Method and device for counting fluorescence proportion information of microfluidic droplet array

Abstract

The invention discloses a method and a device for counting fluorescence proportion information of a microfluidic droplet array, and relates to the technical field of fluorescence intensity recognition, classification and analysis. The method comprises the following steps: acquiring a fluorescence field image and a bright field image of a chip; analyzing the bright field image to obtain the total number of micro-droplets in a preset size range; analyzing the fluorescent field image to obtain the total number of fluorescent microdroplets in a preset size range; and based on the total number of the microdroplets in the preset size range and the total number of the fluorescent microdroplets in the preset size range, calculating the proportion of the fluorescent microdroplets. According to the method and the device, the problems of confusion of fluorescent microdroplets and non-fluorescent microdroplets, abnormal microdroplet mixing statistics in fluorescent microdroplet analysis and the like in the prior art can be solved.

Description

Technical Field

[0001] The present application relates to the technical field of fluorescence intensity recognition classification analysis, and in particular to a method and device for counting fluorescence ratio information of a microfluidic droplet array. Background Art

[0002] As an important technology for high-precision nucleic acid detection in recent years, droplet digital PCR (Polymerase Chain Reaction, ddPCR for short) uses microfluidics technology to achieve sample dispersion and highly independent reactions. Compared with traditional fluorescent quantitative PCR (qPCR) technology, droplet digital PCR can distribute nucleic acid molecules into tens of thousands of independent droplets, each of which is equivalent to a micro-reaction unit. This distribution method dilutes the original sample to the single-molecule level, thereby ensuring the independence of the amplification reaction.

[0003] The core of droplet digital PCR technology lies in the generation of droplets and the separation of reactions. Through the microfluidic chip, the sample and the oil phase are mixed in the microchannel to form droplets of uniform size and stability. These droplets are enclosed in the oil phase to avoid cross-contamination between samples. Each droplet acts as an independent "microreactor" and can perform amplification reactions at the same time. After the amplification is completed, the positive droplets will indicate the presence of the target nucleic acid through fluorescent signals, while the negative droplets maintain the background fluorescence level.

[0004] After amplification, the system captures the fluorescence signal of the droplets through a fluorescence detection device. The fluorescence intensity of positive droplets and negative droplets is significantly different. Combined with the statistical results of fluorescence intensity, the proportion of positive droplets can be estimated. The absolute concentration of the target nucleic acid molecule in the sample can be further calculated through the Poisson distribution model. Unlike traditional qPCR technology that relies on standard curves for relative quantification, droplet digital PCR performs well in gene detection, genetic mutation detection and trace sample analysis with its absolute quantification advantage.

[0005] The following solutions exist in the prior art: (1) A processing method based on fluorescence intensity data, which is used to classify positive and negative droplets in droplet digital PCR. By collecting fluorescent droplet data and preprocessing it, the maximum likelihood estimation is used to fit the fluorescence intensity distribution to a normal distribution, calculate the classification threshold and identify positive droplets. This method does not require a negative control sample and automatically determines the classification threshold, which improves the accuracy of fluorescent droplet classification and the calculation accuracy of nucleic acid concentration.

[0006] However, this method does not take into account the difference in droplet size, and it is easy to mix abnormal droplets that are too large or merged into the statistical data, resulting in deviations in the calculation of nucleic acid concentration and positive ratio statistics, and unable to ensure the accuracy and reliability of the data. This method has the following shortcomings: Insufficient robustness to high-noise environments: This method relies on a fixed classification threshold formula, which is difficult to effectively deal with problems such as optical noise and background interference in complex experimental environments, and may result in misclassification of false positive or false negative droplets.

[0007] Insufficient utilization of droplet morphology information: Classification relies solely on fluorescence intensity data, without fully considering the potential impact of droplet morphological characteristics (such as size and shape) on classification statistics.

[0008] (2) Anti-interference classification method for droplet digital PCR instrument. This method uses image processing technology to convert fluorescence intensity data into binary images, combines support vector machines and optimized classification models, and improves the anti-interference classification ability of fluorescence data by introducing artificial false negative data and adjusting the classification threshold. This method can adapt to the analysis of droplet fluorescence intensity data in complex experimental environments and effectively improves classification accuracy and efficiency.

[0009] However, this method only relies on fluorescence intensity data for classification, and does not fully consider the potential impact of the morphological characteristics of droplets (such as size and shape) on the classification statistics.

[0010] (3) A machine learning-based droplet digital nucleic acid detection device that combines microfluidic chip technology to achieve high-throughput droplet generation, nucleic acid amplification, and fluorescence detection. The fluorescence images are automatically classified and processed through machine learning algorithms to achieve absolute quantitative detection of viral nucleic acids. The device is portable and highly biosafe, suitable for point-of-care testing (POCT) scenarios, and solves the problems of traditional nucleic acid testing, such as long time consumption and complex operation.

[0011] Although the device can perform a certain degree of automated processing on droplets, the grayscale distribution of the background, non-luminescent droplets, and fluorescent droplets in the fluorescent image often overlaps greatly, and it is difficult to achieve accurate three-classification using traditional threshold segmentation. At the same time, interference signals such as chip edges, guide brackets, fluorescent contamination points, and other small impurities are difficult to effectively eliminate using existing technologies, which can easily cause classification confusion and misjudgment, affecting the reliability of absolute quantitative detection.

[0012] Therefore, how to solve the problems of confusion between fluorescent droplets and non-fluorescent droplets in the prior art and statistics of abnormal droplets mixed in fluorescent droplet analysis has become a technical problem that urgently needs to be solved in this field. Summary of the invention

[0013] The purpose of the present application is to provide a method and device for counting fluorescence ratio information of a microfluidic droplet array, which can solve the problems of confusion between fluorescent droplets and non-fluorescent droplets, and statistics of abnormal droplets mixed in fluorescent droplet analysis in the prior art.

[0014] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a method for counting fluorescence ratio information of a microfluidic droplet array, the method for counting fluorescence ratio information of a microfluidic droplet array comprising: Acquire fluorescence field and bright field images of the chip.

[0015] The bright field image is analyzed to obtain the total number of droplets within a preset size range.

[0016] The fluorescence field image is analyzed to obtain the total number of fluorescent droplets within a preset size range.

[0017] Based on the total number of droplets within the preset size range and the total number of fluorescent droplets within the preset size range, the ratio of fluorescent droplets is calculated.

[0018] In a second aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described methods for statistics of fluorescence ratio information of a microfluidic droplet array.

[0019] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for counting fluorescence ratio information of a microfluidic droplet array.

[0020] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for counting fluorescence ratio information of a microfluidic droplet array.

[0021] According to the specific embodiments provided in this application, this application discloses the following technical effects: The present application provides a method and device for counting fluorescence ratio information of a microfluidic droplet array, the method comprising: acquiring a fluorescence field image and a bright field image of a chip; analyzing the bright field image to obtain the total number of droplets within a preset size range; analyzing the fluorescence field image to obtain the total number of fluorescent droplets within a preset size range; and calculating the ratio of fluorescent droplets based on the total number of droplets within the preset size range and the total number of fluorescent droplets within the preset size range. The present application uses an independent image processing module, which can simultaneously count the total number of droplets within the preset size range in a bright field image, and identify and count positive droplets that fluoresce within the preset size range in a fluorescence field image. This separation and analysis method can ensure that different types of droplets are processed and counted in a targeted manner, avoiding classification errors caused by overlapping droplet intensity signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 This is a diagram of the application environment of a method for counting fluorescence ratio information of a microfluidic droplet array in one embodiment of the present application.

[0024] Figure 2 A schematic flow chart of a method for counting fluorescence ratio information of a microfluidic droplet array provided in one embodiment of the present application.

[0025] Figure 3 A bright field image processing flow chart provided for an embodiment of the present application.

[0026] Figure 4 A schematic diagram of a bright field image acquired according to an embodiment of the present application.

[0027] Figure 5 This is an effect diagram of the bright field image after processing provided in one embodiment of the present application.

[0028] Figure 6 A flowchart of fluorescence field image processing provided in one embodiment of the present application.

[0029] Figure 7 An initial labeling map generated by seed points provided in an embodiment of the present application.

[0030] Figure 8 A schematic diagram of the segmentation results of the watershed segmentation algorithm provided in an embodiment of the present application.

[0031] Fig. 9 A schematic diagram of an acquired fluorescence field image provided in an embodiment of the present application.

[0032] Fig.10 This is a diagram showing the effect of fluorescence field image processing provided in one embodiment of the present application.

[0033] Fig.11 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0035] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0036] The problem to be solved by this application is how to solve the following problems existing in the prior art by improving the image recognition algorithm in high-throughput digital PCR droplet detection: 1. How to solve the problem of abnormal droplets mixing into statistical data.

[0037] Currently, in droplet detection, due to uneven size distribution, abnormal droplets that are too large or merged are often mixed into the statistical data, resulting in distorted results. How to accurately count only droplets that meet the preset size range (i.e., the mainstream size range) is an urgent problem to be solved.

[0038] 2. How to reduce the impact of interference signals on classification accuracy.

[0039] Non-target signals such as chip edges, guide brackets and fluorescent contamination points often interfere with the correct identification and classification of droplets. How to automatically identify and eliminate these interfering signals to ensure accurate distinction between positive and negative droplets is a major difficulty in current technology.

[0040] 3. How to avoid confusion caused by insufficient image scene separation.

[0041] Existing methods mostly rely on a single image (such as a fluorescence image only) for detection, which makes it difficult to accurately separate dark droplets from fluorescent droplets at the same time. How to achieve effective separation and independent counting to avoid confusion is also a technical problem that needs to be solved urgently.

[0042] Based on the above problems, the present invention provides a method for counting fluorescence ratio information of a microfluidic droplet array. The method can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the fluorescence field image and the bright field image of the chip to the server 104. After the server 104 receives the fluorescence field image and the bright field image of the chip, the server 104 analyzes the bright field image to obtain the total number of droplets within the preset size range; analyzes the fluorescence field image to obtain the total number of fluorescent droplets within the preset size range; based on the total number of droplets within the preset size range and the total number of fluorescent droplets within the preset size range, the proportion of fluorescent droplets is calculated. The server 104 can feedback the obtained proportion of fluorescent droplets to the terminal 102. In addition, in some embodiments, the method for fluorescence ratio information statistics of microfluidic droplet arrays can also be implemented independently by the server 104 or the terminal 102. For example, the terminal 102 can directly perform fluorescence ratio information statistics on the fluorescence field image and bright field image of the chip, or the server 104 can obtain the fluorescence field image and bright field image of the chip from the data storage system and perform fluorescence ratio information statistics on the fluorescence field image and bright field image of the chip.

[0043] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, and tablet computers. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0044] In an exemplary embodiment, Figure 2 As shown, a method for counting fluorescence ratio information of a microfluidic droplet array is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate, and the following steps are included: S1: Acquire the fluorescence field image and bright field image of the chip.

[0045] S2: Analyze the bright field image to obtain the total number of droplets within a preset size range.

[0046] S3: Analyze the fluorescence field image to obtain the total number of fluorescent droplets within a preset size range.

[0047] S4: Calculate the ratio of the fluorescent droplets based on the total number of droplets within the preset size range and the total number of fluorescent droplets within the preset size range.

[0048] By implementing the above steps S1 to S4, and combining the independent analysis of the bright field image and the fluorescent field image, the present application realizes the efficient identification and statistics of the total number of droplets and the number of fluorescent droplets within the preset size range, and outputs the statistical information of the droplet size (such as the mean, standard deviation, etc.). The present application adopts an independent image processing module, which can simultaneously count the total number of droplets in the bright field image and identify and count the fluorescent positive droplets in the fluorescent field image. Through this separation and analysis method, it is ensured that different types of droplets are processed and counted in a targeted manner, and the classification error caused by the overlap of the droplet intensity signal is avoided. In addition, the present application designs a size screening mechanism for the first time in droplet identification, realizes the dynamic screening of droplets, and thus can analyze droplets more accurately. By defining the preset size range, i.e., the "mainstream size range", through the accurate statistics of the droplet size (such as the median, mean, standard deviation, etc.), droplets with abnormal size (such as too large or too small) are automatically excluded. This mechanism, combined with dynamic size statistics, can automatically adjust the screening criteria according to the droplet characteristics, rather than relying on a fixed size threshold, thereby improving the flexibility and accuracy of screening.

[0049] As an optional implementation, in step S1, it specifically includes: Turn on the switch of the inverted fluorescence microscope, clamp the incubated chip on the loading platform of the inverted fluorescence microscope, start the fluorescence excitation device and the fluorescence emission switch, adjust to the blue laser, rotate the objective lens to adjust to the 10x objective lens, use the coarse focus and fine focus screws to adjust the focal length, switch the lens between the visual and software, and open the image acquisition software ZEN on the computer. Set the exposure time, which is different for each photo. After taking the picture under the fluorescence excitation state, turn off the emission switch, turn on the white light switch, and take a bright field photo at the same position.

[0050] As an optional implementation, Figure 3 As shown, in step S2, the bright field image is analyzed to obtain the total number of droplets within a preset size range, specifically including: S21: converting the bright field image into a grayscale image and performing preprocessing to obtain a preprocessed bright field image.

[0051] S22: performing automatic threshold segmentation on the preprocessed bright field image to obtain a first binary image.

[0052] S23: Filling holes in the first binary image to obtain a binary image of the droplet with a complete outline.

[0053] S24: performing contour detection and ellipse fitting on the binary image of the microdroplet to obtain geometric features of the microdroplet; the geometric features include: center position, lengths of the major axis and the minor axis, and rotation angle.

[0054] S25: performing ellipse screening based on the geometric features of the droplet to obtain a screened fitting ellipse.

[0055] S26: The total number of the selected fitting ellipses is taken as the total number of droplets within the preset size range, and the corresponding geometric features are recorded.

[0056] As an optional implementation, in step S23, specifically including: S231: using a morphological opening operation to remove isolated small noise points in the first binary image, to obtain a first binary image with small noise points removed.

[0057] S232: using a morphological closing operation to fill holes smaller than a preset shape in the first binary image from which small noise points have been removed, to obtain a filled first binary image.

[0058] S233: Using connected region analysis, the padded first binary image is identified to obtain internal holes that are completely surrounded by foreground pixels.

[0059] S234: Use scipy.ndimage.binary_fill_holes to fill the internal holes that are completely surrounded by foreground pixels to obtain a filled area.

[0060] S235: Merging the filled regions to obtain a binary image of the droplet with a complete outline.

[0061] As an optional implementation, in step S24, specifically including: S241: extracting all external contours in the droplet binary image using a contour detection algorithm to obtain a plurality of external contours.

[0062] S242: Determine the outer contour whose contour points exceed the set minimum threshold, and use an ellipse fitting algorithm to fit it to obtain the geometric features of the droplet.

[0063] The purpose of step S2 is to identify and count the droplets that meet the preset size range, i.e., the mainstream size range, through bright field image analysis, and to count the geometric characteristics of the droplets to ensure the accuracy and reliability of the data. The specific implementation process includes the following steps: 1. Image reading and preprocessing: Read the bright field image and convert it into a grayscale image. Then, perform the following preprocessing operations on the image: (1) Contrast-limited adaptive histogram equalization (CLAHE) is used to enhance the local contrast of the image to highlight the edges and morphological features of the droplets.

[0064] (2) Use the Gaussian blur method to reduce the noise of the image, remove background noise and small interference points, and ensure the smoothness of the image.

[0065] 2. Automatic threshold segmentation: The Otsu's Method is used to automatically calculate the optimal threshold to achieve image binarization. It is an existing classic image segmentation method. Its main idea is to automatically select a threshold so that the inter-class variance between the foreground and background after image segmentation is maximized, thereby obtaining the best segmentation effect. According to the calculated threshold, a binary image is generated, in which the droplet area is the foreground and the background area is black.

[0066] It should be noted that the global fixed threshold method (such as manually setting a fixed grayscale threshold) or iterative threshold segmentation method can be used instead of the Otsu binarization algorithm. Although these methods have low flexibility and robustness, they can still achieve foreground and background separation under the conditions of stable lighting and clear background contrast in the experimental environment.

[0067] 3. Hole filling: In this embodiment, the hole filling algorithm is applied to fill the binary image to repair the blank area inside the droplet that may be generated during the thresholding process and ensure the integrity of the droplet outline. The hole filling algorithm is a commonly used image processing technology, which mainly includes the following logical steps: 3.1 Noise removal and morphological processing: 3.11: Use morphological opening operation to remove tiny noise points and avoid isolated pixels affecting subsequent processing.

[0068] 3.12: Perform morphological operations using an elliptical structure element to better fit the shape of the droplet.

[0069] 3.2 Closing operation to fill small holes: The morphological closing operation is applied on the denoised image to fill the small holes caused by segmentation errors and enhance the coherence of the droplet region.

[0070] 3.3 Hole filling in completely enclosed areas: Connected region analysis was used to identify and fill the internal holes that were completely surrounded by the foreground using scipy.ndimage.binary_fill_holes to ensure the integrity of the droplet morphology.

[0071] 3.4 Generate the final optimized image: Combining the morphological operation with the hole-filled image, the inpainted binary image is obtained, making the boundary of the droplet region continuous and complete.

[0072] Through the above steps, the hole filling algorithm can effectively repair the internal defects caused by thresholding in the binary image and ensure the continuity and integrity of the droplet area.

[0073] 4. Contour detection and ellipse fitting: 4.1 First, use a common contour detection algorithm to extract all external contours in the binary image. Here, the cv2.findContours function in the OpenCV library is used, which can identify all connected external areas (i.e. external contours) in the image and return a set of contour points.

[0074] 4.2 Next, for each contour whose number of points exceeds the set minimum threshold (such as 25 points), it is processed by the ellipse fitting algorithm. The ellipse fitting uses the cv2.fitEllipse function of OpenCV, which fits the contour based on the least squares method and returns a tuple containing the following information: 4.21 Center position: coordinates of the center point of the ellipse. 4.22 Axis lengths: A tuple containing the lengths of the major and minor axes (usually expressed as the entire axis length, which can be divided by 2 to get the semi-axis).

[0075] Through the above fitting, the geometric characteristics of the droplet can be extracted, including the center position, the length of the major and minor axes, and the rotation angle. Based on these characteristics, the ellipse area, the ratio of the major and minor axes and other parameters can be calculated later, and the ellipses that meet the conditions can be screened according to the preset area range or the ratio of the major and minor axes, so as to achieve effective identification and quantitative analysis of the droplet area.

[0076] It should be noted that a simple edge detection algorithm (such as the Canny operator) can be used to directly extract the outer contour of the droplet and calculate the area without performing ellipse fitting. Although this method cannot extract geometric characteristics such as the major and minor axes, it can still meet the basic requirements for experiments that only need to count the number of droplets.

[0077] 5. Ellipse screening: In this example, the geometric characteristics of each droplet are first obtained by ellipse fitting. The complete area of ​​the fitted ellipse is calculated using the standard formula ,in a and b Represents the major and minor axes of the ellipse. Since most of the droplets are of similar size, only a few have abnormal sizes due to merging or squeezing, this example uses the median as the reference value for the mainstream size. Specifically: 5.1 First, count the areas of all fitted ellipses and calculate their median.

[0078] 5.2 Area screening: Ellipses with areas between 75% and 125% of the median are then screened out, and ellipses with areas below 75% or above 125% are eliminated to ensure that droplets within the main size range are statistically analyzed and to eliminate the interference of abnormal situations on subsequent analysis.

[0079] 5.3 Screening of the ratio of major axis to minor axis: Based on the axis lengths of the ellipses obtained by fitting them previously, the ratio of the major axis to the minor axis of the ellipse is calculated, and ellipses with a ratio of major axis to minor axis exceeding 1.3 are eliminated.

[0080] 6. Droplet counting and labeling: 6.1 Statistical information such as quantity and area: For each fitted ellipse that passes the screening conditions (the area is within the range of 75% to 125% of the median), record its relevant geometric characteristics. The statistical process mainly includes: 6.11. Quantity statistics: All ellipses that meet the previous screening conditions are counted cumulatively to obtain the total number of identified droplets.

[0081] 6.12. Area calculation: using the standard ellipse area formula ,in a and b Represent the major and minor axes of the ellipses, calculate the complete area of ​​each ellipse, and summarize the area distribution information.

[0082] 6.13. Geometric property recording: For each ellipse, record its center position, major axis length, and minor axis length to facilitate subsequent data statistics and analysis.

[0083] 6.2 Draw the fitted ellipse on the original image: According to the fitting results, use the drawing function (such as cv2.ellipse in OpenCV) to draw each ellipse that meets the screening conditions on the original image. At the same time, mark a small dot or intersection at the center of the ellipse (using cv2.circle or other marking methods) to clearly show the center position of the droplet.

[0084] 6.3 The correspondence between the finally identified droplets and ellipses: 6.31 Fitting ellipses to represent droplet contours: In the binary image, through contour detection and ellipse fitting, it is considered that each ellipse that meets the conditions basically describes the boundary of a droplet.

[0085] 6.32 The interior of the ellipse is the droplet area: Therefore, the fitted ellipse drawn on the original image is not only used to display the detection results, but also serves as a geometric model of the droplet, and its internal area is regarded as the corresponding droplet.

[0086] 6.33 Consistency of statistics and display: The drawn ellipse and the marked center point correspond one-to-one with the information such as the number of droplets, area distribution, center position and axis length obtained from the above statistics to ensure the consistency of detection and subsequent analysis.

[0087] Through the above steps, the processing effect of the bright field image is as follows Figure 5 As shown (the bright field image obtained is as Figure 4 As shown in the figure, it can efficiently identify droplets that meet the size range, remove abnormal droplets and record their size statistics (such as average diameter and standard deviation), and identify and count the total number of droplets that meet the mainstream size range.

[0088] This application proposes for the first time the use of an ellipse fitting algorithm, which can effectively exclude interference areas, has good robustness, and improves the recognition accuracy of droplets. This application applies the ellipse fitting algorithm to droplet signal recognition for the first time. Through this algorithm, interference areas such as chip edges, guide brackets, and fluorescent contamination points can be excluded, avoiding the influence of these areas on the number and intensity signals of droplets. Ellipse fitting can accurately describe droplets whose morphology is close to that of an ellipse, and provide more accurate contour analysis than traditional circular recognition methods. The algorithm's automated interference rejection function significantly improves the accuracy of droplet recognition under complex backgrounds, avoids errors caused by manual threshold setting, and improves the efficiency and accuracy of data processing.

[0089] It should be noted that a simple upper and lower limit screening can be performed based only on the area (such as setting a fixed area range). Although the lack of shape feature constraints may cause some deformed droplets to be mistakenly counted, it can still meet basic statistical needs under experimental conditions with low abnormality rates. When counting the number and area distribution of droplets, statistical characteristics such as standard deviation can be ignored, and only the average value of the total area can be counted to reduce the computational complexity.

[0090] As an optional implementation, Figure 6 As shown, in step S3, the fluorescence field image is analyzed to obtain the total number of fluorescent droplets within a preset size range, specifically including: S31: converting the fluorescence field image into a grayscale image and performing preprocessing to obtain a preprocessed fluorescence field image.

[0091] S32: performing automatic threshold segmentation on the preprocessed fluorescence field image to obtain a second binary image.

[0092] S33: performing hole filling and smoothing processing on the second binary image to obtain an optimized binary image.

[0093] S34: using a watershed segmentation algorithm to segment the optimized binary image to obtain a segmented labeling image; the labeling image includes a plurality of labeling regions, each of which has a unique non-zero label value.

[0094] S35: Determine the area of ​​each marked region according to the number of pixels in each marked region.

[0095] S36: Calculate area distribution characteristics according to the area of ​​each marked region.

[0096] S37: Perform area screening based on the area distribution characteristics to obtain screened areas.

[0097] S38: The total number of instances corresponding to the screened areas is taken as the total number of fluorescent droplets within a preset size range, and the actual area and center position of each fluorescent droplet, as well as the distribution of the fluorescent droplets, are recorded.

[0098] As an optional implementation, in step S33, it specifically includes: S331: Filling holes in the second binary image using a morphological closing operation to obtain a filled image.

[0099] S332: Using an elliptical kernel to smooth the edge of the filled image to obtain an optimized binary image.

[0100] As an optional implementation, in step S34, specifically including: S341: performing distance transform calculation on the foreground in the optimized binary image to obtain the Euclidean distance from each foreground pixel to the nearest background pixel.

[0101] S342: taking the point corresponding to the Euclidean distance greater than a preset threshold as a seed point.

[0102] S343: Performing connected region analysis on the set of seed points, and generating an initial labeling map on an image of the same size as the optimized binary image; the initial labeling map includes a plurality of labeling regions, each of which has a unique non-zero label value.

[0103] S344: Calculate the grayscale gradient of each pixel in the optimized binary image to obtain a grayscale gradient map; S345: Using the grayscale gradient image as the input of the watershed segmentation algorithm, using the initial label image as the initial region label, and taking the seed point as the starting point, executing the watershed segmentation algorithm to obtain a segmented label image.

[0104] This step S3 uses the fluorescence field image to identify, segment and statistically analyze the positive droplets. Through a series of image processing and algorithm optimization, the accurate classification and visualization of fluorescent droplets are ensured, providing reliable data support for digital PCR experiments. The specific process includes the following steps: 1. Image reading and preprocessing: 1.1 Read the fluorescence field image data and convert it into a grayscale image.

[0105] 1.2 Use Gaussian blur to reduce the noise of the image and filter out tiny high-frequency noise to ensure that the edges of the fluorescent droplets are clearly visible.

[0106] 2. Use Otsu binarization algorithm for automatic threshold segmentation: 2.1 Automatically calculate the optimal threshold to distinguish the foreground (fluorescent droplets) from the background.

[0107] 2.2 Use the calculated threshold to binarize the image, mark the fluorescent droplet area as the foreground (value 255), and mark the rest as the background (value 0).

[0108] It should be noted that the global fixed threshold method (such as manually setting a fixed grayscale threshold) or iterative threshold segmentation method can be used instead of the Otsu binarization algorithm. Although these methods have low flexibility and robustness, they can still achieve foreground and background separation under the conditions of stable lighting and clear background contrast in the experimental environment.

[0109] 3. Binarized image optimization, including hole filling and smoothing: 3.1 Morphological closing operation and internal hole filling: The morphological closing operation of the binary image mainly includes two steps: 3.11 Constructing structural elements: Choose appropriate structural elements (such as elliptical or rectangular cores), whose size should be slightly larger than the expected small cavities inside the droplets.

[0110] 3.12 Closing operation: Closing operation is to first dilate and then erode. The dilation operation makes the droplet boundary expand outward, connecting the internal small holes with the surrounding area; then the erosion operation restores the original size while maintaining the connection effect, thereby filling the internal small holes and ensuring the integrity of the entire droplet area.

[0111] 3.2 Elliptical core edge smoothing: In order to further improve the smoothness of the edge of the fluorescent droplet, an elliptical kernel is used to process the edge of the image. The specific steps are: 3.21 Construction of elliptical core: An elliptical structural element is used, and the size and shape of the core are designed according to the droplet characteristics.

[0112] 3.22 Edge smoothing operation: The elliptical kernel is used to perform morphological operations (such as closing or opening operations) on the image, so that the edge of the droplet is locally expanded and smoothed, thereby eliminating jagged or discontinuous edges and ensuring that the fluorescent droplet area is smoother and more complete.

[0113] 4. Watershed segmentation algorithm: 4.1 Generate seed point map and label map: In this embodiment, the distance transform is used to determine the seed points inside the foreground area, and thus generate a label map. The main steps are as follows: 4.11 Distance transformation calculation: Applies a distance transform to the foreground (usually white areas) in a binary image. For each foreground pixel in the image, calculates the Euclidean distance to the nearest background pixel.

[0114] 4.12. Determination of seed points: In the distance map, the local maximum often corresponds to the inner center of the foreground area. By finding the local maximum of the distance map, these points can be selected as seed points. Usually a threshold can be set to ensure that only obvious peaks are selected as seed points.

[0115] 4.13 Generate initial labeling map: Based on the selected seed points, an initial labeling map is generated on an image of the same size as the original image. Figure 7 As shown. In the initial marking diagram: (1) Each seed point is marked with a unique integer value.

[0116] (2) Background areas are usually marked as 0.

[0117] (3) The remaining regions remain unlabeled or are further updated according to subsequent segmentation strategies.

[0118] The generated initial labeling map provides initial seed points for the subsequent region segmentation (watershed segmentation algorithm), ensuring that the segmentation process can accurately divide the foreground area into different target areas.

[0119] 4.2 Obtaining and using grayscale gradient images: The grayscale gradient map is obtained by calculating the grayscale gradient of each pixel in the original image (or the image after preprocessing, such as the binary image after morphological closing operation). Common calculation methods include using the Sobel operator or Canny edge detection. The grayscale gradient map can highlight the areas with significant grayscale changes in the image (i.e., edges), providing boundary information for the watershed segmentation algorithm.

[0120] 4.3 Combination of initial label map and gradient image: In the previous step, the seed points of the foreground area were obtained using the distance transform, and an initial label map was generated, in which each seed point was assigned a unique label value, and the background pixels were marked as 0. When the initial label map is combined with the grayscale gradient map, the grayscale gradient map is used as the input image ("slope map") of the watershed segmentation algorithm, and the initial label map is used as the initial region label. In this way, the watershed segmentation algorithm will use the position of high gradient (edge) in the grayscale gradient map as a "dam" to limit the expansion of the region, and the seed point is used as the starting point of the watershed region.

[0121] 4.4 Segmentation process of watershed segmentation algorithm: The basic idea of ​​the watershed segmentation algorithm is to regard the gradient map as a terrain height map, the seed point represents the initial "pool", and then simulate the water flowing from low-lying areas to high areas. The specific steps include: 4.41. Initialization: Use the previously generated initial label map as the initial watershed region, where different labels represent different region starting points.

[0122] 4.42. Flooding process: Starting from the lowest point in the grayscale gradient map, adjacent pixels are classified into their respective seed regions. When different regions meet, a "dam" is built to prevent water from mixing.

[0123] 4.43. Region segmentation: Finally, the entire image is divided into several regions, each corresponding to a seed point, so as to achieve accurate segmentation of the foreground area (fluorescent droplets) and obtain the final instantiation mark map as shown in Figure 8 shown.

[0124] 5. Area statistics and screening: 5.1 Area extraction: On the final instantiated label map obtained by the watershed segmentation algorithm, regional analysis is performed to identify and count the area of ​​each fluorescent droplet.

[0125] Specifically, skimage.measure.regionprops is used to calculate the number of pixels in each region as the area of ​​the droplet. The area unit of each pixel is determined according to the image resolution or a predefined pixel size.

[0126] 5.2 Calculation of area distribution characteristics: 5.21 Extract area values ​​from all instances and calculate the area distribution.

[0127] 5.22 The median is used as the reference value for the typical area of ​​fluorescent droplets. Because most droplets are of relatively consistent size, occasional outliers (larger or smaller) will have a greater impact on the average value, so the median is used to better reflect the mainstream size.

[0128] 5.3 Set the screening range and instance screening: 5.31 According to the experimental requirements, the relative change threshold of the area is further set, for example, setting the range to ±25%.

[0129] 5.32 Iterate over the area value of each instance. If the area is less than the lower limit or greater than the upper limit, the instance is regarded as an abnormal droplet and marked as background in the label map (for example, the corresponding pixel value is set to 0).

[0130] 5.33 After screening, the remaining examples are fluorescent droplets that meet the requirements and can be used for subsequent statistical and quantitative analysis.

[0131] 6. Visualization and instance allocation: 6.1 Color visualization of fluorescent droplets: A random color map is used to assign unique colors to different fluorescent droplet instances, generating a color segmentation map that clearly shows the boundaries of each droplet, e.g. Fig.10 As shown ( Fig. 9 (a) is a schematic diagram of the acquired fluorescence field image.

[0132] 6.2 Instance Statistics and Labeling: 6.21 Statistical totals and area markings: First, a region labeling algorithm is applied to the segmented labeled image. Here, each connected region (corresponding to a fluorescent droplet instance) is assigned a unique label. Using the skimage.measure.label function, a labeled image can be obtained in which each foreground region has a unique label value, and the total number of fluorescent droplets can be counted.

[0133] 6.22 Extract area and center location information: Use the regionprops function in the skimage.measure module to calculate relevant properties such as area and center position for each region in the labeled image.

[0134] 7. Output results: 7.1 Provide statistical information of fluorescent droplets, including: (1) The total number of fluorescent droplets that conform to the mainstream size.

[0135] (2) The actual area and center position of each droplet.

[0136] (3) Distribution of fluorescent droplets.

[0137] 7.2 Supports dynamic adjustment of thresholds and segmentation parameters to adapt to different experimental conditions and ensure the accuracy and robustness of classification results.

[0138] Through the above steps, fluorescent droplets (positive droplets) that meet the size range can be efficiently identified, achieving accurate identification, segmentation and statistics of fluorescent droplets.

[0139] As an optional implementation, in step S4, the proportion of fluorescent droplets is statistically calculated.

[0140] The processes of step 2 and step 3 can be performed simultaneously, and then based on the total number of counted droplets of mainstream size in the droplet group and the number of fluorescent droplets obtained by the statistics, the ratio of fluorescent droplets (number of fluorescent droplets / number of counted droplets) can be further calculated.

[0141] Through the above process, accurate identification, segmentation and statistics of droplets are achieved, providing key data support for digital PCR experiments and significantly improving the reliability and efficiency of the experiments.

[0142] In summary, the technical solution of this application has the following beneficial effects: 1. Accurately count the number and area distribution of droplets.

[0143] This application processes the bright field and fluorescent field of the image independently, uses adaptive histogram equalization (CLAHE) to enhance local contrast, and combines Gaussian blur noise reduction to ensure that the edges of the droplets are clear and the noise is effectively suppressed. This preprocessing method can accurately extract the contour information of the droplets, laying the foundation for subsequent statistical analysis. Through clear droplet edge separation, the accuracy of droplet recognition is greatly improved, avoiding missed or false detection of droplets due to blurred boundaries.

[0144] 2. Effectively eliminate the interference of abnormal droplets.

[0145] This application uses contour detection and ellipse fitting technology, combined with area screening and major-minor axis ratio screening rules, to effectively identify and eliminate abnormal droplets (such as droplets whose area exceeds the mainstream size range or droplets with distorted shapes). Specifically, the upper and lower limits of droplets within the mainstream area range (i.e., 75%-125% of the average value within the mainstream droplet area range) and the major-minor axis ratio range (<1.3) are set to eliminate abnormal droplets caused by fusion or rupture. By eliminating abnormal droplets, it is ensured that the statistical results are only based on mainstream size droplets that meet the standards, which greatly improves the credibility of the statistical results.

[0146] 3. Dynamically adjust the threshold to adapt to different experimental conditions.

[0147] This application uses the Otsu binarization algorithm to dynamically calculate the optimal threshold of the image, achieving accurate separation of the background and foreground (fluorescent droplet area). In addition, through regional characteristic analysis and dynamic screening rules, the recognition range and classification accuracy of droplets are further optimized. This application can automatically adjust the threshold according to different experimental conditions (such as lighting, sample concentration, etc.) to ensure the robustness of the segmentation effect, thereby improving the reliability of positive droplet statistics.

[0148] 4. The statistical efficiency of droplets is high and the results are statistically significant.

[0149] This application uses efficient image processing algorithms (such as watershed segmentation algorithms and regional labeling analysis) to quickly process high-throughput experimental data containing tens of thousands of droplets. At the same time, combined with statistical analysis of the area, extreme outliers are eliminated to ensure that the results are statistically significant. Specifically, through the area statistics method based on regional labeling, the droplet instances that meet the standards are screened out, and the total number, area distribution and location information are counted. Rapid statistics and accurate analysis of the number of droplets are achieved, while providing scientific basis and statistical significance support for the experimental data.

[0150] 5. Enhanced visualization of fluorescent droplets.

[0151] This application uses random color mapping technology to assign a unique color to each fluorescent droplet and generate a color instance distribution map. This visualization enhancement technology can intuitively display the distribution, morphology and numbering information of the droplets, which is convenient for subsequent experimental analysis and verification. It improves the interpretability and traceability of the fluorescent droplet detection results, making it easier for experimenters to quickly discover potential anomalies or specific patterns.

[0152] 6. Effectively support digital PCR experimental needs.

[0153] This application not only achieves accurate statistics of positive droplets by constructing a comprehensive analysis framework combining bright field and fluorescence field, but also provides multi-dimensional information such as droplet area and distribution, effectively meeting the high-precision and high-efficiency detection requirements in digital PCR experiments. It reduces detection deviations caused by interference factors and provides technical guarantees for the accuracy, stability and reproducibility of experimental data.

[0154] The present application also provides an application scenario, which applies the above-mentioned method for counting the fluorescence ratio information of a microfluidic droplet array. Specifically: The method for counting the fluorescence ratio information of a microfluidic droplet array provided in this embodiment can be applied in a fluorescence intensity recognition classification analysis scenario. The fluorescence intensity recognition classification analysis scenario includes: an image acquisition link, a bright field image analysis link, a fluorescence field image analysis link, and a ratio statistics link of fluorescent droplets; first, the fluorescence field image and the bright field image of the chip are acquired; then, the bright field image is analyzed to obtain the total number of droplets within a preset size range; the fluorescence field image is analyzed to obtain the total number of fluorescent droplets within a preset size range; finally, based on the total number of droplets within the preset size range and the total number of fluorescent droplets within the preset size range, the ratio of fluorescent droplets can be calculated.

[0155] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Fig.11 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the fluorescence field image and the bright field image of the chip. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for statistics of fluorescence ratio information of a microfluidic droplet array is implemented.

[0156] Those skilled in the art will understand that Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0157] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the above-mentioned method embodiments are implemented when the processor executes the computer program.

[0158] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the above-mentioned method embodiments are implemented.

[0159] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the above-mentioned method embodiments are implemented.

[0160] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0161] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0162] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0163] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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.

[0164] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for counting fluorescence ratio information of a microfluidic droplet array, characterized in that: The method for counting fluorescence ratio information of a microfluidic droplet array comprises: Acquire fluorescence field images and bright field images of the chip; Analyzing the bright field image to obtain the total number of droplets within a preset size range; Analyzing the fluorescence field image to obtain the total number of fluorescent droplets within a preset size range; Based on the total number of droplets within the preset size range and the total number of fluorescent droplets within the preset size range, the ratio of fluorescent droplets is calculated.

2. The method for counting fluorescence ratio information of a microfluidic droplet array according to claim 1, characterized in that: The bright field image is analyzed to obtain the total number of droplets within a preset size range, specifically including: Converting the bright field image into a grayscale image and performing preprocessing to obtain a preprocessed bright field image; Performing automatic threshold segmentation on the preprocessed bright field image to obtain a first binary image; Filling holes in the first binary image to obtain a binary image of droplets with complete contours; Performing contour detection and ellipse fitting on the binary image of the microdroplet to obtain geometric features of the microdroplet; the geometric features include: center position, lengths of the major axis and the minor axis, and rotation angle; Performing ellipse screening based on the geometric features of the droplet to obtain a screened fitting ellipse; The total number of the selected fitting ellipses is taken as the total number of droplets within the preset size range, and the corresponding geometric features are recorded.

3. The method for counting fluorescence ratio information of a microfluidic droplet array according to claim 2, characterized in that: Filling holes in the first binary image to obtain a binary image of droplets with complete contours specifically includes: Using a morphological opening operation to remove isolated small noise points in the first binary image, to obtain a first binary image with small noise points removed; Using a morphological closing operation to fill holes smaller than a preset shape in the first binary image from which small noise points have been removed, to obtain a filled first binary image; Using connected region analysis, the padded first binary image is identified to obtain internal holes that are completely surrounded by foreground pixels; Use scipy.ndimage.binary_fill_holes to fill the internal holes completely surrounded by foreground pixels to obtain a filled area; The filled areas are merged to obtain a binary image of the droplet with a complete outline.

4. The method for counting fluorescence ratio information of a microfluidic droplet array according to claim 2, characterized in that: Performing contour detection and ellipse fitting on the droplet binary image to obtain the geometric features of the droplet, specifically including: Using a contour detection algorithm to extract all external contours in the droplet binary image to obtain a plurality of external contours; The outer contour whose number of contour points exceeds the set minimum threshold is determined, and an ellipse fitting algorithm is used to fit it to obtain the geometric characteristics of the droplet.

5. The method for counting fluorescence ratio information of a microfluidic droplet array according to claim 1, characterized in that: The fluorescence field image is analyzed to obtain the total number of fluorescent droplets within a preset size range, specifically including: Converting the fluorescence field image into a grayscale image and preprocessing it to obtain a preprocessed fluorescence field image; Performing automatic threshold segmentation on the preprocessed fluorescence field image to obtain a second binary image; Performing hole filling and smoothing processing on the second binary image to obtain an optimized binary image; The optimized binary image is segmented by a watershed segmentation algorithm to obtain a segmented label map; the label map includes a plurality of label regions, each of which has a unique non-zero label value; Determine the area of ​​each marked region based on the number of pixels within each marked region; According to the area of ​​each marked region, the area distribution characteristic is calculated; Performing area screening based on the area distribution characteristics to obtain screened areas; The total number of instances corresponding to the screened areas is taken as the total number of fluorescent droplets within the preset size range, and the actual area and center position of each fluorescent droplet, as well as the distribution of the fluorescent droplets, are recorded.

6. The method for counting fluorescence ratio information of a microfluidic droplet array according to claim 5, characterized in that: Performing hole filling and smoothing processing on the second binary image to obtain an optimized binary image specifically includes: Filling holes in the second binary image using a morphological closing operation to obtain a filled image; The edge of the filled image is smoothed using an elliptical kernel to obtain an optimized binary image.

7. The method for counting fluorescence ratio information of a microfluidic droplet array according to claim 5, characterized in that: The optimized binary image is segmented using a watershed segmentation algorithm to obtain a segmented labeling map, which specifically includes: Performing distance transformation calculation on the foreground in the optimized binary image to obtain the Euclidean distance from each foreground pixel to the nearest background pixel; The point corresponding to the Euclidean distance being greater than a preset threshold is taken as a seed point; Performing connected region analysis on the set of seed points to generate an initial labeled image on an image of the same size as the optimized binary image; the initial labeled image includes a plurality of labeled regions, each of which has a unique non-zero label value; Calculating the grayscale gradient of each pixel in the optimized binary image to obtain a grayscale gradient map; The grayscale gradient image is used as the input of the watershed segmentation algorithm, the initial label image is used as the initial area label, and the watershed segmentation algorithm is executed with the seed point as the starting point to obtain the segmented label image.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for fluorescence ratio information statistics of a microfluidic droplet array according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for counting fluorescence ratio information of a microfluidic droplet array according to any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for counting fluorescence ratio information of a microfluidic droplet array according to any one of claims 1 to 7 is implemented.

Images