A deep learning assisted microfluidic immunoassay method based on an imaging transcoding system

Abstract

The application relates to a deep learning assisted microfluidic immunoassay method based on an imaging transcoding system. Millimeter-sized microspheres (PS mm -Ab1) coupled with capture antibodies are subjected to an immune reaction with detection antibodies (Ab2-gDNA) coupled with target objects and guide DNA (gDNA) to form a complex and are separated by gravity sedimentation. Ab2-gDNA not participating in the reaction in supernatant activates CbAgo to cut target DNA (tDNA) on a signal probe PS Signal -tDNA-biotin, and the biotin content on the surface of the target DNA is reduced. Subsequently, streptavidin (SA) modified microspheres (PS Filter -SA) are added to form a PS Filter -SA-biotin-tDNA-PS Signal complex, and the complex is filtered through a microfluidic chip. The cut probe is only PS Signal , which enters an imaging area for subsequent detection. The application accurately analyzes the particle size and quantity of the microspheres and associates the particle size and quantity with the types and concentrations of the target objects, so that sensitive and rapid detection of the target objects is realized, and multiple target objects can be simultaneously detected.

Description

Technical Field

[0001] This invention relates to clinical diagnostics and food safety, specifically to a deep learning-assisted microfluidic immunoassay method based on an imaging transcoding system. Background Technology

[0002] In recent years, rapidly deployable point-of-care testing (POCT) platforms have become crucial for disease prevention and treatment. For example, rapid antigen testing, molecular diagnostics, and other rapid testing methods used during the COVID-19 pandemic effectively facilitated the timely isolation and treatment of SARS-CoV-2 infected individuals. However, despite the growing demand for rapid testing, there is still a lack of testing platforms capable of simultaneously detecting multiple viruses. For instance, during the annual influenza season, multiple respiratory viruses often circulate concurrently, including influenza A virus (FLUA), influenza B virus (FLUB), and human parainfluenza virus (HPIV), which frequently cause similar symptoms such as cough and fever and can lead to serious complications such as pneumonia. Multi-target combined detection can effectively reduce the clinical misdiagnosis rate caused by respiratory virus co-infection and alleviate the burden on the healthcare system. However, existing standard POCT methods, such as lateral flow immunoassay (LFIA), electrochemical sensing (ECS), and optical sensing (OS), still face significant challenges in achieving simultaneous multi-target detection. While LFIA offers fast detection speed and low cost, its sensitivity is limited, and it only provides qualitative or semi-quantitative results. ECS methods, though easily miniaturized and integrated, are susceptible to environmental interference and face challenges in simultaneous multi-target detection. OS, while boasting high resolution, high contrast, and high sensitivity, typically requires expensive optical equipment for real-time multi-target detection. Therefore, developing a POCT platform that combines high accuracy, low cost, ease of operation, and multi-target detection capabilities is of great significance for disease prevention and control.

[0003] In their previous work, the inventors proposed a washless biosensing method for on-site multiplex detection (publication number CN118033119A). In this method, an immune reaction occurs between the analyte and millimeter-sized polystyrene microspheres coupled with biorecognition molecules, and micrometer-sized polystyrene microspheres coupled with biorecognition molecules, forming a "millimeter-sized microsphere-target-micrometer-sized microsphere" complex. By adding a reagent that disrupts the chemical bonds between the biorecognition molecules to the complex, the micrometer-sized microspheres are released. Finally, a particle counter is used to detect the micrometer-sized microspheres in the supernatant, and the number and size of the micrometer-sized microspheres are analyzed to determine the concentration of the target analyte. While the particle counter used for signal readout in the above invention is small and portable, it remains expensive, and the destructive solution used to release the micrometer-sized microspheres can affect the detection electrode, thus impacting the detection results. Furthermore, in subsequent research, the inventors proposed a room-temperature Argonaute protein-based multiplex nucleic acid amplification-free digital encoding-decoding detection method (publication number CN118620996A). This method combines biochemical reactions with visualized microsphere counting and artificial intelligence algorithms. Biorecognition molecules corresponding to the analyte are coupled to the surfaces of a carrier and a signal probe, respectively, to perform DNA hybridization reactions and fluorescently encode the nanomagnetic particles. The sample is then photographed using a fluorescence microscope to obtain fluorescence images. Finally, the fluorescence microscopic images of the carrier nanomagnetic particles with fluorescent probes are identified and counted. This method, based on the fluorescence differences at different emission wavelengths, effectively avoids background interference from the solution matrix and enables simultaneous quantification of multiple targets. However, the inverted fluorescence microscope used in this method is relatively large and requires multiple magnetic separation steps, making it suitable for laboratory testing conditions but difficult to implement rapidly on-site. Summary of the Invention

[0004] In view of the problems existing in the prior art, this invention provides a deep learning-assisted microfluidic immunoassay method based on an imaging transcoding system, aiming to meet the needs of point-of-care testing. The method's core technologies are a microfluidic immunoassay platform and a CbAgo-based immunoassay method: micron-sized polystyrene (PS) microspheres of different sizes (such as PS4, PS6, PS...). 10 These were used as multi-target signal probes, and were transmitted via millimeter-scale microspheres (PS). mmThe synergistic effect of gravity separation and microfluidic chip filtration separation achieves a direct correspondence between the number of signal probes and the concentration of the corresponding analyte. Microfluidic chips not only significantly reduce the detection volume but also greatly improve detection sensitivity due to precise reaction control, rapid reaction rates, and minimal reagent consumption. Simultaneously, complex filtration and microsphere signal probe separation are both completed within the chip, requiring no additional operations. With customized artificial intelligence (AI) microsphere recognition software, smartphone imaging devices can acquire multi-probe encoded images of specific areas of the chip and perform multiple AI decoding tasks, including image processing, encoded probe classification and counting, and concentration calculation, thereby rapidly obtaining detection results.

[0005] Building upon this foundation, this invention innovatively combines the nucleic acid cleavage characteristics of CbAgo protein with microsphere microscopy imaging technology to construct a sensitive, rapid, and label-free multi-target detection platform. The core idea is: conjugating millimeter-sized PS microspheres (PS...) to target analytes to capture antibodies. mm -Ab1)(A) first undergoes an immune reaction with the target analyte and the detection antibody complex (Ab2-gDNA) (B) coupled with the guide DNA strand (gDNA), achieving the capture of the target molecule. Due to the gravity of the millimeter-sized microspheres, the complex (PS) formed after the reaction... mm The separation vector A (A-target-B) rapidly settles to the bottom of the solution. Subsequently, the unbound Ab2-gDNA complex (B) in the supernatant activates the targeted cleavage properties of CbAgo on the target DNA strand (tDNA), enabling the targeting of microsphere signal probes (PS) with biotinylated tDNA conjugated to their surfaces. Signal -tDNA-biotin)(C) performs specific cleavage, altering PS Signal The amount of tDNA on the surface of the microspheres is reduced, thereby decreasing the amount of biotin linked to them, and the cleaved probe PS Signal -tDNA-biotin(C) only PS remains Signal Due to the extremely high binding affinity between biotin and streptavidin (SA), microspheres (PS) with streptavidin SA surface-coupled were subsequently added. Filter -SA)(D) and PS Signal -tDNA-biotin(C) binds to form (PS) Filter -SA-biotin-tDNA-PS Signal The Ab2-gDNA complex (CD) is involved. In the presence of the target, the amount of unbound Ab2-gDNA (B) in the supernatant decreases, thereby reducing its activation of CbAgo to cleave PS. SignalThe change in tDNA content in tDNA-biotin (C) ultimately significantly affects PS. Signal -tDNA-biotin (C) and PS Filte The binding efficiency of -SA(D) is ultimately determined after filtering and separation by a microfluidic chip, and the PS Filter -SA-biotin-tDNA-PS Signal The composite (CD) remains in the filter area of ​​the chip, while the remaining PS... Signal The microspheres are then incorporated into the imaging area for subsequent detection. By capturing images of the imaging area within the chip using a smartphone imaging device and performing AI decoding, parameters such as the particle size and quantity of the microspheres can be accurately analyzed and correlated with the type and concentration of the target, thus achieving sensitive and rapid detection of the target analyte. This platform not only meets the needs of point-of-care testing (POCT) but also has broad application prospects in clinical diagnostics, food safety, and rapid on-site detection.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: Millimeter-scale PS microspheres (PS) conjugated to target-capture antibodies mm -Ab1)(A) first undergoes an immune reaction with the target analyte and the detection antibody complex (Ab2-gDNA) (B) coupled with the guide DNA strand (gDNA), achieving the capture of the target molecule. Due to the gravity of the millimeter-sized microspheres, the complex (PS) formed after the reaction... mm The separation vector A (-Ab1-target-Ab2-gDNA) rapidly settles to the bottom of the solution. Subsequently, the unbound Ab2-gDNA complex (B) in the supernatant activates the nucleic acid-directed cleavage properties of CbAgo on the target DNA strand (tDNA), enabling it to target microsphere signal probes (PS) with biotinylated tDNA conjugated to their surfaces. Signal -tDNA-biotin)(C) performs specific cleavage, altering PS Signal The amount of tDNA on the surface of the microspheres is reduced, thereby decreasing the amount of biotin linked to them. Because biotin has a very high binding affinity to streptavidin (SA), microspheres (PS) with streptavidin SA subsequently added to their surface are then... Filter -SA)(D) and PS Signal -tDNA-biotin(C) binds to form (PS) Filter -SA-biotin-tDNA-PS Signal The Ab2-gDNA complex (CD) is involved. In the presence of the target, the amount of unbound Ab2-gDNA (B) in the supernatant decreases, thereby reducing its activation of CbAgo to cleave PS. SignalThe change in tDNA content in tDNA-biotin (C) ultimately significantly affects PS. Signal -tDNA-biotin (C) and PS Filter The binding efficiency of -SA(D) is ultimately determined after filtering and separation by a microfluidic chip, and the PS Filter -SA-biotin-tDNA-PS Signal The composite (CD) remains in the filter area of ​​the chip, while the remaining PS... Signal The microspheres are then incorporated into the imaging area for subsequent detection. Images are captured and AI-decoded using a smartphone imaging device, precisely analyzing parameters such as particle size and quantity, and correlating these parameters with the type and concentration of the target, thus achieving sensitive and rapid detection. Importantly, this technology can simultaneously detect multiple targets by designing different microsphere signal probes, biorecognition molecules, gDNA, and tDNA. This platform not only meets the needs of point-of-care testing (POCT) but also has broad application prospects in clinical diagnostics, food safety, and rapid on-site detection. Specifically, A deep learning-assisted microfluidic immunoassay method based on an imaging transcoding system, the method comprising the following steps: 1) The target analyte is coupled with millimeter-sized microspheres (PS) containing capture antibodies. mm -Detection antibodies Ab1 and gDNA markers react with Ab2-gDNA to form "PS". mm The "-Ab1-target-Ab2-gDNA" complex, which is in PS mm Under the influence of gravity, it quickly settles to the bottom of the container, and the unreacted Ab2-gDNA is suspended in the supernatant. The unreacted Ab2-gDNA in the supernatant is taken for subsequent detection. 2) Unreacted Ab2-gDNA from the supernatant was added to the CbAgo solution, followed by the addition of microsphere signal probes PS with biotinylated tDNA conjugated to the Ab2-gDNA and CbAgo mixture. Signal -tDNA-biotin, to obtain a mixture; gDNA can activate the nucleic acid cleavage properties of CbAgo, thereby enabling CbAgo to cleave PS. Signal Specific cleavage of tDNA on tDNA-biotin, resulting in PS after cleavage Signal -tDNA-biotin only has PS remaining Signal It enters the imaging area for subsequent detection; 3) Add microspheres PS with streptavidin SA surface-modified and coupled to the above mixture. Filter -SA, due to the binding of SA and biotin, forms PS Filter-SA-biotin-tDNA-PS Signal complex; 4) Pass the mixture obtained in step 3) above into the microfluidic chip. The filtration area of ​​the chip only allows PS to pass through. Signal Through, PS Signal The image passes through the filtering area and enters the imaging area. 5) Use the constructed portable imaging device to take pictures of the imaging area inside the chip to obtain microsphere images, and identify and count the particle size and number of microspheres in the images; 6) Plot the logarithm of the target standard stock solution on the x-axis, and compare the PS values ​​of different particle sizes. Signal Using the quantity as the ordinate, calculate the concentration of the target analyte to complete the detection of the target analyte.

[0007] Preferably, the millimeter-sized microspheres PS mm The separation carrier 1, used for sedimentation separation, has a diameter ranging from 0.5 mm to 2 mm, more preferably 1 mm; the microspheres PS have a diameter of... Filter Separation carrier 2, used for filtration separation, has a diameter ranging from 40 μm to 100 μm, more preferably 70 μm; the PS Signal It is a signal probe used to encode the target object. Each target object corresponds to a microsphere of a certain size. Different sizes of signal probes are selected according to the type of target object, with a diameter range of 1 μm-20 μm.

[0008] Preferably, the method is used for the simultaneous detection of multiple target objects.

[0009] More preferably, the method is used for the simultaneous detection of three target objects, with the microsphere signal probe PS... Signal PS in tDNA-biotin Signal The sizes were selected as 4, 6, and 10 μm, respectively, corresponding to PS4-tDNA1-biotin, PS6-tDNA2-biotin, and PS... 10 -tDNA3-biotin; The specific process is as follows: Using three different Photoshop versions mm -Ab1 and its corresponding Ab2-gDNA1, Ab2-gDNA2, Ab2-gDNA3, the PS complex after the immune response mm -Ab1-Target 1-Ab2-gDNA1, PS mm -Ab1-target 2-Ab2-gDNA2, PS mm -Ab1-Target 3-Ab2-gDNA3 after PS mmAfter sedimentation and separation, unreacted Ab2-gDNA1, Ab2-gDNA2, and Ab2-gDNA3 activate CbAgo via gDNA to specifically cleave different tDNAs (tDNA1, tDNA2, tDNA3), reducing the specific cleavage of the corresponding signal probe complexes PS4-tDNA1-biotin, PS6-tDNA2-biotin, and PS... 10 -tDNA3-biotin and PS Filter -SA binding efficiency, the cleaved signal probe only retains PS4 and PS 6、 PS 10 The probes enter the imaging area for subsequent detection, enabling simultaneous identification and counting of microspheres of multiple sizes. The size of the microsphere probes corresponds to the type of target analyte, and the number of microsphere probes corresponds to the concentration of the target analyte.

[0010] Preferably, in step 1), the capture antibody Ab1 and the detection antibody Ab2 are biorecognition molecules targeting large molecular targets. When the target is a small molecule, they can be replaced with the capture antibody Ab and the complete antigen BSA. The small molecule target is coupled with millimeter-sized microspheres PS with the capture antibody. mm -A competitive immune response occurs among the complete antigens BSA and gDNA labeled with Ab and gDNA; The microspheres are any one of polystyrene latex microspheres, polybutadiene latex microspheres, polyisoprene latex microspheres, or polyacrylic acid latex microspheres; the gDNA and tDNA are single-stranded nucleic acids capable of base pairing; the biotin and streptavidin SA can bind to form a stable complex.

[0011] Preferably, the millimeter-sized microspheres PS mm The settling velocity is greater than PS Filter and PS Signal The settling velocity.

[0012] Preferably, the biorecognition molecule includes, but is not limited to, analyte-specific antibodies and analyte complete antigens, detection antibodies and capture antibodies, antibodies and antigens, antigens and antibodies, DNA capture probes and DNA detection probes, DNA detection probes and DNA capture probes, bacteriophages and antibodies, antibodies and bacteriophages, and bacteriophages and peptides. The biorecognition molecule can only specifically bind to the target and cannot bind to other target substances.

[0013] Preferably, in step 4), the microfluidic chip includes three functional regions: a mixing region, a filtering region, and an imaging region; the mixed solution is fully mixed in the mixing region before entering the filtering region; The outlet of the filtering region contains a ring array of micropillars, through which the filtered signal probe PS SignalEntering the imaging region for subsequent imaging and identification; the micropillar size does not affect the liquid flow rate, ranging from 100-400 μm, with 300 μm being more preferred; the micropillar spacing should ensure that only PS is allowed. signal The micrometer size is 20-50 μm, with 30 μm being a further preferred size, and the micrometer size does not aggregate.

[0014] Preferably, in step 4), the shear stress in the mixing region is 0.01-0.02 Pa, preferably 0.013 Pa; the fluid velocity in the filtering region and the imaging region is 0.5-1.5 mm / s, preferably 1 mm / s; Preferably, in step 5), the portable imaging device consists of a light source, an adjustable aperture, a focusing mechanism, a lens, a microfluidic chip, a smartphone, and a 3D printed shell, and is powered by a 12V DC power supply; the imaging area of ​​the microfluidic chip is illuminated by the light source, and the light is magnified 400 times by a lens with a diameter of 4 mm and a thickness of 2 mm before being projected onto the smartphone image sensor. In step 5), the portable imaging device is responsible for capturing images of the microspheres inside the chip to achieve rapid analysis. The imaging process is as follows: the microfluidic chip is placed on the imaging device, and the chip position is adjusted so that the imaging area is aligned with the smartphone lens; the image is ensured to be clear by adjusting the aperture size and the focal length above the light source; each sample is imaged three times, and the average number of microspheres in each sample is calculated to ensure the accuracy and reliability of the data.

[0015] The present invention has the following beneficial effects: 1. By combining a CbAgo protein-mediated specific nucleic acid cleavage system with immunological techniques, different gDNA and tDNA chains can be designed independently, eliminating the need for nucleic acid amplification of the target analyte. A large amount of gDNA is chemically coupled to the surface of proteins such as antibodies. A single antigen-antibody binding event triggers a significant change in the amount of gDNA. This method serves as a bridge between immunology and molecular biology, solving the false-positive problem in nucleic acid amplification caused by traditional CbAgo methods while simultaneously amplifying the signal. The intermediate-temperature (37°C) nucleic acid cleavage characteristics of CbAgo protein are combined with the specificity of antigen and antibody, enabling a synergistic effect between the two at the optimal reaction temperature (37°C).

[0016] 2. By combining signal probes encoding PS microspheres of different particle sizes with the nucleic acid cleavage characteristics of CbAgo protein, and with each PS microsphere corresponding to a separately designed tDNA, a highly sensitive, rapid, and label-free multiplex detection platform was constructed to achieve parallel detection of multiple targets.

[0017] 3. A multifunctional microfluidic chip integrating mixing, filtering, and imaging was designed. After the aforementioned reaction, microsphere signal probes of different sizes are automatically mixed and separated within the designed microfluidic chip. This operation is more standardized, avoiding the aggregation and signal probe loss problems caused by subjective operation in traditional separation methods (magnetic separation, centrifugation). This significantly improves the efficiency and accuracy of multi-target detection. The microfluidic chip can filter and separate microspheres of different sizes to ensure the accuracy of subsequent imaging and identification. The integrated design of this chip requires no additional external operation, significantly reducing detection time and human error, and improving the practicality and reliability of the point-of-care testing platform in on-site diagnosis.

[0018] 4. A portable smartphone imaging device was developed to capture high-resolution images of microspheres within a chip, enabling rapid analysis without additional equipment, greatly enhancing the portability and practicality of the detection system. The accompanying microsphere identification software, specifically designed to identify microsphere signal probes, avoids signal interference caused by insufficient filtering leading to the entry of large protein molecules into the imaging area; it can also effectively distinguish between bubbles and microsphere aggregations. By adjusting the aperture size and the focal length above the light source, image clarity is ensured, guaranteeing the accuracy and reliability of the data. The compact design and intelligent functions of this imaging device demonstrate broad application prospects in resource-constrained environments, including clinical diagnosis, food safety, and rapid on-site detection. Attached Figure Description

[0019] Figure 1 A schematic diagram of a deep learning-assisted microfluidic immunoassay method based on an imaging transcoding system.

[0020] Figure 2 Simulation analysis of mixing, filtering and separation regions of microfluidic chips.

[0021] Figure 3 The functions and dimensions of each part of a microfluidic chip.

[0022] Figure 4 Schematic diagram of a portable smartphone imaging device.

[0023] Figure 5 Practical verification of the mixing, filtering, and separation regions of microfluidic chips.

[0024] Figure 6 Three respiratory viruses were detected simultaneously using a commercially available colloidal gold method. Detailed Implementation

[0025] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments, but the embodiments of the present invention are not limited thereto. Unless otherwise specified, the reagents, methods, and equipment used in the present invention are conventional reagents, methods, and equipment in this technical field. Test methods in the following embodiments that do not specify specific experimental conditions are generally performed according to conventional experimental conditions or experimental conditions recommended by the manufacturer. Unless otherwise specified, the reagents and raw materials used in the present invention can be obtained commercially or by conventional methods.

[0026] Example 1: Explanation of Test Materials and Related Terminology Carboxyl-modified polystyrene microspheres (diameters 4, 6, 10, 70 µm; 10 mg / mL): purchased from Thermo Fisher Scientific.

[0027] Carboxyl-modified polystyrene microspheres (1 mm in diameter): purchased from Suzhou Zhiyi Microsphere Technology Co., Ltd.

[0028] The recombinant nucleoprotein of influenza B virus (CSB-YP361330IJK) and two rabbit polyclonal antibodies (CSB-YP361330LA01IJK and CSB-PA356318HA01IJU) were purchased from Wuhan Huamei Biotechnology Co., Ltd.

[0029] Anti-H1N1 influenza virus nucleocapsid protein antibody (ab104870), anti-influenza A virus nucleoprotein antibody [C43] (ab128193), recombinant influenza A hemagglutinin protein (His tag) strain California / 07 / 2009 / H1N1 (ab217662), anti-parainfluenza virus type III antibody (ab28584), anti-parainfluenza virus type III hemagglutinin antibody [B289M] (ab252769), and natural parainfluenza virus type III protein (ab274662) were purchased from Abogen (Shanghai) Trading Co., Ltd.

[0030] PBS buffer (10 mM, pH=7.4): Take 8.00 g NaCl, 0.20 g KCl, 0.20 g KH2PO4 and 2.90 g Na2HPO4·12H2O and dilute to volume in a 1000 mL volumetric flask, then shake well.

[0031] MES buffer (0.1 M, pH=6.0): Dissolve 21.325 g of MES in deionized water and bring the volume to 1000 mL to obtain solution A; dissolve 4 g of NaOH in deionized water and bring the volume to 1000 mL to obtain solution B; mix 1000 mL of solution A and 400 mL of solution B and shake well.

[0032] PBST, MEST: Add 0.05% Tween-20 to the prepared PBS or MES buffer.

[0033] Example 2: A Deep Learning-Assisted Microfluidic Immunoassay Method Based on an Imaging Transcoding System In this invention, we developed a deep learning-assisted microfluidic immunoassay method based on an imaging transcoding system. This method can be used for rapid on-site detection of multiple pathogenic factors. The core technologies of this method are a microfluidic immunoassay platform and a CbAgo-based immunoassay method. First, the target analyte is coupled with millimeter-sized polystyrene microspheres (PS) containing capture antibodies. mm The antibody (Ab1) and the detection antibody (Ab2-gDNA) labeled with gDNA are mixed to carry out an immune reaction, forming "PS". mm The "Ab1-target-Ab2-gDNA" complex was rapidly deposited at the bottom of the container due to the gravity of the millimeter-sized microspheres, while the unreacted Ab2-gDNA remained suspended in the supernatant. Figure 1 a). Unreacted Ab2-gDNA from the supernatant was added to the CbAgo protein solution, followed by the addition of microsphere signal probes (PS) with biotinylated tDNA conjugated to the mixture. Signal -tDNA-biotin). In this process, gDNA has the property of activating CbAgo protein nucleic acid cleavage, and can activate CbAgo to PS. Signal The tDNA on the surface of the tDNA-biotin is specifically cleaved. The cleaved probe PS Signal -tDNA-biotin only has PS remaining Signal This cutting action significantly reduces PS Signal -tDNA-biotin microspheres coupled with streptavidin SA (PS) Filter The binding efficiency of -SA), the remaining PS Signal -tDNA-biotin and PS Filter -SA combines to form PS Filter -SA-biotin-tDNA-PS Signal complex ( Figure 1 b). Subsequently, the above mixture is injected into the microfluidic chip. The microfluidic chip design includes three functional regions: a mixing region, a filtering region, and an imaging region. Figure 1 c). In the mixing zone, the solution is thoroughly mixed before entering the filtration zone. The outlet of the filtration zone is equipped with a ring array of micropillars with a diameter of 300 micrometers and a central axis spacing of 30 micrometers, allowing only microspheres with a diameter smaller than this size to pass through, thereby achieving the desired effect on PS. SignalEffective separation. These filtered PS Signal It then enters the imaging area, ready for subsequent imaging and recognition. Figure 1 d).

[0034] In the imaging step, a portable smartphone imaging device is used to photograph the imaging area within the microfluidic chip, acquiring images of the microspheres. This imaging device consists of a light source, adjustable aperture, focusing mechanism, lens, microfluidic chip, smartphone, and 3D-printed casing, and is powered by a 12V DC power supply. Figure 4 The imaging area of ​​the microfluidic chip is illuminated by a light source. The light is magnified 400 times by a 4 mm diameter, 2 mm thick lens and projected onto the smartphone's image sensor. The smartphone imaging device captures high-resolution images of the microspheres within the chip, enabling rapid analysis without additional equipment. During imaging, the microfluidic chip is placed on the imaging device, ensuring precise alignment between the imaging area and the smartphone lens. Image clarity is ensured by adjusting the aperture size and the focal length above the light source. Each sample is imaged three times, and the average number of PS microspheres in each sample is calculated to guarantee data accuracy and reliability. Finally, the acquired microsphere images are identified and counted using developed deep learning microsphere recognition software. A linear relationship between the target analyte concentration and the number of microsphere signal probes of different particle sizes is established using the logarithm of the target analyte standard stock solution as the x-axis and the number of microsphere signal probes of different particle sizes as the y-axis, thereby achieving quantitative detection of the target analyte.

[0035] This technical solution can achieve simultaneous detection of multiple targets by designing different antibodies (Abs), guide DNA (gDNA), and tDNA. For example, three different PSs can be used for three different target analytes. mm -Ab1 and its corresponding Ab2-gDNA1, Ab2-gDNA2, Ab2-gDNA3, the PS complex after the immune response mm -Ab1-Target 1-Ab2-gDNA1, PS mm -Ab1-target 2-Ab2-gDNA2, PS mm -Ab1-Target 3-Ab2-gDNA3 after PS mm After sedimentation and separation, unreacted Ab2-gDNA1, Ab2-gDNA2, and Ab2-gDNA3 activate CbAgo via gDNA to specifically cleave different tDNAs (tDNA1, tDNA2, tDNA3), reducing the corresponding PS4-tDNA1-biotin, PS6-tDNA2-biotin, and PS... 10 -tDNA3-biotin and PS FilterThe method leverages the binding efficiency of SA to achieve simultaneous identification and counting of microspheres of various sizes. It captures images of microspheres of different sizes using a smartphone imaging device and utilizes microsphere recognition software to accurately analyze the particle size and quantity of the microspheres, thereby achieving sensitive and rapid detection of multiple targets.

[0036] Example 3: Modification of millimeter-scale and micrometer-scale microspheres using biorecognition molecules 1. Activation of microspheres (1) Take 2 mg of PS microspheres (average diameter of 4, 6, 10, 70 μm) into a centrifuge tube, wash twice with 500 μL MEST (10 mM MES, 0.05% Tween 20, pH 6.0), centrifuge (10000 rpm, 6 min) and remove the supernatant. (2) Prepare a 5 mg / mL EDC solution and a 5 mg / mL NHS solution using 10 mM MES (pH 6.0); (3) Add 100 μL EDC (5 mg / mL) and 50 μL NHS (5 mg / mL) to centrifuge tubes containing PS microspheres, respectively, and use a vortex mixer to mix and fully suspend the PS microspheres. Dilute with MES to 500 μL, place on a rotary mixer, and activate at 37 ℃ for 30 min. (4) Centrifuge (10000 rpm, 6 min), remove the supernatant, and wash 3 times with 500 μL MEST; Through the above steps, the carboxyl groups on the surface of the PS microspheres have been activated.

[0037] 2. Coupling of PS microspheres with biorecognition molecules With PS Signal The coupling process is illustrated using the example of microspheres and biotinylated tDNA. Similar methods can be used for other biorecognition molecules, including influenza A virus, parainfluenza virus, and fungal toxin antibodies.

[0038] (1) Add 10 μM of biotinylated tDNA (tDNA-biotin) to the activated PS Signal The reaction was carried out in microspheres at 37°C for 3 hours. tDNA-biotin and PS Signal Microspheres are covalently bonded through active ester groups to generate PS Signal -tDNA-biotin complex; (2) After the reaction is complete, wash the microspheres with PBST (PBS+Tween) buffer to remove unreacted tDNA-biotin; (3) Centrifuge (10000 rpm, 6 min), remove the supernatant, add 500 μL of PBST (pH 7.4) containing 1% BSA, resuspend the PS microspheres, place them on a rotary mixer, and seal at 37℃ for 30 min. (4) Centrifuge (10000 rpm, 6 min), remove the supernatant, and wash 3 times with 500 μL PBST; (5) The obtained biotinylated tDNA modified PS Signal Microspheres (PS) Signal The tDNA-biotin was resuspended in 1 mL PBST (pH 7.4, containing 0.02% NaN3 and 0.5% BSA) and stored at 4 °C.

[0039] 3. PS Filter Coupling of microspheres with streptavidin (SA) (1) To the activated PS Filter 50 μg of streptavidin (SA) was added to the microsphere solution and reacted at 37°C for 3 hours, so that SA could bind to the surface of the microspheres through covalent bonds. (2) After the reaction is complete, wash the microspheres with PBST (PBS+Tween) buffer to remove unreacted SA; (3) Centrifuge (10000 rpm, 6 min), remove the supernatant, add 500 μL of PBST (pH 7.4) containing 1% BSA, resuspend the PS microspheres, place them on a rotary mixer, and seal at 37℃ for 30 min. (4) Centrifuge (10000 rpm, 6 min), remove the supernatant, and wash 3 times with 500 μL PBST; (5) The obtained streptavidin SA PS microspheres (PS Filter The 0.5% BSA was resuspended in 1 mL PBST (pH 7.4, containing 0.02% NaN3 and 0.5% BSA) and stored at 4 °C.

[0040] Example 3: Simulation Analysis of the Mixing, Filtering, and Separation Regions of a Microfluidic Chip To achieve optimal chip performance, this invention used COMSOL finite element simulation software to analyze fluid mixing efficiency, flow velocity in the filtration zone, and shear force. First, the mixing effect in the mixing zone was simulated to verify its effectiveness in achieving uniform fluid mixing. Extending the straight channel to the left and right sides and adding multiple right-angle turns increases the overall channel length and the collision time of particles in the solution. Figure 2a). This change in flow direction within the reaction zone elongates the fluid streamlines, enhancing the mixing effect through cross-permeation. Subsequent simulations of the fluid concentration within the reaction zone show that the concentration at the outlet tends to stabilize, while the concentration at the inlet exhibits fluctuations. Figure 2 (b) This result indicates that the fluid achieves thorough mixing upon passing through the mixing zone. After thorough mixing, the fluid is guided to the filtration zone for separation. The array of micropillars in this zone is designed with a diameter of 300 μm and a spacing of 30 μm, allowing only smaller coded probes to pass through. Figure 3 This design ensures that the encoded probe can be effectively separated from the reaction carrier, guaranteeing the accuracy of subsequent imaging and detection.

[0041] To investigate the effects of different conditions on the filtration process, we performed computational fluid dynamics (CFD) simulations. These simulations evaluated the fluid velocity distribution within the filtration and imaging regions. The results show that different flow velocities significantly affect fluid stability. Figure 2 c). Among the tested flow rates, 1 mm / s was found to be the optimal flow rate. At this flow rate, the velocity remained relatively stable in both regions, with only slight fluctuations near the channel boundary. Figure 2 d). Furthermore, particle tracking of 10 μm particles showed that at this flow rate, particles could pass through the array of micropillars without agglomeration. Simulation results confirmed that the fluid was effectively stabilized within the micropillar region, supporting the consistency of fluid behavior during filtration. Figure 2 e). Furthermore, the low shear stress (0.013 Pa) ensured effective mixing of the microsphere mixture, prevented aggregation, and maintained channel patency. These simulation results validate the efficiency of the microfluidic chip in mixing, separation, and filtration processes, providing a theoretical basis for its application in biosensing.

[0042] Example 4 Practical verification of mixing, filtering and separation regions of microfluidic chips like Figure 5 As shown in Figure a, the portable smartphone imaging device designed in this study is easy to transport and install, making it ideal for field applications. The device measures approximately 160 mm (length) × 100 mm (width) × 118 mm (height) and weighs approximately 285 grams, ensuring portability and low cost. The accompanying portable imaging device includes a reagent kit specifically designed for point-of-care testing (POCT), encompassing a filter membrane, syringe, reagents, microfluidic chip, and sampling needle. These components are customized for on-site diagnostic procedures, facilitating seamless integration with the imaging device. Figure 5 b).

[0043] To practically validate the chip, this study compared the separation performance of filtration with that of centrifugation. Scanning electron microscopy (SEM) images showed that centrifugation resulted in more PS4 microspheres and PS4 microspheres regardless of the presence of the target analyte. 70 Carrier combination ( Figure 5 c, e). Furthermore, an automated particle counter was used to evaluate the number of PS microsphere probes in the complexes containing and without the target analyte under both separation methods. Statistical analysis showed a significant difference in the number of PS4 microspheres between samples containing and without the target analyte (c, e). Figure 5 d, f). This advantage is attributed to the fact that filtration reduces protein aggregation and nonspecific binding, problems that are typically caused by high gravity during centrifugation. (PS complex) 70 -SA-biotin-tDNA-PS4, PS 70 -SA-biotin-tDNA-PS6, PS 70 -SA-biotin-tDNA-PS 10 SEM image analysis showed the occurrence of the reaction. Figure 5 g). Simultaneously, optical microscopy confirmed that the coded probe was filtered out after the separation reaction, while the complex remained within the filtration area. Figure 5 Therefore, these experimental results and characterization data validate the feasibility of this method and lay the foundation for the accurate detection of multiple targets.

[0044] Example 5 Taking the detection of respiratory viruses as an example, this method can be used to simultaneously detect influenza A virus (FLUA), influenza B virus (FLUB), and human parainfluenza virus (HPIV) and establish standard curves and linear ranges. The experiments in this embodiment were conducted using the experimental steps described in Example 2 above. The conditions for simultaneous detection of the three respiratory viruses were set as follows: Simultaneous detection of multiple targets can be achieved by designing different antibodies (Ab1, Ab2), guide DNA (gDNA1, gDNA2, gDNA3), and target DNA (tDNA1, tDNA2, tDNA3).

[0045] Target 1 (FLUA): and its corresponding PS mm -Ab1 (targeting FLUA) and Ab2-gDNA1 (targeting FLUA) trigger an immune response. Ab2-gDNA1 from the supernatant is used to activate CbAgo's specific cleavage of tDNA1, reducing the interaction between PS4-tDNA-biotin and PS... 70 -SA binding efficiency.

[0046] Target 2 (FLUB): and its corresponding PSmm -Ab1 (targeting FLUB) and Ab2-gDNA2 (targeting FLUB) trigger an immune response. Ab2-gDNA2 from the supernatant is used to activate CbAgo's specific cleavage of tDNA2, reducing the interaction between PS6-tDNA2-biotin and PS... 70 -SA binding efficiency.

[0047] Target 3 (HPIV): and its corresponding PS mm -Ab1 (for HPIV) and Ab2-gDNA3 (for HPIV) trigger an immune response. Ab2-gDNA3 from the supernatant is used to activate CbAgo's specific cleavage of tDNA3, reducing PS. 10 -tDNA3-biotin and PS 70 -SA binding efficiency.

[0048] (1) Millimeter-sized polystyrene microspheres (PS) conjugated with capture antibodies mm -Ab1 (two antibodies for each target), gDNA-labeled detection antibodies (Ab2-gDNA1, Ab2-gDNA2, Ab2-gDNA3, concentration 25 μg / mL, volume 50 μL) and target viral recombinant proteins (FLUA, FLUB, HPIV) were mixed at a 1:1:1 ratio. The mixture was incubated at 37°C for 30 minutes to allow the target analytes to react with PS. mm -Ab1 and Ab2-gDNA form "PS" mm The "Ab1-target-Ab2-gDNA" complex was used. Due to the gravity of the millimeter-sized microspheres, the complex rapidly settled to the bottom of the container, while the unreacted Ab2-gDNA remained suspended in the supernatant. The settled complex was separated from the supernatant by gravity sedimentation, and the unreacted Ab2-gDNA in the supernatant was used for subsequent steps.

[0049] (2) Add the unreacted Ab2-gDNA from the supernatant to the solution containing CbAgo protein, adjusting the final volume to 50 μL; add PS4-tDNA-biotin, PS6-tDNA-biotin, and PS to the Ab2-gDNA and CbAgo mixture. 10 -tDNA-biotin, concentration 1 mg / mL, volume 100 μL, incubated at 37°C for 30 minutes, gDNA activates the nucleic acid cleavage properties of CbAgo protein, realizing CbAgo's ability to cleave PS4-tDNA-biotin, PS6-tDNA-biotin, and PS... 10 The tDNA on the surface of the tDNA-biotin is specifically cleaved, releasing the cleavage products.

[0050] (3) Add PS microspheres (PS) to the above mixture. 70 A mixture of streptavidin SA (2 mg / mL, 100 μL) was prepared. Due to the high affinity between streptavidin SA and biotin, a "PS" (penicillin-based) compound was formed. 70 -SA-biotin-tDNA-PS4, PS 70 -SA-biotin-tDNA-PS6, PS 70 -SA-biotin-tDNA-PS 10 "Compound. Incubate at 25°C for 5 minutes to ensure complete binding."

[0051] (4) The above mixture is injected into the microfluidic chip. After the mixture is fully mixed, it enters the filtration zone. The filtration zone only allows polystyrene microspheres (PS4, PS6, PS7, PS8) with diameters of 4 μm, 6 μm, and 10 μm. 10 This process enables efficient size screening and purification. After filtration, PS4, PS6, and PS... 10 The microspheres enter the imaging area, ready for subsequent imaging and recognition.

[0052] (5) Use the constructed portable smartphone imaging device to take pictures of the imaging area inside the microfluidic chip to obtain microsphere images. Each sample is imaged three times, and the average number of PS microspheres in each sample is calculated to ensure the accuracy and reliability of the data.

[0053] Table 1. Nucleic acid sequences corresponding to the three target substances

[0054] Example 6 When using the colloidal gold method for detection, colloidal gold-labeled virus-specific antibodies are immobilized on the conjugation pad of the test strip. When a respiratory sample containing three viral antigens is added to the sample application area, the viral antigens in the sample bind to the colloidal gold-labeled antibodies to form a complex. This complex migrates along the nitrocellulose membrane via chromatography. When it reaches the detection line, a pre-embedded virus-specific antibody captures the complex, causing the colloidal gold to aggregate and develop color. Three drops of each sample are added to the corresponding test strip, incubated for 15 minutes, and then photographed. The result can be determined by visually observing the color bands. The detection results of this method are as follows: Figure 6 As shown.

[0055] A comparison with the two conventional methods described above demonstrates that the deep learning-assisted microfluidic immunoassay method based on a smartphone imaging transcoding system, as described in this invention, combines the specificity of immunological antigen-antibody binding, avoids false positives caused by nucleic acid amplification, and exhibits extremely high sensitivity and specificity. The microfluidic chip ensures that only the signal probe passes through the filtering area, and the developed AI microsphere recognition software can accurately analyze interfering microspheres (aggregates, bubbles, impurities), greatly improving the reliability of this method. The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The embodiments and features described in these embodiments can be arbitrarily combined without conflict. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A deep learning-assisted microfluidic immunoassay method based on an imaging transcoding system, characterized in that: The method includes the following steps: 1) The target analyte is coupled with millimeter-sized microspheres (PS) containing capture antibodies. mm -Detection antibodies Ab1 and Ab2-gDNA markers react with Ab2-gDNA to form "PS". mm The "-Ab1-target-Ab2-gDNA" complex, which is in PS mm Under the influence of gravity, it quickly settles to the bottom of the container, and the unreacted Ab2-gDNA is suspended in the supernatant. The unreacted Ab2-gDNA in the supernatant is taken for subsequent detection. 2) Unreacted Ab2-gDNA in the supernatant was added to the CbAgo solution, followed by the addition of microsphere signal probes PS with different particle sizes of biotinylated tDNA conjugated to the Ab2-gDNA and CbAgo mixture. Signal -tDNA-biotin, to obtain a mixture; gDNA can activate the nucleic acid cleavage properties of CbAgo, thereby enabling CbAgo to cleave PS. Signal Specific cleavage of tDNA on tDNA-biotin, resulting in PS after cleavage Signal -tDNA-biotin only has PS remaining Signal It enters the imaging area for subsequent detection; 3) Add microspheres PS with streptavidin SA surface-modified and coupled to the above mixture. Filter -SA, due to the binding of SA and biotin, forms PS Filter -SA-biotin-tDNA-PS Signal complex; 4) Pass the mixture obtained in step 3) above into the microfluidic chip. The filtration area of ​​the chip only allows PS to pass through. Signal Through, PS Signal The image passes through the filtering area and enters the imaging area. 5) Use the constructed portable imaging device to take pictures of the imaging area inside the chip to obtain microsphere images, and identify and count the particle size and number of microspheres in the images; 6) Plot the logarithm of the target standard stock solution on the x-axis, and compare the PS values ​​of different particle sizes. Signal Using the quantity as the ordinate, calculate the concentration of the target analyte to complete the detection of the target analyte; The method is used for the simultaneous detection of multiple targets.

2. The method according to claim 1, characterized in that: The microspheres PS mm The separation carrier 1, used for sedimentation separation, has a diameter ranging from 0.5 mm to 2 mm; the microspheres PS Filter The separation carrier 2, used for filtration separation, has a diameter ranging from 40 μm to 100 μm; the PS Signal It is a signal probe used to encode the target object. Each target object corresponds to a microsphere of a certain size. Different sizes of signal probes are selected according to the type of target object, with a diameter range of 1 μm-20 μm.

3. The method according to claim 2, characterized in that: The microspheres PS mm The diameter is 1 mm; the microspheres PS Filter The diameter is 70μm.

4. The method according to claim 1, characterized in that: The method is used for the simultaneous detection of three target objects, with the microsphere signal probe PS Signal PS in tDNA-biotin Signal The sizes were selected as 4, 6, and 10 μm, respectively, corresponding to complexes PS4-tDNA1-biotin, PS6-tDNA2-biotin, and PS... 10 -tDNA3-biotin; The specific process is as follows: Using three different Photoshop versions mm -Ab1 and its corresponding Ab2-gDNA1, Ab2-gDNA2, Ab2-gDNA3, the PS complex after the immune response mm -Ab1-Target 1-Ab2-gDNA1, PS mm -Ab1-target 2-Ab2-gDNA2, PS mm -Ab1-Target 3-Ab2-gDNA3 after PS mm After sedimentation and separation, unreacted Ab2-gDNA1, Ab2-gDNA2, and Ab2-gDNA3 activate CbAgo via gDNA to specifically cleave different tDNAs, namely tDNA1, tDNA2, and tDNA3, thereby reducing the corresponding signal probe complexes PS4-tDNA1-biotin, PS6-tDNA2-biotin, and PS... 10 -tDNA3-biotin and PS Filter -SA binding efficiency, the cleaved signal probe only retains PS4 and PS 6、 PS 10 The probes enter the imaging area for subsequent detection, enabling simultaneous identification and counting of microspheres of multiple sizes. The size of the microsphere probes corresponds to the type of target analyte, and the number of microsphere probes corresponds to the concentration of the target analyte.

5. The method according to claim 1, characterized in that: In step 1), the capture antibody Ab1 and the detection antibody Ab2 are biorecognition molecules targeting the large molecular target. When the target is a small molecule, they can be replaced with the capture antibody Ab and the complete antigen BSA. The small molecule target is coupled with millimeter-sized microspheres PS with the capture antibody. mm -A competitive immune response occurs among the complete antigens BSA and gDNA labeled with Ab and gDNA; The microspheres are any one of polystyrene latex microspheres, polybutadiene latex microspheres, polyisoprene latex microspheres, or polyacrylic acid latex microspheres; the gDNA and tDNA are single-stranded nucleic acids capable of base pairing; the biotin and streptavidin SA can bind to form a stable complex.

6. The method according to claim 1, characterized in that: The microspheres PS mm The settling velocity is greater than PS Filter and PS Signal The settling velocity.

7. The method according to claim 1, characterized in that: In step 4), the microfluidic chip contains three functional regions: a mixing region, a filtering region, and an imaging region; the mixed solution is fully mixed in the mixing region before entering the filtering region; The outlet of the filtering region contains a ring array of micropillars, through which the filtered signal probe PS Signal Entering the imaging region for subsequent imaging and identification; the micropillar size does not affect the liquid flow rate, with a diameter range of 100-400 μm, and the micropillar spacing should ensure that only PS is allowed. signal It passes through without aggregation, and the range is 20-50 μm.

8. The method according to claim 1, characterized in that: In step 4), the micropillar diameter is 300 μm and the micropillar spacing is 30 μm.

9. The method according to claim 8, characterized in that: In step 4), the shear stress in the mixing region is 0.01-0.02 Pa; the fluid velocity in the filtering region and the imaging region is 0.5-1.5 mm / s.

10. The method according to claim 9, characterized in that: In step 4), the shear stress in the mixing region is 0.013 Pa; the fluid velocity in the filtering region and the imaging region is 1 mm / s.

11. The method according to claim 1, characterized in that: In step 5), the portable imaging device consists of a light source, an adjustable aperture, a focusing mechanism, a lens, a microfluidic chip, a smartphone, and a 3D printed shell, and is powered by a DC mobile power supply; the imaging area of ​​the microfluidic chip is illuminated by the light source, and the light is magnified by the lens and projected onto the smartphone image sensor; In step 5), the portable imaging device is responsible for capturing images of the microspheres inside the chip to achieve rapid analysis. The imaging process is as follows: the microfluidic chip is placed on the imaging device, and the chip position is adjusted to align the imaging area with the smartphone lens. By adjusting the aperture size and the focal length above the light source, the image is ensured to be clear.

Images