A droplet digital enzyme-linked immunosorbent assay method, device and related medium

Abstract

The present invention discloses a droplet digital enzyme-linked immunosorbent assay method, device and related medium, the method comprising: obtaining a droplet image; identifying the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet; classifying the droplets according to the number of magnetic beads to obtain multiple droplet types; for the target droplets in each droplet type, statistically obtaining the positive probability and negative probability of the target droplet; based on the positive probability and negative probability, constructing a Poisson distribution and likelihood function of the average enzyme on the magnetic beads, obtaining the number of the average enzyme according to the Poisson distribution and likelihood function, and then outputting the number of the average enzyme as a measurement result. The present invention uses Poisson distribution and likelihood function to perform statistical analysis on different types of droplets, which can improve the accuracy and applicability of the analysis and test results of the enzyme-linked immunosorbent assay, thereby reducing the difficulty of downstream analysis.

Description

Technical Field

[0001] The present invention relates to the field of biological testing technology, and in particular to a droplet digital enzyme-linked immunosorbent assay method, device and related medium. Background Art

[0002] Accurate and sensitive biomarker detection plays an important role in clinical diagnosis. Traditional detection methods, such as enzyme-linked immunosorbent assay (ELISA), chemiluminescent immunoassay (CLIA) and lateral flow assay, can form immune complexes based on specific antigen-antibody binding to generate detection signals. However, when detecting ultra-low concentrations of biomarker targets, these detection methods are insufficient in terms of accuracy and detection limit. Digitizing traditional detection methods can effectively improve detection sensitivity and accuracy. Droplet digital ELISA (ddELISA) has been successfully applied to low-concentration protein detection, and its detection limit is 1000 times higher than that of standard ELISA. However, during the droplet generation process, the number of magnetic beads encapsulated in each droplet follows a Poisson distribution. Digital ELISA requires that only one magnetic bead can be encapsulated in a droplet. If more than one magnetic bead is encapsulated, the probability of the presence of multiple immune complexes increases, thereby increasing the probability of a positive result, which will lead to inaccurate analysis results. Therefore, the existing droplet digital ELISA chooses not to analyze droplets other than droplets encapsulating one magnetic bead, which limits the scope of application of droplet digital ELISA.

[0003] Therefore, how to improve the accuracy and application scope of the analysis and test results of the enzyme-linked immunosorbent assay is a problem that those skilled in the art need to solve. Summary of the invention

[0004] The embodiments of the present invention provide a droplet digital enzyme-linked immunosorbent assay method, apparatus, computer equipment and storage medium, aiming to improve the accuracy and applicability of analysis and test results of enzyme-linked immunosorbent assay.

[0005] In a first aspect, an embodiment of the present invention provides a droplet digital enzyme-linked immunosorbent assay method, comprising:

[0006] Acquire a droplet image;

[0007] The droplet image is identified by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet;

[0008] Classifying the droplets into different types according to the number of magnetic beads to obtain multiple droplet types;

[0009] For the target droplets in each droplet type, the positive probability and negative probability of the target droplets are statistically obtained;

[0010] Based on the positive probability and the negative probability, a Poisson distribution and a likelihood function about the average enzyme on the magnetic beads are constructed, and the number of the average enzyme is obtained according to the Poisson distribution and the likelihood function, and then the number of the average enzyme is output as a measurement result.

[0011] In a second aspect, an embodiment of the present invention provides a droplet digital enzyme-linked immunosorbent assay device, characterized in that it includes:

[0012] An image acquisition unit, used for acquiring a droplet image;

[0013] An image recognition unit, used to recognize the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet;

[0014] A type classification unit, used for classifying the droplets according to the number of the magnetic beads to obtain a plurality of droplet types;

[0015] A probability statistics unit, for obtaining a positive probability and a negative probability of a target droplet in each droplet type;

[0016] The determination output unit is used to construct a Poisson distribution and a likelihood function about the average enzyme on the magnetic beads based on the positive probability and the negative probability, and obtain the number of the average enzyme according to the Poisson distribution and the likelihood function, and then output the number of the average enzyme as a determination result.

[0017] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the droplet digital enzyme-linked immunosorbent assay method as described in the first aspect is implemented.

[0018] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the droplet digital enzyme-linked immunosorbent assay method as described in the first aspect is implemented.

[0019] The embodiment of the present invention provides a droplet digital enzyme-linked immunosorbent assay method, device, computer equipment and storage medium, the method comprising: obtaining a droplet image; identifying the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet; classifying the droplets according to the number of magnetic beads to obtain multiple droplet types; for the target droplets in each droplet type, statistically obtaining the positive probability and negative probability of the target droplet; based on the positive probability and negative probability, constructing a Poisson distribution and likelihood function about the average enzyme on the magnetic beads, and obtaining the number of the average enzyme according to the Poisson distribution and likelihood function, and then outputting the number of the average enzyme as the assay result. The embodiment of the present invention uses Poisson distribution and likelihood function to perform statistical analysis on different types of droplets, overcomes the limitation of Poisson distribution on digital droplet enzyme-linked immunosorbent assay, improves the accuracy and applicability of the analysis and test results of the enzyme-linked immunosorbent assay, and thus reduces the difficulty of downstream analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0021] Figure 1 A schematic diagram of a droplet digital enzyme-linked immunosorbent assay method provided in an embodiment of the present invention;

[0022] Figure 2 A schematic block diagram of a droplet digital enzyme-linked immunosorbent assay device provided in an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of the principle of a droplet digital enzyme-linked immunosorbent assay method provided by an embodiment of the present invention;

[0024] Figure 4 A schematic diagram of the image recognition principle in a droplet digital enzyme-linked immunosorbent assay method provided by an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of the first experimental result of a droplet digital enzyme-linked immunosorbent assay method provided by an embodiment of the present invention;

[0026] Figure 6 A schematic diagram of a second experimental result of a droplet digital enzyme-linked immunosorbent assay method provided by an embodiment of the present invention;

[0027] Figure 7A schematic diagram of the third experimental result of a droplet digital enzyme-linked immunosorbent assay method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0030] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0031] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0032] See below Figure 1 The embodiment of the present invention provides a droplet digital enzyme-linked immunosorbent assay method, which specifically includes: steps S101 to S105.

[0033] S101, acquiring a droplet image;

[0034] S102, identifying the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet;

[0035] S103, classifying the droplets into types according to the number of the magnetic beads to obtain a plurality of droplet types;

[0036] S104, for each type of target droplets, obtaining a positive probability and a negative probability of the target droplets by statistics;

[0037] S105. Based on the positive probability and the negative probability, construct a Poisson distribution and a likelihood function about the average enzyme on the magnetic beads, obtain the number of the average enzyme according to the Poisson distribution and the likelihood function, and then output the number of the average enzyme as a measurement result.

[0038] In this embodiment, a droplet image is first acquired, and then image recognition is performed on it to obtain the number of droplets in the droplet image and the number of magnetic beads in each droplet, and then the droplets are divided into different types according to the number of magnetic beads, for example, a droplet containing one magnetic bead is a first type, a droplet containing two magnetic beads is a second type, a droplet containing three or more magnetic beads is a third type, and a droplet containing zero magnetic beads (i.e., an empty droplet containing no magnetic beads) is a fourth type, etc. For the classified droplets, the probability of being positive and the probability of being negative are counted, and then the Poisson distribution and likelihood function are constructed to obtain the average number of enzymes on the droplet magnetic beads, thereby completing the enzyme-linked immunosorbent assay.

[0039] This embodiment uses Poisson distribution and likelihood function to perform statistical analysis on different types of droplets, overcomes the limitations of Poisson distribution on digital droplet ELISA, improves the accuracy and applicability of ELISA analysis results, and thus reduces the difficulty of downstream analysis.

[0040] In one embodiment, the step S101 includes:

[0041] Magnetic beads are used as solid phase carriers, and the capture antibody, antigen, detection antibody, biotin, streptavidin and HRP enzyme are connected to the magnetic beads to form an immune sandwich structure;

[0042] The immune sandwich structure is generated into droplets through a microfluidic chip;

[0043] Incubating the droplets, and loading them into the chamber of the observation chip after the incubation is completed;

[0044] Fluorescence images and bright field images are taken of the droplets in the observation chip chamber to obtain droplet fluorescence images and droplet bright field images, and the droplet fluorescence images and droplet bright field images are combined into the droplet image.

[0045] In this embodiment, when acquiring the droplet image, the solid phase carrier is first set as magnetic beads, and the capture antibody is connected to the magnetic beads, and then the antigen, detection antibody and biotin are added, and finally SA-hsHRP (streptavidin and HRP enzyme) is added to form an immune sandwich structure. Then, the magnetic beads and substrates with the immune sandwich structure are used to generate droplets through a microfluidic chip. Next, the droplets are controlled to incubate at room temperature for 10 minutes, and then injected into the observation chip for imaging observation and preservation.

[0046] Specifically, combined Figure 3In the process of magnetic bead coupling antibody, 50 μL magnetic beads were washed three times with PBST buffer (phosphate Tween buffer, pH=7.3, 0.01% Tween20 (Tween 20)) and then resuspended in 250 μL activation buffer (5 mM BBS, pH=7.2, 0.01% Tween20), followed by adding 16.67 μL EDC·HCL (1-ethyl-(3-dimethylaminopropyl) carbodiimide hydrochloride, 75.6 mg / mL) and Sulfo·NHS (N-hydroxysulfosuccinimide, 54 mg / mL) and ultrasonically mixed for 5 min, washed once with coupling buffer (5 mM BBS, pH=8.0, 0.01% Tween20) and resuspended in 250 μL BBS buffer (boric acid buffer, pH=8.0), followed by adding 40 μL capture antibody, mixing and incubating overnight in a 4°C refrigerator. After the reaction, 16.67 μL 10% casein and 0.8 μL ethanolamine were added to the mixed solution and rotated at 37°C for 30 min, followed by washing three times with coupling buffer and once with PBST buffer. The magnetic beads coupled with antibodies were resuspended in 500 μL PBS buffer (pH = 7.3, 1% BSA (bovine serum albumin), 0.1% PEG8000 (polyethylene glycol 8000), 0.01% Tween20), mixed and placed in a 4°C refrigerator for rotation and blocking overnight. Finally, the beads were washed three times with PBST buffer and resuspended in 50 μL PBS buffer (pH = 7.3, 0.1% BSA, 0.05% Tween-20) for use.

[0047] During the antibody biotinylation process, the detection antibody was diluted to 2 mg / mL with DPBS (Dulbecco's phosphate buffer), 53 μL was taken and then 1.41 μL of 50 mM NHS-PEG12-Biotin (NHS-PEG12-biotin) was added, and the reaction was shaken at room temperature for 30 minutes. After the reaction was completed, it was stored in a 4°C refrigerator for later use.

[0048] During the immune reaction, 2.5 μL of magnetic beads coated with antibodies were taken into a centrifuge tube, and 80,000 magnetic beads were used for each sample for ddELISA determination. The protein standard or sample was serially diluted to the required concentration with reaction buffer (PBS, pH=7.30, 0.1% Casein, 0.01% Tween20), and the biotinylated detection antibody and SA-hsHRP (high-sensitivity streptavidin-horseradish peroxidase) were diluted to the required concentration with reaction buffer. 100 μL of sample and 100 μL of biotinylated detection antibody were pipetted into a centrifuge tube containing magnetic beads, mixed appropriately and incubated in a 37°C oven for 30 min. After the reaction, the magnetic beads were enriched with a magnetic rack and the supernatant was removed. After washing with PBST several times, 200 μL of SA-hsHRP was added, mixed appropriately and incubated in a 37°C oven for 10 min. After the reaction, the immunomagnetic bead complex was obtained. The magnetic beads were enriched with a magnetic rack and the supernatant was removed. After washing the magnetic beads, they were resuspended in 10 μL of PBS buffer and stored in a centrifuge tube for later use.

[0049] During droplet generation and imaging, take 10μL Amplite Red (Amplite, red HRP fluorescent substrate) substrate (250×, i.e. 250 times) and dilute it to 16× (i.e. 16 times), add 1.5μL 100mM H2O2, mix well and set aside. Take a 2.5mL syringe as the oil phase, absorb the droplet generation oil and connect the PE tube, manually push the oil to the tube mouth, and load the syringe onto the syringe pump. Then insert the tube into the oil phase inlet of the droplet generation chip. Take two 1mL syringes as the water phase, absorb a certain amount of 7500 oil and connect the needle and PE tube (polyethylene tube), then load the syringe onto the second syringe pump, and use the syringe pump to push the 7500 oil until the oil is about 1 cm away from the tube mouth. Vortex the centrifuge tube containing the immunomagnetic bead complex for more than 30s to resuspend the magnetic beads. Then, the tubes connected to the two aqueous syringes were inserted into the PCR tubes containing the immunomagnetic bead mixture and the substrate solution, and the syringe pump was used to suck the 10 μL immunomagnetic bead complex solution back into the tube as much as possible, and an equal volume of substrate solution was sucked at the same time. Then, the tubes were inserted into the two aqueous inlets of the droplet generation chip. The droplet generation chip was placed on the microscope stage, and the droplet generation process was observed by a high-speed camera. The flow rate of the aqueous phase was 240 μL / h, and the flow rate of the oil phase was 1200 μL / h. After the flow rate was set, the syringe pump was turned on. After the droplets were stably generated, the pipette tip was inserted into the outlet to collect the droplets. After collecting for 2 min 30 s, the droplets were transferred to the PCR tube and incubated in the dark for 10 min. The droplets were then loaded into the two chambers of the observation chip and imaged with a fluorescence microscope: in the first step, the droplet fluorescence image was taken using the TRITC channel to identify the positive droplets; in the second step, the droplet bright field image was taken to identify the droplets and magnetic beads. The total imaging time for both chambers was approximately 4 min.

[0050] In one embodiment, the step S102 includes:

[0051] Performing image recognition on the droplet bright field image using a YOLOv5 network model to obtain all droplets in the droplet image;

[0052] The YOLOv5 network model is used to perform image recognition on each droplet to obtain the number of magnetic beads in each droplet.

[0053] Furthermore, the step S102 further includes:

[0054] Based on the number of droplets in the droplet bright field image and the number of magnetic beads in the droplets, the average fluorescence intensity of the empty droplets in the droplet fluorescence image is calculated; wherein the empty droplets are droplets that do not contain magnetic beads;

[0055] Obtaining a pixel average value and a standard deviation of the empty droplets based on the fluorescence intensity mean;

[0056] The pixel average value and standard deviation of the empty droplets are compared with a preset threshold to obtain a comparison result, and the droplets are determined to be positive or negative according to the comparison result, thereby obtaining the number of positive droplets and the number of negative droplets.

[0057] Combination Figure 4 ,This embodiment first identifies all the droplets in the droplet bright field image, and then identifies the magnetic beads in the droplets, such as Figure 4 As shown in part (a) of the figure. The number and coordinate information of the droplets and the number of magnetic beads in the droplets are recorded at the same time, and then the droplets are classified according to the number of magnetic beads in the droplets. For example, the droplets are divided into droplets containing one magnetic bead, droplets containing two magnetic beads, droplets containing three or more magnetic beads, and empty droplets containing no magnetic beads.

[0058] Image recognition in this embodiment is mainly completed by two YOLOv5 runs. The training set data of the droplets is annotated using Labelimg software, with a total of 90 images annotated. The annotated training images are randomly divided into a training set and a validation set at a ratio of 2:1, and trained 300 times using a GPU graphics processor. The training set of magnetic beads uses the same method as above, with a total of 350 images annotated. After multiple rounds of training, the number of droplets in the training images and the number of magnetic beads they contain are obtained. Figure 4 As shown in parts (b) to (d) of the figure. For the fluorescence image, the bounding box of the droplet on the fluorescence image is obtained based on the droplet coordinate information detected by the bright field image. The bounding box is reduced by 1 / 8, and the mean fluorescence intensity of each empty droplet is calculated first. Then, the pixel mean and standard deviation of all empty droplets are calculated using the mean of each empty droplet, and the mean±5std is used as the threshold to judge the positive and negative nature of the droplet. That is, when the pixel mean exceeds the preset mean±5 times the standard deviation, the droplet is judged to be positive, otherwise the droplet is judged to be negative. The number of positive and total numbers of different types of droplets are obtained through image analysis.

[0059] In a specific embodiment, taking the YOLOv5 network model to perform image recognition on the droplet bright field image to obtain all the droplets in the droplet image as an example, the method specifically includes:

[0060] Inputting the droplet bright field image into the YOLOv5 network model, and using the input end of the YOLOv5 network model to sequentially perform mosaic data enhancement, adaptive image scaling, and adaptive anchor frame calculation on the droplet bright field image;

[0061] The droplet bright field image is reduced in information loss and learning features through the Focus structure and CSP structure in the Backbone end of the YOLOv5 network model;

[0062] A feature map is predicted for the droplet bright field image through the FPN structure and the PAN structure in the Neck end of the YOLOv5 network model, and all droplets in the droplet bright field image are determined based on the feature map.

[0063] In one embodiment, the step S105 includes:

[0064] A Poisson distribution is constructed based on the positive probability of the target droplet according to the following formula:

[0065] P i,+ =1-(1-q) i

[0066] Among them, P i，+ represents the positive probability of i magnetic beads in the target droplet, q represents the probability of at least one average enzyme on the magnetic bead, q = 1-e -λ , λ represents the average number of enzymes on the magnetic beads;

[0067] And construct a Poisson distribution based on the negative probability of the target droplet according to the following formula:

[0068] P i,- =(1-q) i

[0069] Among them, P i，- It indicates the negative probability that there are i magnetic beads in the target droplet.

[0070] In this embodiment, since it is necessary to analyze the probability of positive droplets in different types of droplets (different numbers of magnetic beads encapsulated in the droplets), after the immune complex is formed on the microsphere and the microsphere is encapsulated in the microsphere, assuming that the average enzyme (AEB) on each magnetic bead is λ, the probability q that there is at least one enzyme on the magnetic bead follows the Poisson distribution:

[0071] q=1-e -λ

[0072] Therefore, the positive probability and negative probability of a droplet containing i magnetic beads can be recorded as P i，+ and P i，- , the calculation formulas of the two are as follows:

[0073] P i,+ =1-(1-q) i

[0074] P i,- =(1-q) i

[0075] Therefore, after obtaining the positive probability and negative probability of the target droplet, the probability q of having at least one enzyme can be derived according to the above formula, and the average number of enzymes λ on each magnetic bead can be further obtained.

[0076] In one embodiment, the step S105 further includes:

[0077] The likelihood function is constructed as follows:

[0078]

[0079] Among them, L(λ; n i , m i ) represents the likelihood function, λ represents the average number of enzymes on the magnetic beads, ni represents the number of positive droplets containing i magnetic beads, and mi represents the number of negative droplets containing i magnetic beads.

[0080] Furthermore, the step S105 further includes:

[0081] According to the following formula, the likelihood function is maximized to obtain the average number of enzymes:

[0082]

[0083] in, represents the average number of enzymes on the beads.

[0084] Combined with the positive probability P of the droplet containing i magnetic beads mentioned above i，+ and negative probability P i，- , and the number of positive droplets and negative droplets containing i magnetic beads obtained by deep learning technology, the likelihood function can be constructed:

[0085]

[0086] The value of parameter λ that maximizes the likelihood function is considered the most likely value of the average enzyme AEB, namely:

[0087]

[0088] In the specific calculation process, the logarithm of the likelihood function is taken:

[0089]

[0090] For the convenience of writing, let 1-q=p, then:

[0091]

[0092] Since the proportion of droplets containing more than 5 magnetic beads is very low (0.2%), the first 5 terms of the function are retained:

[0093] lnL(λ;n i ,m i )=n1 ln(1-p)+m1lnp

[0094] +n2 ln(1-p 2 )+2·m2 lnp

[0095] +n3ln(1-p 3 )+3·m3lnp

[0096] +n4ln(1-p 4 )+4·m4lnp

[0097] +n5 ln(1-p 5 )+5·m5 lnp

[0098] Combine similar terms in the formula:

[0099] lnL(λ;n i ,m i )=n1ln(1-p)+m1 lnp

[0100] +n2ln(1-p 2 )+2·m2lnp

[0101] +(n3+n4+n5)ln(1-p 3 )+3·(m3+m4+m5)lnp

[0102] +ε p

[0103] in,

[0104]

[0105] Since the proportion of droplets containing more than 3 magnetic beads is only 1.6%, that is, εp is much smaller than the first 6 items, εp is ignored for the convenience of calculation, and the derivative of lnL is set to 0:

[0106]

[0107] In an actual application scenario, the number of magnetic beads in a droplet is set to N1, of which the number of luminescent beads is n1; the number of magnetic beads in a droplet is set to N2, of which the number of luminescent beads is n2; the number of magnetic beads in a droplet is set to N3, of which the number of luminescent beads is n3, wherein N1, n1, N2, n2, N3, and n3 are obtained by the aforementioned image recognition and other steps. When the probability of satisfying the conditions of the number of droplets and the luminescent droplets in the application scenario is the largest, the calculated q is the probability value of forming a "sandwich structure" on the magnetic beads before the droplets are generated.

[0108] There is only one magnetic bead in the droplet, and the probability that the droplet emits light is:

[0109]

[0110] There are only two magnetic beads in the droplet, and the probability that the droplet emits light is:

[0111]

[0112] If there are three or more magnetic beads in a droplet, the probability that the droplet will glow is:

[0113]

[0114] Construct the likelihood function:

[0115] L(λ)=P1·P2·P3

[0116] Taking the logarithm we get:

[0117] ln(L(λ))=n1ln[1-(1-q)]+n2ln[1-(1-q) 2 ]+n3ln[1-(1-q) 3 ]+[(N1-n1)+2(N2-n2)+3(N3-n3)]ln(1-q)

[0118] Taking the derivative and setting it to 0 gives:

[0119]

[0120] Then we can get q by Poisson distribution q = 1-e -λ , the average enzyme λ was calculated.

[0121] In one application scenario, the optimization of the assay method provided in this embodiment is explained from the number of magnetic beads and the sample volume. Figure 5 The sample volume and the number of magnetic beads have a certain influence on the number and type of droplets generated. It is necessary to explore the sample volume and the number of magnetic beads. Figure 5In parts (a) and (b), when the liquid volume is reduced from 20 μL to 10 μL, the proportion of empty droplets decreases from 86.68% to 58.09%. When the final liquid volume is 10 μL, the single-bead (i.e., containing one magnetic bead), double-bead, and multi-bead droplet rates reach approximately 29.74%, 9.31%, and 2.85%, respectively. At the same time, satisfactory droplet generation stability is obtained, and the number of droplets is sufficient for digital analysis. On the other hand, combined with Figure 5 In parts (c) and (d), when the number of magnetic beads increases from 50k to 70k, the proportion of empty droplets drops sharply from 82.04% to 59.86%. Continuing from 80k to 100k, it can be seen that with the increasing number, the results of bead and droplet counting are similar. This example uses 80k as the starting number because too many magnetic beads sometimes cause microchannels to block on the droplet generation chip. Finally, the experiments were all conducted with a sample volume of 10μL and a number of magnetic beads of 80k.

[0122] Combination Figure 6 In terms of the ratio of biotin to detection antibody, the ratio of biotin to detection antibody has a crucial influence on the detection of the signal, so it is necessary to explore its ratio. This embodiment designs three conditions of 5:1, 10:1 and 20:1 for verification. Figure 6 As shown in parts (a) to (d) of the experiment, the experimental results show that the ratio of 20:1 has the highest signal-to-noise ratio and the best effect. On the other hand, as the detection antibody concentration increases from 20ng / mL to 400ng / mL, the AEB level and sensitivity of the detection method also increase. Subsequent analysis uses 400ng / mL of detection antibody, which has higher performance and stability.

[0123] Combination Figure 7 From the perspective of protein detection and calibration curve drawing, the proportion of positive droplets gradually increased in experiments with higher protein concentrations. This trend is more obvious in the case of two or more microbeads, because the possibility of forming immune complexes in multi-microbead systems is increasing. Figure 7As shown in parts (a) to (c) of the figure, when the analyte concentration increased from 4fg / mL to 2.5pg / mL, the number of negative droplets containing magnetic beads dropped sharply from 97.79% to 22.70%. Near the upper limit of iddELISA detection (2.5pg / mL), the proportions of positive 1, 2, and more droplets increased to 69.12% and 79.15%, respectively. According to the proposed maximum likelihood estimate, the AEB values ​​of the five antigen concentrations ranging from 4fg / mL to 2.5pg / mL were calculated and used to construct a calibration curve (R2=0.9960). By optimizing the reagents and data processing algorithms, the LOD (limit of detection) was calculated to be 0.71fg / mL (15.78aM) and the LOQ (limit of quantification) was 2.80fg / mL (62.22aM) using the average AEB value of the blank analysis plus 3 and 10 standard deviations, respectively.

[0124] In general, the assay method provided in this embodiment can quantitatively measure protein biomarkers at aM concentration. Based on the basic principles of maximum likelihood estimation and Poisson distribution, an improved mathematical model for calculating AEB is proposed, which integrates the positive counts and negative counts of 0, 1, 2 and more droplets. For the droplet detection system, the droplet encapsulated magnetic beads, reagents and image processing are optimized to increase the sensitivity by 200,000 times compared with traditional ELISA and 10 to 100 times compared with other digital ELISAs. The detection limit of this method is 0.71fg / mL, and the CVs range is 6.55% to 13.84%. At the same time, the assay method does not require the manufacture of microwells or complex equipment, and shows advantages in the quantification of ultra-low concentration biomarkers.

[0125] Figure 2 A schematic block diagram of a droplet digital enzyme-linked immunosorbent assay device 200 provided in an embodiment of the present invention, the device 200 comprises:

[0126] An image acquisition unit 201, used to acquire a droplet image;

[0127] An image recognition unit 202 is used to recognize the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet;

[0128] A type classification unit 203, configured to classify the droplets according to the number of magnetic beads to obtain a plurality of droplet types;

[0129] A probability statistics unit 204 is used to obtain a positive probability and a negative probability of a target droplet in each droplet type by statistics;

[0130] The determination output unit 205 is used to construct a Poisson distribution and a likelihood function about the average enzyme on the magnetic beads based on the positive probability and the negative probability, obtain the number of the average enzyme according to the Poisson distribution and the likelihood function, and then output the number of the average enzyme as a determination result.

[0131] In one embodiment, the image acquisition unit 201 includes:

[0132] The structure forming unit is used to set the magnetic beads as a solid phase carrier, and connect the capture antibody, the antigen, the detection antibody, biotin, streptavidin and the HRP enzyme to the magnetic beads to form an immune sandwich structure;

[0133] A droplet generation unit, used to generate droplets from the immune sandwich structure through a microfluidic chip;

[0134] A droplet loading unit, used to incubate the droplets and load them into the chamber of the observation chip after the incubation is completed;

[0135] The image capturing unit is used to capture fluorescence images and bright field images of the droplets in the observation chip chamber, respectively, to obtain droplet fluorescence images and droplet bright field images, and to combine the droplet fluorescence images and droplet bright field images into the droplet image.

[0136] In one embodiment, the image recognition unit 202 includes:

[0137] A first recognition unit is used to perform image recognition on the droplet bright field image using a YOLOv5 network model to obtain all droplets in the droplet image;

[0138] The second recognition unit is used to perform image recognition on each droplet using the YOLOv5 network model to obtain the number of magnetic beads in each droplet.

[0139] In one embodiment, the image recognition unit 202 further includes:

[0140] an intensity calculation unit, configured to calculate the average fluorescence intensity of the empty droplets in the droplet fluorescence image based on the number of droplets in the droplet bright field image and the number of magnetic beads in the droplets; wherein the empty droplets are droplets that do not contain magnetic beads;

[0141] A difference acquisition unit, used for acquiring a pixel average value and a standard deviation of the empty droplet based on the fluorescence intensity mean value;

[0142] The number determination unit is used to compare the pixel average value and standard deviation of the empty droplets with a preset threshold value to obtain a comparison result, and determine whether the droplet is positive or negative according to the comparison result, thereby obtaining the number of positive droplets and the number of negative droplets.

[0143] In one embodiment, the measurement output unit 205 includes:

[0144] The first construction unit is used to construct a Poisson distribution based on the positive probability of the target droplet according to the following formula:

[0145] P i,+ =1-(1-q) i

[0146] Among them, P i，+ represents the positive probability of i magnetic beads in the target droplet, q represents the probability of at least one average enzyme on the magnetic bead, q = 1-e -λ , λ represents the average number of enzymes on the magnetic beads;

[0147] The second construction unit is used to construct a Poisson distribution based on the negative probability of the target droplet according to the following formula:

[0148] P i,- =(1-q) i

[0149] Among them, P i，- It indicates the negative probability that there are i magnetic beads in the target droplet.

[0150] In one embodiment, the measurement output unit 205 further includes:

[0151] A function construction unit is used to construct the likelihood function according to the following formula:

[0152]

[0153] Among them, L(λ; n i , m i ) represents the likelihood function, λ represents the average number of enzymes on the magnetic beads, n i represents the number of positive droplets containing i magnetic beads, m i represents the number of negative droplets containing i magnetic beads.

[0154] In one embodiment, the measurement output unit 205 further includes:

[0155] The function maximization unit is used to maximize the likelihood function according to the following formula to obtain the average enzyme quantity:

[0156]

[0157] in, represents the average number of enzymes on the beads.

[0158] Since the embodiments of the apparatus part correspond to the embodiments of the method part, please refer to the description of the embodiments of the method part for the embodiments of the apparatus part, which will not be repeated here.

[0159] The embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed, the steps provided in the above embodiment can be implemented. The storage medium may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0160] The embodiment of the present invention also provides a computer device, which may include a memory and a processor, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, the steps provided in the above embodiment may be implemented. Of course, the computer device may also include various network interfaces, power supplies and other components.

[0161] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the various embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of this application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of this application.

[0162] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

Claims

1. A droplet digital enzyme-linked immunosorbent assay method, It is characterized in that include: Acquire a droplet image; The droplet image is identified by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet; Classifying the droplets into different types according to the number of magnetic beads to obtain multiple droplet types; For the target droplets in each droplet type, the positive probability and negative probability of the target droplets are statistically obtained; Based on the positive probability and the negative probability, construct a Poisson distribution and a likelihood function about the average enzyme on the magnetic beads, obtain the number of the average enzyme according to the Poisson distribution and the likelihood function, and then output the number of the average enzyme as a measurement result; The method of constructing a Poisson distribution and a likelihood function of the average enzyme on the magnetic beads based on the positive probability and the negative probability, and obtaining the number of the average enzyme according to the Poisson distribution and the likelihood function, comprises: According to the following formula (1), a Poisson distribution is constructed based on the positive probability of the target droplet: P i,+ =1-(1-q) i (1) Among them, P i，+ represents the positive probability of i magnetic beads in the target droplet, q represents the probability of at least one average enzyme on the magnetic bead, q = 1-e -λ , λ represents the average number of enzymes on the magnetic beads; And according to the following formula (2), a Poisson distribution is constructed based on the negative probability of the target droplet: P i,- =(1-q) i (2) Among them, P i，- represents the negative probability that there are i magnetic beads in the target droplet; The method further comprises constructing a Poisson distribution and a likelihood function of the average enzyme on the magnetic beads based on the positive probability and the negative probability, and obtaining the number of the average enzyme according to the Poisson distribution and the likelihood function, and further comprising: The likelihood function is constructed according to the following formula (3): Among them, L(λ; n i , m i ) represents the likelihood function, λ represents the average number of enzymes on the magnetic beads, n i represents the number of positive droplets containing i magnetic beads, m i represents the number of negative droplets containing i magnetic beads.

2. The droplet digital enzyme-linked immunosorbent assay method according to claim 1, It is characterized in that The step of acquiring a droplet image comprises: Magnetic beads are used as solid phase carriers, and the capture antibody, antigen, detection antibody, biotin, streptavidin and HRP enzyme are connected to the magnetic beads to form an immune sandwich structure; The immune sandwich structure is generated into droplets through a microfluidic chip; Incubating the droplets, and loading them into the chamber of the observation chip after the incubation is completed; Fluorescence images and bright field images are taken of the droplets in the observation chip chamber to obtain droplet fluorescence images and droplet bright field images, and the droplet fluorescence images and droplet bright field images are combined into the droplet image.

3. The droplet digital enzyme-linked immunosorbent assay method according to claim 2, It is characterized in that The method of identifying the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet includes: Performing image recognition on the droplet bright field image using a YOLOv5 network model to obtain all droplets in the droplet image; The YOLOv5 network model is used to perform image recognition on each droplet to obtain the number of magnetic beads in each droplet.

4. The droplet digital enzyme-linked immunosorbent assay method according to claim 2, It is characterized in that The method further includes: identifying the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet; Based on the number of droplets in the droplet bright field image and the number of magnetic beads in the droplets, the average fluorescence intensity of the empty droplets in the droplet fluorescence image is calculated; wherein the empty droplets are droplets that do not contain magnetic beads; Obtaining a pixel average value and a standard deviation of the empty droplets based on the fluorescence intensity mean; The pixel average value and standard deviation of the empty droplets are compared with a preset threshold to obtain a comparison result, and the droplets are determined to be positive or negative according to the comparison result, thereby obtaining the number of positive droplets and the number of negative droplets.

5. A droplet digital enzyme-linked immunosorbent assay device, It is characterized in that include: An image acquisition unit, used for acquiring a droplet image; An image recognition unit, used to recognize the droplet image by deep learning technology to obtain all droplets in the droplet image and the number of magnetic beads contained in each droplet; A type classification unit, used for classifying the droplets according to the number of the magnetic beads to obtain a plurality of droplet types; A probability statistics unit, for obtaining a positive probability and a negative probability of a target droplet in each droplet type; A determination output unit, for constructing a Poisson distribution and a likelihood function about the average enzyme on the magnetic beads based on the positive probability and the negative probability, and obtaining the number of the average enzyme according to the Poisson distribution and the likelihood function, and then outputting the number of the average enzyme as a determination result; The measurement output unit comprises: The first construction unit is used to construct a Poisson distribution based on the positive probability of the target droplet according to the following formula (1): P i,+ =1-(1-q) i (1) Among them, P i，+ represents the positive probability of i magnetic beads in the target droplet, q represents the probability of at least one average enzyme on the magnetic bead, q = 1-e -λ , λ represents the average number of enzymes on the magnetic beads; The second construction unit is used to construct a Poisson distribution based on the negative probability of the target droplet according to the following formula (2): P i,- =(1-q) i (2) Among them, P i，- represents the negative probability that there are i magnetic beads in the target droplet; The measurement output unit also includes: A function construction unit is used to construct the likelihood function according to the following formula (3): Among them, L(λ; n i , m i ) represents the likelihood function, λ represents the average number of enzymes on the magnetic beads, n i represents the number of positive droplets containing i magnetic beads, m i represents the number of negative droplets containing i magnetic beads.

6. A computer device, It is characterized in that The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the droplet digital enzyme-linked immunosorbent assay method according to any one of claims 1 to 4 when executing the computer program.

7. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the droplet digital enzyme-linked immunosorbent assay method according to any one of claims 1 to 4 is implemented.

Images