Microdroplet dividing and clustering method, device, medium and system

Abstract

The invention relates to a droplet dividing and clustering method, device, medium and system, the method is applied to a droplet-based digital nucleic acid amplification quantitative analysis system, and the method comprises the following steps: reading original scatter data of each droplet in a droplet fluorescence image, including position and brightness; acquiring a brightness distribution density condition based on the original scatter data; screening reference micro-droplets according to the brightness distribution density condition; performing light field correction on the microdroplet fluorescence image according to the reference microdroplet to obtain actual scatter data of the microdroplet; and dividing and clustering the microdroplets in the microdroplet fluorescence image according to the actual scatter data. According to the method, light field estimation and correction are carried out in real time completely based on the micro-droplet image, an additional special calibration object or a reference target is not needed, the method has higher robustness and adaptability to light field changes, brightness data of the micro-droplet are fully and effectively restored, and the presentation effect of the brightness of the micro-droplet data is improved.

Description

[0001] Cross-references to related applications

[0002] This invention claims priority to Chinese patent application CN202411560659.2, filed on November 4, 2024, entitled “Method, Apparatus, Medium and System for Droplet Partitioning and Clustering”, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This disclosure relates to the field of image-based digital PCR technology, and particularly to a method, apparatus, medium, and system for droplet partitioning and clustering. Background Technology

[0004] Image-based digital PCR (Polymerase Chain Reaction) systems suffer from uneven light fields during imaging. Observing droplet fluorescence images, this unevenness causes brightness differences across different regions. This difference leads to significant fluctuations in fluorescence intensity among the same type of negative or positive droplets in different areas of the image. Furthermore, it can cause overlap in fluorescence intensity between negative and positive droplets in different regions. Both phenomena make it difficult to accurately represent droplet fluorescence intensity and hinder the classification and clustering of positive and negative droplets.

[0005] Some related technologies employ thresholding before optical field correction, but this method suffers from distortion after correction due to incorrect thresholding. Other related technologies use fixed markers or reference targets for correction to form correction parameters. However, this method may not work well when the actual situation differs from the pre-calibration method, and in extreme cases, it may even lead to incorrect correction. Summary of the Invention

[0006] This disclosure provides a method, apparatus, medium, and system for droplet segmentation and clustering, so as to accurately present the brightness information of droplets and facilitate the segmentation and clustering of droplets.

[0007] In a first aspect, embodiments of this disclosure provide a droplet partitioning and clustering method, applied to a droplet-based digital nucleic acid amplification and quantitative analysis system, including:

[0008] Read the raw scatter data of each droplet in the droplet fluorescence image, including its position and brightness;

[0009] Based on the original scatter data, the brightness distribution density is obtained;

[0010] Based on the brightness distribution density, a benchmark microdroplet is selected;

[0011] The fluorescence image of the microdroplet is corrected for light field based on the reference microdroplet to obtain the actual scatter data of the microdroplet.

[0012] The droplets in the droplet fluorescence image are divided and clustered based on the actual scatter data.

[0013] In some exemplary embodiments, obtaining the brightness distribution density based on the original scatter data includes:

[0014] The original scatter data is converted into a droplet scatter heatmap to characterize the brightness distribution density.

[0015] In some exemplary embodiments, converting the raw scatter data into a droplet scatter heatmap includes:

[0016] The microdroplet fluorescence image is divided into multiple grids, and all grids are sorted and numbered. All microdroplets in each grid form a subarray, and the subarrays are connected to form a microdroplet array.

[0017] Using the index of the droplets in the droplet array as the horizontal axis and the brightness of the droplets as the vertical axis, the droplet array is mapped onto a target image of a set size to form a droplet scatter heatmap.

[0018] In some exemplary embodiments, the droplet fluorescence image is divided into multiple grids, all grids are sorted and numbered, all droplets in each grid form a subarray, and the subarrays are connected to form a droplet array, including:

[0019] The microdroplet fluorescence image is divided into multiple grids;

[0020] Sort and number all grids in an S-shaped order;

[0021] The grid in which each droplet is located is determined based on its position.

[0022] Create a subarray for each grid, and add each droplet to the subarray of the grid it belongs to;

[0023] Connect all subarrays according to the corresponding grid number order to form a droplet array.

[0024] In some exemplary embodiments, the microdroplet array is mapped onto a target image of a set size, using the index of the microdroplet in the array as the x-axis and the brightness of the microdroplet as the y-axis, to form a microdroplet scatter heatmap, including:

[0025] Using the index of the droplet in the droplet array as the x-axis and the brightness of the droplet as the y-axis, the x-axis is mapped to the value in the width direction of the target image, and the y-axis is mapped to the value in the height direction of the target image. The gray value of the mapped position of the droplet on the target image is incremented by 1. The mapped target image is then normalized and converted into an image with a gray value range of 0 to 255 to obtain a droplet scatter heatmap.

[0026] In some exemplary embodiments, screening benchmark droplets based on the brightness distribution density includes:

[0027] According to the preset reference microdroplet mask size conditions, the target area in the microdroplet scatter thermal map that meets the preset conditions is segmented;

[0028] If there is only one target region, that target region is determined as the reference microdroplet mask;

[0029] When there is more than one target region, the lowermost target region in the vertical position relationship is determined as the reference microdroplet mask;

[0030] Extract the reference microdroplets that fall into the reference microdroplet mask.

[0031] In some exemplary embodiments, performing optical field correction on the microdroplet fluorescence image based on the reference microdroplet to obtain the actual scatter data of the microdroplet includes:

[0032] The microdroplet fluorescence image is divided into multiple grids;

[0033] The brightness of each grid is calculated based on the brightness of the reference microdroplet, and the brightness of all grids is used as elements to form a light field estimation matrix;

[0034] Calculate the mean brightness of the light field estimation matrix, and based on the mean brightness and each element of the light field estimation matrix, calculate a correction field matrix with the same size as the light field estimation matrix;

[0035] Based on the correction field matrix and the brightness value of the droplets in each grid, the actual brightness of the droplets is calculated to obtain the actual scatter data.

[0036] In some exemplary embodiments, calculating the brightness of each grid based on the brightness of the reference droplet includes:

[0037] For a first grid containing a reference droplet, the average brightness of the reference droplets within the first grid is calculated and used as the brightness of the first grid; for a second grid where no reference droplet exists, the brightness of the second grid is determined based on the brightness of the grid containing the reference droplet that is closest to the second grid.

[0038] In some exemplary embodiments, a first calculation formula is used to calculate the mean brightness of the light field estimation matrix; the first calculation formula is as follows:

[0039]

[0040] In the formula, base represents the mean brightness of the light field estimation matrix, w and h represent the number of columns and rows of the light field estimation matrix, and L[i,j] represents the light field estimation matrix;

[0041] Based on the mean brightness and each element of the light field estimation matrix, a second calculation formula is used to calculate the correction coefficients, and the correction coefficients corresponding to each element of the light field estimation matrix form a correction field matrix; the second calculation formula is as follows:

[0042]

[0043] In the formula, alpha represents the correction coefficient, and value represents each element of the light field estimation matrix.

[0044] In some exemplary embodiments, the actual brightness of the droplets is calculated using a third formula based on the correction field matrix and the brightness value of the droplets within each grid; the third formula is as follows:

[0045] cal value =value*alpha

[0046] In the formula, cal value This represents the actual brightness of the droplet.

[0047] In some exemplary embodiments, the step of segmenting and clustering droplets in the droplet fluorescence image based on the actual scatter data includes:

[0048] Calculate the brightness ratio between each microdroplet and its neighboring reference microdroplet based on the actual scatter data;

[0049] Based on the brightness ratio of the reference droplet and the brightness ratio of the non-reference droplet, corresponding kernel density curves are generated respectively. The generated kernel density curves are superimposed to obtain the brightness ratio density curve.

[0050] Peak information is extracted from the brightness ratio density curve, and droplets in the droplet fluorescence image are divided and clustered according to the peak information to obtain negative droplet clusters and / or positive droplet clusters.

[0051] In some exemplary embodiments, calculating the brightness ratio between each droplet and a reference droplet in its neighborhood includes:

[0052] Traverse each droplet in the droplet array, taking the position of the current droplet in the droplet fluorescence image as the center, and a set multiple of the current droplet radius as the initial radius, gradually expand the radius to iteratively search for reference droplets in the neighborhood of the current droplet. Stop the search when the number of reference droplets found reaches a specified threshold or the set number of iterations is reached.

[0053] After the search stops, if a reference droplet is found, the ratio between the brightness of the current droplet and the median or mean brightness of all found reference droplets is calculated as the brightness ratio between the current droplet and the reference droplets in its neighborhood. If no reference droplet is found, the ratio between the brightness of the current droplet and the median or mean brightness of all found reference droplets is calculated as the brightness ratio between the current droplet and the reference droplets in its neighborhood.

[0054] In some exemplary embodiments, peak information is extracted from the brightness ratio density curve, and droplets in the droplet fluorescence image are divided and clustered based on the peak information to obtain negative droplet clusters and / or positive droplet clusters, including:

[0055] Extract peak information from the brightness ratio density curve;

[0056] The extracted peak information is filtered and merged;

[0057] According to the classification requirements of singleton PCR detection or multiplex PCR detection, the first m peaks in the brightness ratio density curve are selected, and the brightness ratio corresponding to the valley point is determined based on the peak information of the first m peaks. The boundary line between negative droplet clusters and positive droplet clusters is determined based on the brightness ratio.

[0058] The boundary lines are used to divide and cluster droplets in the droplet fluorescence image to obtain negative droplet clusters and / or positive droplet clusters.

[0059] Secondly, embodiments of this disclosure provide a droplet partitioning and clustering device, applied to a droplet-based digital nucleic acid amplification and quantitative analysis system, comprising:

[0060] The reading module is used to read the raw scatter data of each droplet in the droplet fluorescence image, including its position and brightness;

[0061] The acquisition module is used to acquire the brightness distribution density based on the original scatter data;

[0062] The screening module is used to screen reference microdroplets based on the brightness distribution density.

[0063] The correction module is used to perform optical field correction on the fluorescence image of the microdroplet based on the reference microdroplet to obtain the actual scatter data of the microdroplet.

[0064] The segmentation and clustering module is used to segment and cluster droplets in the droplet fluorescence image based on the actual scatter data.

[0065] Thirdly, embodiments of this disclosure provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0066] Fourthly, embodiments of this disclosure provide a droplet-based digital nucleic acid amplification and quantitative analysis system, comprising:

[0067] A droplet generation device for generating multiple droplets based on a sample;

[0068] A nucleic acid amplification temperature control device is used to perform nucleic acid amplification reactions on the plurality of microdroplets; and

[0069] A product signal acquisition device is used to acquire the product signal after the nucleic acid amplification reaction;

[0070] The droplet generation device includes:

[0071] An open container used to store multiple generated droplets and provide a site for nucleic acid amplification reactions;

[0072] A microchannel located above the open container, the microchannel being used to load the nucleic acid amplification reaction solution to be detected, has openings at both ends;

[0073] A vibrating device for driving the microchannel to reciprocate left and right below the liquid surface of the open container to continuously generate multiple microdroplets;

[0074] A fluorescence imaging detection device for capturing images of microdroplet fluorescence; and

[0075] A controller for implementing the steps of the method described in the first aspect.

[0076] The droplet segmentation and clustering method, apparatus, medium, and system disclosed herein obtain the brightness distribution density based on the original scatter data of each droplet in a droplet fluorescence image; a baseline droplet is selected based on the brightness distribution density; the light field of the droplet fluorescence image is corrected based on the baseline droplets to obtain the actual scatter data of the droplets; and the droplets in the droplet fluorescence image are segmented and clustered based on the actual scatter data. This method is based entirely on the real-time light field estimation and correction of the droplet image itself, without the need for additional special calibration materials or reference targets. It has stronger robustness and adaptability to changes in the light field, fully and effectively restores the brightness data of the droplets, improves the presentation effect of the droplet brightness data, and is easier to implement and facilitate human-computer interaction for threshold-based droplet positive / negative classification methods. Attached Figure Description

[0077] The present disclosure will be described in more detail below based on embodiments and with reference to the accompanying drawings:

[0078] Figure 1 A flowchart illustrating a droplet partitioning and clustering method provided in an embodiment of this disclosure;

[0079] Figure 2 This is a schematic diagram of grid division and numbering provided in an embodiment of the present disclosure;

[0080] Figure 3 A schematic diagram comparing the positive and negative distinguishing properties of droplets in a droplet scatter plot using global sorting and S-shaped sorting methods, provided for embodiments of this disclosure;

[0081] Figure 4 This is a schematic diagram of the microdroplet scatter plot thermal mapping provided in the embodiments of this disclosure;

[0082] Figure 5 This is a schematic diagram of a droplet scatter heatmap and a baseline droplet segmentation result provided in an embodiment of this disclosure;

[0083] Figure 6 A schematic diagram of a reference microdroplet provided in an embodiment of this disclosure;

[0084] Figure 7 This is a schematic diagram of mesh generation and light field estimation provided in an embodiment of the present disclosure;

[0085] Figure 8 A schematic diagram of the optical field and correction field provided in the embodiments of this disclosure;

[0086] Figure 9 A schematic diagram of droplet data before and after correction provided in an embodiment of this disclosure;

[0087] Figure 10 This is a schematic diagram of the luminance ratio density curve provided in an embodiment of the present disclosure;

[0088] Figures 11 to 13 These are schematic diagrams illustrating the data and correction effects for three different degrees of illumination influence provided in the embodiments of this disclosure.

[0089] In the accompanying drawings, the same parts are referred to by the same reference numerals, and the drawings are not drawn to scale. Detailed Implementation

[0090] To enable those skilled in the art to better understand the technical solutions of this disclosure, and to fully understand and implement the process of how this disclosure applies technical means to solve technical problems and achieve corresponding technical effects, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. The embodiments of this disclosure and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort should fall within the protection scope of this disclosure.

[0091] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0092] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0093] Image-based digital PCR systems suffer from uneven light field during imaging. Observations at the droplet fluorescence image level reveal variations in brightness across different regions. This unevenness leads to significant fluctuations in fluorescence data for the same type of negative or positive droplets in different image regions. Furthermore, it can cause overlap between negative and positive droplets in different image regions. Both phenomena hinder the accurate representation of droplet fluorescence data and make it difficult to differentiate between positive and negative droplets.

[0094] Several factors contribute to uneven light field in droplet fluorescence images of digital PCR systems. These primarily include uneven light field in the light source itself, the light source center not being in the center of the imaging field, the temperature control stability of the nucleic acid amplification temperature control device, and the light reflection of reaction reagents and consumables. Light field calibration for image-based digital PCR systems presents the following challenges: ① Due to the aforementioned factors causing uneven light field, the real-time light field exhibits significant uncertainty, making calibration using fixed calibrators or reference targets difficult to adapt to complex light field changes; ② Because different types of droplets exhibit inconsistent brightness under different lighting conditions, it is difficult to directly characterize the light field of different regions using image blocks with theoretically consistent brightness, as in other scenarios. The brightness of droplet fluorescence images in digital PCR systems is directly affected by sample concentration, i.e., the number of different anion and cation droplets.

[0095] Therefore, correcting light field inhomogeneity and restoring true fluorescence intensity is of great significance for data analysis in digital PCR systems.

[0096] Example 1

[0097] Figure 1 This is a flowchart illustrating a droplet partitioning and clustering method provided in an embodiment of this disclosure. Figure 1 As shown, a droplet partitioning and clustering method is applied to a droplet-based digital nucleic acid amplification and quantitative analysis system. The method includes:

[0098] Step 101: Read the raw scatter data of each droplet in the droplet fluorescence image, including its position and brightness.

[0099] In some examples, droplet fluorescence images are acquired, and visual artificial intelligence algorithms are used to locate the pixel center coordinates (cx, cy) and radius of each droplet in the droplet fluorescence image. Based on the pixel center coordinates and radius, a circular region is determined in the droplet fluorescence image. The pixel gray values ​​within this circular region are counted, and the fluorescence intensity of the current droplet, i.e., the brightness gray, is calculated based on the pixel gray values. This yields scatter data of each droplet in the droplet fluorescence image, which includes both position and brightness.

[0100] Step 102: Based on the original scatter data, obtain the brightness distribution density.

[0101] In some examples, the brightness distribution density is obtained based on the original scatter data, further including:

[0102] Step 102a: Convert the original scatter data into a droplet scatter heatmap to characterize the brightness distribution density.

[0103] Light field correction requires observing targets with approximately uniform brightness to obtain information about the non-uniformity of the light field in different regions. However, real-time estimation of the light field in droplet fluorescence images is affected by factors such as sample concentration (different proportions of positive and negative droplets) and target type (various droplet brightness levels). Directly statistically analyzing the brightness of different regions in the droplet fluorescence image is insufficient for appropriate light field estimation. Therefore, this method converts droplet scatter data into a droplet scatter heatmap, and then extracts regions with relatively high thermal intensity from the heatmap, i.e., the type of droplets with a higher proportion in the image, as the baseline droplets. The brightness of this type of droplet in different regions of the droplet fluorescence image is then used to characterize the light field.

[0104] In some examples, the raw scatter data is converted into a droplet scatter heatmap, which further includes:

[0105] Step 102a-1: Divide the droplet fluorescence image into multiple grids, sort and number all grids, and form a subarray with all droplets in each grid. Connect the subarrays to form a droplet array.

[0106] In some examples, the droplet fluorescence image is divided into multiple grids, all grids are sorted and numbered, all droplets in each grid form a subarray, and the subarrays are connected to form a droplet array, further including:

[0107] Step 102a-11: Divide the droplet fluorescence image into multiple grids.

[0108] Considering the global unevenness of the light field in droplet fluorescence images, but the relative uniformity in small local areas, this embodiment divides the droplet image into an n×n grid according to its spatial location. The value of n can be determined based on the actual situation. The larger the value of n, the denser the grid and the more uniform the local illumination. However, the value of n should not be too small, as a small value of n may result in fewer droplets within the grid area, which would be detrimental to reflecting the trend of uneven light field. The value of n can be determined based on experimental results, for example, n=15.

[0109] Step 102a-12: Sort and number all grids in an S-shaped order.

[0110] Based on the grid division, all grids are further sorted and numbered. In this embodiment, starting from the upper left corner of the droplet fluorescence image and ending at the lower right corner, all grids are numbered in an S-shaped order along the horizontal direction. The purpose of sorting and numbering all grids in an S-shaped order is to ensure that droplets that are both negative or both positive have as continuous a fluorescence intensity (brightness) data as possible, thereby avoiding discontinuities caused by uneven light fields on the left and right sides of the droplet fluorescence image, and facilitating subsequent extraction of the reference droplet mask and reference droplet. In some examples, the grids are numbered as: 0, 1, 2, ... The grid division and numbering are as follows: Figure 2 As shown.

[0111] Step 102a-13: Determine the grid in which each droplet is located based on its position.

[0112] Steps 102a-14: Create a subarray for each grid, and add each droplet to the subarray of the grid in which it is located. Figure 2 The numbers in the image represent sorting numbers. The left image shows the fluorescence image of the microdroplets, and the right image shows the grid division and numbering results. Subarrays are created for each grid according to these numbers. Each microdroplet is added to the subarray of its corresponding grid, and the microdroplets within each grid's subarray can be sorted in any way.

[0113] Steps 102a-15: Connect all subarrays according to the corresponding grid number order to form a droplet array.

[0114] Global sorting directly ranks droplets according to their center coordinates (x, y) or (cy, x) in the fluorescence image. However, droplets at the same x or y coordinate may be located on the left or right sides or top or bottom of the image, introducing uneven light field into the grid division and making the data somewhat chaotic, which is not conducive to the differentiation of positive and negative droplets. Compared with global sorting, the sorting method in this embodiment can, to some extent, suppress the influence of different grid regions on the positive and negative droplet properties. Figure 3 This paper presents a comparison of the positive and negative droplet differentiation in droplet scatter plots using global sorting and the S-shaped sorting method of this embodiment for the same droplet fluorescence image. The left image shows the positive and negative droplet differentiation results of global sorting in the droplet scatter plot, while the right image shows the positive and negative droplet differentiation results of the S-shaped sorting method in the droplet scatter plot.

[0115] Step 102a-2: Using the index of the droplets in the droplet array as the x-axis and the brightness of the droplets as the y-axis, map the droplet array onto a target image of a set size to form a droplet scatter heatmap.

[0116] In order to better extract the baseline droplets from the scatter data and avoid interference from some abnormal discrete points, this embodiment converts the scatter data into a droplet scatter heatmap based on the scatter distribution density.

[0117] In some examples, the droplet array is mapped onto a target image of a set size, using the droplet's index in the droplet array as the x-axis and the droplet's brightness as the y-axis, to form a droplet scatter heatmap, including:

[0118] Using the index of the droplet in the droplet array as the x-axis and the brightness of the droplet as the y-axis, the x-axis is mapped to the value in the width direction of the target image, and the y-axis is mapped to the value in the height direction of the target image. The gray value of the mapped position of the droplet on the target image is incremented by 1. The mapped target image is then normalized and converted into an image with a gray value range of 0 to 255 to obtain a droplet scatter heatmap.

[0119] In this embodiment, the index of the subarray of the grid containing the droplets is used as the x-axis, and the brightness of the droplets is used as the y-axis. The droplet scatter data of the droplet array is mapped to a target image of size (width, height) to form a droplet scatter heatmap. The mapping process involves mapping the x-axis (index) of the droplet scatter data of the droplet array to the width of the target image and the y-axis (brightness) to the height of the target image. The size of the target image can be specified as needed, for example, the width of the target image is 4 times its height.

[0120] In some examples, droplet scatter data (x droplet ,y droplet ) mapped to image coordinates (x image ,y image The formula for calculating ) is as follows:

[0121] x image =int((x droplet -x min ) / (x max -x min )·(width-1))

[0122] y image =int((y droplet -y min ) / (y max -y min )·(height-1))

[0123] In the formula, (x min ,y min (x) represents the minimum coordinate of the target image. max ,y max ) represents the maximum coordinate value of the target image.

[0124] After the above mapping transformation, each droplet scatter plot can be mapped to a specific location in the target image, mapping the gray value corresponding to that specific location of the droplet scatter plot data +1. After all droplet scatter plot data is mapped, in order to ensure good visibility of the droplet scatter plot heatmap under different sample concentrations and different droplet numbers, that is, to reflect the relative distribution density of different anion and cation types of droplets, rather than absolute density values, this embodiment uses the max-min normalization method to normalize the mapped image, and then converts it into an image with a gray scale range of 0-255. The schematic diagram of the droplet scatter plot heatmap mapping is shown below. Figure 4 As shown, the left image is a droplet scatter heatmap, and the right image is a pseudo-color image of the mapped droplet scatter heatmap.

[0125] Step 103: Select benchmark droplets based on brightness distribution density.

[0126] In some examples, benchmark droplets are selected based on their brightness distribution density, further including:

[0127] Step 103a: According to the preset reference microdroplet mask size conditions, segment out the target area in the microdroplet scatter thermal map that meets the preset conditions;

[0128] Step 103b: If there is only one target region, determine that target region as the reference microdroplet mask;

[0129] Step 103c: If there is more than one target region, the target region at the bottom in the vertical position relationship is determined as the reference microdroplet mask;

[0130] Step 103d: Extract the reference microdroplets that have fallen into the reference microdroplet mask.

[0131] Theoretically, without the influence of uneven light field, the brightness of a certain type of droplet (positive or negative) in an image should be consistent in the droplet fluorescence image. Based on this assumption, a certain type of high-density droplet can be extracted from the droplet scatter heatmap as a reference droplet, and the light field can be approximated by these reference droplets.

[0132] In this embodiment, regions with higher thermal intensity are segmented from the droplet scatter plot and used as reference droplet masks. Droplets falling into the reference droplet mask region are extracted from the droplet array and used as reference droplets. Higher thermal intensity can be determined by satisfying certain preset thermal conditions, which are not specifically limited here. When segmenting regions with higher thermal intensity from the droplet scatter plot according to preset mask length and width conditions and using them as reference droplet masks, if there are multiple target regions that meet preset size conditions, the lowest target region is selected according to the vertical position relationship of the multiple reference droplet masks. This prioritizes the use of clustered droplets with lower brightness, as these droplets have more stable brightness and are better able to reflect the global trend of light field non-uniformity. Using this screening method, when the densities of negative and positive droplets are not significantly different, the mask containing the negative droplets will be selected. The brightness of negative droplets is more stable and less susceptible to the influence of other factors such as target amplification or reagent differences. Therefore, negative droplets are more helpful in reflecting the global trend of uneven light field, while positive droplets may be affected by insufficient amplification or reagent instability, resulting in uneven brightness that is local and individual rather than global.

[0133] The droplets that fall into the reference droplet mask are extracted from the droplet array and used as reference droplets. Extraction of reference droplets can be performed using image segmentation methods, such as traditional image segmentation algorithms, or semantic segmentation or instance segmentation methods based on convolutional neural networks. In one example, the result is as follows: Figure 5 The diagram shows the droplet scatter plot and the baseline droplet segmentation results. The left image is the droplet scatter plot, and the right image is the baseline droplet segmentation results. A corresponding schematic diagram of the baseline droplet is shown below. Figure 6 As shown, the green dots represent reference droplets.

[0134] Step 104: Perform optical field correction on the fluorescence image of the microdroplet based on the reference microdroplet to obtain the actual scatter data of the microdroplet.

[0135] In some examples, optical field correction is performed on the droplet fluorescence image based on a reference droplet to obtain the actual scatter plot data of the droplet, further including:

[0136] Step 104a: Divide the droplet fluorescence image into multiple grids.

[0137] Considering the global unevenness of the light field in the image, but the relative uniformity in small local areas, this embodiment divides the droplet fluorescence image into an n×n grid according to its spatial location. The value of n depends on the actual situation; the larger the value of n, the denser the grid and the more uniform the local illumination. However, it should not be too small, as a small value may contain fewer droplets in the grid area, potentially introducing local noise and failing to reflect the trend of uneven light field. In one example, n = 25.

[0138] Step 104b: Calculate the brightness of each grid based on the brightness of the reference droplet, and form a light field estimation matrix using the brightness of all grids as elements.

[0139] In some examples, the brightness of each grid is calculated based on the brightness of a reference droplet, further including:

[0140] For the first grid cell containing a reference droplet, the average brightness of the reference droplet within the first grid cell is calculated and used as the brightness of the first grid cell; for the second grid cell containing no reference droplet, the brightness of the second grid cell is determined based on the brightness of the grid cell closest to the second grid cell containing the reference droplet.

[0141] Assuming the array of brightness values ​​of droplets falling into a certain first grid is G, and the brightness of that grid is gl, then the formula for calculating the brightness of the first grid is:

[0142]

[0143] In the formula, n represents the number of elements in array G. The brightness is calculated for all grids, forming an n×n light field estimation matrix. Each element of this matrix approximately represents the brightness value of a grid in the droplet fluorescence image. Since some regions of the droplet fluorescence image lack a reference droplet, the brightness of the corresponding grid cannot be calculated using the above formula. For these grids, the brightness value of the nearest grid containing the reference droplet is used for filling, or other similar interpolation algorithms are employed. To avoid abnormal brightness values, filtering algorithms can be used to filter out abrupt changes in brightness or to smooth the light field, depending on the specific circumstances. In one example, the grid division and light field estimation are as follows: Figure 7 As shown, the left figure is a schematic diagram of grid division, and the right figure is a schematic diagram of light field estimation.

[0144] Step 104c: Calculate the mean brightness of the light field estimation matrix. Based on the mean brightness and each element of the light field estimation matrix, calculate the correction field matrix with the same size as the light field estimation matrix.

[0145] To correct the light field to a relatively uniform level, this embodiment calculates the mean brightness of the light field estimation matrix. Based on the mean brightness and each element of the light field estimation matrix obtained in the previous step, a correction field matrix with the same size as the light field estimation matrix is ​​calculated, thereby achieving the correction of the brightness of droplets in the light field. First, the mean brightness base of the light field estimation matrix is ​​calculated. Then, for each element value value in the light field estimation matrix L, the correction coefficient alpha is calculated to form a correction field matrix A with the same size as the light field matrix.

[0146] In some examples, the first calculation formula is used to calculate the mean brightness of the light field estimation matrix.

[0147] The first calculation formula is as follows:

[0148]

[0149] In the formula, base represents the mean brightness of the light field estimation matrix, w and h represent the number of columns and rows of the light field estimation matrix, and L[i,j] represents the light field estimation matrix.

[0150] In some examples, a second calculation formula is used to calculate correction coefficients based on the mean brightness and each element of the light field estimation matrix, forming a correction field matrix with the correction coefficients corresponding to each element of the light field estimation matrix.

[0151] The second calculation formula is as follows:

[0152]

[0153] In the formula, alpha represents the correction coefficient, and value represents each element of the light field estimation matrix.

[0154] In one example, the light field and the correction field are as follows: Figure 8 As shown, the left image is the light field, and the right image is the correction field.

[0155] Step 104d: Based on the calibration field matrix and the brightness value of the droplets in each grid, calculate the actual brightness of the droplets to obtain the actual scatter data.

[0156] Based on the aforementioned calibration field and reference droplet grid division strategy, all microdroplets falling into each grid region are calibrated using the calibration coefficients at the corresponding positions in the calibration field.

[0157] In some examples, the actual brightness of the droplets is calculated using a third formula based on the correction field matrix and the brightness value of the droplets within each grid. The third formula is as follows:

[0158] cal value =value*alpha

[0159] In the formula, cal value This represents the actual brightness of the droplet.

[0160] The correction was performed on all droplets according to the third calculation formula to obtain the corrected actual brightness.

[0161] In one example Figure 9 The diagram illustrates the droplet data before and after correction based on the third calculation formula mentioned above. The left image shows the original brightness value of the droplet before correction, and the right image shows the actual brightness after correction. It can be seen that the unevenness of the light field is significantly improved after correction.

[0162] Step 105: Divide and cluster the droplets in the droplet fluorescence image based on the actual scatter data.

[0163] In some examples, droplets in droplet fluorescence images are segmented and clustered based on actual scatter data, further including:

[0164] Step 105a: Calculate the brightness ratio between each droplet and its neighboring reference droplet based on the actual scatter data.

[0165] This embodiment determines the relative intensity of the current droplet and the reference droplet in its neighborhood by calculating the brightness ratio between each droplet and the reference droplet in its neighborhood. This avoids the influence of the numerical difference in absolute brightness between negative and positive droplets under the influence of uneven light field on the classification of negative and positive droplets.

[0166] In some examples, the brightness ratio between each droplet and a reference droplet in its neighborhood is calculated, including:

[0167] Traverse each droplet in the droplet array, taking the current droplet's position in the droplet fluorescence image as the center and a set multiple of the current droplet's radius as the initial radius, and gradually expand the radius to iteratively search for reference droplets in the neighborhood of the current droplet. Stop the search when the number of reference droplets found reaches a specified threshold or the set number of iterations is reached.

[0168] After the search stops, if a reference droplet is found, the ratio between the brightness of the current droplet and the median or mean brightness of all found reference droplets is calculated as the brightness ratio between the current droplet and the reference droplets in its neighborhood. If no reference droplet is found, the ratio between the brightness of the current droplet and the median or mean brightness of all found reference droplets is calculated as the brightness ratio between the current droplet and the reference droplets in its neighborhood.

[0169] Step 105b: Generate corresponding kernel density curves based on the brightness ratio of the reference droplet and the brightness ratio of the non-reference droplet respectively, and superimpose the generated kernel density curves to obtain the brightness ratio density curve.

[0170] In this embodiment, the brightness ratio of the reference droplet and the brightness ratio of the non-reference droplet are used as input data to perform kernel density estimation, generating corresponding kernel density curves. These kernel density curves are then superimposed to form the final brightness ratio density curve. Since the reference droplet occupies a large proportion, generating separate kernel density curves for the reference and non-reference droplets before superimposing them during the brightness ratio density curve generation better reflects the lower proportion of the non-reference droplets, thus better handling samples with extremely high or low concentrations.

[0171] In some examples, the kernel density estimation function used for kernel density estimation is as follows:

[0172]

[0173] In the formula, N represents the number of data points, i.e., the total number of luminance ratio data points, w represents the bandwidth of the kernel density estimation, and x i Let represent the i-th data point, x represent the x-coordinate of the density curve to be estimated, and G represent the kernel function used for kernel density estimation. For example, a one-dimensional Gaussian kernel function can be used as shown in the following formula:

[0174]

[0175] In the formula, x represents the input data, μ represents the mean of the Gaussian kernel function, and σ represents the standard deviation of the Gaussian kernel function. After kernel density estimation, the curve needs to be filtered using a Gaussian filtering method to remove minor local jitter; and normalization processing is also required.

[0176] The formula for normalization is as follows:

[0177]

[0178] In the formula, P(i) represents the result of the normalization process, that is, the area under the kernel density curve is normalized. i This represents the value at the i-th position on the kernel density curve.

[0179] In one example, the luminance-to-density ratio curve is as follows: Figure 10 As shown, the dashed lines represent the kernel density curves of the reference droplet and the non-reference droplet, respectively, and the black solid lines represent the final brightness ratio density curve obtained by superimposing the two. The valleys and peaks of the brightness ratio density curve are marked on the curve.

[0180] Step 105c: Extract peak information from the brightness ratio density curve, and classify and cluster droplets in the droplet fluorescence image according to the peak information to obtain negative droplet clusters and / or positive droplet clusters.

[0181] In some examples, peak information is extracted from the brightness ratio density curve, and droplets in the droplet fluorescence image are divided and clustered based on the peak information to obtain negative droplet clusters and / or positive droplet clusters, further including:

[0182] Step 105c-1: Extract peak information from the brightness ratio density curve.

[0183] Step 105c-2: Filter and merge the extracted peak information.

[0184] Step 105c-3: According to the requirements for singleton PCR detection or multiplex PCR detection, select the first m peaks in the brightness ratio density curve, determine the brightness ratio corresponding to the valley point based on the peak information of the first m peaks, and determine the boundary line between negative droplet clusters and positive droplet clusters based on the brightness ratio.

[0185] Step 105c-4: Use the boundary lines to divide and cluster the droplets in the droplet fluorescence image to obtain negative droplet clusters and / or positive droplet clusters.

[0186] Compared to methods in related technologies, the method of this embodiment can bring the following beneficial effects:

[0187] 1. Unlike offline calibration or reference target correction methods, the method in this embodiment is based entirely on real-time light field estimation and correction of the droplet image itself, without the need for additional special calibration materials or reference targets, and has stronger robustness and adaptability to changes in the light field.

[0188] 2. The method in this embodiment corrects the data from the principle level, effectively restores the brightness data of the droplets, improves the presentation effect of the droplet data brightness, and makes it easier to implement the threshold-based droplet yin-yang classification method and human-computer interaction.

[0189] 3. The method in this embodiment corrects the brightness data of microdroplets in a way that differs from data correction methods based on the positive and negative classification of microdroplets. This method can fully restore the data and ensure its authenticity and validity.

[0190] In one example Figure 11 , Figure 12 , Figure 13 The data and correction effects for three different levels of illumination influence are presented respectively. It can be seen that the method in this embodiment has good adaptability in different scenarios. Figure 11 , Figure 12 , Figure 13 The left image in the figure shows the original droplet scatter data, the middle image shows the corrected actual scatter data, and the right image shows the light field estimation results.

[0191] Example 2

[0192] This disclosure provides a droplet partitioning and clustering device for use in a droplet-based digital nucleic acid amplification and quantitative analysis system, comprising:

[0193] The reading module 201 is used to read the raw scatter data of each droplet in the droplet fluorescence image, including its position and brightness.

[0194] The acquisition module 202 is used to obtain the brightness distribution density based on the original scatter data.

[0195] The screening module 203 is used to screen benchmark droplets based on the brightness distribution density.

[0196] The correction module 204 is used to perform optical field correction on the microdroplet fluorescence image based on the reference microdroplet to obtain the actual scatter data of the microdroplet.

[0197] The segmentation and clustering module 205 is used to segment and cluster droplets in the droplet fluorescence image based on the actual scatter data.

[0198] For a detailed implementation of the above modules, please refer to Example 1, which will not be repeated here.

[0199] Example 3

[0200] Based on the above embodiments, this embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps of the method in Embodiment 1 above.

[0201] In some embodiments of this example, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0202] In some embodiments of this example, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0203] The processor may include, but is not limited to, one or more processors or microprocessors. Each processor may be implemented as an Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor, or other electronic component, for executing the methods in the above embodiments.

[0204] Computer-readable storage media can be implemented by any type of volatile or non-volatile storage device or a combination thereof. Computer-readable storage media may include, but are not limited to, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, and computer storage media (e.g., hard disks, floppy disks, solid-state drives, removable disks, CD-ROMs, DVD-ROMs, Blu-ray discs, etc.).

[0205] Computer-readable storage media may also store at least one computer-executable program / instruction, such as computer-readable instructions. Computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Computer-readable storage media may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, a non-transitory computer-readable storage medium may be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0206] In addition, the computer device may include (but is not limited to) a data bus, an input / output (I / O) bus, a display, and input / output devices (e.g., keyboard, mouse, speakers, etc.).

[0207] The processor can communicate with external devices via the I / O bus through wired or wireless networks.

[0208] In one embodiment, the at least one computer-executable instruction may also be compiled into or comprise a software product / computer program product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods in the embodiments described herein.

[0209] Example 4

[0210] This disclosure provides a droplet-based digital nucleic acid amplification and quantitative analysis system, comprising:

[0211] A droplet generation device for generating multiple droplets based on a sample;

[0212] A nucleic acid amplification temperature control device is used to perform nucleic acid amplification reactions on the plurality of microdroplets; and

[0213] A product signal acquisition device is used to acquire the product signal after the nucleic acid amplification reaction;

[0214] The droplet generation device includes:

[0215] An open container used to store multiple generated droplets and provide a site for nucleic acid amplification reactions;

[0216] A microchannel located above the open container, the microchannel being used to load the nucleic acid amplification reaction solution to be detected, has openings at both ends;

[0217] A vibrating device for driving the microchannel to reciprocate left and right below the liquid surface of the open container to continuously generate multiple microdroplets;

[0218] A fluorescence imaging detection device for capturing images of microdroplet fluorescence; and

[0219] A controller for implementing the steps of the method in Embodiment 1.

[0220] The digital nucleic acid amplification and quantitative analysis system of this embodiment reads the original scatter data of each droplet in the droplet fluorescence image through a controller. Based on the original scatter data, it obtains the brightness distribution density and then selects benchmark droplets. The system then performs light field correction on the droplet fluorescence image based on the benchmark droplets to obtain the actual scatter data of the droplets. Finally, it classifies and clusters the droplets in the fluorescence image based on the actual scatter data. Through light field correction before classification and clustering, the unevenness of the light field is significantly improved, allowing the droplet brightness to be better presented and restoring the true brightness, facilitating the differentiation of positive and negative droplets. Furthermore, this system does not require additional special calibration materials or reference targets and has stronger robustness and adaptability to changes in the light field.

[0221] In one example, the method for capturing microdroplet fluorescence images is as follows: after nucleic acid amplification, an excitation light source emits excitation light directly above the well plate corresponding to the generated microdroplets, and then the microdroplets in the corresponding well plate are photographed to obtain microdroplet fluorescence images. The light source remains on throughout the photographing process. Monochromatic multiplex detection obtains a single microdroplet fluorescence image through one fluorescence channel, and detects multiple targets (target molecules) using different fluorescence intensities.

[0222] In the embodiments provided in this disclosure, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0223] It should be noted that, in this disclosure, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element limited by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0224] While the embodiments disclosed herein are as described above, the foregoing content is merely for the purpose of facilitating understanding of this disclosure and is not intended to limit this disclosure. Any person skilled in the art to which this disclosure pertains may make any modifications and changes in form and detail of the implementation without departing from the spirit and scope of this disclosure; however, the scope of patent protection of this disclosure shall still be determined by the scope defined in the appended claims.

Claims

1. A droplet partitioning and clustering method, applied to a droplet-based digital nucleic acid amplification and quantitative analysis system, characterized in that, include: Read the raw scatter data of each droplet in the droplet fluorescence image, including its position and brightness; Based on the original scatter data, the brightness distribution density is obtained; Based on the brightness distribution density, a benchmark microdroplet is selected; The fluorescence image of the microdroplet is corrected for light field based on the reference microdroplet to obtain the actual scatter data of the microdroplet. The droplets in the droplet fluorescence image are divided and clustered based on the actual scatter data.

2. The method according to claim 1, characterized in that, Based on the original scatter data, the brightness distribution density is obtained, including: The original scatter data is converted into a droplet scatter heatmap to characterize the brightness distribution density.

3. The method according to claim 2, characterized in that, Converting the original scatter data into a droplet scatter heatmap includes: The microdroplet fluorescence image is divided into multiple grids, and all grids are sorted and numbered. All microdroplets in each grid form a subarray, and the subarrays are connected to form a microdroplet array. Using the index of the droplets in the droplet array as the horizontal axis and the brightness of the droplets as the vertical axis, the droplet array is mapped onto a target image of a set size to form a droplet scatter heatmap.

4. The method according to claim 3, characterized in that, The microdroplet fluorescence image is divided into multiple grids, and all grids are sorted and numbered. All microdroplets in each grid form a subarray, and these subarrays are connected to form a microdroplet array, including: The microdroplet fluorescence image is divided into multiple grids; Sort and number all grids in an S-shaped order; The grid in which each droplet is located is determined based on its position. Create a subarray for each grid, and add each droplet to the subarray of the grid it belongs to; Connect all subarrays according to the corresponding grid number order to form a droplet array.

5. The method according to claim 3, characterized in that, Using the index of the droplets in the droplet array as the x-axis and the brightness of the droplets as the y-axis, the droplet array is mapped onto a target image of a set size to form a droplet scatter heatmap, including: Using the index of the droplet in the droplet array as the x-axis and the brightness of the droplet as the y-axis, the x-axis is mapped to the value in the width direction of the target image, and the y-axis is mapped to the value in the height direction of the target image. The gray value of the mapped position of the droplet on the target image is incremented by 1. The mapped target image is then normalized and converted into an image with a gray value range of 0 to 255 to obtain a droplet scatter heatmap.

6. The method according to claim 2, characterized in that, Based on the brightness distribution density, a baseline microdroplet is selected, including: According to the preset reference microdroplet mask size conditions, the target area in the microdroplet scatter thermal map that meets the preset conditions is segmented; If there is only one target region, that target region is determined as the reference microdroplet mask; When there is more than one target region, the lowermost target region in the vertical position relationship is determined as the reference microdroplet mask; Extract the reference microdroplets that fall into the reference microdroplet mask.

7. The method according to claim 1, characterized in that, Based on the reference microdroplet, the fluorescence image of the microdroplet is subjected to optical field correction to obtain the actual scatter plot data of the microdroplet, including: The microdroplet fluorescence image is divided into multiple grids; The brightness of each grid is calculated based on the brightness of the reference microdroplet, and the brightness of all grids is used as elements to form a light field estimation matrix; Calculate the mean brightness of the light field estimation matrix, and based on the mean brightness and each element of the light field estimation matrix, calculate a correction field matrix with the same size as the light field estimation matrix; Based on the correction field matrix and the brightness value of the droplets in each grid, the actual brightness of the droplets is calculated to obtain the actual scatter data.

8. The method according to claim 7, characterized in that, The brightness of each grid is calculated based on the brightness of the reference droplet, including: For a first grid containing a reference droplet, the average brightness of the reference droplets within the first grid is calculated and used as the brightness of the first grid; for a second grid where no reference droplet exists, the brightness of the second grid is determined based on the brightness of the grid containing the reference droplet that is closest to the second grid.

9. The method according to claim 7, characterized in that, The mean luminance of the light field estimation matrix is ​​calculated using the first calculation formula; the first calculation formula is as follows: In the formula, base represents the mean brightness of the light field estimation matrix, w and h represent the number of columns and rows of the light field estimation matrix, and L[i,j] represents the light field estimation matrix; Based on the mean brightness and each element of the light field estimation matrix, a second calculation formula is used to calculate the correction coefficients, and the correction coefficients corresponding to each element of the light field estimation matrix form a correction field matrix; the second calculation formula is as follows: In the formula, alpha represents the correction coefficient, and value represents each element of the light field estimation matrix.

10. The method according to claim 9, characterized in that, Based on the correction field matrix and the brightness value of the droplets within each grid, the actual brightness of the droplets is calculated using a third formula; the third formula is as follows: cal value =value*alpha In the formula, cal value This represents the actual brightness of the droplet.

11. The method according to claim 1, characterized in that, The step of classifying and clustering droplets in the droplet fluorescence image based on the actual scatter data includes: Calculate the brightness ratio between each microdroplet and its neighboring reference microdroplet based on the actual scatter data; Based on the brightness ratio of the reference droplet and the brightness ratio of the non-reference droplet, corresponding kernel density curves are generated respectively. The generated kernel density curves are superimposed to obtain the brightness ratio density curve. Peak information is extracted from the brightness ratio density curve, and droplets in the droplet fluorescence image are divided and clustered according to the peak information to obtain negative droplet clusters and / or positive droplet clusters.

12. The method according to claim 11, characterized in that, The calculation of the brightness ratio between each droplet and a reference droplet in its neighborhood includes: Traverse each droplet in the droplet array, taking the position of the current droplet in the droplet fluorescence image as the center, and a set multiple of the current droplet radius as the initial radius, gradually expand the radius to iteratively search for reference droplets in the neighborhood of the current droplet. Stop the search when the number of reference droplets found reaches a specified threshold or the set number of iterations is reached. After the search stops, if a reference droplet is found, the ratio between the brightness of the current droplet and the median or mean brightness of all found reference droplets is calculated as the brightness ratio between the current droplet and the reference droplets in its neighborhood. If no reference droplet is found, the ratio between the brightness of the current droplet and the median or mean brightness of all found reference droplets is calculated as the brightness ratio between the current droplet and the reference droplets in its neighborhood.

13. The method according to claim 11, characterized in that, Peak information is extracted from the brightness ratio density curve. Based on the peak information, droplets in the droplet fluorescence image are divided and clustered to obtain negative droplet clusters and / or positive droplet clusters, including: Extract peak information from the brightness ratio density curve; The extracted peak information is filtered and merged; According to the classification requirements of singleton PCR detection or multiplex PCR detection, the first m peaks in the brightness ratio density curve are selected, and the brightness ratio corresponding to the valley point is determined based on the peak information of the first m peaks. The boundary line between negative droplet clusters and positive droplet clusters is determined based on the brightness ratio. The boundary lines are used to divide and cluster droplets in the droplet fluorescence image to obtain negative droplet clusters and / or positive droplet clusters.

14. A droplet partitioning and clustering device, applied to a droplet-based digital nucleic acid amplification and quantitative analysis system, characterized in that, include: The reading module is used to read the raw scatter data of each droplet in the droplet fluorescence image, including its position and brightness; The acquisition module is used to acquire the brightness distribution density based on the original scatter data; The screening module is used to screen reference microdroplets based on the brightness distribution density. The calibration module is used to perform optical field correction on the fluorescence image of the microdroplet based on the reference microdroplet to obtain the actual scatter data of the microdroplet; The segmentation and clustering module is used to segment and cluster droplets in the droplet fluorescence image based on the actual scatter data.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program performs the steps of the method according to any one of claims 1 to 13.

16. A droplet-based digital nucleic acid amplification and quantitative analysis system, characterized in that, include: A droplet generation device for generating multiple droplets based on a sample; A nucleic acid amplification temperature control device is used to perform nucleic acid amplification reactions on the multiple microdroplets; and A product signal acquisition device is used to acquire the product signal after the nucleic acid amplification reaction; The droplet generation device includes: An open container used to store multiple generated droplets and provide a site for nucleic acid amplification reactions; A microchannel located above the open container, the microchannel being used to load the nucleic acid amplification reaction solution to be detected, has openings at both ends; A vibrating device for driving the microchannel to reciprocate left and right below the liquid surface of the open container to continuously generate multiple microdroplets; A fluorescence imaging detection device for capturing images of microdroplet fluorescence; and A controller for implementing the steps of the method according to any one of claims 1 to 13.

Images