AI recognition training and detection method for high-sensitivity detection of polymeric particles

Abstract

In the AI recognition training and detection method for high-sensitivity detection of the polymeric particles, a sample is prepared, and the sample comprises the polymeric particles; the agglomerated microparticles are formed by aggregating microparticles; shooting an image of the sample; obtaining an agglomerated microparticle image, and marking the agglomerated microparticle image to obtain a marked agglomerated microparticle image; the identifier comprises an identifier for a single particle, and the identifier comprises an identifier for an agglomerated particle body containing n particles; carrying out AI training by using the identified agglomerated microparticle image to obtain an agglomerated microparticle identification AI feature data set; the aggregated microparticle recognition AI feature data set is matched with AI software, and aggregated microparticles in the image or non-aggregated microparticles are recognized; and calculating the total number or total area of the agglomerated microparticles in the sample image.

Description

Technical Field

[0001] This application belongs to the technical field of in vitro diagnostic immunology, and particularly relates to an AI recognition training and detection method for highly sensitive polymer microparticles. 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 level changes can help doctors diagnose diseases, monitor treatment effects, or predict disease progression.

[0003] The affinity reaction is crucial for the normal function of the immune system because it enables the immune system to recognize and eliminate pathogens, foreign substances, and abnormal cells in the body. In addition, the affinity reaction is also widely used in laboratory techniques and clinical diagnosis. For example, techniques such as ELISA (enzyme-linked immunosorbent assay) and immunohistochemistry utilize the affinity reaction between antibodies and antigens to detect specific molecules or cells.

[0004] In the analytical 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 analysis, and chemiluminescence immunoassay.

[0005] 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 No. "CN201410197209.1" integrates immunomagnetic enrichment and visualization detection. It is simple to operate, requires a short time for the entire detection process, does not require special instrument equipment, and the method of the present invention has low requirements for the purity of antibodies, which can greatly reduce the detection cost. However, this patent only reveals that "the aggregation degree of the aggregates is positively correlated with the concentration of the biomacromolecules or microorganisms", and "determine the concentration of the biomacromolecules or microorganisms in the sample to be detected according to the aggregation degree of the aggregates, specifically: obtain a photo of the aggregates, and determine the concentration of the biomacromolecules or microorganisms in the sample to be detected by quantifying the gray value of the photo", rather than accurately calculating the concentration. The method proposed in this patent application can at most give a qualitative conclusion and cannot give a quantitative measurement result, so it cannot be applied to medical detection.

[0006] Because, during the process of taking photos of the test samples, for the same type of sample, under different thickness conditions, the grayscale values of the taken photos are different. Therefore, the patent application with the application number "CN201410197209.1" cannot accurately measure the content of the target substance in the sample. This type of technology is not mature and cannot be practically applied in the industry. Especially in medical application scenarios that require accurate measurement, it cannot meet the 5% measurement accuracy required for veterinary tests, and when applied to human medical tests, it is even more impossible to achieve a 3% measurement accuracy. Especially for the detection of samples with extremely low target substance content, the detection particles in the test samples basically do not aggregate, and the grayscale values of the photos basically do not change. This application method cannot measure samples with low content.

[0007] The Chinese patent application number "CN202180041737.6" has the application name "Aggregation Induction Assay for Improving Sensitivity". In this application, it is proposed that "a system, device, and method for rapidly and accurately measuring analyte particle-binding-induced reporter particle aggregation are provided. In the presence of analyte particles of interest, the reporter particles form aggregates, and the average particle size thereof increases as the analyte concentration increases. Based on the analysis of the average particle size determined from the sample frames, the presence and / or concentration of the analyte can be determined", and a "calibration curve" is proposed. This "calibration curve" has "horizontal axis = SARS-CoV-2 antibody (mg / dL), vertical axis = aggregation size (pixels)", establishing the relationship between the average aggregation area and the target analyte. After being verified by a large number of experiments, when the change in the content of the target substance fluctuates by more than 20%, the measurement deviation of this method is huge, and it has no application value for medical measurement. This patent application is essentially the same as the Chinese patent application number "CN201410197209.1" with the application name "A Method for Detecting Biological Macromolecules or Microorganisms", and it does not show how to accurately calculate the content of the target substance.

[0008] Facts have proved that there is no direct correlation between the average size of aggregated microparticles and the content of the target substance. It is affected by various factors such as the length of time for preparing the test sample, the magnitude of the vibration force, and the temperature during preparation. It is not scientific and has no application value to use the average size of aggregated microparticles to measure the content of the target substance.

[0009] Under modern technological conditions, it is impossible to achieve absolute uniformity in detecting the diameters of particles. At the same time, the level of specification consistency of particle diameters directly determines the price of the detected particles. Agglomerated microparticles formed with different diameter sizes cannot be distinguished in ordinary grayscale images. Estimating the particle size based on binary images cannot accurately evaluate the concentration of the target object, especially in a low-concentration measurement environment. For example, the volume of two aggregated detected particles with small diameters is smaller than that of a single large detected particle. Therefore, accurate measurement cannot be achieved solely based on the area size of the agglomerated microparticles, especially in low-concentration conditions, and quantitative measurement cannot be carried out. Summary of the Invention

[0010] Through a large number of experiments, repeated inferences, and demonstrations, the applicant has discovered that the concentration of the target detection object is directly related to the number of agglomerated microparticles and also to the degree of binding of the detected microparticles. The degree of binding needs to be obtained through experimental measurement. Among them, the accurate identification of the number of agglomerated microparticles is the basis for highly sensitive detection of aggregated microparticles. The applicant proposes that through a large amount of AI training, the AI software directly identifies a large number of particles. Through a large amount of AI calculations, an AI feature data set of the images of the detected microparticles and the agglomerated microparticles is obtained. The AI feature data set cooperates with the AI software to identify the agglomerated microparticles or the non-agglomerated microparticles in the image. Different-diameter detected microparticles within a certain range can accurately identify individual microparticles or agglomerated microparticles, making it possible to detect highly accurate aggregated microparticles.

[0011] An AI recognition training method for highly sensitive detection of aggregated microparticles includes sample preparation, where the sample includes agglomerated microparticles; the agglomerated microparticles are formed by the aggregation of microparticles; taking an image of the sample; obtaining an image of the agglomerated microparticles, and marking the image of the agglomerated microparticles to obtain a marked image of the agglomerated microparticles; the marking includes marking single microparticles and marking agglomerated microparticles containing n microparticles; using the marked image of the agglomerated microparticles for AI training to obtain an AI feature data set for identifying agglomerated microparticles; the AI feature data set for identifying agglomerated microparticles is used to cooperate with the AI software to identify the agglomerated microparticles in the image.

[0012] It can be to mark the image of the agglomerated microparticles formed by the aggregation of two microparticles to obtain a marked image of the two-particle agglomerated microparticles. It can be to mark the image of the agglomerated microparticles formed by the aggregation of three microparticles to obtain a marked image of the three-particle agglomerated microparticles. It can be to mark the image of the agglomerated microparticles with more than a set number of microparticles to obtain a marked image of the agglomerated microparticles with more than the set number of microparticles.

[0013] It can be that the image of the sample is taken by a microscope camera. It can be that the image of the sample is taken by an image sensor attached to the sample container, and the sample container is made of a transparent material, and the sample is contained in the sample container.

[0014] It can be that detection particles are added to transparent glue, and the transparent glue adsorbs two or more detection particles to form an aggregated particle body; it can be that the sample is a serum sample, and the surface of the detection particles includes an antigen or an antibody; the antibody or antigen in the serum sample binds to the detection particles to form an aggregated particle body; it can be that the sample is a blood sample, and the surface of the detection particles includes an antigen or an antibody; the antibody or antigen in the blood sample binds to the detection particles to form an aggregated particle body; it can be that the detection particles include detection particle A and detection particle B, the surface of detection particle A includes an antigen or an antibody, the surface of detection particle B includes an antigen or an antibody, and detection particle A and detection particle B adsorb each other to form an aggregated particle body; it can be that the detection particles include detection particle A and detection particle B, detection particle A is a magnetic particle, and detection particle A can aggregate with each other.

[0015] It can be that the detection particles include detection particle A and detection particle B, detection particle A is a magnetic particle, detection particle A can attract and aggregate with each other, and detection particle A can attract and aggregate with detection particle B.

[0016] A high-sensitivity polymer particle detection method, in which detection particles are added to a sample to form a detection sample; the analyte in the detection sample binds to the detection particles to form a particle binding body; the particle binding bodies aggregate to form an aggregated particle body; an image of the sample is taken; a sample image is obtained; the aggregated particle body recognition AI feature dataset is used in cooperation with AI software to identify the aggregated particle bodies or the unaggregated particles in the image; the total number or total area of the aggregated particle bodies in the sample image is calculated.

[0017] It can be that the surface of the detection particle includes an antigen or an antibody, and the particle-particle conjugate is an antigen-antibody binding; it can be that the surface of the detection particle includes a protein or an enzyme, and the particle-particle conjugate is an affinity binding; it can be that the surface of the detection particle includes two antigens or antibodies, and a fluorescent group or a quenching group is modified on the antigen or antibody. When the antigen or antibody is attracted and aggregated by the same antigen or antibody, the fluorescent group is quenched and no longer emits fluorescence. The more the aggregation, the less the fluorescence; it can be that the surface of the detection particle includes two antigens or antibodies, and the two antigens or antibodies are modified with fluorescent groups. When the two antigens or antibodies are attracted and aggregated by the same antigen or antibody, the fluorescent group excites fluorescence. The more the aggregation, the more the fluorescence; it can be that the sample is irradiated with a light source, and the light source is a white light or a blue light source; it can be that the sample is excited with a fluorescent light source to excite the antigen / antibody modified with the fluorescent group to obtain fluorescence emission; it can be that the aggregated particulate is a single unaggregated particle or a particulate conjugate with a number less than the set number; the content of the analyte is equal to the maximum value of the particle content minus the total content of the aggregated particulate in the sample image; the volume of the detection sample corresponding to the captured image is VT, and the unit volume content of the aggregated particulate = the total content of the aggregated particulate / VT; the area of the aggregated particulate does not include the aggregated particulate with the number of aggregated particulates greater than or equal to the set number.

[0018] It can be that the volume of the detection sample corresponding to the captured image is VT, and the unit volume content of the cells = the total number of cells / VT.

[0019] It can be that the aggregated particulate is formed by the aggregation of particulate conjugates with a number greater than the set number; according to the total number, total volume, and / or total area of the aggregated particulate, the total content of the analyte is calculated by an empirical formula; the volume of the detection sample corresponding to the captured image is VT, and the volume content of the antigen-antibody detection sample = the total content of the analyte / VT; the area of the aggregated particulate does not include the aggregated particulate with the number of aggregated particulates less than the set number.

[0020] It can be that the sample volume is obtained according to the sample area and sample height corresponding to the captured image; the set number is equal to 3 or 2.

[0021] It can be that the sample includes body fluid or excrement; it can be that the sample includes serum, plasma, whole blood, saliva, local body fluid effusion; it can be that the sample includes urine, feces; the feces are diluted feces; the detection particle can be a polymer particle, a polymer particle or a magnetic bead particle; the diameter of the detection particle is greater than 0.1 micrometer; the calculation formula for the analyte concentration is M = A × Sn - B, where Sn is the area of the aggregated particulate and M is the analyte concentration per unit volume.

[0022] A data storage device stores all or part of the program code for executing the above method; stores all or part of the data of the above method. A detection device is used to execute part or all of the above method or store all or part of the data of the above method.

[0023] The technical effects of the above technical solutions include: using the labeled aggregated microsome images for AI training to obtain an AI feature dataset for identifying aggregated microsomes; being able to effectively support the detection of corresponding finer components and improving the detection efficiency with the help of AI.

[0024] The technical effects of the above technical solutions include: single-particle labeled images; two-particle aggregated microsome images; three-particle aggregated microsome images; labeled images of aggregated microsomes with more than a set number of particles; expanding the AI feature dataset for identifying aggregated microsomes, increasing the scope of objects for AI recognition, and improving the detection efficiency.

[0025] The technical effects of the above technical solutions include: sample captured images, which can be compatible with various image acquisition methods as long as the objects in the images are clear.

[0026] The technical effects of the above technical solutions include: there are various forms of aggregated microsomes, improving the acquisition efficiency of AI training samples.

[0027] The technical effects of the above technical solutions include: being able to be compatible with serum samples and blood samples, reducing the requirements for samples and having better compatibility.

[0028] The technical effects of the above technical solutions include: the combination of detecting particle A and detecting particle B increases the types of aggregated microsomes and expands the recognition scope.

[0029] The technical effects of the above technical solutions include: obtaining an AI feature dataset for identifying aggregated microsomes through AI training and AI image recognition methods; greatly reducing the detection cost of the object to be detected and being able to achieve detection without complex optical system design.

[0030] The technical effects of the above technical solutions include: through AI training and AI image recognition methods, and making the object to be detected that is invisible under a conventional magnification microscope in the form of aggregated microsomes, being able to perform efficient quantitative analysis operations in the form of images. Turning the impossible into engineering realizable, enabling the classification and recognition of more components based on the form of aggregated microsomes, and allowing the industrial application of AI technology in this field.

[0031] The technical effects of the above technical solutions include: the surface of the detection particle includes an antigen or an antibody, and quantitative analysis related to the antigen or antibody components can be achieved based on images.

[0032] The technical effects of the above technical solution include: the forms of aggregated microparticles include the binding between detection microparticles, the binding of antibodies or antigens to detection microparticles, and the binding of cells to microparticles. There are many types, which are suitable for different application scenarios.

[0033] The technical effects of the above technical solution include: it can realize the detection and analysis of various substances such as antigens or antibodies, proteins or enzymes, and is an efficient solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the steps of an AI recognition training method for highly sensitive detection of aggregated microparticles;

[0035] Figure 2 It is a micrograph in which the detection microparticles are relatively uniform without the target substance;

[0036] Figure 3 It is a schematic diagram of a single microparticle;

[0037] Figure 4 It is a schematic diagram of an aggregated microparticle with two aggregated microparticles;

[0038] Figure 5 It is a micrograph in which the detection microparticles are aggregated with the target substance;

[0039] Figure 6 It is a schematic diagram of an aggregated microparticle with three aggregated microparticles;

[0040] Figure 7 It is a schematic diagram of an aggregated microparticle with more than the set number of microparticles;

[0041] Figure 8 It is a schematic diagram of the sample preparation process;

[0042] Figure 9 It is a schematic diagram of the sample preparation process;

[0043] Figure 10 It is a schematic diagram of the sample preparation process;

[0044] Figure 11 It is a schematic diagram of the sample preparation process;

[0045] Figure 12 It is a schematic diagram of the identification of the binding of detection microparticles and cells;

[0046] Figure 13 It is a schematic diagram of the identification of the binding of detection microparticles and cells;

[0047] Figure 14 It is a schematic diagram of a highly sensitive target substance detection method;

[0048] Figure 15Schematic diagram of a high-sensitivity target substance detection method;

[0049] Figure 16 Schematic diagram of the steps of a target substance detection method based on particle aggregation;

[0050] Figure 17 Schematic diagram of the substance to be detected that is invisible in the sample;

[0051] Figure 18 Schematic diagram of the structures of traditional monoclonal antibodies, heavy chain antibodies, and nanobodies;

[0052] Figure 19 Schematic diagram of antibodies or antigens coated on particles such as latex or magnetic beads;

[0053] Figure 20 Schematic diagram of the coated particles adsorbing and aggregating with each other;

[0054] Figure 21 Schematic diagram of obtaining aggregated microparticles through AI image recognition;

[0055] Figure 22 Schematic diagram of obtaining an image through binarization of the image recognized by AI.

[0056] Figure 23 The aggregated microparticles recognized by AI;

[0057] Figure 24 Schematic diagram of taking an array image and the detection area corresponding to one image;

[0058] Figure 25 Top view schematic diagram of a detection chip;

[0059] Figure 26 Cross-sectional view schematic diagram of the detection cavity and the sample injection channel inside the detection chip. Detailed implementation manners

[0060] The following further details the content of this application in conjunction with each attached drawing. It should be noted that the following is an illustration of the preferred embodiments of the present invention and does not constitute any limitation to the present invention. The illustration of the preferred embodiments of the present invention is only for the illustration of the general principle of the present invention. The numbers such as "first", "second", "A", and "B" involved in the present invention are only for the convenience of illustration and do not represent the order relationship in time or space. The combinations of letters and numbers such as "TA", "TB", and "H" involved in the present invention are only for the convenience of illustration, and the specific meanings are determined by the specific words they represent.

[0061] Such as Figure 1, in some embodiments, in sample preparation, the sample includes aggregated microsomes; the aggregated microsomes are formed by aggregation of microsomes with microsomes; an image of the sample is taken; an image of the aggregated microsomes is obtained, the image of the aggregated microsomes is labeled to obtain a labeled image of the aggregated microsomes; the labeled image of the aggregated microsomes is used for AI training to obtain an AI feature dataset for recognizing aggregated microsomes; the AI feature dataset for recognizing aggregated microsomes is used in cooperation with an AI software to recognize the aggregated microsomes in the image.

[0062] Such as Figure 3 , in some embodiments, a single microsome image is labeled to obtain a labeled image of a single labeled microsome. Such as Figure 4 , in some embodiments, an image of an aggregated microsome formed by aggregation of two microsomes is labeled to obtain a labeled image of an aggregated microsome of two aggregated microsomes.

[0063] Such as Figure 6 , in some embodiments, an image of an aggregated microsome formed by aggregation of three microsomes is labeled to obtain a labeled image of an aggregated microsome of three aggregated microsomes. Such as Figure 5 And Figure 7 , in some embodiments, in the case of an image of an aggregated microsome with more than a set number of microsomes, the image is labeled to obtain a labeled image of an aggregated microsome with more than a set number of aggregated microsomes.

[0064] In some embodiments, the image of the sample is taken by a microscope camera. In some embodiments, the image of the sample is taken by attaching an image sensor to a sample container, the sample container is made of a transparent material, and the sample is contained in the sample container.

[0065] In some embodiments, a transparent glue is added to the detection microsomes, and the transparent glue adsorbs two or more detection microsomes to form aggregated microsomes.

[0066] Such as Figure 8 , in some embodiments, the sample is a serum sample, and the surface of the detection microsomes includes an antigen or an antibody; the antibody or antigen in the serum sample binds to the detection microsomes to form aggregated microsomes.

[0067] Such as Figure 9 , in some embodiments, the sample is a blood sample, and the surface of the detection microsomes includes an antigen or an antibody; the antibody or antigen in the blood sample binds to the detection microsomes to form aggregated microsomes.

[0068] Such as Figure 10 , in some embodiments, the detection microsomes include detection microsome A and detection microsome B, the surface of detection microsome A includes an antigen or an antibody, the surface of detection microsome B includes an antigen or an antibody, and detection microsome A and detection microsome B adsorb each other to form aggregated microsomes.

[0069] Such as Figure 11, in some embodiments, the detected particles include detected particle A, detected particle B, and detected particle C. The surface of detected particle A includes an antigen or an antibody, the surface of detected particle B includes an antigen or an antibody, and the surface of detected particle C includes an antigen or an antibody. Detected particle A and detected particle B adsorb to each other to form an aggregated particulate body.

[0070] Such as Figure 12 , in some embodiments, the sample contains cells, an image of the sample is taken; a cell image is obtained, the cell image is identified, and an identified cell image is obtained; the cells include any one or more of red blood cells, white blood cells, and platelets; the identified cell image is identified to obtain an identified cell image; the identified cell image is used for AI training to obtain an aggregated particulate body recognition AI feature dataset; the aggregated particulate body recognition AI feature dataset is used in cooperation with an AI software to identify the cells in the image.

[0071] Such as Figure 12 And Figure 13 , in some embodiments, the sample contains cells, the surface of the detected particle includes an antigen or an antibody, and the antigen or antibody can bind to the antigen or antibody on the cell surface to form a cell-particle conjugate; an image of the sample is taken; a cell-particle conjugate image is obtained, the cell-particle conjugate image is identified, and an identified cell-particle conjugate image is obtained; the identified image is used for AI training to obtain an aggregated particulate body recognition AI feature dataset; the aggregated particulate body recognition AI feature dataset is used in cooperation with an AI software to identify the cell-particle conjugates in the image.

[0072] Figure 12 And Figure 13 Among them, the larger ones are cells and the smaller ones are detected particles. Such as Figure 12 And Figure 13 , in some embodiments, the cell-particle conjugates bind to each other to form a cell-particle aggregate; an image of the sample is taken; a cell-particle aggregate image is obtained, the cell-particle aggregate image is identified, and an identified cell-particle aggregate image is obtained; the identified image is used for AI training to obtain an aggregated particulate body recognition AI feature dataset; the aggregated particulate body recognition AI feature dataset is used in cooperation with an AI software to identify the cell-particle conjugates in the image.

[0073] In an embodiment of a high-sensitivity polymer particle detection method, detected particles are added to a sample to form a detection sample; the analyte in the detection sample binds to the detected particles to form particle conjugates; the particle conjugates aggregate to form aggregated particulate bodies; an image of the sample is taken; a sample image is obtained; the aggregated particulate body recognition AI feature dataset is used in cooperation with an AI software to identify the aggregated particulate bodies in the image; the total number, total area, and / or total volume of the aggregated particulate bodies in the sample image are calculated.

[0074] In some embodiments, the surface of the detection microparticles comprises an antigen or an antibody, and the microparticle conjugate is an antigen-antibody binding. In some embodiments, the surface of the detection microparticles comprises a protein or an enzyme, and the microparticle conjugate is an affinity binding.

[0075] In some embodiments, the surface of the detection microparticles comprises two antigens or antibodies, and the antigen or antibody is modified with a fluorescent group or a quenching group. When the antigen or antibody is attracted and aggregated by the same antigen or antibody, the fluorescent group is quenched and no longer emits fluorescence. The more aggregated, the less fluorescence.

[0076] In some embodiments, the surface of the detection microparticles comprises two antigens or antibodies, and the two antigens or antibodies are modified with fluorescent groups. When the two antigens or antibodies are attracted and aggregated by the same antigen or antibody, the fluorescent groups emit fluorescence. The more aggregated, the more fluorescence. In some embodiments, the sample is irradiated with a light source, and the light source is a white light or blue light source. In some embodiments, the sample is excited with a fluorescent light source to excite the antigen / antibody modified with the fluorescent group to obtain fluorescence emission.

[0077] In some embodiments, the aggregated microparticles are single unaggregated microparticles or microparticle conjugates with a number less than a set number; the content of the analyte is equal to the maximum value of the microparticle content minus the total content of the aggregated microparticles in the sample image; the volume of the detection sample corresponding to the captured image is VT, and the unit volume content of the aggregated microparticles = the total content of the aggregated microparticles / VT; the area of the aggregated microparticles does not include the aggregated microparticles with the number of aggregated microparticles greater than or equal to the set number.

[0078] Such as Figure 14 , in some embodiments, the aggregated microparticles are formed by the aggregation of microparticle conjugates with a number greater than a set number; the total content of the analyte is calculated by an empirical formula based on the total number, total volume, and / or total area of the aggregated microparticles; the volume of the detection sample corresponding to the captured image is VT, and the volume content of the antigen-antibody detection sample = the total content of the analyte / VT; the area of the aggregated microparticles does not include the aggregated microparticles with the number of aggregated microparticles less than the set number. The corresponding detection sample volume VT can be the volume V obtained by multiplying the image area and the sample spreading height H. Such as Figure 15 , in some embodiments, the sample concentration can be transformed, and the corresponding detection sample volume VT after transformation can be N times the volume V.

[0079] The sample volume is obtained based on the sample area and sample height corresponding to the captured image; the set number is equal to 3 or 2.

[0080] In some embodiments, the sample includes body fluids or excreta. In some embodiments, the sample includes serum, plasma, whole blood, saliva, local body fluid effusion. In some embodiments, the sample includes urine, feces; the feces are diluted feces.

[0081] In some embodiments, the detection particles are polymer particles, polymer particles or magnetic bead particles. In some embodiments, the diameter of the detection particles is greater than 0.1 micron.

[0082] In some embodiments, the calculation formula for the concentration of the analyte is M = A × Sn - B, where Sn is the area of the aggregated particulate matter, and M is the concentration of the analyte per unit volume. The concentration of the analyte per unit volume is the quantity or content of the analyte in a unit volume.

[0083] The calculation of the concentration of the analyte can be converted from any one of the total number, total volume, and total area of the aggregated particulate matter. Figure 14 Among them, the number of aggregated particulate matter per unit volume is calculated from the number of aggregated particulate matter, and then converted into the quantity or content of the analyte in a unit volume.

[0084] A computing and processing device is used to run all or part of the above method; the memory of the computing and processing device includes all or part of the data of the above method. A data storage device stores all or part of the program code for executing the above method; stores all or part of the data of the above method. A detection device is used to execute part or all of the above method; or stores all or part of the data of the above method.

[0085] As Figure 16 , a detection method for detecting a target substance based on particle aggregation includes adding detection particles to a sample; the target substance in the sample binds to the detection particles to form a particle conjugate; the particle conjugates aggregate to form aggregated particulate matter; taking an image of the sample under a microscope; analyzing the image to obtain the number, area or volume of the aggregated particulate matter, and obtaining the quantity or content of the target substance according to the binding degree between the target substance and the detection particles.

[0086] As Figure 17 , the target substance in the sample, such as substances like proteins, small molecules, amino acids, sugars, enzymes, etc., substances that cannot be observed or are difficult to observe under an optical microscope, cannot be directly observed and detected with an ordinary optical microscope.

[0087] As Figure 18 , in an antigen-antibody detection system, specific antibodies react with the substance to be detected, but the antibodies and antigens cannot be directly observed and detected with an ordinary optical microscope. Traditional monoclonal antibodies, heavy chain antibodies, and nanobodies cannot be directly observed under a traditional microscope.

[0088] As Figure 19 , antibodies or antigens are coated on particles such as latex or magnetic beads. The diameter of the particles is greater than 0.1 μm, and the particles can be directly observed and detected with an ordinary optical microscope.

[0089] As Figure 20, the encapsulated microparticles are adsorbed to each other or aggregated to form large clusters of aggregated microparticles. By observing the number, area, or volume of the aggregated microparticles under a microscope, information on the quantity or content of the target substance can be obtained.

[0090] For example Figure 2 , for the actually photographed detection microparticles, there is no target substance in the sample, and the detection microparticles are relatively evenly distributed, with only a very small number of microparticles aggregating into two small clusters or so. For large-sized detection microparticles (such as 1000nm magnetic bead detection microparticles), when there is no target substance in the sample or the content is extremely low, the detection microparticles have good dispersibility.

[0091] For example Figure 5 , for the actually photographed detection microparticles, the sample contains the target substance, and some microparticles aggregate into clusters. As the concentration of the corresponding target substance increases, the individually dispersed detection microparticles aggregate to produce aggregated detection microparticles, and the aggregated detection microparticles are clearly identifiable after aggregation.

[0092] For example Figure 23 , during detection, detection microparticles with a fixed concentration are added, the number of aggregated detection microparticles is identified, and the antigen concentration is quantified. By identifying and counting the microparticles that aggregate into clusters, and based on the binding degree between the microparticles and the target substance, the quantity or content of the target substance can be calculated. The binding degree between the microparticles and the target substance is calculated based on the particle size, the amount of coated antibody, and the degree of aggregation.

[0093] The detection method described above; the microparticles can be polymer microparticles or magnetic bead microparticles. The polymer microparticles can be polystyrene microspheres, i.e., latex microparticles. The polystyrene microspheres, i.e., latex microparticles, adsorb proteins. Through adsorption, the invisible microparticles can expand to visible microparticles, and by detecting the number and size of the visible microparticles, the amount of protein can be detected.

[0094] Various modifications can be carried out on the surface of polystyrene. For example, hydrophilic modification. The main surface modification groups of the polymer microparticles are functional groups such as polysaccharides, acrylamides, polyvinyl alcohols, and polyamines. After modification, the polystyrene can selectively adsorb different target substances. Therefore, the present invention can not only be applied to the detection of antibodies and antigens, but also to the detection of various target substances or non-target substances.

[0095] The detection method described above; the detection microparticles adsorb or conjugate antibodies or antigens. The detection of antibodies and antigens is a very important detection item. The modification methods or coating methods of magnetic bead microspheres in various existing technologies can all be applied to the modification and coating of the microparticles in this application. The modified or coated microparticles can selectively aggregate when encountering the target substance to be detected.

[0096] The above detection method; the sample includes body fluids or excreta; the body fluids include serum, plasma, whole blood, saliva, local body fluid accumulation, and the excreta include urine and feces; the feces are diluted feces.

[0097] The above detection method; according to the sample area and sample height corresponding to the captured image, the sample volume is obtained, and the unit volume content of the aggregated microparticles is calculated based on the number of aggregated microparticles and the sample volume.

[0098] Such as Figure 24 , during the microscopic imaging process, the field of view is relatively small, and the volume corresponding to each imaging is very small. By selecting the sample area and sample height corresponding to the area of the image, the volume calculated for each imaging can be obtained. By increasing the number of captured images, a larger sample volume can be obtained, improving the detection accuracy.

[0099] The above detection method; according to the binding degree between the target analyte and the detection microparticles and the unit volume content of the aggregated microparticles, the unit volume content of the target analyte is calculated.

[0100] Such as Figure 24 , by identifying the aggregated microparticles in the sample volume, the content or quantity per unit volume corresponding to the sample volume can be calculated, and the detection result can be converted into a measurement index in the prior art, such as virus content, the unit volume content of a specific protein, etc.

[0101] The above detection method; irradiate the aggregated microparticles with excitation light, and the aggregated microparticles are excited by the excitation light to emit fluorescence, and the number, area, volume, or fluorescence intensity of the aggregated microparticles emitting fluorescence is obtained.

[0102] Such as Figure 25 , a detection chip for detecting antibody-antigen, including a sample injection channel 1010, a detection cavity 1020, and a sample injection port 1011; the detection cavity is used to hold the sample; one end of the sample injection channel is communicated with the detection cavity; the other end of the sample injection channel is communicated with the sample injection port, and the sample injection port is communicated with the external atmosphere; the upper and lower parts of the detection cavity include transparent windows, and external illumination light can enter the detection cavity through the transparent windows; when placed horizontally, the sample injection port is higher than the top of the detection cavity; the sample includes detection microparticles, and the detection microparticles are adsorbed or conjugated with antibodies or antigens; the target analyte in the sample binds to the detection microparticles to form a microparticle conjugate; the microparticle conjugates aggregate to form aggregated microparticles; through the transparent window, an image of the sample can be captured; the number, area, or volume of the aggregated microparticles is obtained.

[0103] Such as Figure 26, A cross-sectional schematic diagram of the detection chip detecting cavity and the injection channel. The sample is accommodated inside the detection cavity 1120. The injection channel 1110 is connected to the injection port 1111. The injection port 1111 is higher than the upper surface 1121 of the detection cavity, so that the sample will not overflow from the injection port 1111. At the same time, due to the internal pressure of the liquid, the air will be automatically discharged.

[0104] The upper and lower surfaces of the detection cavity are transparent windows, which can introduce light to obtain an image of the agglomerated microparticles inside.

[0105] The above detection chip; the polymer microparticles include polystyrene microspheres, i.e., latex microparticles.

[0106] According to the corresponding sample area in the captured image and the height of the detection cavity, the sample volume is obtained. The height H of the detection cavity is used to calculate the sample volume. Sample volume = the corresponding sample area in the image × the height H of the detection cavity.

[0107] The above detection chip; according to the binding degree between the target analyte and the microparticles and the unit volume content of the agglomerated microparticles, the unit volume content of the target analyte is calculated. The unit volume content of the agglomerated microparticles is calculated according to the number of agglomerated microparticles and the sample volume. The number of agglomerated microparticles is obtained from the image; the unit volume content of the agglomerated microparticles = the number of agglomerated microparticles ÷ the sample volume; the unit volume content of the target analyte = the unit volume content of the agglomerated microparticles × the binding degree between the target analyte and the microparticles.

[0108] Such as Figure 24 , During the process of taking pictures, multiple pictures are taken to improve the sample volume. In the captured images, all or part are selected. The area of the selected image corresponds to an area inside the chip that is 0.3 mm long, 0.2 mm wide, and the height is 0.4 mm. The height is determined by the height of the internal cavity of the chip.

[0109] Furthermore, according to the binding degree between the target analyte and the microparticles and the unit volume content of the agglomerated microparticles, the unit volume content of the target analyte is calculated.

[0110] In a certain concentration range, the binding degree between the target analyte and the microparticles is relatively fixed. With the unit volume content of the agglomerated microparticles, multiplying by the binding degree gives the unit volume content of the target analyte. The binding degree of the microparticles is obtained by calibration experiments.

[0111] The sample can be serum, and the target analyte can be various viruses, bacteria, antibodies caused by various pathogens, and the microparticles are coated with the corresponding antigens.

[0112] A reagent for antibody-antigen detection, comprising detection microparticles, with the detection microparticles adsorbing or conjugating antibodies or antigens; during the detection process, the reagent is mixed with a sample; the target analyte in the sample binds to the detection microparticles to form microparticle conjugates; the microparticle conjugates aggregate to form aggregated microparticle bodies; an image of the aggregated microparticle bodies is obtained, and by analyzing the number of aggregated microparticle bodies in the image, the amount of antibodies or antigens in the sample is obtained.

[0113] The detection microparticles can be added to various detection liquids, and the concentration of the detection microparticles can be pre-prepared so that the detection microparticles are dispersed in the detection liquid. During the test, it is added according to the volume ratio, which is convenient for the detection process. There are detection microparticles in the reagent, which can serve as the target for focusing of the optical microscopy system to assist in focusing. The detection microparticles can be polymer microparticles or magnetic bead microparticles. The polymer microparticles can be polystyrene microspheres, i.e., latex microparticles. The diameter of the detection microparticles can be greater than 0.1 μm. The diameter of the detection microparticles can be 0.3 - 3 μm. If the diameter is too small, such as less than 0.1 μm, it cannot be seen under a conventional microscope magnification of 40 times. If it is greater than 3 μm, the antigen-antibody binding force cannot cause it to aggregate, and it is difficult to aggregate, so detection cannot be carried out.

[0114] Such as Figure 21 , a calculation method for detecting the content of a target substance based on the microparticle aggregation image, including selecting one or more microscopic test pictures as the calculation selection pictures; selecting all or part of them as the test image area in the calculation selection pictures; obtaining the test sample volume according to the test image area and the sample height; obtaining the number or area or volume of the aggregated microparticle bodies in the test image area; obtaining the number or content of the target substance according to the number or area or volume information of the aggregated microparticle bodies; dividing the number or content of the target substance by the test sample volume to obtain the content or number per unit volume of the target substance.

[0115] The method for calculating the sample volume can be to obtain the test image area by multiplying the pixel area of the image sensor by the number of pixel points in the test image area; obtaining the test sample area by dividing the test image area by the microscope magnification; multiplying the test sample area by the test sample height to obtain the test sample volume.

[0116] Obtaining the number or area or volume of the aggregated microparticle bodies in the test image area can be to obtain the number or area or volume of the aggregated microparticle bodies through AI image recognition.

[0117] Such as Figure 22, photos of the target substance at three concentrations taken with the detection device of "Shenzhen Anlv Medical Technology Co., Ltd.". Picture A corresponds to a very low content of the target analyte in the sample, Picture B contains a small amount of the target analyte, and Picture C contains a relatively high concentration of the target analyte. First, the detection particles and agglomerated microsomes are identified, and binarization processing is performed on the identified partial positions to obtain pictures A1, B1, and C1. It can be clearly seen that the area of the agglomerated microsomes changes with the content of the target analyte.

[0118] Through AI image recognition, the number or area or volume of non-agglomerated microsomes in the test image area is measured. The more the number or area or volume of non-agglomerated microsomes recognized, the less the content of the target substance.

[0119] The target substance can be a substance with antigenic activity, and the substance with antigenic activity includes any one of proteins, nucleic acids, and polysaccharides.

[0120] The target substance can also be a substance with a cellular structure, and the substance with a cellular structure includes any one of platelets, blood parasites, and red blood cells.

[0121] Although the present invention is described and illustrated according to preferred embodiments and several alternative options, 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. An AI recognition training method for highly sensitive detection of aggregated microparticles, characterized in that, Including, Sample preparation, where the sample includes aggregated microsomes; The aggregated microsomes are formed by the aggregation of microsomes with microsomes; Taking an image of the sample; obtaining an image of the aggregated microsomes, and identifying the image of the aggregated microsomes to obtain an identified image of the aggregated microsomes; The identification includes identifying single microsomes, and the identification includes identifying aggregated microsomes containing n microsomes; Using the identified image of the aggregated microsomes for AI training to obtain an AI feature dataset for identifying aggregated microsomes; The AI feature dataset for identifying aggregated microsomes is used in cooperation with AI software to identify aggregated microsomes in an image.

2. The AI recognition training method for highly sensitive detection of aggregated microparticles according to claim 1, characterized in that, Including any one or more of the following technical features: Feature TA1: Identifying an image of aggregated microsomes formed by the aggregation of two microsomes to obtain an identified image of the two-aggregated microsomes; Feature TA2: Identifying an image of aggregated microsomes formed by the aggregation of three microsomes to obtain an identified image of the three-aggregated microsomes; Feature TA3: Identifying an image of aggregated microsomes with more than a set number of microsomes to obtain an identified image of the aggregated microsomes with more than the set number of microsomes; Feature TA5: The image of the sample is taken using a microscopic camera; Feature TA6: The image of the sample is taken by attaching an image sensor to the sample container, the sample container is made of a transparent material, and the sample is contained in the sample container.

3. The AI recognition training method for highly sensitive detection of aggregated microparticles according to claim 1, characterized in that Including any one or more of the following technical features: Feature TB1: Adding detection microsomes to a transparent glue, and the transparent glue adsorbs two or more detection microsomes to form aggregated microsomes; Feature TB2: The sample is a serum sample, and the surface of the detection microsomes includes an antigen or an antibody; an antibody or an antigen in the serum sample binds to the detection microsomes to form aggregated microsomes; Feature TB3: The sample is a blood sample, and the surface of the detection microsomes includes an antigen or an antibody; an antibody or an antigen in the blood sample binds to the detection microsomes to form aggregated microsomes; Feature TB4: The detection microsomes include detection microsome A and detection microsome B, the surface of detection microsome A includes an antigen or an antibody, the surface of detection microsome B includes an antigen or an antibody, and detection microsome A and detection microsome B adsorb each other to form aggregated microsomes; Feature TB5: The detection microsomes include detection microsome A and detection microsome B, detection microsome A is a magnetic particle, and detection microsome A can aggregate with each other. Feature TB6: The detection microsomes include detection microsome A and detection microsome B, detection microsome A is a magnetic particle, detection microsome A can attract and aggregate with each other, and detection microsome A can attract and aggregate with detection microsome B.

4. A high-sensitivity detection method for polymeric microparticles, characterized in that Detection microsomes are added to the sample to form a detection sample; The analyte in the detection sample binds to the detection microsomes to form a microparticle conjugate; The microparticle conjugates aggregate to form aggregated microsomes; Taking an image of the sample; obtaining a sample image; Using the AI feature dataset for identifying aggregated microsomes in cooperation with AI software to identify aggregated microsomes or unaggregated microsomes in the image; Calculating the total number or total area of the aggregated microsomes in the sample image.

5. The high-sensitivity polymer particle detection method according to claim 4, wherein Any one or more of the following features; Feature TA10: The surface of the detection particle includes an antigen or an antibody, and the particle and the particle conjugate are antigen-antibody binding; Feature TA20: The surface of the detection particle includes a protein or an enzyme, and the particle and the particle conjugate are affinity binding; Feature TA40: The surface of the detection particle includes two antigens or antibodies, and the antigen or antibody is modified with a fluorescent group or a quenching group. When the antigen or antibody is attracted and aggregated by the same antigen or antibody, the fluorescent group is quenched and no longer emits fluorescence. The more aggregated, the less fluorescence; Feature TA41: The surface of the detection particle includes two antigens or antibodies, and the two antigens or antibodies are modified with fluorescent groups. When the two antigens or antibodies are attracted and aggregated by the same antigen or antibody, the fluorescent groups emit fluorescence. The more aggregated, the more fluorescence; Feature TA50: The sample is irradiated with a light source, and the light source is a white light or a blue light source; Feature TA60: The sample is excited with a fluorescent light source to excite the antigen / antibody modified with a fluorescent group to obtain fluorescence emission.

6. The highly sensitive polymeric particle detection method according to claim 7, wherein The aggregated particulate matter is unaggregated single particles or particulate conjugates less than a set number; The content of the analyte is equal to the maximum value of the particle content minus the total content of the aggregated particulate matter in the sample image; The volume of the detection sample corresponding to the captured image is VT, and the unit volume content of the aggregated particulate matter = the total content of the aggregated particulate matter / VT; The area of the aggregated particulate matter does not include aggregated particulate matter with the number of aggregated particles greater than or equal to the set number.

7. The highly sensitive polymeric particle detection method according to claim 8, wherein The volume of the detection sample corresponding to the captured image is VT, and the unit volume content of the cells = the total number of cells / VT.

8. The highly sensitive polymeric particle detection method according to claim 7, wherein The aggregated particulate matter is formed by the aggregation of particulate conjugates greater than a set number; The total content of the analyte is calculated by an empirical formula based on the total number, total volume, and / or total area of the aggregated particulate matter; The volume of the detection sample corresponding to the captured image is VT, and the volume content of the antibody-antigen detection sample = the total content of the analyte / VT; The area of the aggregated particulate matter does not include aggregated particulate matter with the number of aggregated particles less than the set number.

9. The highly sensitive polymeric particle detection method according to claim 6, wherein The sample volume is obtained based on the sample area and sample height corresponding to the captured image; the set number is equal to 3 or 2.

10. The high-sensitivity aggregated particle detection method according to claim 9, wherein Any one or more of the following features; Feature TB1: The sample includes body fluid or excrement; Feature TB2: The sample includes serum, plasma, whole blood, saliva, local body fluid effusion; Feature TB3: The sample includes urine, feces; the feces are diluted feces; Feature TB4: The detection particles are polymer particles, polymer particles or magnetic bead particles; Feature TB5: The diameter of the detection particle is greater than 0.1 micrometer; Feature TB6: The formula for calculating the concentration of the analyte to be detected is M = A × Sn - B, where Sn is the area of the aggregated microsomes and M is the concentration of the analyte to be detected per unit volume.

11. A data storage device, characterized in that, it includes any one of the following technical features: TG1: Store all or part of the program code for executing the method described in any one of claims 1 to 10; TG2: Store all or part of the data of the method described in any one of claims 1 to 10.

12. A detection device, characterized in that, it is used to execute part or all of the method described in any one of claims 1 to 10; or store all or part of the data of the method described in any one of claims 1 to 10.

Images