Quantitative immunodetection, algorithm acquisition and AI training method

Abstract

Quantitative immunodetection based on a microscopic image for detecting particle aggregation, algorithm acquisition and an AI training method thereof; respectively marking the images of the non-aggregated detection particles and aggregated particles in the image, carrying out AI model training to obtain an AI model feature data set, and identifying the non-aggregated detection particles and aggregated particles in the image in cooperation with an AI calculation model; preparing an immune sample with set concentration; obtaining a detection signal value based on the number or area of the non-aggregated detection particles and the aggregated particles; acquiring detection signal values corresponding to a plurality of groups of immune samples with different concentrations; and fitting detection signal values corresponding to the immune samples with different concentrations to obtain a quantitative immunodetection fitting formula or establish an immune substance concentration lookup table. During detection, a detection signal value is obtained based on the number or area of the non-aggregated detection particles and the aggregated particles; calculating by using a fitting formula to obtain the content of immune components in the sample; or searching by using an immune substance concentration lookup table to obtain the content of immune components in the sample.

Description

Technical Field

[0001] The present application belongs to the field of in vitro diagnostic immunology technology, and in particular to quantitative immunoassay based on microscopic images of particle aggregation detection, AI training and algorithm acquisition methods thereof. Background Art

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

[0003] Affinity reactions are essential for the normal function of the immune system because they enable the immune system to recognize and eliminate pathogens, foreign matter, and abnormal cells in the body. In addition, affinity reactions are also widely used in laboratory techniques and clinical diagnosis. For example, techniques such as ELISA (enzyme-linked immunosorbent assay) and immunohistochemistry use the affinity reaction between antibodies and antigens to detect specific molecules or cells.

[0004] There are many different technical paths for quantitative and qualitative analysis of trace immune markers. Mature technologies include enzyme-linked immunosorbent assay, radioimmunoassay, immunoturbidimetric assay, immunofluorescence assay, and chemiluminescence immunoassay.

[0005] Chinese patent application number "CN201410197209.1" "A method for detecting biological macromolecules or microorganisms" proposes a method for detecting biological macromolecules or microorganisms. The patent with application number "CN201410197209.1" integrates immunomagnetic enrichment and visual detection, and is simple to operate. However, the patent only reveals that "the degree of agglomeration of aggregates is positively correlated with the concentration of biological macromolecules or microorganisms", "the concentration of biological macromolecules or microorganisms in the sample to be detected is determined according to the degree of agglomeration of aggregates, specifically: obtain a photo of the aggregates, and determine the concentration of biological macromolecules or microorganisms in the sample to be detected by quantifying the grayscale value of the photo", which is not an accurately calculated concentration. The method proposed in this patent application can only give a qualitative conclusion at most, but cannot give a quantitative measurement result, and cannot be applied to medical testing.

[0006] Chinese patent application number "CN202180041737.6" is named "Induced Agglomeration Assay for Improving Sensitivity". In this application, a "system, device and method for quickly and accurately measuring the aggregation of reporter particles induced by analyte particle binding is proposed. In the presence of analyte particles of interest, the reporter particles form agglomerates, and their average particle size increases with the increase of analyte concentration. Based on the analysis of the average particle size determined from the sample frame, the presence and / or concentration of the analyte can be determined", and a "calibration curve" is proposed. The "calibration curve" is composed of "horizontal axis = SARS-CoV-2 antibody (mg / dL, vertical axis = agglomeration size (pixel)", which establishes the relationship between the average agglomeration area and the target detection object. After a large number of experimental verifications, this method has a huge measurement deviation when the target content fluctuates by more than 20%, and has no application value for medical measurement. It has been proved that the average agglomerated particle size has no direct correlation with the target content. It is affected by many factors such as the length of time the test sample is configured, the vibration force, the temperature during configuration, etc. It is unscientific to use the average agglomerated particle size to measure the target content, and its application value is not high.

[0007] With the development of image analysis technology, especially AI technology, image recognition and classification are becoming more and more accurate. It has become possible to detect antigen-antibody or affinity reaction based on image recognition technology of aggregated particles. To measure accurately, the accuracy of the fitting formula is the key.

[0008] Technology Name: Antibody: refers to a protective protein produced by the body due to the stimulation of antigens. 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 the blood and other body fluids of vertebrates and on the cell membrane surface of their B cells. Antibodies can recognize a unique feature of a specific foreign object, which is called an antigen.

[0009] Affinity reaction: It is essential for the normal function of the immune system because it enables the immune system to recognize and eliminate pathogens, foreign matter and abnormal cells in the body. Affinity reaction is widely used in laboratory techniques and clinical diagnosis. For example, ELISA (enzyme-linked immunosorbent assay), immunofluorescence, chemiluminescence and immunohistochemistry all use the affinity reaction between antibodies and antigens to detect specific molecules or cells. Summary of the invention

[0010] The detection of antigens and antibodies based on the image recognition technology of aggregated particles uses the aggregation phenomenon of micron-level detection particles to quantitatively measure the content of the target at the molecular level. It is like using a balloon to detect the amount of oil in a room. It relies on the oil being adsorbed on the balloon and then aggregating with each other. This is very difficult and the accuracy of modeling is the key. The existing technology requires accurate calculation of the volume of the room to obtain the content of the target substance.

[0011] In this application, the applicant proposes that, during fitting, a relationship is established between the ratio relationship between the detected particles and the aggregated particles and the concentration of the target object, thereby avoiding the influence of the accuracy of the room capacity calculation on the accuracy of the target object concentration detection.

[0012] Objective evaluation parameters are obtained by measuring the total area of ​​aggregated particles or the number of unclustered test particles. For example, if the total number of balloons aggregated by oil pollution in a room or a single unclustered balloon is measured, a fitting formula can be established based on objective and stable evaluation parameters.

[0013] The solution to the above technical problems in the present application is a method for obtaining a quantitative immunoassay algorithm based on the aggregation of detection particles, wherein detection particles are added to an immune sample to form a microscopic sample, and antibodies or antigens are present on the surface of the detection particles; a microscopic image is obtained by photographing; the number or area of ​​unaggregated detection particles in the image is identified; the number or area of ​​aggregated granules in the image is identified; a detection signal value is obtained based on the number or area of ​​unaggregated detection particles and the number or area of ​​aggregated granules; the detection signal values ​​corresponding to multiple groups of immune samples of different concentrations are obtained; the detection signal values ​​corresponding to the above immune samples of different concentrations are used to fit the quantitative immunoassay fitting formula; or a lookup table of immune substance concentrations is established using the detection signal values ​​corresponding to the above immune samples of different concentrations. Some of the detection particles in the microscopic sample are aggregated to form aggregated granules.

[0014] The microscopic examination sample may be prepared by adding detection particles to an immune sample of known concentration; and the concentrations of the multiple groups of immune samples of different concentrations are set in a stepwise manner.

[0015] It can be that the AI ​​model feature data set cooperates with the AI ​​computing model to identify the number or area of ​​unclustered detection particles in the image; It can be that the AI ​​model feature data set is combined with the AI ​​computing model to identify the number or area of ​​aggregated particles in the image.

[0016] It can be that the detection signal value or the inverse of the signal value = (number of aggregated particles×coefficient A1) / (number of single particles×coefficient B1), coefficient A1 is equal to the average number of particles contained in the aggregated particles, and coefficient B1 is the adjustment coefficient.

[0017] It can be that the detection signal value or the inverse of the signal value = (area of ​​aggregated particles×coefficient A3) / (area of ​​single particle×coefficient B3); coefficient A3 is the conversion coefficient for converting the total area of ​​aggregated particles into the area of ​​a single layer of particles, and coefficient B3 is the adjustment coefficient.

[0018] The detection signal value or the reciprocal of the signal value needs to be added with a bias coefficient.

[0019] It can be that the total number or total area of ​​two detection particle aggregation images is MB1; it can be that the total number or total area of ​​three to five detection particle aggregation images is MB2; it can be that the total number or total area of ​​six to ten detection particle aggregation images is MB3; it can be that the total number or total area of ​​more than ten detection particle aggregation images is MB4; it can be that the total number or total area of ​​one detection particle image is MA1; it can be that a concentration immune sample is photographed to obtain multiple microscopic images, and a detection signal value A is calculated for each microscopic image, and the detection signal value corresponding to the concentration immune sample is equal to the average value of the detection signal value A; it can be that the microscopic sample is placed in a detection cavity for photographing, and the height of the detection cavity is greater than 30um and less than 600um; it can be that the detection particles in the microscopic sample have a concentration greater than 10ug / mL and less than 5000 ug / mL.

[0020] It can be that the detection signal value or the reciprocal of the detection signal value = (coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4) / (coefficient D1×MA1); the detection signal value or the reciprocal of the detection signal value needs to add coefficient Q5, coefficient Q5 is the bias coefficient; MA1, MB1, MB2, MB3, MB4 are the total number of corresponding aggregated detection particles; coefficient D1, coefficient C1, coefficient C2, coefficient C3, coefficient C4 are the average number of particles in the corresponding aggregated detection particles.

[0021] It can be that the detection signal value or the reciprocal of the detection signal value = (coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4) / (coefficient D1×MA1); the detection signal value or the reciprocal of the detection signal value needs to add coefficient Q5, and coefficient Q5 is a bias coefficient. MA1, MB1, MB2, MB3, and MB4 are the total areas of the corresponding aggregated detection particles; coefficient D1, coefficient C1, coefficient C2, coefficient C3, and coefficient C4 are the adjustment coefficients of the corresponding aggregated detection particles.

[0022] The solution to the above-mentioned technical problem in the present application can also be a quantitative immunoassay method based on the aggregation of detection particles, comprising: adding detection particles to an immune sample to form a microscopic sample, wherein antibodies or antigens are present on the surface of the detection particles; photographing to obtain a microscopic image; identifying the number or area of ​​unaggregated detection particles in the image; identifying the number or area of ​​aggregated particles in the image; identifying the number or area of ​​unaggregated detection particles in the image; identifying the number or area of ​​aggregated particles in the image; obtaining a detection signal value based on the number or area of ​​unaggregated detection particles and the number or area of ​​aggregated particles; using the fitting formula obtained by the above method to calculate the content of immune components in the sample; or using the immune substance concentration lookup table obtained by the above method to find the content of immune components in the sample.

[0023] It can be that the AI ​​model feature data set is combined with the AI ​​computing model to identify the number or area of ​​unclustered detection particles in the image; it can be that the AI ​​model feature data set is combined with the AI ​​computing model to identify the number or area of ​​aggregated particles in the image.

[0024] It can be that the search obtains the content of immune components in the sample and uses an interpolation algorithm to obtain the concentration of the immune substance.

[0025] The solution to the above-mentioned technical problems in the present application can also be a quantitative immunoassay AI training method based on detection particle aggregation, comprising: adding detection particles to an immune sample to form a microscopic sample, wherein antibodies or antigens are present on the surface of the detection particles; photographing to obtain a microscopic image; labeling the image of non-aggregated detection particles in the image with an identification name A; labeling the image of aggregated particles in the image with an identification name B; using the labeled images to train an AI model to obtain an AI model feature data set; the AI ​​model feature data set is combined with an AI computing model to identify non-aggregated detection particles in the image; the AI ​​model feature data set is combined with an AI computing model to identify aggregated particles in the image; the AI ​​model feature data set is used to obtain a fitting formula for quantitative immunoassay; or the AI ​​model feature data set is used for quantitative immunoassay.

[0026] It can be two images for detecting particle aggregation, with the image labeled with the identification name B1; it can be three to five images for detecting particle aggregation, with the image labeled with the identification name B2; it can be six to ten images for detecting particle aggregation, with the image labeled with the identification name B3; it can be more than ten images for detecting particle aggregation, with the image labeled with the identification name B4; it can be one image for detecting particles, with the image labeled with the identification name LA1; it can be that the microscopic image includes multiple microscopic images, and each microscopic image calculates a detection signal value A, and the detection signal value is equal to the average value of the detection signal value A.

[0027] It may be that the microscopic sample is placed in a detection cavity for photographing, and the height of the detection cavity is greater than 30um and less than 600um.

[0028] The concentration of the detection particles in the microscopic examination sample may be greater than 10 ug / mL and less than 5000 ug / mL.

[0029] The solution to the above technical problem in the present application can also be a detection device or a computing processing device, which is used to run all or part of the above method; the memory of the detection device or the computing processing device includes all or part of the data of all or part of the above method.

[0030] The solution to the above technical problem in the present application may also be 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.

[0031] The technical effects of the above technical solution include: the AI ​​model feature data set cooperates with the AI ​​computing model to identify unaggregated detection particles and aggregated particles in the image; it greatly improves the detection efficiency of detection particles and aggregated particles, and greatly reduces the cost compared to traditional optical absorption detection. It does not require the setting of a precise photoelectric detection system, and only requires microscopic photography to complete immune detection, which is a major breakthrough in this field.

[0032] The technical effects of the above technical solution include: obtaining a fitting formula for quantitative immunoassay based on the AI ​​model feature data set or directly using it for quantitative immunoassay; the quantitative process can eliminate the requirement for high accuracy of the cavity and improve the accuracy of quantitative calculations. Traditional quantification is based on a certain sample size, and accurate quantitative analysis results can only be obtained if the sample volume is accurately quantified. In this application, accurate quantitative analysis can be achieved by obtaining relative data of unaggregated detection particles and aggregated particles based on the AI ​​model feature data set. The requirements for cavity accuracy are reduced, the redundancy of cavity capacity accuracy is higher, and the scope of application is wider.

[0033] The technical effects of the above technical solution include: one detection particle image, two detection particle aggregation images, three to five detection particle aggregation images, six to ten detection particle aggregation images, more than ten detection particle aggregation images, and multi-layer hierarchical detection, which can improve the detection accuracy of different types of aggregations, provide a deeper detection basis for accurate quantitative calculations, and expand the available materials for subsequent signal value calculations. The data in the AI ​​model feature data set accumulated by the quantitative immune detection AI training method is detailed enough to support subsequent more accurate immune project detection.

[0034] The technical effects of the above technical scheme include: a method for obtaining a quantitative immune detection algorithm, obtaining the number or area of ​​unaggregated detection particles and the number or area of ​​aggregated particles based on an AI calculation model; establishing basic rules for immune samples of different concentrations, and displaying the rules in the form of fitting formulas and substance concentration lookup tables to facilitate subsequent concentration detection of target immune substances.

[0035] The technical effects of the above technical solution include: the calculation of the detection signal value integrates the conditions of various aggregated particles and single detection particles; and the corresponding detection signal value calculation method can be selected according to different target detection substances.

[0036] The technical effects of the above technical scheme include: quantitative immunoassay method, which identifies non-aggregated detection particles and aggregated granules based on the AI ​​calculation model, and calculates the detection signal value based on their number or area, and establishes a new concentration detection mechanism. The new mechanism can obtain the concentration of the target immune substance based on the detection signal value by fitting formulas or looking up tables.

[0037] The technical effects of the above technical solution include: as long as the height of the detection cavity is within a suitable range and the above calculation mechanism is effective, the quantitative analysis result of the target immune substance can be obtained, and the target immune substance content per unit volume can be output. The target immune substance content can be the quantity per unit volume or the mass per unit volume.

[0038] The technical effects of the above technical solution include: a quantitative immunoassay method, as long as the concentration of the detection particles is within a suitable range and the above calculation mechanism is effective, the quantitative analysis results of the target immune substance can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flowchart of the quantitative immune detection AI training method; Figure 2 is a flowchart of a method for obtaining a quantitative immunoassay algorithm; Figure 3 is a schematic diagram of the process of the quantitative immunoassay method; Figure 4 is a schematic diagram of the process of the quantitative immunoassay method; Figure 5 It is a microscopic image obtained from a low-concentration immune sample; Figure 6 These are microscopic images obtained from low- and medium-concentration immune samples; Figure 7 These are microscopic images obtained from medium and high concentration immune samples; Figure 8 It is a microscopic image obtained from a high-concentration immune sample; Fig. 9It is a schematic diagram of the annotation of aggregated particles in the microscopic image; Fig.10 It is a schematic diagram of the annotation of aggregated particles in the microscopic image; Fig.11 is a schematic diagram of a single detected particle in a microscopic image; Fig.12 is a schematic diagram of aggregated granules formed by double-aggregation detection particles in a microscopic image; Fig.13 is a schematic diagram of aggregated particles formed by a small number of aggregated test particles in a microscopic image; Fig.14 is a schematic diagram of aggregated granules formed by medium-amount aggregation detection particles in a microscopic image; Fig.15 is a schematic diagram of aggregated granules formed by a large number of aggregated test particles in a microscopic image; Fig.16 It is a table that shows the number of aggregated particles and detected particles with different labels and their corresponding signal values; Fig.17 It is a schematic diagram of the straight line equation fitting; Fig.18 It is a schematic diagram of the four-parameter equation fitting; Fig.19 is a microscopic image corresponding to a concentration sample; Fig. 20 is a microscopic image corresponding to a concentration sample; Fig.21 is a microscopic image corresponding to a concentration sample. DETAILED DESCRIPTION

[0040] The content of this application is further described in detail below in conjunction with the accompanying drawings. 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 an illustration of the general principles of the present invention. The numbers such as "first", "second" and "A" and "B" involved in the present invention are only for the convenience of explanation and do not represent the order relationship in time or space. The combination of letters and numbers "TA", "TB" and "H" involved in the present invention are only for the convenience of explanation, and the specific meaning is determined by the specific words referred to.

[0041] In the prior art, in the quantitative immunoassay method for detecting particle aggregation based on microscopic images, single immunoassay particles or aggregated immunoassay particles in the image are identified, and the non-aggregation ratio or aggregation ratio is calculated based on the number of single particles or aggregated particles at a fixed volume, and finally converted into the target concentration.

[0042] This method requires several parameters to be fixed to obtain accurate results, including 1. the ratio of sample to reagent, 2. the concentration of immunoassay particles in the reagent, 3. the thickness of the reaction container, i.e. the container that carries the test sample, 4. the area of ​​the microscopic image, and 5. the affinity and distribution uniformity of the immunoassay particles. Since the target substance usually tested by the immune response is usually in a state of low concentration, in order to improve its detection accuracy, the volume accuracy of the container cavity or the test sample involved in the calculation must be sufficient; high volume accuracy requirements require the thickness accuracy of the reaction container or the container that carries the test sample.

[0043] Although it is now possible to manufacture high-precision cavities, the manufacturing cost of high-precision cavities rises sharply when the accuracy requirements reach above the micron level. How to reduce the accuracy requirements for cavity height or thickness through methodological efforts while maintaining good measurement accuracy is a technical problem to be solved.

[0044] In the present application, aggregated granules and unaggregated detection particles are obtained based on microscopic image recognition; aggregated granules are aggregated detection particles; immune detection particles are detection particles. Single immune detection particles and aggregated immune detection particles can be identified, and different labels are divided according to the different aggregation degrees of aggregated particles. The number or area of ​​immune detection particles with different aggregation degrees is compared with the number or area of ​​single immune detection particles to obtain a ratio value, and then the ratio is used to convert the target concentration. This can reduce the accuracy requirements for the thickness of the reaction container, reduce the disturbance of the target detection concentration caused by the disturbance of the container volume or height, and improve the performance and stability of the detection.

[0045] like Figure 1 , an embodiment of an AI training method for quantitative immunoassay based on detection particle aggregation, comprising: adding detection particles to a sample to form a detection sample; having antibodies or antigens on the surface of the detection particles; taking images to obtain microscopic images of aggregated particles or detection particles; annotating the images of non-aggregated detection particles in the image with an identification name A; annotating the images of aggregated particles in the image with an identification name B; aggregated particles are aggregated detection particles; using the annotated images to train an AI model to obtain an AI model feature data set; the AI ​​model feature data set is combined with an AI computing model to identify non-aggregated detection particles in the image; the AI ​​model feature data set is combined with an AI computing model to identify aggregated particles in the image; the AI ​​model feature data set is used to obtain a fitting formula for quantitative immunoassay; or the AI ​​model feature data set is used for quantitative immunoassay.

[0046] like Fig. 9 and Fig.10 , is a schematic diagram of the annotation of detection particles and aggregated particles in a microscopic image, in which different labels correspond to different aggregated particles or detection particles.

[0047] In some embodiments of the quantitative immunoassay AI training method, two images of detected particle aggregation are labeled with the identification name B1; three to five images of detected particle aggregation are labeled with the identification name B2; six to ten images of detected particle aggregation are labeled with the identification name B3; more than ten images of detected particle aggregation are labeled with the identification name B4; and one image of detected particle aggregation is labeled with the identification name LA1.

[0048] Figures 11 to 15 It is a partial schematic diagram. Fig.11 is a schematic diagram of a single detected particle in a microscopic image; Fig.12 is a schematic diagram of aggregated granules formed by double-aggregation detection particles in a microscopic image; Fig.13 is a schematic diagram of aggregated particles formed by a small number of aggregated test particles in a microscopic image; Fig.14 is a schematic diagram of aggregated granules formed by medium-amount aggregated test particles in a microscopic image; Fig.15 It is a schematic diagram of aggregated granules formed by a large number of aggregated detection particles in a microscopic image.

[0049] In some embodiments, the microscopic image includes multiple microscopic images, and a detection signal value A is calculated for each microscopic image, and the detection signal value is equal to the average value of the detection signal values ​​A.

[0050] like Figure 2 In an embodiment of a method for acquiring a quantitative immunoassay algorithm based on the aggregation of detection particles, the method includes preparing an immune sample of a set concentration; the immune sample includes detection particles, and antibodies or antigens are present on the surface of the detection particles; some of the detection particles in the immune sample aggregate to form aggregated particles; taking an image to obtain a microscopic image of the aggregated particles or the detection particles; using an AI model feature data set in conjunction with an AI calculation model to identify the number or area of ​​unaggregated detection particles in the image; using an AI model feature data set in conjunction with an AI calculation model to identify the number or area of ​​aggregated particles in the image; obtaining a detection signal value based on the number or area of ​​unaggregated detection particles and the number or area of ​​aggregated particles; obtaining detection signal values ​​corresponding to multiple groups of immune samples of different concentrations; fitting the detection signal values ​​corresponding to the above-mentioned immune samples of different concentrations to obtain a quantitative immunoassay fitting formula; or establishing an immune substance concentration lookup table using the detection signal values ​​corresponding to the above-mentioned immune samples of different concentrations.

[0051] The microscopic examination sample is prepared by adding detection particles to an immune sample of known concentration; the concentrations of multiple groups of immune samples of different concentrations are set according to step concentrations.

[0052] The AI ​​model feature data set is used in conjunction with the AI ​​computing model to identify the number or area of ​​unclustered detection particles in the image; the AI ​​model feature data set is used in conjunction with the AI ​​computing model to identify the number or area of ​​aggregated particles in the image.

[0053] It can be that the detection signal value or the inverse of the signal value = (the number of aggregated particles × coefficient A1) / (the number of single particles × coefficient B1), coefficient A1 is equal to the average number of particles contained in the aggregated particles, and coefficient B1 is the adjustment coefficient. Coefficient A1 is equal to the average number of particles contained in the aggregated particles, and coefficient B1 is generally 1. The value range of coefficient A1 is 2-5.

[0054] The detection signal value or the reciprocal of the signal value may be: (area of ​​aggregated particles × coefficient A3) / (area of ​​a single particle × coefficient B3); coefficient A3 is a conversion coefficient for converting the total area of ​​aggregated particles into the area of ​​a single layer of particles, and coefficient B3 is an adjustment coefficient. The value range of coefficient A3 is 1 to 3; generally 1. Coefficients B1 and B3 are obtained experimentally.

[0055] The total area of ​​aggregated particles and the total area of ​​individual particles are obtained through images. This method does not need to obtain the volume of the sample corresponding to the image.

[0056] The detection signal value or the reciprocal of the signal value needs to be added with a bias coefficient, which is obtained experimentally according to different test systems.

[0057] The number of aggregated particles × coefficient A1 = total number of particles - number of single particles without aggregation. The total number of particles can be equal to the volume multiplied by the particle concentration, but this method requires obtaining the volume of the sample corresponding to the image. In actual detection, the operation is difficult and requires accurate acquisition of the image area and the corresponding sample cavity height. The number of aggregated particles is obtained through the image and then multiplied by the coefficient A1. This method does not need to obtain the volume of the sample corresponding to the image.

[0058] like Figures 5 to 8 , are microscopic images obtained from samples at 4 different concentration states; Figure 5 It is a low-concentration immune sample; Figure 6 It is a low to medium concentration immune sample; Figure 7 It is a medium to high concentration immune sample; Figure 8 It is a high-concentration immune sample; as can be seen from the figure, the number and area of ​​unaggregated detection particles and aggregated particles in the image are different for samples of different concentrations. The difference can be seen based on the naked eye and manual counting. Accurate calculation and modeling also require the use of AI calculation models to identify the number or area of ​​unaggregated detection particles and aggregated particles in the image.

[0059] like Figures 19 to 21 , three groups of microscopic images corresponding to samples with different concentrations are given. The microscopic images of target samples with different concentrations may include multiple microscopic images; a detection signal value A is calculated for each microscopic image, and the detection signal value is equal to the average value of the detection signal value A.

[0060] like Figures 19 to 21 , AI performs aggregated particle or detection particle identification on all microscopic images, calculates the signal value corresponding to each microscopic image based on the AI ​​identification result, and then gives each image a signal value label. The signal values ​​of all images are averaged to obtain the average signal value of the entire group of images. The average signal value is then used to convert the corresponding target substance concentration. Figures 19 to 21 The signal values ​​corresponding to the three pictures are 0, 5, and 10 respectively.

[0061] The number or area of ​​aggregated particles is calculated by ratio to the number or area of ​​single detection particles, eliminating the requirement for accurate calculation of the sample volume involved in the calculation. Based on the above ratio, the relationship between the ratio and the concentration of the target immune substance is established. When establishing a fitting formula or looking up a table, the ratio between the number or area of ​​aggregated particles and the number or area of ​​single detection particles is obtained based on the conditions of multiple known target immune substance concentrations. Because the above reactions all occur in the same volume, only the ratio is calculated to obtain the concentration of the immune target substance. The process of calculating the sample volume concentration involved in the calculation is avoided, and the algorithm steps are also simplified.

[0062] In some embodiments of the quantitative immunoassay algorithm acquisition method, the total number or total area of ​​two detection particle aggregation images is MB1; two detection particles aggregate to form an aggregated particle body. The total number or total area of ​​three to five detection particle aggregation images is MB2; three to five detection particles aggregate to form an aggregated particle body. The total number or total area of ​​six to ten detection particle aggregation images is MB3; six to ten detection particles aggregate to form an aggregated particle body. The total number or total area of ​​more than ten detection particle aggregation images is MB4; more than one detection particle aggregates to form an aggregated particle body. The total number or total area of ​​one detection particle image is MA1.

[0063] In some embodiments of the quantitative immunoassay algorithm acquisition method, the detection signal value = (coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4) / (coefficient D1×MA1); it can be that the detection signal value or the reciprocal of the detection signal value = (coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4) / (coefficient D1×MA1); the detection signal value or the reciprocal of the detection signal value needs to add coefficient Q5, and coefficient Q5 is a bias coefficient; MA1, MB1, MB2, MB3, and MB4 are the total number of corresponding aggregated detection particles; coefficient D1, coefficient C1, coefficient C2, coefficient C3, and coefficient C4 are the average number of particles in the corresponding aggregated detection particles. In general, coefficient D1=1, coefficient C1=2, coefficient C2=3.5, coefficient C3=7.5, and coefficient C4=10.5.

[0064] In some embodiments of the quantitative immunoassay algorithm acquisition method, the detection signal value=(coefficient D1×MA1) / (coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4).

[0065] It can be that the detection signal value or the inverse of the detection signal value = (coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4) / (coefficient D1×MA1); the detection signal value or the inverse of the detection signal value needs to add coefficient Q5, and coefficient Q5 is the bias coefficient. MA1, MB1, MB2, MB3, and MB4 are the total areas of the corresponding aggregated detection particles; coefficient D1, coefficient C1, coefficient C2, coefficient C3, and coefficient C4 are the adjustment coefficients of the corresponding aggregated detection particles. In general, coefficient D1=1, coefficient C1=1.2, coefficient C2=1.5, coefficient C3=2.1, and coefficient C4=3.2. Because the average number of superimposed areas is 2-3 layers, the adjustment coefficient is around 1-3.

[0066] In other embodiments, the detection signal value can also be calculated by the following method: detection signal value = (number of single detection particles × coefficient D1) / (number of double-aggregated detection particles × coefficient C1 + number of small-amount aggregated detection particles × coefficient C2 + number of medium-amount aggregated detection particles × coefficient C3 + number of large-amount aggregated detection particles × coefficient C4). Coefficient C1 is the coefficient corresponding to double-aggregated detection particles. Coefficient C2 is the coefficient corresponding to small-amount aggregated detection particles; small-amount aggregation can be three to five detection particles aggregated. Coefficient C3 is the coefficient corresponding to medium-amount aggregated detection particles; medium-amount aggregation can be six to ten detection particles aggregated. Coefficient C4 is the coefficient corresponding to large-amount aggregated detection particles; large-amount aggregation can be more than ten detection particles aggregated. Depending on different detection items, the coefficients may be different. The boundary between small-amount aggregation, medium-amount aggregation and large-amount aggregation can be adjusted according to actual conditions.

[0067] In other embodiments, the detection signal value can also be calculated by the following method: detection signal value = (area of ​​a single detection particle × coefficient D1) / (area of ​​double-aggregated detection particles × coefficient C1 + area of ​​small-amount aggregated detection particles × coefficient C2 + area of ​​medium-amount aggregated detection particles × coefficient C3 + area of ​​large-amount aggregated detection particles × coefficient C4). Coefficient C1 is the coefficient corresponding to double-aggregated detection particles. Coefficient C2 is the coefficient corresponding to small-amount aggregated detection particles; small-amount aggregation is the aggregation of three to five detection particles. Coefficient C3 is the coefficient corresponding to medium-amount aggregated detection particles; medium-amount aggregation is the aggregation of six to ten detection particles. Coefficient C4 is the coefficient corresponding to large-amount aggregated detection particles; large-amount aggregation is the aggregation of more than ten detection particles. The coefficients may be different depending on different detection items.

[0068] For the same test item, the coefficients D1, C1, C2 and C3 used in particle number calculations and the coefficients D1, C1, C2 and C3 used in particle area calculations may have different specific values.

[0069] In other embodiments, the detection signal value may also be calculated by the following method: detection signal value = single detection particle pixel area / aggregate detection particle pixel area; aggregate detection particle pixel area = MB1+MB2+MB3+MB4.

[0070] In other embodiments of the quantitative immune detection algorithm acquisition method, the division of aggregated particles can be more detailed or more rough, and can be set according to the actual detection requirements of the target immune substance to be detected. In high-precision situations, it can be set more detailed, and in situations where the precision requirements are not high, it can be set to be rough to save computing resources.

[0071] like Figure 3 , an embodiment of a quantitative immunoassay method based on detection particle aggregation, comprising: adding detection particles to a sample to form a detection sample; antibodies or antigens are present on the surface of the detection particles; some of the detection particles in the detection sample aggregate to form aggregated particles; taking images to obtain microscopic images of aggregated particles or detection particles; using an AI model feature data set in conjunction with an AI computing model to identify the number or area of ​​unaggregated detection particles in the image; using an AI model feature data set in conjunction with an AI computing model to identify the number or area of ​​aggregated particles in the image; obtaining a detection signal value based on the number or area of ​​unaggregated detection particles and the number or area of ​​aggregated particles; using the above method to obtain a fitting formula to calculate the content of immune components in the sample; or using the above method to obtain an immune substance concentration lookup table to obtain the content of immune components in the sample.

[0072] In some embodiments, the search obtains the content of immune components in the sample and uses an interpolation algorithm to obtain the concentration of the immune substance.

[0073] The test sample or immune sample is placed in a test cavity for photographing, and the height of the test cavity is greater than 30um and less than 600um.

[0074] The detection particles in the detection sample or immune sample have a concentration greater than 10 ug / mL and less than 5000 ug / mL. Different target detection objects can be matched with corresponding detection particle concentrations.

[0075] The concentration of the detected particles can be 10ug / mL~100ug / mL; it can also be 50ug / mL~80ug / mL.

[0076] The concentration of the detected particles can be 100ug / mL to 1000ug / mL; it can also be 150ug / mL to 800ug / mL. It can also be 200ug / mL to 300ug / mL; it can also be 400ug / mL to 500ug / mL; it can also be 500ug / mL to 600ug / mL; it can also be 600ug / mL to 700ug / mL; it can also be 650ug / mL to 750ug / mL; it can also be 800ug / mL to 900ug / mL.

[0077] The concentration of the detected particles can be 1000ug / mL to 5000ug / mL; 1500ug / mL to 4500ug / mL; 2000ug / mL to 3000ug / mL; 4000ug / mL to 5000ug / mL; 2500ug / mL to 3500ug / mL; and 3600ug / mL to 4700ug / mL.

[0078] 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.

[0079] 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.

[0080] In some embodiments, the steps include: image acquisition - AI identification of labels in the image - label statistics - signal value calculation - standard curve - concentration. First, the image acquisition skills: the formed component analyzer takes a microscopic photo of the sample chip and collects the required microscopic magnification image; AI label recognition: the immune version of AI will automatically identify the immune label in the image, and then count the identified labels to obtain the label data; signal value calculation: the number of labels is calculated according to the formula built into the instrument, and the number of labels is converted into the signal value of the sample test; concentration: after obtaining the signal value, the concentration of the sample test can be calculated according to the standard curve of signal value-concentration built into the instrument.

[0081] Immune AI automatically identifies labels on images and obtains data corresponding to each label, including number and area. Labels include: CSM1 (single particle), CSCMD1 (2-3 aggregated particles), CSCMT1 (4-10 small aggregates of particles), CMCM1 (medium aggregates), and CBCM1 (large aggregates). Calculate according to the label data obtained by AI recognition: signal value = (coefficient n1×CSCMD1+coefficient n2×CSCMT1+coefficient n3×CMCM1+coefficient n4×CBCM1) / (coefficient n5×CSM1); where coefficient n1, coefficient n2, coefficient n3, coefficient n4, and coefficient n5 are set and updated through the main program of the instrument; in the above formula, CSM1, CSCMD1, CSCMT1, CMCM1, and CBCM1 can be either quantity or area. It should be noted that the letters and numbers after the coefficients in this application are only for identification. In different applications, the specific values ​​of the coefficients with the same letter and number numbers are different. In one embodiment, it may be (2×CSCMD1+3×CSCMT1+5×CMCM1+10×CBCM1) / (1×CSM1). The specific coefficients may be reset to corresponding values ​​according to the test requirements of different projects.

[0082] Concentration value calculation: Concentration conversion is performed based on the signal value obtained by AI recognition and calculation: Instrument standard curve preparation: The signal value of the fixed value sample can be obtained by measuring the sample with known fixed value through the standard machine of the formed component analyzer. With the fixed value concentration as X and the test signal value as Y, the standard curve of each batch of reagents can be obtained through a straight line or a four-parameter fitting equation. The fixed value sample means the concentration of the target detection object with known fixed value.

[0083] The fitting method can be a linear equation: Y=kx+b, where Y is the signal value, X is the concentration value, and b is the intercept of the equation. When calculating the fitting equation, k and b in the above equation can be obtained by fitting according to a set of known concentration X values. In addition to linear fitting, corresponding fitting formulas are established according to the characteristics of different target substances, such as quadratic fitting, polynomial fitting or other fitting formulas.

[0084] The fitting method can also be a four-parameter equation: Y=(AD) / (1+(X / C)^B)+D, where Y is the signal value, X is the concentration value, A: the asymptote estimate on the curve; D: the asymptote estimate under the curve; B: the slope of the curve; C: the dose corresponding to half the maximum binding. The calibration curve is built into the formative analyzer, and the signal value of the clinical samples tested later can be converted to the corresponding test concentration through the calibration curve.

[0085] like Fig.16The table shown shows the number of aggregated particles (CSCMD1, CSCMT1, CMCM1, CBCM1) and detection particles (CSM1) with various labels at different detection sample concentrations, and their corresponding signal values ​​in one embodiment.

[0086] like Fig.17 , through the straight line equation fitting: Y=kx+b, the straight line fitting equation y=0.0661x+0.02 (R2=0.9533) is obtained.

[0087] like Fig.18 , four-parameter equation: Y=(AD) / (1+(X / C)^B)+D, and the four-parameter equation Y=(6.0829-0.2444) / (1+(X / 35.1735)^(-2.0931))+(0.2444); the calibration curve is built into the formed analyzer: linear equation K=0.0661, b=0.02; or four-parameter equation: A=6.0829, B=-2.0931, C=35.1735, D=0.2444; the signal value of the clinical sample tested subsequently can be converted to the corresponding test concentration through the calibration curve.

[0088] Although the present invention is illustrated and described according to the preferred embodiment and several alternatives, the invention is not limited by the specific description in this specification. Other additional replacement or equivalent components can also be used to practice the present invention.

Claims

1. A method for obtaining a quantitative immunoassay algorithm based on detecting particle aggregation, characterized in that: include: Adding detection particles to the immune sample to form a microscopic sample, the detection particles have antibodies or antigens on their surfaces; Take photos to obtain microscopic images; Identify the number or area of ​​unaggregated detection particles in the image; Identify the number or area of ​​aggregated particles in the image; Obtaining a detection signal value based on the number or area of ​​unaggregated detection particles and the number or area of ​​aggregated particles; Obtain detection signal values ​​corresponding to multiple groups of immune samples with different concentrations; The quantitative immune detection fitting formula is obtained by fitting the detection signal values ​​corresponding to the immune samples of different concentrations; Alternatively, a lookup table of immune substance concentrations is established using the detection signal values ​​corresponding to the immune samples of different concentrations.

2. The method for obtaining a quantitative immunoassay algorithm based on detecting particle aggregation 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 multiple groups of immune samples with different concentrations are set according to step concentrations.

3. The method for obtaining a quantitative immunoassay algorithm based on detecting particle aggregation according to claim 1, characterized in that: The AI ​​model feature data set is used in conjunction with the AI ​​computing model to identify the number or area of ​​unaggregated detection particles in the image; The AI ​​model feature data set is combined with the AI ​​computing model to identify the number or area of ​​aggregated particles in the image.

4. The method for obtaining a quantitative immunoassay algorithm based on detecting particle aggregation according to claim 1, characterized in that: Includes any of the following characteristics: Characteristic TC10: The detection signal value or the reciprocal of the signal value = (the number of aggregated particles × coefficient A1) / (the number of single particles × coefficient B1), coefficient A1 is equal to the average number of particles contained in the aggregated particles, and coefficient B1 is the adjustment coefficient; Feature TC30: The detection signal value or the reciprocal of the signal value = (area of ​​aggregated particles × coefficient A3) / (area of ​​a single particle × coefficient B3); coefficient A3 is the conversion coefficient of the total area of ​​aggregated particles to the area of ​​a single layer of particles, and coefficient B3 is the adjustment coefficient; Feature TC40: A bias coefficient needs to be added to the detection signal value or the reciprocal of the signal value.

5. The method for obtaining a quantitative immunoassay algorithm based on detecting particle aggregation according to claim 1, characterized in that: Includes one or more of the following features: Feature TB10: The total number or total area of ​​the two detected particle aggregation images is MB1; Feature TB20: The total number or total area of ​​three to five detected particle aggregation images is MB2; Feature TB30: The total number or total area of ​​six to ten detected particle aggregation images is MB3; Feature TB40: The total number or total area of ​​more than ten detected particle aggregation images is MB4; Feature TB50: The total number or total area of ​​a detected particle image is MA1; Feature TB60: A plurality of microscopic images are obtained by photographing an immune sample of one concentration, and a detection signal value A is calculated for each microscopic image, and the detection signal value corresponding to the immune sample of the concentration is equal to the average value of the detection signal value A; Feature TB70: The microscopic sample is placed in a detection cavity for photographing, and the height of the detection cavity is greater than 30um and less than 600um; Feature TB80: The concentration of the detection particles in the microscopic sample is greater than 10 ug / mL and less than 5000 ug / mL.

6. The method for obtaining a quantitative immunoassay algorithm based on detecting particle aggregation according to claim 5, characterized in that: One of the following features: Feature TD10: the detection signal value or the reciprocal of the detection signal value=(coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4) / (coefficient D1×MA1); Feature TD30: The detection signal value or the reciprocal of the detection signal value needs to be added with a coefficient Q5, where the coefficient Q5 is a bias coefficient; MA1, MB1, MB2, MB3, and MB4 are the total numbers of the corresponding aggregated detection particles; Coefficient D1, coefficient C1, coefficient C2, coefficient C3, and coefficient C4 are the average numbers of particles in the corresponding aggregated detection particles.

7. The method for obtaining a quantitative immunoassay algorithm based on detecting particle aggregation according to claim 5, characterized in that: One of the following features: Feature TD10: the detection signal value or the reciprocal of the detection signal value=(coefficient C1×MB1+coefficient C2×MB2+coefficient C3×MB3+coefficient C4×MB4) / (coefficient D1×MA1); Feature TD30: The detection signal value or the reciprocal of the detection signal value needs to be added with a coefficient Q5, where the coefficient Q5 is a bias coefficient; MA1, MB1, MB2, MB3, and MB4 are the total areas of the corresponding aggregated detection particles; Coefficient D1, coefficient C1, coefficient C2, coefficient C3, and coefficient C4 are adjustment coefficients for the corresponding aggregated detection particles.

8. A quantitative immunoassay method based on detecting particle aggregation, characterized in that: include: Adding detection particles to the immune sample to form a microscopic sample, the detection particles have antibodies or antigens on their surfaces; Take photos to obtain microscopic images; Identify the number or area of ​​unaggregated detection particles in the image; Identify the number or area of ​​aggregated particles in the image; Identify the number or area of ​​unaggregated detection particles in the image; Identify the number or area of ​​aggregated particles in the image; Obtaining a detection signal value based on the number or area of ​​unaggregated detection particles and the number or area of ​​aggregated particles; Using the fitting formula obtained by the method of claim 1, the content of immune components in the sample is calculated; Or use the immune substance concentration lookup table obtained by the method of claim 1 to find the immune component content in the sample.

9. The quantitative immunoassay method based on detecting particle aggregation according to claim 8, characterized in that: The AI ​​model feature data set is used in conjunction with the AI ​​computing model to identify the number or area of ​​unaggregated detection particles in the image; The AI ​​model feature data set is combined with the AI ​​computing model to identify the number or area of ​​aggregated particles in the image.

10. The quantitative immunoassay method based on detecting particle aggregation according to claim 8, characterized in that: The search obtains the content of immune components in the sample, and uses an interpolation algorithm to obtain the concentration of the immune substance.

11. A quantitative immunoassay AI training method based on detecting particle aggregation, characterized in that: include: Adding detection particles to the immune sample to form a microscopic sample, the detection particles have antibodies or antigens on their surfaces; Take photos to obtain microscopic images; Annotate the image of the unaggregated detection particles with an identification name A; Annotate the image of aggregated particles with an identification name B; Use the annotated images to train the AI ​​model and obtain the AI ​​model feature data set; The AI ​​model feature data set, in conjunction with the AI ​​computing model, can identify unaggregated detection particles in an image; The AI ​​model feature data set, in conjunction with the AI ​​computing model, can identify aggregated particles in an image; The AI ​​model feature data set is used to obtain a fitting formula for quantitative immunoassay; Or the AI ​​model feature dataset is used for quantitative immune detection.

12. The quantitative immunoassay AI training method based on detecting particle aggregation according to claim 11, characterized in that: Includes one or more of the following features: Feature TA10: two images of detected particle aggregation, with the image annotated with the identification name B1; Feature TA20: three to five images of detected particle aggregation, with the image annotated with the identification name B2; Feature TA30: six to ten images of detected particle aggregation, with image annotation and identification name B3; Feature TA40: more than ten detected particle aggregation images, with image annotation and identification name B4; Feature TA50: a detected particle image with the image annotated with the identification name LA1; Feature TA60: The microscopic image includes a plurality of microscopic images, each microscopic image is calculated to obtain a detection signal value A, and the detection signal value is equal to an average value of the detection signal values ​​A; Feature TA70: The microscopic sample is placed in a detection cavity for photographing, and the height of the detection cavity is greater than 30um and less than 600um; Feature TA80: The concentration of the detection particles in the microscopic sample is greater than 10 ug / mL and less than 5000 ug / mL.

13. 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 12; The memory of the detection device or the computing and processing device includes all or part of the data of the above-mentioned method.

14. 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 12; or storing all or part of the data of the method described in any one of claims 1 to 12.

Images