Immune quantitative detection AI model training and detection method

Abstract

According to the immune quantitative detection AI model training and detection method, detection particles are added into a sample to form a microscopic sample, the sample comprises components to be detected, and antibodies or antigens exist on the surfaces of the detection particles; shooting the microscopic examination sample to obtain a microscopic image; identifying the microscopic image by using the concentration of the component to be detected to obtain an identified microscopic image; training an AI model by using the identified microscopic image to obtain an AI model feature data set or a trained AI model; the AI model feature data set is matched with the AI calculation model, so that the concentration of a to-be-detected component in a to-be-detected sample can be identified; or the trained AI model can identify the concentration of a to-be-detected component in a to-be-detected sample. The AI model feature data set obtained by the method is matched with an AI calculation model or a trained AI model, and a microscopic image is recognized to obtain the concentration of the component to be detected.

Description

Technical Field

[0001] This application belongs to the technical field of in vitro diagnostic immunology, and particularly relates to an immunometric detection and AI training method for detecting particle aggregation in microscopic images. Background Art

[0002] In vitro diagnostic immunology is a method for diagnosing diseases by detecting specific immune markers in body fluids. These immune markers can be antibodies, antigens, immunoglobulins, etc. Detecting their presence or changes in levels can help doctors diagnose diseases, monitor treatment effects, or predict disease progression.

[0003] In the techniques for quantitatively and qualitatively analyzing trace immune markers, there are multiple different technical paths. The mature techniques include enzyme-linked immunosorbent assay, radioimmunoassay, immunoturbidimetry, immunofluorescence assay, and chemiluminescent immunoassay.

[0004] Chinese Patent Application No. "CN201410197209.1", "A Method for Detecting Biomacromolecules or Microorganisms", proposes a method for detecting biomacromolecules or microorganisms. The patent with application number "CN201410197209.1" integrates immunomagnetic enrichment and visualization detection, and is easy to operate. However, this patent reveals that "the degree of aggregation of aggregates is positively correlated with the concentration of biomacromolecules or microorganisms", and "determine the concentration of biomacromolecules or microorganisms in the sample to be detected according to the degree of aggregation of aggregates. Specifically: obtain a photo of the aggregates, and determine the concentration of biomacromolecules or microorganisms in the sample to be detected by quantifying the gray value of the photo".

[0005] In the prior art, in the process of establishing a fitting formula through the aggregation degree and aggregation morphology of aggregated particles, a large amount of experimental data and a complex mathematical modeling process are required. How to extract features to establish the mathematical relationship between the image and the component to be detected requires a large number of experiments and modeling analyses, and there has been no good method in the industry. Therefore, the technology for obtaining the concentration of the substance to be detected based on the image analysis of aggregated particles is difficult to be applied industrially. Even the fitting formula obtained through a large amount of experimental data is difficult to achieve a high degree of accuracy.

[0006] Technical Name: Antibody: It refers to a protective protein produced by the body due to the stimulation of an antigen. It (immunoglobulin is not just an antibody) is a large Y-shaped protein secreted by plasma cells (effector B cells) and used by the immune system to identify and neutralize foreign substances such as bacteria and viruses. It is only found in body fluids such as the blood of vertebrates and on the cell membranes of their B cells. An antibody can recognize a unique feature of a specific foreign substance, and this foreign target is called an antigen. Summary of the Invention

[0007] After a large number of experimental verifications in this application, it is found that an AI model can be directly trained with pictures of calibrated concentrations, enabling the AI to directly identify the concentration of the target substance in the test sample. This avoids the complex derivation process of the fitting formula and replaces the mathematical modeling process with AI computing power. After a large number of experimental verifications, this method can achieve the accuracy required for medical detection. It has broad prospects for industrial application and is a major breakthrough in the field of immunoassay technology.

[0008] The method for solving the above technical problems in this application is an AI model training method for immunoquantitative detection, which is used to obtain a trained AI model or an AI model feature dataset; detection particles are added to the sample to form a microscopic examination sample, the sample includes the component to be detected, and antibodies or antigens are on the surface of the detection particles; the microscopic examination sample is photographed to obtain a microscopic image; the microscopic image is labeled with the concentration of the component to be detected to obtain a labeled microscopic image; the labeled microscopic image is used to train the AI model to obtain an AI model feature dataset or a trained AI model; the AI model feature dataset cooperates with the AI calculation model to be able to identify the concentration of the component to be detected in the sample to be detected; or the trained AI model can identify the concentration of the component to be detected in the sample to be detected.

[0009] The microscopic images are multiple microscopic images taken of the same sample. After each microscopic image is labeled, AI model training is carried out.

[0010] The microscopic examination sample is added to detection cavities of the same height or different heights, and the microscopic examination sample in the detection cavity is photographed to obtain a microscopic image.

[0011] The microscopic examination sample is obtained by adding detection particles to an immune sample with a known concentration; the concentrations of different-concentration immune samples in a group of training samples are set according to step concentrations.

[0012] The different-concentration immune samples include training sample group A, and training sample group A includes a 1-fold unit concentration sample and a 2-fold unit concentration sample.

[0013] The different-concentration immune samples include training sample group B, and training sample group B includes a 1-fold unit concentration sample and a 10-fold unit concentration sample.

[0014] The different-concentration immune samples include training sample group C, and training sample group C includes more than two training sample subgroups. The training sample subgroups include training sample subgroup C1 and training sample subgroup C2; training sample subgroup C1 includes multiple training samples with different concentrations in magnitude K1; training sample subgroup C2 includes multiple training samples with different concentrations in magnitude K2, and magnitude K2 is greater than magnitude K1.

[0015] In the AI model training method for immunoassay quantification, it is possible that the sample is serum, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, bacteria, and fungi. It is possible that the sample is a whole blood dilution, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, and cells. It is possible that the sample is urine, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, cells, and bacteria. It is possible that the sample is feces, and the detection components include any one or more of proteins, protozoa, eggs, cells, and microorganisms.

[0016] An immunoassay quantification method includes: adding detection microparticles to a sample to form a microscopic examination sample, where the sample includes components to be detected, and the surfaces of the detection microparticles have antibodies or antigens; photographing the microscopic examination sample to obtain a microscopic image; using the AI model feature dataset obtained by the above method in combination with an AI calculation model to identify the microscopic image to obtain the concentration of the component to be detected; or using the trained AI model obtained by the above method to identify the microscopic image to obtain the concentration of the component to be detected.

[0017] The microscopic image includes microscopic image A1 and microscopic image A2; identifying microscopic image A1 to obtain the concentration W1 of the component to be detected; identifying microscopic image A1 to obtain the concentration W2 of the component to be detected; the concentration of the component to be detected is equal to the average of the concentration W1 of the component to be detected and the concentration W2 of the component to be detected.

[0018] In the immunoassay quantification method, it is possible that the sample is serum, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, bacteria, and fungi.

[0019] It is possible that the sample is a whole blood dilution, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, and cells.

[0020] It is possible that the sample is urine, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, cells, and bacteria.

[0021] It is possible that the sample is feces, and the detection components include any one or more of proteins, protozoa, eggs, cells, and microorganisms.

[0022] A detection device or a computing and processing device is used to run all or part of the above method; the memory of the detection device or the computing and processing device includes all or part of the data of all or part of the above method.

[0023] A data storage device stores all or part of the program code for executing the above method; or stores all or part of the data of the above method.

[0024] The technical effects of the above technical solution include: Through a large number of experimental verifications in this application, it is found that the AI model can be directly trained with pictures of calibrated concentrations, enabling the AI to directly recognize the concentration of the target substance in the test sample. This avoids the complex derivation process of the fitting formula and replaces the mathematical modeling process with AI computing power. After a large number of experimental verifications, this method can achieve the accuracy required for medical detection. It has broad prospects for industrial application and is a major breakthrough in the field of immunoassay technology.

[0025] The technical effects of the above technical solution include: The AI model feature data set, in cooperation with the AI computing model or the trained AI model, can recognize the concentration of the component to be detected corresponding to the image and directly establish the association between the image and the concentration of the component to be detected; it greatly simplifies the intermediate image processing, recognition, and intermediate calculation processes; and it greatly improves the detection efficiency.

[0026] The technical effects of the above technical solution include: Compared with traditional optical absorption detection, the cost is greatly reduced. There is no need to set up a precise optoelectronic detection system, and immunoassay can be completed only by being able to achieve microscopic photography, which is a major breakthrough in this field.

[0027] The technical effects of the above technical solution include: Multiple microscopic images taken of the same sample are used for AI model training, that is, for samples of the same concentration, the AI is repeatedly trained with multiple pictures to enable the AI to extract the relationship between the concentration and the key elements of the graph, which can improve the efficiency and accuracy of recognition.

[0028] The technical effects of the above technical solution include: In the prior art quantitative analysis technology, it is necessary to calculate the sample volume according to the image area and the height of the sample accommodation cavity, and the accuracy of the sample volume has a great impact on the accuracy of the detection parameters. In this application, microscopic images are obtained by photographing microscopic examination samples in detection cavities of different heights for AI training, enabling the AI to have the ability to recognize the concentration of the detection sample in different cavity heights, reducing the accuracy requirements for the sample accommodation device, and improving the accuracy and robustness of the detection.

[0029] The technical effects of the above technical solution include: Detection particles are added to the immunoassay samples of known concentration; the concentrations of multiple groups of immunoassay samples of different concentrations are set according to step concentrations, which can generate sufficient characteristic pictures.

[0030] The technical effects of the above technical solution include: The training sample group A includes samples with 1-fold unit concentration and 2-fold unit concentration, and the sample concentration interval is set relatively close, which can be used in occasions where high concentration resolution is required.

[0031] The technical effects of the above technical solution include: the training sample group B includes 1 times unit concentration samples and 10 times unit concentration samples, and the sample concentration intervals are set relatively close, which can be used in occasions with a large concentration measurement range.

[0032] The technical effects of the above technical scheme include: training sample group C1 and training sample group C2; training sample group C1 includes multiple training samples of different concentrations in magnitude K1; training sample group C2 includes multiple training samples of different concentrations in magnitude K2, and magnitude K2 is greater than magnitude K1; training the two groups of samples at the same time can improve the range and accuracy of concentration measurement at the same time.

[0033] The technical effects of the above technical solution include: the concentration of the component to be detected can be obtained directly based on the sample microscopic image and the training data set or model with concentration labels. It is an efficient method for detecting the concentration of the component to be detected and has opened up a new path.

[0034] The technical effects of the above technical solution include: the concentration of the component to be detected is equal to the average value of the concentration W1 of the component to be detected and the concentration W2 of the component to be detected, and the average calculation reduces the measurement error caused by the shooting deviation.

[0035] The technical effects of the above technical solution include: the samples include a variety of types, the adaptability is good, and it can be used for almost all body fluid-related immune detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a schematic diagram of the process of quantitative immune detection AI training method Figure 1 ; Figure 2 This is a schematic diagram of the process of quantitative immune detection AI training method Figure 2 ; Figure 3 This is a schematic diagram of the process of quantitative immune detection AI training method Figure 3 ; Figure 4 is a schematic diagram of the process of the quantitative immunoassay method; Figure 5 This is a microscopic image obtained with a 1x concentration immune sample; Figure 6 This is a microscopic image obtained with a 2-fold concentration of immune sample; Figure 7 This is a microscopic image obtained with a 3-fold concentration of immune sample; Figure 8 This is a microscopic image obtained with a 4-fold concentration of immune sample; Figure 9 This is a schematic diagram of Group A of training samples used for AI training; Figure 10 This is a schematic diagram of training sample group B used for AI training; Figure 11 It is a schematic diagram of multiple groups of training samples used for AI training. Specific implementation manners

[0037] The following further details the content of the present application in conjunction with each attached drawing. It should be noted that the following is a description of the preferred embodiments of the present invention and does not constitute any limitation to the present invention. The description of the preferred embodiments of the present invention is only for the description of the general principles of the present invention. The numbers such as "first", "second", "A", and "B" involved in the present invention are only for the convenience of description and do not represent the sequence relationship in time or space. The combinations of letters and numbers "TA", "TB", and "H" involved in the present invention are only for the convenience of description, and the specific meanings are determined by the specific words they represent.

[0038] In the prior art, the detection of immune substances is mostly based on the detection of optical absorption, and a precise optoelectronic detection system needs to be set up, resulting in high costs for design, maintenance, and use.

[0039] The detection of antigen and antibody based on the image recognition technology of aggregated microparticles uses the aggregation phenomenon of micron-sized detection particles to quantitatively measure the content of molecular-level target substances.

[0040] After years of research, the applicant has invested a large amount of manpower and material resources. Through experimental verification, it is obtained that the detection method directly uses AI to recognize the images of aggregated microparticles, and then based on the number and area of individual detection microparticles and aggregated microparticle bodies recognized by microscopic photography, and then calculates the concentration of the target detected substance through an empirical formula. For example, in the formed component analyzer independently developed by the applicant, the above method is also adopted. Since it is necessary to recognize individual detection microparticles and aggregated microparticle bodies, it consumes a lot of computing power.

[0041] In the method of the present application, detection microparticles are added to the sample to form a microscopic examination sample. The sample includes the component to be detected, and the surface of the detection microparticles has antibodies or antigens. It is also based on the change in the microscopic image caused by the antigen-antibody binding characteristics in the sample. However, instead of using AI to recognize each local single detection microparticle and microparticle aggregate or aggregated microparticle body, during AI training, the concentration of the target substance is directly marked, and AI can be trained to directly obtain the content of the component to be detected according to the image. Using the computing power of AI to avoid the calculation of the fitting formula enables the technology of obtaining the concentration of the component to be detected based on the image analysis of aggregated particles to be industrially applied, which is a major breakthrough in the field of immunoassay technology.

[0042] The solution in the present application can complete immunoassay as long as microscopic photography can be achieved, which is a major breakthrough in the field. Compared with the situation of establishing a fitting formula by recognizing the details such as individual detection microparticles and aggregated microparticle bodies based on microscopic photography, directly marking the microscopic image with the concentration of the component to be detected gives full play to the advantages of AI.

[0043] As Figure 1 , an AI model training method for immunoquantitative detection is used to obtain a trained AI model or an AI model feature dataset; detection particles are added to a sample to form a microscopic examination sample, the sample includes a component to be detected, and antibodies or antigens are on the surface of the detection particles; a microscopic image of the microscopic examination sample is taken; the microscopic image is labeled with the concentration of the component to be detected to obtain a labeled microscopic image; the labeled microscopic image is used to train the AI model to obtain an AI model feature dataset or a trained AI model; the AI model feature dataset is combined with an AI calculation model to be able to identify the concentration of the component to be detected in a sample to be detected; or the trained AI model is able to identify the concentration of the component to be detected in a sample to be detected.

[0044] In some embodiments, the microscopic images are multiple microscopic images taken of the same sample. After each microscopic image is labeled, AI model training is performed. Figure 5 is a microscopic image obtained from an immune sample with a concentration of 1 times; Figure 6 is a microscopic image obtained from an immune sample with a concentration of 2 times; Figure 7 is a microscopic image obtained from an immune sample with a concentration of 3 times; Figure 8 is a microscopic image obtained from an immune sample with a concentration of 4 times; all can be used for AI model training.

[0045] Generally, the microscopic examination sample is added to a detection cavity with the same height, and a microscopic image of the microscopic examination sample in the detection cavity is taken. Alternatively, as Figure 3 , the microscopic examination sample is added to detection cavities with different heights, and a microscopic image of the microscopic examination sample in the detection cavity is taken.

[0046] As Figure 2 , the microscopic examination sample is obtained by adding detection particles to an immune sample with a known concentration; the concentrations of a group of immune samples with different concentrations are set according to stepwise concentrations.

[0047] As Figure 9 , the immune samples with different concentrations include a training sample group A, and the training sample group A includes a 1-fold unit concentration sample, a 2-fold unit concentration sample, a 3-fold unit concentration sample, and a 4-fold unit concentration sample.

[0048] As Figure 10 , the immune samples with different concentrations include a training sample group B, and the training sample group B includes a 1-fold unit concentration sample, a 10-fold unit concentration sample, a 100-fold unit concentration sample, and a 1000-fold unit concentration sample.

[0049] As Figure 11 , the immune samples with different concentrations include a training sample group C, and the training sample group C includes more than two training sample subgroups, and the training sample subgroups include a training sample group C1, a training sample group C2, a training sample group C3, and a training sample group C4.

[0050] The training sample set C1 includes multiple training samples with different concentrations in magnitude K1, namely, a 1-fold unit concentration sample and a 5-fold unit concentration sample. The multiple different concentrations can be more.

[0051] The training sample set C2 includes multiple training samples with different concentrations in magnitude K2, namely, a 10-fold unit concentration sample and a 50-fold unit concentration sample. Magnitude K2 is greater than magnitude K1.

[0052] The training sample set C3 includes multiple training samples with different concentrations in magnitude K3, namely, a 100-fold unit concentration sample and a 500-fold unit concentration sample. Magnitude K3 is greater than magnitude K2.

[0053] The training sample set C4 includes multiple training samples with different concentrations in magnitude K4, namely, a 1000-fold unit concentration sample and a 5000-fold unit concentration sample. Magnitude K4 is greater than magnitude K3.

[0054] Such as Figure 4 , an immunometric detection method includes: adding detection microparticles to a sample to form a microscopic examination sample, where the sample includes the component to be detected, and the surface of the detection microparticles has an antibody or antigen; taking a microscopic image of the microscopic examination sample; using the AI model feature dataset obtained by the above method in combination with an AI calculation model to identify the microscopic image to obtain the concentration of the component to be detected; or using the trained AI model obtained by the above method to identify the microscopic image to obtain the concentration of the component to be detected.

[0055] The microscopic image includes microscopic images A1 and A2; identifying microscopic image A1 to obtain the concentration W1 of the component to be detected; identifying microscopic image A1 to obtain the concentration W2 of the component to be detected; the concentration of the component to be detected is equal to the average value of the concentration W1 of the component to be detected and the concentration W2 of the component to be detected.

[0056] In some embodiments, the sample is serum, and the detected components include any one or more of protein, lipid, polysaccharide, small molecule hormone, bacteria, and fungi.

[0057] In some embodiments, the sample is a whole blood dilution, and the detected components include any one or more of protein, lipid, polysaccharide, small molecule hormone, and cells.

[0058] In some embodiments, the sample is urine, and the detected components include any one or more of protein, lipid, polysaccharide, small molecule hormone, cells, and bacteria.

[0059] In some embodiments, the sample is feces, and the detected components include any one or more of protein, protozoa, eggs, cells, and microorganisms.

[0060] A detection device or a computing and processing device for running all or part of the above method; the memory of the detection device or the computing and processing device includes all or part of the data of all or part of the above method.

[0061] A data storage device that stores all or part of the program code for executing the above method; or stores all or part of the data of the above method.

[0062] Although the present invention is described and illustrated according to preferred embodiments and several alternative solutions, the invention is not limited by the specific descriptions in this specification. Other additional alternatives or equivalent components can also be used to practice the present invention.

Claims

1. A method for training an AI model for quantitative immune detection, characterized in that: Used to obtain a trained AI model or an AI model feature dataset; Adding detection microparticles to a sample to form a microscopic sample, the sample includes components to be detected, and the detection microparticles have antibodies or antigens on their surfaces; photographing the microscopic sample to obtain a microscopic image; marking the microscopic image with the concentration of the component to be detected to obtain a marked microscopic image; Training an AI model with the labeled microscopic image to obtain an AI model feature data set or a trained AI model; The AI ​​model feature data set, combined with the AI ​​computing model, can identify the concentration of the components to be tested in the samples to be tested; Or the trained AI model can identify the concentration of the component to be detected in the sample to be detected.

2. The immune quantitative detection AI model training method according to claim 1, characterized in that: The microscopic images are multiple microscopic images taken of the same sample. After each microscopic image is marked, AI model training is performed.

3. The immune quantitative detection AI model training method according to claim 1, characterized in that: The microscopic inspection samples are added into the inspection cavities at the same height or different heights, and the microscopic inspection samples in the inspection cavities are photographed to obtain microscopic images.

4. The immune quantitative detection AI model training method according to claim 1, characterized in that: The microscopic sample is prepared by adding detection particles to an immune sample of known concentration; The concentrations of immune samples with different concentrations in a set of training samples are set according to step concentrations.

5. The immune quantitative detection AI model training method according to claim 4, characterized in that: The immune samples with different concentrations include a training sample group A, and the training sample group A includes a 1-fold unit concentration sample and a 2-fold unit concentration sample.

6. The immune quantitative detection AI model training method according to claim 4, characterized in that: The immune samples with different concentrations include a training sample group B, and the training sample group B includes 1-fold unit concentration samples and 10-fold unit concentration samples.

7. The immune quantitative detection AI model training method according to claim 4, characterized in that: The immune samples with different concentrations include a training sample group C, the training sample group C includes more than two training sample groups, and the training sample groups include a training sample group C1 and a training sample group C2; The training sample C1 group includes a plurality of training samples of different concentrations in the magnitude K1; The training sample group C2 includes a plurality of training samples of different concentrations in a level K2, where the level K2 is greater than the level K1.

8. The immune quantitative detection AI model training method according to any one of claims 1 to 7, characterized in that: Includes one or more of the following features: Feature TB10: The sample is serum, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, bacteria, and fungi; Feature TB20: The sample is a dilution of whole blood, and the detection components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, and cells; Feature TB30: The sample is urine, and the detected components include any one or more of proteins, lipids, polysaccharides, small molecule hormones, cells, and bacteria; Feature TB40: The sample is feces, and the detection components include any one or more of proteins, protozoa, eggs, cells, and microorganisms.

9. An immunoquantitative detection method, characterized in that: include: Adding detection microparticles to a sample to form a microscopic sample, the sample includes components to be detected, and the detection microparticles have antibodies or antigens on their surfaces; Photograph the microscopic specimen to obtain a microscopic image; Use the AI ​​model feature data set obtained by the method described in any one of claims 1 to 8 in conjunction with the AI ​​computing model to identify the microscopic image and obtain the concentration of the component to be detected; Alternatively, the trained AI model obtained by the method described in any one of claims 1 to 7 is used to identify the microscopic image and obtain the concentration of the component to be detected.

10. The immunoquantitative detection method according to claim 9, characterized in that: The microscopic images include microscopic image A1 and microscopic image A2; Recognize the microscopic image A1 to obtain the concentration W1 of the component to be detected; recognize the microscopic image A1 to obtain the concentration W2 of the component to be detected; The concentration of the component to be detected is equal to the average value of the concentration W1 of the component to be detected and the concentration W2 of the component to be detected.

11. A detection device or a computing device, characterized in that: Used to perform all or part of the method according to any one of claims 1 to 10; The memory of the detection device or the computing and processing device includes all or part of the data of the method according to any one of claims 1 to 10.

12. A data storage device, characterized in that: Storing all or part of the program code for executing the method described in any one of claims 1 to 10; or storing all or part of the data of the method described in any one of claims 1 to 10.

Images