Single molecule rapid detection method and system
By combining bright-field focusing mode and deep learning model, fast and accurate single-molecule detection is achieved, solving the problems of slow detection speed and low accuracy in existing technologies, and realizing efficient single-molecule detection.
Patent Information
- Application Number
- CN202511824587.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-05
- Publication Date
- 2026-03-17
AI Technical Summary
Existing single-molecule detection technologies suffer from problems such as system complexity, high cost, slow speed, poor accuracy, and high false positive rate, making it difficult to meet the demand for low-cost, high-throughput, and rapid detection.
By employing a bright-field focusing mode combined with a deep learning model, and through autofocus, image acquisition, and Poisson distribution statistics, rapid and accurate single-molecule detection is achieved.
It achieves rapid, high-precision, and absolutely quantitative single-molecule detection, overcoming the shortcomings of slow point-by-point scanning speed and inaccurate quantification in traditional wide-field imaging, and realizing rapid end-to-end optimization from focusing to analysis.
Smart Images
Figure CN121678616A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of microfluidic chip technology, and in particular to a method and system for rapid detection of single molecules. Background Technology
[0002] Single-molecule detection technology is a rapidly developing biomedical detection technology in recent years, which can achieve high-sensitivity and high-resolution detection and analysis of individual biomolecules or chemical molecules.
[0003] Existing high-sensitivity immunoassay methods are mostly based on laser confocal point-by-point scanning methods using PMT fluorescence intensity detection. While this method offers high sensitivity, it suffers from drawbacks such as system complexity and high cost, making it difficult to meet the demands of industrial, research, and clinical diagnostics for low-cost, high-throughput, and rapid detection. Wide-field fluorescence imaging technologies based on CCD and CMOS cameras offer high speed, but face multiple challenges at the single-molecule detection level: low signal-to-noise ratio: weak single-molecule signals are difficult to distinguish from strong background noise; focusing: focusing under the fluorescence channel is difficult due to weak signals, susceptibility to photobleaching, and interference from impurities, resulting in slow speed, poor accuracy, and high false positive rates, severely impacting the accuracy of subsequent imaging and quantification; identification difficulties: accurately identifying thousands of real single-molecule signal points from complex image backgrounds and eliminating false positives caused by dust, impurities, and non-specific adsorption.
[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention
[0005] Based on this, this application provides a rapid single-molecule detection method and system, which realizes rapid, high-precision, and absolute quantitative single-molecule detection, overcomes the shortcomings of slow point-by-point scanning speed and inaccurate quantification of traditional wide-field imaging, and achieves rapid end-to-end optimization from focusing to analysis.
[0006] To achieve the above objectives, embodiments of this application provide a rapid single-molecule detection method, comprising:
[0007] The microarray chip to be tested, which has undergone in-situ reaction, is placed on the detection stage of the detection system;
[0008] The detection system is powered on with a bright field light source;
[0009] The fluorescent camera activates its autofocus function and completes autofocus within 1 second;
[0010] The fluorescent camera locks the focus position parameters;
[0011] The detection system shuts down the bright field light source and turns on the fluorescence excitation light source;
[0012] Control the fluorescence camera to perform exposure and acquire fluorescence images;
[0013] The fluorescence image is input into a deep learning model to identify and output the location information of all single-molecule fluorescent spots in the fluorescence image;
[0014] Count the number of light spots falling into each unit of the microarray chip;
[0015] Based on the Poisson distribution principle, the number of single-molecule spots obtained by statistics is used as the direct counting result to calculate the absolute concentration or absolute copy number.
[0016] Preferably, based on the Poisson distribution principle, the number of single-molecule spots obtained statistically is used as the direct counting result. Calculating the absolute concentration or absolute copy number further includes the following steps:
[0017] The quantitative analysis results of the entire microarray chip are output via a mobile terminal.
[0018] Preferably, the deep learning model is a trained algorithm model.
[0019] Preferably, the deep learning model includes a convolutional neural network for image features and a graph neural network for processing spatial relationships.
[0020] Preferably, before placing the microarray chip to be tested, which has undergone in-situ reaction, on the detection stage of the detection system, the following steps are also included:
[0021] Coarse adjustments are achieved by moving the Z-axis over a wide range.
[0022] Preferably, the fluorescence camera's autofocus function, which completes autofocus within 1 second, includes the following steps:
[0023] Use array structures or grid positioning points in the microarray chip as the focus target;
[0024] Precise focusing is achieved by controlling the lens assembly inside the fluorescent camera to move in minute, rapid, and vibration-free movements.
[0025] Preferably, the microarray chip includes a microarray gene chip, a microarray protein chip, a microfluidic chip, and an organ-on-a-chip.
[0026] Preferably, the mobile terminal includes a PC, an industrial control computer, a mobile phone, a tablet, or a handheld analyzer.
[0027] Preferably, it is used for the detection of microspheres, nucleic acids, and fluorescent groups.
[0028] This application also provides a single-molecule rapid detection system for implementing the above-described single-molecule rapid detection method, comprising:
[0029] The light source combines excitation light and white light, the objective lens, the CMOS camera with autofocus, and the dichroic mirror and filter that are compatible with the fluorescent microspheres.
[0030] The single-molecule rapid detection method provided by this invention has the following advantages and beneficial effects:
[0031] This application combines efficient optical design, a high-sensitivity camera, bright-field structure focusing, deep learning image recognition, and a Poisson distribution statistical model to achieve rapid, high-precision, and absolutely quantitative single-molecule detection. It overcomes the shortcomings of slow point-by-point scanning speed and inaccurate quantification in traditional wide-field imaging, and achieves rapid end-to-end optimization from focusing to analysis. Attached Figure Description
[0032] Figure 1 This is a schematic flowchart of an embodiment of the single-molecule rapid detection method of this application.
[0033] Figure 2 This is an example image of bright-field focusing in the single-molecule rapid detection method of this application.
[0034] Figure 3 This is a fluorescence field diagram of the same focal plane in the single-molecule rapid detection method of this application.
[0035] Figure 4 This is the original fluorescence field image from the single-molecule rapid detection method in this application.
[0036] Figure 5 This is a fluorescence spot recognition diagram in the single-molecule rapid detection method of this application.
[0037] Figure 6 This is a flowchart illustrating the single-molecule rapid detection method of this application for the detection of microspheres, nucleic acids, and fluorescent groups.
[0038] Figure 7 This is a fluorescence imaging image of the single-molecule rapid detection method of this application for the detection of microspheres, nucleic acids and fluorescent groups. Detailed Implementation
[0039] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings, which illustrate preferred embodiments of the application. However, this application may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that the disclosure of this application will be thorough and complete.
[0040] It should be noted that when a component is considered to be "connected" to another component, it can be directly connected to and integrated with the other component, or there may be an intervening component present. The term "mounted" and similar expressions used in this document are for illustrative purposes only.
[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.
[0042] In Example 1, as Figure 1 As shown, a rapid single-molecule detection method is provided, which includes:
[0043] S10. Place the microarray chip to be tested, which has undergone in-situ reaction, on the testing stage of the detection system.
[0044] S11. The detection system turns on the bright field light source;
[0045] S12. The fluorescent camera starts the autofocus function and completes autofocus within 1 second;
[0046] S13, Fluorescent camera lock focus position parameters;
[0047] S14. The detection system turns off the bright field light source and turns on the fluorescence excitation light source;
[0048] S15. Control the fluorescence camera to perform exposure and acquire fluorescence images;
[0049] S16. Input the fluorescence image into the deep learning model to identify and output the position information of all single-molecule fluorescent spots in the fluorescence image;
[0050] S17. Count the number of light spots falling into each unit on the microarray chip;
[0051] S18. Based on the Poisson distribution principle, the number of single-molecule spots obtained by statistics is used as the direct counting result to calculate the absolute concentration or absolute copy number.
[0052] In specific implementation, S18, based on the Poisson distribution principle, uses the statistically obtained number of single-molecule spots as the direct counting result, and the calculation of absolute concentration or absolute copy number also includes the following steps:
[0053] S19. Output the quantitative analysis results of the entire microarray chip through the mobile terminal.
[0054] For samples with flat surfaces but minute height and material differences, such as microarray chips and microfluidic chips, this method employs coaxial light imaging technology to ensure image clarity and contrast. It achieves clear imaging and differentiation of minute height differences (at the μm level).
[0055] This method employs a bright-field focusing mode, utilizing the stable physical structure of the microarray chip itself (such as the array structure and grid positioning points) as the focusing target. After achieving precise focusing and bright-field imaging, the focal plane is locked, and the image is switched to fluorescence mode for imaging. The chip structure in bright-field mode exhibits high signal intensity, good contrast, is not bleached, and is unaffected by phototoxicity, providing an extremely stable and reliable input source for the focusing algorithm. Relying on fluorescence signal intensity for focusing can easily lead to misinterpretation of strong fluorescence generated by dust, impurities, or non-specific adsorption points as the focal plane, resulting in focusing errors. Of course, under certain conditions, such as a stable number of microspheres in each imaging region, the bright-field imaging step can be skipped.
[0056] This application employs a Z-axis movement + autofocus camera to replace the traditional Z-axis focusing platform, achieving a coarse-fine focusing strategy. The coarse-adjustment Z-axis platform handles large-scale movement, while fine-adjustment is accomplished by the minute, rapid, and vibration-free movement of the camera's internal lens groups. If the chip flatness is sufficient, Z-axis movement is unnecessary after the device has undergone factory coarse adjustment. This method significantly shortens focusing time and completely avoids image jitter that may be caused by mechanical movement, thus achieving unprecedented focusing speed and stability, which is particularly beneficial for rapid batch inspection.
[0057] This application employs advanced machine learning models (primarily deep learning models) for single-molecule recognition, effectively distinguishing real signals from background noise and eliminating interference from dust, impurities, and other contaminants, significantly reducing the system's false positive rate. The machine learning model, after training, is used to identify single-molecule spots in fluorescence images. The deep learning model primarily includes convolutional neural networks (CNNs) for image feature processing and graph neural networks (GNNs) for handling spatial relationships. Of course, other applicable traditional and deep learning algorithms can also be used.
[0058] This application achieves absolute quantification by counting the number of single-molecule spots and then directly calculating based on the Poisson distribution principle. The output is converted from a volatile "analog signal" to a stable "digital signal," achieving true absolute quantification whose accuracy is no longer affected by variables such as fluorescence efficiency and uneven illumination.
[0059] The entire detection process of this application is automated, realizing a fast and fully automated workflow from bright field autofocus, fluorescence channel switching, image acquisition, intelligent model recognition, quantitative analysis of Poisson distribution to report generation. The detection time for each interval can be shortened to less than 10 seconds.
[0060] In specific implementation, S12, the fluorescence camera starts the autofocus function and completes autofocus within 1 second, including the following steps:
[0061] S121. Use the array structure or grid positioning points in the microarray chip as the focus target;
[0062] S122. Control the lens group inside the fluorescence camera to perform small, rapid, and vibration-free movement, so as to achieve precise focusing.
[0063] In specific implementation, the microarray chip includes a microarray gene chip, a microarray protein chip, a microfluidic chip, and an organ chip.
[0064] In specific implementation, the mobile terminal includes a PC, an industrial control computer, a mobile phone, a tablet, and a handheld analyzer.
[0065] This application also provides a single-molecule rapid detection system for implementing the above-mentioned single-molecule rapid detection method, which includes: a light source that combines excitation light and white light, a CMOS camera with an objective lens of 20x / 0.75NA, 4k resolution, and an automatic focusing function, a dichroic mirror and a filter that are compatible with fluorescent microspheres. The imaging principle of this detection system is the imaging principle of coaxial light. When taking pictures, white light or excitation light enters from the light source entrance, is reflected by the dichroic mirror and vertically irradiates the sample surface of the microfluidic chip. The vertically reflected light returns along the original path, passes through the beam splitter and is received by the camera, thereby forming a high-contrast bright-field image.
[0066] The bright-field focusing example diagram obtained through this application (as shown in Figure 2 ), the co-focal plane fluorescence field diagram (as shown in Figure 3 ), in the large model to identify bright spots, the original fluorescence field diagram (as shown in Figure 4 ), the fluorescence spot identification diagram (as shown in Figure 5 ), the test data of different antigen concentration gradients of two projects, IL-6 and cTnI, through this application are shown in the following table.
[0067]
[0068] According to this table, in the IL-� project, the signal at 0 pg / ml is less than 100, the sensitivity is as low as 20 fg / ml, the detection range is 0 - 1000 pg / ml, and there is a good gradient reactivity.
[0069] In the cTnI project, the signal at 0 pg / ml is less than 100, the sensitivity is as low as 500 fg / ml, the detection range is 0 - 1000 pg / ml, and there is a good gradient reactivity.
[0070] From the above data, it can be seen that in-situ reactions can perform multi-index detections, and each project has good gradient reactivity and fg-level sensitivity.
[0071] In specific implementation, this method is not limited to immunofluorescence detection, and can also be used for the detection of microspheres, nucleic acids, and fluorescent groups. <When applied to the detection of microspheres, nucleic acids, and fluorescent groups, in situ hybridization is used, eliminating the need for PCR. The surface of the microspheres can no longer be coated with antibodies; instead, nucleic acid probes (a known sequence of DNA or RNA) can be grafted onto them. Like antibodies, these probes can specifically capture complementary target nucleic acid sequences in solution through base pairing. The detection process is as follows: Figure 6 As shown, its fluorescence imaging is as follows Figure 7 As shown.
[0073] In summary, the single-molecule rapid detection method and system provided in this application achieve rapid, high-precision, and absolute quantitative single-molecule detection by combining efficient optical design, high-sensitivity camera, bright-field structure focusing, deep learning image recognition, and Poisson distribution statistical model. It overcomes the shortcomings of slow point-by-point scanning speed and inaccurate quantification of traditional wide-field imaging, and realizes rapid end-to-end optimization from focusing to analysis.
[0074] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0075] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A single molecule rapid detection method, characterized by, The method comprises the following steps: placing the microarray chip subjected to in-situ reaction on a detection table of a detection system; the detection system turns on a bright field light source; the fluorescence camera starts an automatic focusing function and completes automatic focusing within 1s; the fluorescence camera locks the focusing position parameters; the detection system turns off the bright field light source and turns on a fluorescence excitation light source; the fluorescence camera is controlled to expose and collect a fluorescence image; the fluorescence image is input into a deep learning model to identify and output position information of all single-molecule fluorescent spots in the fluorescence image; the number of fluorescent spots falling into each unit on the microarray chip is counted; based on the Poisson distribution principle, the number of single-molecule spots counted is taken as a direct counting result to calculate absolute concentration or absolute copy number.
2. The single molecule rapid detection method according to claim 1, wherein, Based on the Poisson distribution principle, the number of single-molecule spots counted is taken as a direct counting result to calculate absolute concentration or absolute copy number further comprises the following steps: outputting the quantitative analysis result of the entire microarray chip through a mobile terminal.
3. The single molecule rapid detection method of claim 1, wherein, The deep learning model is a trained algorithm model.
4. The single molecule rapid detection method of claim 1, wherein, The deep learning model comprises a convolutional neural network for image features and a graph neural network for processing spatial relationships.
5. The single molecule rapid detection method of claim 1, wherein, Before placing the microarray chip subjected to in-situ reaction on the detection table of the detection system, the following step is further included: coarse adjustment is realized by moving the Z table in a large range.
6. The single molecule rapid detection method according to claim 5, wherein, The fluorescence camera starts an automatic focusing function and completes automatic focusing within 1s, which comprises the following steps: array structures or grid positioning points in the microarray chip are taken as focusing targets; the lens group inside the fluorescence camera is controlled to move in a small, fast and vibration-free manner, so as to realize accurate focusing.
7. The single molecule rapid detection method of claim 1, wherein, The microarray chip comprises a microarray gene chip, a microarray protein chip, a microfluidic chip and an organ chip.
8. The single molecule rapid detection method of claim 2, wherein, The mobile terminal comprises a PC, an industrial computer, a mobile phone, a tablet and a handheld analyzer.
9. The single molecule rapid detection method of claim 1, wherein, The detection is used for microspheres, nucleic acids and fluorescent groups.
10. A single molecule rapid detection system for carrying out the single molecule rapid detection method according to any one of claims 1 to 9, characterized by, It comprises: a light source combining excitation light and white light, an objective lens, a CMOS camera with an automatic focusing function, a dichroic mirror and a filter matched with fluorescent microspheres.