based on the PCR historical data, and the basic knowledge base is optimized using the image-text conversion function of the multimodal large model to obtain a structured knowledge base.

[0086] Specifically, PCR historical data includes historical documents and historical analysis records related to PCR experiments. The contents of these documents are integrated and text is organized to establish a basic knowledge base. The text organization process can be implemented based on various programs, models, and logic programming that can integrate and analyze texts. This step is not limited to the multimodal large model. At the same time, PCR historical data also includes PCR curve data that have been analyzed for abnormalities. For PCR curve data, the advantages of the image-text conversion function of the multimodal large model and its multimodal processing capabilities can be used to extract and convert PCR curve data to obtain textual image features. As a supplement, the basic knowledge base is expanded and updated, the content of the basic knowledge base is optimized, and a structured knowledge base is obtained.

[0087] Furthermore, in an exemplary embodiment, Figure 4 As shown, the PCR historical data includes document data and historical image data, and step 302 includes steps 402 to 406 .

[0088] Step 402: input the document data into a large language model to extract a basic knowledge base.

[0089] Among them, the basic knowledge base includes an abnormal image feature library and an abnormal cause solution library. The basic knowledge base is a library obtained by integrating document data. According to the prompting words (keywords or retrieval entries), the abnormal judgment steps and functions in the document data for the PCR experiment are divided, and the document data can be sorted and split into two knowledge bases, namely the abnormal image feature library and the abnormal cause solution library. The abnormal image feature library integrates the image features that these abnormal curves may have when abnormal PCR curve data is obtained from the PCR experiment, such as curve jitter. The abnormal cause solution library integrates different abnormal incentives and the corresponding solutions to avoid or mitigate abnormalities in the case of abnormalities in the PCR experiment. Further, the abnormal incentives corresponding to the abnormal PCR curve data can also be integrated into the abnormal image feature library to establish a correlation between the abnormal cause and the abnormal feature, making the basic knowledge base more complete.

[0090] Specifically, a large language model is used to process the document data. The document data is the content of the document type, and the acquisition method of the document data is not unique. It can be through an APP, a remote interface, a web page, etc. that can directly interact with the large language model, manually or automatically import the text knowledge related to the abnormal judgment of the PCR experiment as one of the document data; it can also be to access a private document library, such as a laboratory document library, a private enterprise document space, etc., and through traversal and retrieval of the content related to the abnormal judgment of the PCR experiment as one of the document data; it can also be to use a network search plugin to connect to the HTML web page of the Internet, etc., and extract the network resources related to the abnormal judgment of the PCR experiment as one of the document data.

[0091] Among them, the large language model is a complex neural network that can be trained based on a large amount of data sets, capable of capturing and simulating the complexity and diversity of language. Using neural network architectures such as Transformer, through unsupervised learning and fine-tuning and other technologies, it realizes the understanding and generation of natural language. The large language model is abbreviated as LLM, which can analyze information such as words, grammar, and context in the text data and learn language rules. When generating text, the model can also predict the next content of the sentence according to the learned knowledge, so as to generate coherent and relevant text. In addition, the model can also respond in a reasonable manner according to the conversation content to achieve natural language interaction.

[0092] Step 404, perform image-text conversion on the historical image data based on the multimodal large model, generate problem image knowledge, and incorporate the problem image knowledge into the abnormal image feature library to update the abnormal image feature library.

[0093] Among them, the historical image data includes historical PCR curve data and corresponding analysis results, that is, whether the historical PCR curve data is abnormal, and the abnormal reasons for the historical PCR curve data determined to be abnormal, etc.

[0094] Specifically, use the multimodal function of the multimodal large model, utilize its image-text conversion function, perform image-text conversion on the historical PCR curve data that has been discriminated in the historical image data, utilize the multimodal understanding ability, sort out the corresponding abnormal image characteristics, and generate problem image knowledge. Optionally, when inputting the historical PCR curve data into the multimodal large model, a prompt can also be added to improve the accuracy of generating problem image knowledge by the multimodal large model.

[0095] The problem image knowledge includes the image features of the abnormal historical PCR curve data. After obtaining the problem image knowledge, use the problem image knowledge as supplementary knowledge to supplement the abnormal image feature library, so that the abnormal image feature library includes not only the image features obtained from the document data but also the image features in the problem image knowledge. Optionally, it can be to fill the abnormal image feature library with the problem image knowledge, or to correct or adjust some of the image features in the abnormal image feature library with the problem image knowledge, or to fuse the problem image knowledge with some of the image features in the abnormal image feature library.

[0096] Step 406, determine the structured knowledge base according to the updated abnormal image feature library and the abnormal reason solution library.

[0097] After updating the abnormal image feature library with the problem image knowledge, merge the updated abnormal image feature library and the abnormal reason solution library, which is equivalent to partially updating the basic knowledge base, making the content in the knowledge base more comprehensive and accurate, and determining to establish a structured knowledge base.

[0098] In this embodiment, by separately storing the image features and the cause analysis knowledge, and separately establishing the abnormal image feature library and the abnormal reason solution library, on the one hand, a structured and step-by-step workflow process can be carried out in the subsequent discrimination steps of the application; on the other hand, by disassembling according to the features of the document data and the historical image data, and splitting the knowledge base according to the abnormal discrimination features and solution methods, the weight of a single document or knowledge can be reduced in the case of a large amount of input data, and the subjectivity of each knowledge in the knowledge base can be diluted.

[0099] Furthermore, in this embodiment, by utilizing the image-text conversion function of the multimodal large model, which is different from traditional machine learning that requires a large amount of historical image data for learning and training to build a feature engineering, this embodiment does not need to model the PCR image data or manually construct features. Using the multimodal function of the multimodal large model, only a small amount of historical image data is needed to generate problem image knowledge, and then combined with the document data of various types of literature, the curve characteristics corresponding to the abnormal PCR curve data can be sorted out more accurately, reducing the demand for PCR curve data. In this way, the multimodal large model can also automatically and efficiently generate abnormal discrimination knowledge for PCR experiments from the perspective of historical PCR curve data, and then integrate it with the abnormal image feature library generated from the document data, without the need for manual description of image abnormality rules, improving the intelligence and the efficiency of knowledge base establishment.

[0100] Step 304: Solidify the structure of the multimodal large model according to the structured knowledge base to customize and obtain a discrimination model.

[0101] After obtaining the structured knowledge base, it is also necessary to solidify the knowledge sorted out in the structured knowledge base into the structure of the multimodal large model, so that the operation logic of the multimodal large model can be based on the structured knowledge base, and the multimodal large model is adjusted to a specialized customized model for abnormal discrimination of PCR curve data, that is, a discrimination model.

[0102] Specifically, the solidification process of the discrimination model is not unique. In an exemplary embodiment, as Figure 5 shown, step 304 includes steps 502 to 504.

[0103] Step 502: Convert the structured knowledge base into a Q&A document in a retrieval-augmented generation manner.

[0104] Specifically, first connect the multimodal large model to the established structured knowledge base. Based on the content established in the structured knowledge base, use the retrieval-augmented generation method to convert the structured knowledge base into a RAG knowledge base, and establish connections for each piece of knowledge in the form of questions and corresponding answers, that is, a Q&A document. For example, it can be asking about the image features of PCR curve data under a certain abnormal reason, and there is a corresponding image feature answer. Among them, Retrieval-Augmented Generation (RAG) is a technology that uses information from proprietary data sources to supplement text generation. In this embodiment, the proprietary data source is the structured knowledge base, including the updated abnormal image feature library and abnormal cause solution library in the structured knowledge base.

[0105] The RAG technology generally includes two stages: retrieving context-related information and using the retrieved knowledge to guide the generation process. It can use a structured knowledge base to generate more accurate and context-aware responses, and by optimizing the way of context integration, reduce redundancy and duplicate content, and ensure that the relevance of the retrieved information to the generation task is properly balanced.

[0106] Step 504: Vectorize and save the Q&A document, and input it into the multi-modal large model to obtain a discrimination model.

[0107] Specifically, after obtaining the Q&A document, use a large language model to vectorize the Q&A document and input it into the multi-modal large model, which can serve as the knowledge base of the multi-modal large model, adjust and solidify the logical structure of the multi-modal large model to obtain a customized discrimination model. Vectorization processing refers to the process of converting non-numerical data (such as text, images, audio, etc.) into numerical data (i.e., vectors). This conversion enables the data to be effectively processed and learned by machine learning and deep learning models. Its principle is to map the original data into a low-dimensional vector space, and reflect the internal relationship between data through the calculation of vector similarity.

[0108] Exemplarily, the ways of vectorizing the text of the Q&A document include but are not limited to One-hot encoding, Bag of Words, TF-IDF (Term Frequency-Inverse Document Frequency), and Word Embedding, etc.

[0109] Furthermore, in an exemplary embodiment, as Figure 6 shown, based on the discrimination model established in the previous step based on the Q&A document, when the processor executes step 204, the discrimination model will execute steps 602 to 604.

[0110] Step 602: Obtain the data to be discriminated, extract the keywords of the data to be discriminated, and vectorize the keywords.

[0111] Specifically, the discrimination model obtains the input data to be discriminated, extracts keywords from the data to be discriminated, and then vectorizes the keywords. The step of extracting keywords can be obtained by means of prompt. The vectorization processing can correspond to the previous vectorization processing of the Q&A document, making the matching degree of the two after vectorization processing higher.

[0112] Step 604: Based on the vector similarity matching between the vectorized keywords and the vectorized and saved Q&A document, output image analysis data.

[0113] The discrimination model uses the vectorized keywords to perform vector similarity matching in the vectorized saved Q&A documents, so as to retrieve relevant information such as knowledge strongly related to the keywords in the Q&A documents, enhance the discrimination accuracy of the discrimination model, and then output the retrieved relevant knowledge as image analysis data.

[0114] Exemplarily, in an exemplary embodiment, as Figure 7 shown, the image analysis data includes an anomaly determination representing the PCR anomaly discrimination result, and step 604 may include step 702: performing image feature matching based on the vectorized keywords and the vectorized saved Q&A documents to obtain the anomaly determination.

[0115] Specifically, to ensure the accuracy of discrimination, the vectorized saved Q&A documents are called in the form of a workflow, and the task of PCR anomaly interpretation is disassembled based on methods such as Chain of Thought (COT). The discrimination model performs image-text conversion on the input data to be discriminated according to the vectorized keywords, and obtains the image features of the data to be discriminated described in text. Then, the image features of the data to be discriminated are matched with the anomaly image feature library in the vectorized saved Q&A documents to determine whether the PCR curve data represented by the data to be discriminated is abnormal PCR curve data, and the obtained judgment result is the anomaly determination.

[0116] Optionally, when the PCR curve data represented by the data to be discriminated is abnormal PCR curve data, the anomaly determination represents that the data to be discriminated is abnormal; when the PCR curve data represented by the data to be discriminated is not abnormal PCR curve data, the anomaly determination represents that the data to be discriminated is normal. Further, when the data to be discriminated is normal, the discrimination model can directly output this anomaly determination as the image analysis data.

[0117] Further, in an exemplary embodiment, as Figure 7 shown, the image analysis data further includes an anomaly resolution instruction representing the anomaly cause and the anomaly resolution measure. After step 702, the discrimination module further includes step 704: when the anomaly determination represents that the data to be discriminated is abnormal, performing solution feature matching based on the anomaly determination and the vectorized saved Q&A documents to obtain the anomaly resolution instruction.

[0118] Specifically, in the case where the anomaly determination represents that the data to be discriminated is abnormal, the discrimination model further analyzes the solution to the anomaly, and performs solution feature matching on the anomaly cause solution library in the vectorized saved Q&A documents in combination with the anomaly determination. Optionally, in the case where the anomaly determination represents that the data to be discriminated is abnormal, the anomaly determination may further include the inducement causing the anomaly, and the discrimination model may perform solution feature matching based on the inducement and the vectorized saved Q&A documents.

[0119] Furthermore, before performing feature matching, it is also possible to further refine keywords (or prompts), and then perform feature matching based on the keywords (or prompts) and the vectorized question and answer documents. The feature matching can be vectorized matching, and the terms or texts determined based on the matching degree are the solutions to the PCR curve anomaly, that is, the indication of anomaly resolution.

[0120] It should be noted that the cause of the anomaly can also be obtained when solving the feature matching, and the entries or texts corresponding to the cause and the anomaly resolution indication can be one or more. If there is only one, it is the entry or text with the highest matching degree. If there are multiple entries or texts, the entries or texts are arranged in order of matching degree.

[0121] In this embodiment, the data to be judged is processed in a workflow manner, and the PCR abnormality judgment task is disassembled in a thought chain manner, so that the judgment process does not simply classify PCR, but outputs various characteristics and problems of PCR curve data in steps, and then locates the possible cause of the problem, and finally provides a solution to the problem, thereby realizing a complete PCR curve data abnormality analysis.

[0122] In an exemplary embodiment, step 304 may also include other steps, such as Figure 8 As shown, step 304 includes step 802: fine-tuning the multimodal large model using the structured knowledge base as fine-tuning corpus to obtain a discriminant model.

[0123] Specifically, there is not only one way to solidify the structured knowledge base in the multimodal large model. The structured knowledge base can be used as the fine-tuning corpus of the multimodal large model, and the multimodal large model can be directly fine-tuned to obtain a discriminant model. Among them, the fine-tuning corpus refers to the data set used in the process of model fine-tuning. Fine-tuning is an important step in machine learning, especially deep learning. It involves further training and adjusting the model based on the pre-trained model using a task-specific data set to improve the performance of the model on specific tasks.

[0124] It should be noted that step 304 and step 802 can be selected arbitrarily, either one of them or both of them can be performed in combination, and both can be used as a way to build a customized discriminant model.

[0125] In order to better understand the above solution, a detailed explanation is given below in conjunction with a specific embodiment.

[0126] In one embodiment, a discriminant model is first established in the following manner: document data is entered into a large language model to extract a basic knowledge base, historical image data is converted into text based on a multimodal large model, problem image knowledge is generated, and problem image knowledge is incorporated into an abnormal image feature library, the abnormal image feature library is updated, and a structured knowledge base is determined based on the updated abnormal image feature library and abnormal cause solution library. The structured knowledge base is converted into a question-and-answer document in a retrieval-enhanced generation manner, the question-and-answer document is vectorized and saved, and entered into a multimodal large model to obtain a discriminant model; or the structured knowledge base is used as a fine-tuning corpus to fine-tune the multimodal large model to obtain a discriminant model.

[0127] In the PCR image analysis method, the processor obtains PCR experimental data and converts the PCR experimental data into data to be judged, and the data to be judged is the PCR curve data to be judged. Then the data to be judged is connected to the judgment model to execute the workflow. The specific process is as follows Figure 9 As shown. With the prompt "The above is an image of a PCR device, please describe the characteristics of the image", enter node 1: describe the image characteristics of the data to be judged, and obtain image feature "A"; then with the prompt "The above is the characteristics of a PCR image, please combine the knowledge base to judge the abnormal conditions that the image may meet" enter node 2: PCR abnormality judgment, and judge whether the data to be judged is abnormal.

[0128] When the image is judged as a normal image, the normal mark is directly output as image analysis data without subsequent processing. When the image is judged as an abnormal image, combined with the corresponding knowledge of the structured knowledge base in the discrimination model, the abnormal judgment is obtained as "abnormal question 1: Q1; corresponding image performance A1; abnormal question 2: Q2; corresponding image performance A2...", where A1, A2... belong to A.

[0129] Then, with the prompt "The above is the problem and image performance corresponding to a PCR abnormality curve graph. Please analyze the cause of the problem in combination with the knowledge base and provide solutions to each problem", enter node 3: solution generation, and generate abnormality resolution instructions as "reason / solution W1; reason / solution W2". Integrate the abnormality resolution instructions and abnormality judgment, and output the judgment results of the PCR image as image analysis data.

[0130] In this embodiment, the PCR interpretation ability of scientific researchers or experimental personnel is extracted and learned by using an LLM, and the multi-modal ability of the multi-modal large model is combined to automatically generate interpretation rules for the labeled abnormal PCR images. Since the knowledge source integrates various historical professional documents, the Internet, and the basic knowledge of the base model, it can eliminate the subjectivity of a single information source to the greatest extent. At the same time, the interpretation of this solution utilizes the powerful general reasoning ability of the multi-modal large model, and without a large number of training samples, it can achieve few-shot PCR abnormal interpretation. The established discrimination model provides a complete automated process, not just simply classifying PCR anomalies, but gradually outputting the problems, causes, and corresponding solutions of PCR through means such as COT (Chain of Thought), enabling the discrimination model to accurately analyze the data to be discriminated, so as to obtain and output accurate and reliable image analysis data, and improving the accuracy of abnormal discrimination of PCR images.

[0131] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages, and these steps or stages do not necessarily have to be executed at the same time, but can be executed at different times, and the execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0132] Based on the same inventive concept, the embodiments of the present application also provide a PCR image analysis device for implementing the PCR image analysis method involved above. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the PCR image analysis device provided below can refer to the limitations on the PCR image analysis method in the above text, and will not be repeated here.

[0133] In an exemplary embodiment, as Figure 10 shown, a PCR image analysis device is provided, including: a data acquisition module 1020, an input module 1040, and an output module 1060, where:

[0134] The data acquisition module 1020 is used to acquire the data to be discriminated;

[0135] An input module 1040 for inputting data to be discriminated into a discrimination model to obtain image analysis data; the discrimination model is established based on a multi-modal large model and PCR historical data; the discrimination model is used to extract and analyze features of the data to be discriminated and output image analysis data for characterizing the PCR anomaly discrimination result; in the case where the PCR anomaly discrimination result indicates that the data to be discriminated is abnormal, the image analysis data further includes an anomaly resolution instruction characterizing the cause of the anomaly and the solution to the anomaly.

[0136] An output module 1060 for storing the image analysis data and outputting the image analysis data to a terminal.

[0137] In one embodiment, the PCR image analysis device further includes a model establishment module for establishing a basic knowledge base based on PCR historical data and optimizing the basic knowledge base using the text-image conversion function of the multi-modal large model to obtain a structured knowledge base. The discrimination model is customized by structurally solidifying the multi-modal large model according to the structured knowledge base.

[0138] In one embodiment, the PCR historical data includes document data and historical image data, and the model establishment module is further configured to input the document data into a large language model to extract a basic knowledge base; the basic knowledge base includes an abnormal image feature library and an abnormal cause solution library. The historical image data is subjected to text-image conversion based on the multi-modal large model to generate problem image knowledge, and the problem image knowledge is incorporated into the abnormal image feature library to update the abnormal image feature library. The structured knowledge base is determined according to the updated abnormal image feature library and abnormal cause solution library.

[0139] In one embodiment, the model establishment module is further configured to convert the structured knowledge base into a Q&A document in a retrieval-augmented generation manner, vectorize and save the Q&A document, and input it into the multi-modal large model to obtain the discrimination model.

[0140] In one embodiment, when the input module 1040 is executed, the discrimination model established by the model establishment module is used to obtain the data to be discriminated, extract keywords of the data to be discriminated, perform vectorization processing on the keywords, and perform vector similarity matching based on the vectorized keywords and the vectorized and saved Q&A document to output the image analysis data.

[0141] In one embodiment, the image analysis data includes an anomaly determination characterizing the PCR anomaly discrimination result, and the discrimination model is further configured to perform image feature matching based on the vectorized keywords and the vectorized and saved Q&A document to obtain the anomaly determination.

[0142] In one of the embodiments, the image analysis data also includes an exception resolution indication that characterizes the cause of the exception and the exception resolution measures. The discrimination model is also used to obtain an exception resolution indication by matching solution features based on the exception judgment and the vectorized question and answer document when the exception judgment characterizes that the data to be judged is abnormal.

[0143] In one embodiment, the data input module 1020 is also used to obtain PCR experimental data and convert the PCR experimental data into data to be determined.

[0144] Each module in the above-mentioned PCR image analysis device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0145] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 11 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a PCR image analysis method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0146] Those skilled in the art will understand that Figure 11The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0147] In one embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0148] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0149] In one embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0150] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0151] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in the present application.

[0152] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A PCR image analysis method, characterized in that, The method includes: Obtain the data to be discriminated; Input the data to be discriminated into the discrimination model to obtain image analysis data; the discrimination model is established based on a multimodal large model and PCR historical data; the discrimination model is used to extract and analyze the features of the data to be discriminated and output the image analysis data for characterizing the PCR abnormal discrimination result; when the PCR abnormal discrimination result indicates that the data to be discriminated is abnormal, the image analysis data further includes an abnormality resolution instruction for characterizing the cause of the abnormality and the solution to the abnormality; Store the image analysis data and output the image analysis data to the terminal.

2. The method according to claim 1, wherein The method further includes: Establish a basic knowledge base based on the PCR historical data and optimize the basic knowledge base using the text-image conversion function of the multimodal large model to obtain a structured knowledge base; Customize the discrimination model by structurally solidifying the multimodal large model according to the structured knowledge base.

3. The method according to claim 2, wherein The PCR historical data includes document data and historical image data. The step of establishing a basic knowledge base based on the PCR historical data and optimizing the basic knowledge base using the text-image conversion function of the multimodal large model to obtain a structured knowledge base includes: Input the document data into a large language model and extract the basic knowledge base; the basic knowledge base includes an abnormal image feature library and an abnormal cause solution library; Perform text-image conversion on the historical image data based on the multimodal large model, generate problem image knowledge, incorporate the problem image knowledge into the abnormal image feature library, and update the abnormal image feature library; Determine the structured knowledge base according to the updated abnormal image feature library and the abnormal cause solution library.

4. The method according to claim 2, characterized in that The step of customizing the discrimination model by structurally solidifying the multimodal large model according to the structured knowledge base includes: Convert the structured knowledge base into a Q&A document in a retrieval-augmented generation manner; Vectorize and save the Q&A document and input it into the multimodal large model to obtain the discrimination model.

5. The method according to claim 4, characterized in that, In the step of inputting the data to be discriminated into the discrimination model to obtain image analysis data, the discrimination model is used to: Obtain the data to be discriminated, extract the keywords of the data to be discriminated, and perform vectorization processing on the keywords; Perform vector similarity matching based on the vectorized keywords and the vectorized and saved Q&A document, and output the image analysis data.

6. The method according to claim 5, characterized in that, The image analysis data includes an abnormality determination for characterizing the PCR abnormal discrimination result. The step of performing vector similarity matching based on the vectorized keywords and the vectorized and saved Q&A document and outputting the image analysis data includes: Perform image feature matching based on the vectorized keywords and the vectorized and saved Q&A document to obtain an abnormality determination.

7. The method according to claim 6, wherein The image analysis data further includes an abnormality resolution instruction for characterizing the cause of the abnormality and the solution to the abnormality. After obtaining the abnormality determination by performing image feature matching based on the vectorized keywords and the vectorized and saved Q&A document, the discrimination model is further used to: When the anomaly determination characterizes that the data to be discriminated is anomalous, perform a solution feature match based on the anomaly determination and the vectorized saved Q&A documents to obtain an anomaly resolution indication.

8. The method according to claim 1, wherein The obtaining of the data to be discriminated includes: Obtain the original output signal of the PCR device and convert the original output signal into image data of a fixed format and size; the image data is the data to be discriminated.

9. A PCR image analysis device, characterized in that, The device includes: A data acquisition module for acquiring data to be discriminated; An input module for inputting the data to be discriminated into a discrimination model to obtain image analysis data; the discrimination model is established based on a multi-modal large model and PCR historical data; the discrimination model is used to perform feature extraction and feature analysis on the data to be discriminated and output the image analysis data for characterizing the PCR anomaly discrimination result; in the case where the PCR anomaly discrimination result characterizes that the data to be discriminated is anomalous, the image analysis data further includes an anomaly resolution indication characterizing the anomaly cause and the anomaly resolution measure; An output module for storing the image analysis data and outputting the image analysis data to a terminal.

10. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images