Gene chip microbead detection method and device, electronic equipment and readable storage medium

Abstract

The invention relates to a gene chip microbead detection method and device, electronic equipment and a readable storage medium, and the method comprises the steps: obtaining a target microbead rectangular region image which is formed by splicing a plurality of sub-images; processing the target micro-bead rectangular region image by using a target detection model to obtain micro-bead related parameters; a target matrix is constructed according to the micro-bead related parameters and the target micro-bead rectangular area, a micro-bead detection result is obtained, and the target matrix stores the micro-bead related parameters according to the arrangement condition of micro-beads in the target micro-bead rectangular area. According to the method, the target detection model obtained through deep learning training identifies the target microbead rectangular region image, the corresponding microbead related parameters are obtained, and the target matrix is constructed in combination with the microbead related parameters and the target microbead rectangular region, so that a clear and complete detection result can be output while microbead positioning and brightness identification can be rapidly and accurately completed.

Description

Technical Field

[0001] The present invention relates to the field of biochip technology, and in particular to a gene chip microbead detection method, device, electronic equipment and readable storage medium. Background Art

[0002] The micro beads are distributed in a long rectangular area Rect of the chip (the length of the long side W is more than twice the length of the short side H), and the distribution positions of the micro beads are distributed in a rectangular grid or a regular hexagonal honeycomb.

[0003] The existing solution is to use a line scan camera. The laser light source emits light in a specific wavelength range to excite the fluorescence of the microbeads. The line scan camera collects images. The laser light source and the camera move synchronously along the long side of the rectangle, collecting fluorescence signals of one frequency band each time they move.

[0004] There are usually two methods for locating microbeads on the acquired images: 1. Using an iterative search algorithm based on neighbor relationships to traverse the microbead positions; 2. Projection method.

[0005] When using a line scan camera, the image is scanned line by line, and the fluorescence excitation time is short, which requires a high power laser light source, high cost, and short service life. Because the diameter of the beads is small, the lens magnification is high, the depth of field is small, and the line scan camera scans a large area, which requires the camera movement line to be highly perpendicular to the chip surface. Otherwise, the object distance changes greatly, exceeding the depth of field range, and part of the captured image is blurred. In addition, the flatness of the chip surface itself is required to be high, which makes the hardware cost high and the equipment debugging difficult.

[0006] The iterative algorithm is time-consuming and prone to repeated calculations. The projection method requires the microbead distribution to be horizontal and vertical, that is, the microbead matrix must be rotated, otherwise the projection method positioning deviation will be large. Summary of the invention

[0007] In order to solve the above technical problems, the embodiments of the present application provide a gene chip microbead detection method, device, electronic device and readable storage medium that can accurately and quickly complete microbead positioning and light and dark classification identification. The specific scheme is as follows: In a first aspect, the present application provides a gene chip microbead detection method, comprising: Acquire a target microbead rectangular region image, wherein the target microbead rectangular region image is formed by splicing a plurality of sub-images, and the sub-images cover a portion of the target microbead rectangular region; Processing the target microbead rectangular area image using a target detection model to obtain microbead related parameters, wherein the target detection model is a model obtained based on deep learning training, and the microbead related parameters include microbead position, microbead light and dark classification results, and microbead area grayscale mean; A target matrix is ​​constructed according to the microbead-related parameters and the target microbead rectangular area to obtain a microbead detection result, wherein the target matrix stores the microbead-related parameters according to the arrangement of the microbeads in the target microbead rectangular area.

[0008] According to a specific implementation of the embodiment of the present application, the step of acquiring a rectangular area image of a target microbead includes: Dividing the target microbead rectangular area into multiple sub-areas to be photographed, wherein adjacent sub-areas to be photographed include a preset overlapping area, and the target microbead rectangular area includes a first side and a second side, and the length of the first side is greater than the length of the second side; Acquire an image of the sub-area to be photographed to obtain a plurality of sub-images; The plurality of sub-images are subjected to a preset stitching process to obtain the rectangular region image of the target microbeads.

[0009] According to a specific implementation of the embodiment of the present application, the step of performing a preset stitching process on the multiple sub-images to obtain the target microbead rectangular area image includes: For two adjacent sub-images, a first adjacent region of one sub-image is determined as a template region, and a second adjacent region of the other sub-image is determined as a matched region; Searching for a position matching the template region in the matched region until a target matching position is obtained, wherein the target matching position is a position where the similarity between the matched region and the template region is the greatest; Moving the other sub-image according to the target matching position and discarding redundant partial images until the pixel points of the template area are consistent with those of the matched area; When any two adjacent sub-images are stitched together, the stitched image is determined as the target microbead rectangular area image.

[0010] According to a specific implementation of the embodiment of the present application, before using the target detection model to process the target microbead rectangular area image to obtain microbead related parameters, the method also includes: Acquiring different gene chip images, wherein the gene chip images include microbeads; Annotating the gene chip image with microbead-related parameters; According to the annotated gene chip image, a target detection network or a segmentation network is used to perform model training to obtain the target detection model.

[0011] According to a specific implementation of the embodiment of the present application, the target matrix is ​​constructed according to the microbead related parameters and the target microbead rectangular area to obtain the microbead detection result, including: Acquire a target affine matrix according to the position coordinates of the target microbead rectangular area, wherein the distribution state of the target affine matrix is ​​a horizontal and vertical state; Constructing a virtual matrix according to the microbead distribution in the target microbead rectangular area, wherein the number of rows and columns of the virtual matrix corresponds to the number of microbeads in the microbead distribution in the row direction and the column direction; The target affine matrix and the virtual matrix are combined to map the microbead related parameters to each element of the virtual matrix to obtain the microbead detection result, wherein one element corresponds to one microbead.

[0012] According to a specific implementation of the embodiment of the present application, the step of obtaining a target affine matrix according to the position coordinates of the target microbead rectangular area includes: Extracting four boundary lines of the target microbead rectangular area according to a preset iterative fitting algorithm; Connecting the four boundary lines to obtain an edge matrix; Performing an affine transformation on the edge matrix to obtain the target affine matrix.

[0013] According to a specific implementation of the embodiment of the present application, mapping the microbead related parameters to each element of the virtual matrix to obtain the microbead detection result includes: Initializing the virtual matrix and filling the target element with preset content, wherein the target element is the element in the virtual matrix corresponding to the position of the microbead in the target affine matrix; The unupdated elements in the virtual matrix are traversed, and the corresponding micro-bead related parameters are filled in for each unupdated element mapping, wherein the unupdated elements are filled with the preset content.

[0014] In a second aspect, the present application provides a gene chip microbead detection device, comprising: An acquisition module, used for acquiring a rectangular area image of a target microbead, wherein the rectangular area image of the target microbead is formed by splicing a plurality of sub-images, and the sub-images cover a part of the rectangular area of ​​the target microbead; A detection module, used to process the target microbead rectangular area image using a target detection model to obtain microbead related parameters, wherein the target detection model is a model obtained based on deep learning training, and the microbead related parameters include microbead position, microbead light and dark classification results, and microbead area grayscale mean; The output module is used to construct a target matrix according to the microbead related parameters and the target microbead rectangular area to obtain the microbead detection results, wherein the target matrix stores the microbead related parameters according to the arrangement of the microbeads in the target microbead rectangular area.

[0015] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the gene chip microbead detection method described in the first aspect.

[0016] In a fourth aspect, an embodiment of the present application further provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the gene chip microbead detection method described in the first aspect.

[0017] In summary, the present embodiment provides a gene chip microbead detection method, device, electronic device and readable storage medium, including: obtaining a target microbead rectangular area image, wherein the target microbead rectangular area image is formed by splicing a plurality of sub-images; using a target detection model to process the target microbead rectangular area image to obtain microbead related parameters; constructing a target matrix according to the microbead related parameters and the target microbead rectangular area to obtain microbead detection results, wherein the target matrix stores the microbead related parameters according to the arrangement of the microbeads in the target microbead rectangular area. The present application uses a target detection model obtained through deep learning training to identify the target microbead rectangular area image, obtain the corresponding microbead related parameters, and construct a target matrix in combination with the microbead related parameters and the target microbead rectangular area, which can quickly and accurately complete microbead positioning and light and dark identification while outputting clear and complete detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, 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 disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of a gene chip microbead detection method provided in an embodiment of the present application; Figure 2 A schematic diagram of the steps for obtaining a rectangular area image of a target microbead according to an embodiment of the present application; Figure 3 A schematic diagram of the steps of performing preset stitching processing on multiple sub-images according to an embodiment of the present application; Figure 4 A schematic diagram of the steps for training a target detection model according to an embodiment of the present application; Figure 5 A schematic diagram of the steps for obtaining microbead detection results in an embodiment of the present application; Figure 6 A schematic diagram of the steps for obtaining a target affine matrix in an embodiment of the present application; Figure 7 A schematic diagram of the steps of mapping microbead-related parameters in a virtual matrix according to an embodiment of the present application; Figure 8 A schematic diagram of a rectangular area of ​​target microbeads provided in an embodiment of the present application; Fig. 9 A schematic diagram of multiple sub-images acquired by an area array camera according to an embodiment of the present application; Fig.10 Schematic diagram of the application of the preset splicing process for multiple sub-images in the embodiment of the present application Fig.11 This is a schematic diagram of an application of marking microbeads in a rectangular area of ​​microbeads according to an embodiment of the present application; Fig.12 This is a schematic diagram of an application for calculating a target affine matrix according to an embodiment of the present application; Fig.13 A schematic diagram of a device module of a gene chip microbead detection device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0021] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0022] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0023] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0024] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0025] In related technologies, microbeads on gene chips usually refer to tiny spheres with specific sequence DNA fragments coupled to their surfaces, which are used in gene chip technology. These microbeads play a key role in gene chips. As carriers of probes, they can hybridize with DNA or RNA samples to be tested, thereby realizing the detection of information such as gene expression and gene mutation.

[0026] In actual application, microbeads are arranged in long strips on gene chips, which can make full use of the chip area, improve detection throughput, and facilitate sample addition and hybridization reaction operations. The long strip distribution of microbeads makes signal detection more concentrated. In related technologies, laser scanners and other equipment are often used to detect the positioning and light and dark classification of microbeads on gene chips.

[0027] In the related art, the steps of detecting the position of microbeads on the gene chip include: extracting the DNA or RNA sample to be detected and labeling it. Then, the labeled sample is hybridized with the microbeads on the gene chip. The laser light source of the line scan camera is controlled to emit light in a specific wavelength range to excite the fluorescence of the microbeads, and then the line scan camera is used to collect images. The laser light source and the camera move synchronously along the long side of the long rectangle of the gene chip, and the fluorescence signal of one frequency band is collected once.

[0028] However, when using a line scan camera, the image is scanned line by line, and the fluorescence excitation time is short, which requires a high power laser light source, high cost, and short service life. In addition, due to the small diameter of the beads, the lens magnification is high, the depth of field is small, and the line scan camera scans a large area, requiring the camera movement line to be highly perpendicular to the chip surface. Otherwise, the object distance will change greatly, exceeding the depth of field range, and the captured image will be regionally blurred. In addition, the flatness of the chip surface itself is required to be high, which makes the hardware cost high and the equipment debugging difficult.

[0029] To solve the aforementioned problems, an embodiment of the present application provides a gene chip microbead detection method, which replaces the line scanner with an area scan camera, collects images of microbeads distributed in long strips on the gene chip through the area scan camera, and quickly completes microbead positioning and light and dark classification through a deep learning neural network model, which can improve the efficiency of gene chip microbead detection while reducing equipment costs.

[0030] refer to Figure 1 The present application provides a gene chip microbead detection method, comprising the following steps: S101, obtaining a target microbead rectangular region image, wherein the target microbead rectangular region image is formed by splicing a plurality of sub-images, and the sub-images cover a portion of the target microbead rectangular region.

[0031] In this embodiment, the target microbead rectangular area refers to a rectangular area on the gene chip that includes microbeads arranged in a strip form. Figure 8 As shown, the microbeads are distributed in a rectangular area of ​​the chip, and the length of the long side W of the rectangular area of ​​the chip may be 2 times or more than 2 times the length of the short side H. It should be noted that the length of the long side W is greater than the length of the short side H, and the specific length difference may be determined according to the distribution of microbeads in actual application scenarios.

[0032] In practical applications, the distribution positions of the microbeads can be distributed in a rectangular grid, for example Figure 8 (B) Partial image. The distribution of microbeads can also be in the form of a regular hexagonal honeycomb, for example Figure 8 (A) Part of the image is shown. The length of the long side of the rectangular area is , the short side length is .

[0033] In this embodiment, multiple sub-images in the rectangular area of ​​the target microbeads are collected based on the area array camera, and then the target microbead rectangular area image is obtained by splicing the multiple sub-images. In actual application, the target microbead rectangular area is divided into multiple sub-areas according to the grid. The area array camera shoots the sub-areas one by one, and focuses and shoots in sequence, so that a clearer image can be obtained, which provides guarantee for subsequent microbead position detection and light and dark classification detection.

[0034] It should be noted that the grid division method of the sub-area and the specific shooting method of the area array camera can be customized according to the needs of the actual application scenario. For example, the target micro-bead rectangular area can be divided into a multi-grid area with two rows and multiple columns, each grid covering a part of the target micro-bead rectangular area, and adjacent grids also include overlapping areas. The area array camera can obtain images of all sub-areas at one time, or it can obtain images of each sub-area in sequence according to the arrangement order of the sub-areas.

[0035] S102, using the target detection model to process the target microbead rectangular area image to obtain microbead related parameters, wherein the target detection model is a model obtained based on deep learning training, and the microbead related parameters include the microbead position, the microbead light and dark classification results, and the grayscale mean of the microbead area.

[0036] In this embodiment, the target detection model is used to make predictions based on the input microbead rectangular area image, and output the microbead position (center point coordinates) in the image, the microbead brightness and darkness classification results, and the grayscale mean of the microbead area.

[0037] Deep learning (DL) is a branch of machine learning (ML). DL builds and trains deep neural network models to learn and extract features from data to achieve automated processing and decision-making of complex tasks. In this embodiment, deep learning training can be performed based on a large number of gene chip images containing microbeads to obtain a target detection model.

[0038] In a specific embodiment, the training target detection model can be trained based on the object detection (Object Detection) or segmentation network method. The steps of training the target detection model include: 1. Collecting sample data, the sample data includes a gene chip image with microbeads, and adding annotation information to the gene chip image with microbeads, and the annotation information includes microbead related parameters. 2. Selecting a suitable pre-trained model according to the needs of the actual application scenario, the pre-trained model includes YOLOv5, Faster R-CNN, Mask R-CNN and other models, among which YOLOv5 and Faster R-CNN are pre-trained models corresponding to target detection, and Mask R-CNN is a pre-trained model corresponding to the segmentation network. 3. Configure the environment for model training and install the deep learning framework required for the actual application scenario. 4. Load and pre-process the data. 5. Define training parameters such as loss function, optimizer, hyperparameters, and perform model training based on the training parameters and pre-trained model. 6. Perform performance verification on the trained target detection model to verify the effectiveness of the target detection model. Based on the above steps, a target detection model can be trained that can quickly detect and accurately output microbead-related parameter detection results.

[0039] This embodiment completes the detection of microbead related parameters through the target detection model, has accurate positioning and can simultaneously determine the light and dark classification results, has high parallel efficiency and fast speed, and does not need to consider the requirement of the projection method that the rectangular area image of the microbead must be straightened, which can greatly improve the efficiency of positioning detection and light and dark classification detection.

[0040] S103, constructing a target matrix according to the microbead related parameters and the target microbead rectangular area to obtain the microbead detection results, wherein the target matrix stores the microbead related parameters according to the arrangement of the microbeads in the target microbead rectangular area.

[0041] In this embodiment, after completing the detection of microbead-related parameters, a target matrix that can intuitively understand the microbead detection results is further constructed based on the microbead-related parameters and the arrangement of microbeads in the target microbead rectangular area. The arrangement of microbeads in the target microbead rectangular area can be referred to Figure 8 As shown, from top to bottom, the rows are numbered 1, 2, 3, 4, and from left to right, the columns are numbered 1, 2, 3, 4. It should be noted that the arrangement of the microbeads in the target microbead rectangular area and the numbering method can be adjusted according to the needs of the actual application scenario, and are not limited to the directions illustrated in the above examples.

[0042] In this embodiment, the target matrix includes multiple elements, and the elements corresponding to the microbeads in the target microbead rectangular area image store microbead-related parameters corresponding to the microbeads, so that the microbead detection results can be intuitively understood through the target matrix, and subsequent analysis and application can be performed.

[0043] Based on the above steps, this embodiment provides a gene chip microbead detection method, which realizes image acquisition of the rectangular area image of the target microbead through an array camera, which can improve the image acquisition accuracy while reducing the equipment cost and use cost of image acquisition. The target detection model using deep learning can learn the various geometric forms of microbeads, accurately locate them, and directly judge the brightness of the microbeads based on the pixel value of the microbead position and the pixels in the adjacent area, which has a higher accuracy rate than extracting the pixel value features of the microbeads to judge the brightness. It can also provide more intuitive detection results, which is convenient for the subsequent use and detection display of the gene chip microbead detection results.

[0044] According to a specific implementation of the embodiment of the present application, Figure 2 As shown, the target microbead rectangular area image is obtained, including: S201, dividing the target microbead rectangular area into multiple sub-areas to be photographed, wherein adjacent sub-areas to be photographed include a preset overlapping area, and the target microbead rectangular area includes a first side and a second side, and the length of the first side is greater than the length of the second side.

[0045] S202, acquiring an image of the sub-area to be photographed to obtain a plurality of sub-images.

[0046] S203, performing a preset stitching process on the multiple sub-images to obtain a target microbead rectangular area image.

[0047] In this embodiment, it is assumed that the camera's field of view is a rectangle. ,like Fig. 9 As shown, the rectangle The two sides of and short side In this embodiment, the first side is the long side , the second side is the short side .

[0048] In the process of area division, the long side is divided into Columns, divide the short side into equal intervals It should be noted that, when performing region division in this embodiment, the target microbead rectangular region is divided into a plurality of sub-regions to be photographed having the same size. In this embodiment, the two sides corresponding to the sub-regions to be photographed are and .

[0049] like Fig. 9 As shown, the adjacent sub-areas to be photographed include a preset overlapping area, and the preset overlapping area includes two types, namely, a first overlapping area and a second overlapping area, wherein the two sides corresponding to the first overlapping area are and The two edges corresponding to the second overlapping area are and .

[0050] In actual application, the calculation formula for the number of rows and columns of area division is: in, is the number of columns, is the number of rows, is the length of the long side of the target microbead rectangular area, is the short side length of the target microbead rectangular area, is the length of the long side of the sub-area to be photographed, is the short side length of the sub-area to be photographed, is the length of the preset overlapping area in the long direction of the target microbead rectangular area, It is the length of the preset overlapping area in the short side direction of the target microbead rectangular area.

[0051] After the sub-regions to be photographed are divided, the image acquisition device can be controlled to acquire images of the sub-regions to be photographed to obtain multiple sub-images. The image acquisition device can use an array camera or other photographic device that can capture sub-images of each sub-region to be photographed, and the image acquisition device can be replaced according to the needs of the actual application scenario. After acquiring multiple sub-images, the images are spliced ​​based on the similar parts between the multiple sub-images to obtain a target microbead rectangular area image including a complete target microbead rectangular area.

[0052] Based on the above steps, the gene chip microbead detection method provided in this embodiment uses an array camera to shoot the sub-areas to be photographed one by one, which can reduce the requirements for the flatness of the gene chip surface and the requirements for the verticality between the image acquisition device and the chip plane. The light source can also be replaced with a low-cost and long-life LED light source.

[0053] According to a specific implementation of the embodiment of the present application, Figure 3 As shown, multiple sub-images are spliced ​​in a preset manner to obtain a rectangular area image of a target microbead, including: S301 : For two adjacent sub-images, a first adjacent region of one sub-image is determined as a template region, and a second adjacent region of the other sub-image is determined as a matched region.

[0054] S302, searching for a position matching the template region in the matched region until a target matching position is obtained, wherein the target matching position is a position where the similarity between the matched region and the template region is the greatest.

[0055] S303, moving another sub-image according to the target matching position and discarding the redundant part of the image until the pixel points of the template area are consistent with the pixel points of the matched area.

[0056] S304, when any two adjacent sub-images are stitched together, determining the stitched image as the target microbead rectangular area image.

[0057] In this embodiment, during the stitching process, any two adjacent sub-images may be stitched together in a template matching manner.

[0058] like Fig.10 As shown in part (A) of , images of two adjacent sub-areas to be photographed are captured to obtain a first sub-image sub1 and a second sub-image sub2.

[0059] like Fig.10 As shown in part (B) of , the right sub-image of the first sub-image sub1 (the right box part) can be taken as the matched area, where the width of the right sub-image of the first sub-image sub1 is , Gao Wei and > ,in, is the standard size of the preset overlapping area. Take the left sub-image of the second sub-image sub2 (the left box part) as the template area, where the width of the left sub-image of the second sub-image sub2 is , Gao Wei ,and , .

[0060] Based on the template area, search and match in the matched area, and search for all the sizes of width and , Gao Wei The similarity between the searched area and the template area is calculated according to the following formula: in, is the similarity, and is the pixel coordinate in the template area, and is the pixel coordinate of the matched area. In actual application, the target matching position ( ) is the similarity The largest location, Indicates that the convolution kernel is locally offset ( , ), Indicates the matched image at coordinates The pixel value of .

[0061] Target matching position ( ) in the first sub-image sub1 corresponds to the upper left corner of the template area in the second sub-image sub2. Based on the above correspondence, the second sub-image sub2 is moved and spliced ​​with the first sub-image sub1 so that each pixel in the overlapping area is basically consistent, that is, Fig.10 As shown in the image (C) of FIG. It should be noted that the movement operation of the sub-image corresponding to the template area can be left-right movement or up-down movement, and the specific splicing movement operation can be set according to the needs of the actual application scenario.

[0062] like Fig.10 As shown in the image (C), after the stitching process of the two sub-images is completed, the image with consistent pixels within the dotted frame can be retained, and the image in the redundant area (i.e., the image outside the dotted frame) can be discarded to obtain a complete stitched image.

[0063] Based on the above steps, the multiple sub-images obtained by dividing the regions and capturing the sub-regions in the above steps can be stitched together one by one using the above two-by-two stitching method, and finally an accurate rectangular region image of the target microbeads can be obtained.

[0064] According to a specific implementation of the embodiment of the present application, Figure 4 As shown, before using the target detection model to process the target microbead rectangular area image and obtain microbead related parameters, the method also includes: S401, acquiring different gene chip images, wherein the gene chip images include microbeads.

[0065] S402, annotating the gene chip image with microbead-related parameters.

[0066] S403, according to the annotated gene chip image, a target detection network or a segmentation network is used to perform model training to obtain a target detection model.

[0067] In this embodiment, by collecting images of different gene chips collected on various instruments, sufficient sample images are prepared for deep learning. Among them, the image parameters such as brightness and resolution of different images can also be different. By collecting a large number of images of gene chips with different image parameters, this embodiment can improve the prediction accuracy of the subsequent deep learning framework model.

[0068] After collecting enough sample images, the sample images are labeled, such as Fig.11 As shown, a rectangular frame, a center point or a circular frame is used for labeling, wherein the labeling information includes the center point coordinates, the classification of bright beads or dark beads, and the grayscale mean of the beads. It should be noted that during the labeling process, the center point coordinates can be directly calculated during the labeling process. Fig.11 As shown, deep learning detects the center point of the microbead and determines the light and dark classification of the microbead. The black dots represent the microbead, and the square box represents the rectangular box where the detection model predicts the position of the microbead.

[0069] Use target detection or segmentation network for model training. It should be noted that the steps of completing model training based on deep learning can refer to the deep learning training technology in the relevant technology. The target detection model trained in this embodiment can locate the microbeads in the picture containing microbeads and output the microbead position and light and dark classification results.

[0070] Based on the above steps, the gene chip microbead detection method proposed in this embodiment constructs a target detection model through deep learning, and then completes microbead positioning and brightness judgment based on the target detection model. The parallel detection is efficient and fast, and there is no position requirement for image acquisition of the rectangular area of ​​the target microbead. It can reduce the cost of microbead detection while improving the efficiency of microbead detection.

[0071] According to a specific implementation of the embodiment of the present application, Figure 5 As shown, a target matrix is ​​constructed according to microbead related parameters and the target microbead rectangular area to obtain microbead detection results, including: S501, obtaining a target affine matrix according to the position coordinates of the target microbead rectangular area, wherein the distribution state of the target affine matrix is ​​a horizontal and vertical state.

[0072] S502, constructing a virtual matrix according to the microbead distribution in the target microbead rectangular area, wherein the number of rows and columns of the virtual matrix corresponds to the number of microbeads in the microbead distribution in the row direction and the column direction.

[0073] S503, combining the target affine matrix and the virtual matrix, mapping the microbead related parameters to each element of the virtual matrix, and obtaining the microbead detection result, wherein one element corresponds to one microbead.

[0074] In this embodiment, during the detection of microbead-related parameters in the target microbead rectangular area, there is no need to adjust the placement position of the gene chip, and the placement position of the gene chip is not required to be horizontal or vertical. Therefore, the collected target microbead rectangular area may appear as follows: Fig.12 The skewed state is shown in the image of part (B). The target affine matrix is ​​calculated according to the position coordinates of the target microbead rectangular area. The target affine matrix is ​​as follows: Fig.12 As shown in the image of part (C), it can effectively cooperate with the virtual matrix to complete the mapping of microbead related parameters and obtain more intuitive microbead detection results.

[0075] According to a specific implementation of the embodiment of the present application, Figure 6 As shown, the target affine matrix is ​​obtained according to the position coordinates of the target microbead rectangular area, including: S601, extracting four boundary lines of the target microbead rectangular area according to a preset iterative fitting algorithm.

[0076] S602, connecting four boundary lines to obtain an edge matrix.

[0077] S603, performing affine transformation on the edge matrix to obtain a target affine matrix.

[0078] In this embodiment, if Fig.12 As shown in (A), (B) and (C), the four boundary lines of the target microbead rectangular area are first fitted according to the preset iterative fitting algorithm, where the intersection points of the four boundary lines are A, B, C and D. The four boundary lines are connected to obtain the edge matrix ABCD. Then, the edge matrix ABCD is affine transformed to obtain the target affine matrix.

[0079] In this embodiment, the specific steps of the preset iterative fitting algorithm include: First, extract the microbead region The minimum coordinate of the center point of the leftmost microbead .

[0080] Second, perform initialization processing to screen all the micro-bead point coordinates detected by the target detection model and screen out those that meet the requirements. The center point coordinates of ,in, > 0. Set the maximum number of iterations N, the distance threshold d, and the minimum number of inliers k.

[0081] Third, start the iterative process. Randomly select two points from all data points as sample points; calculate the equation of a straight line based on these two sample points; calculate the distance from all data points to the straight line, and if the distance is less than the threshold d, mark the point as an inlier; if the size of the current inlier set is larger than the best inlier set previously recorded, update the best model and its corresponding inlier set.

[0082] Fourth, determine whether the termination condition is met. When the maximum number of iterations N is reached or enough internal points are found (the number reaches k), the iteration stops.

[0083] Fifth, output the result. Use the internal point set corresponding to the best model to refit the straight line and get the final straight line equation ax+b=0.

[0084] Sixth, iterate again and reduce The value of is initialized repeatedly and a straight line fitting is performed. If the variation range of a and b is less than the preset threshold δ for 2 or more times in a row, the iteration is terminated, and the last straight line equation is output to obtain the corresponding edge line.

[0085] Based on the above steps, four boundary lines of the target microbead rectangular area can be extracted, including the upper boundary line, the lower boundary line, the left boundary line and the right boundary line. The four vertices of the four boundary lines are calculated, such as Fig.12 As shown, an upper boundary line AB, a lower boundary line DC, a left boundary line AD and a right boundary line BC can be obtained.

[0086] Use affine transformation to map the ABCD quadrilateral to a standard rectangle , and calculate the target affine matrix .

[0087] Based on the above steps, assuming that the photographed area is n times larger than a single field of view, the distortion effect will be magnified n times, which will cause microbead positioning deviation. Affine transformation mapping to a standard rectangle can correct the optical distortion caused by the tilt of the chip and reduce positioning deviation.

[0088] According to a specific implementation of the embodiment of the present application, Figure 7 As shown, the microbead related parameters are mapped to each element of the virtual matrix to obtain the microbead detection results, including: S701, initializing a virtual matrix, and filling a target element with preset content, wherein the target element is an element in the virtual matrix corresponding to a microbead position in a target affine matrix; S702, traversing the unupdated elements in the virtual matrix, and mapping and filling corresponding micro-bead related parameters for each unupdated element, wherein the unupdated elements are filled with preset content.

[0089] In this embodiment, it is assumed that the microbeads are arranged in OK, Columns (numbered starting from 1), arranged as follows Figure 8 (A) As shown in the image, the construction OK, The virtual matrix IM of the column, the storage content of each position can be expressed as , which includes light and dark classification , center point coordinates , gray value of beads .

[0090] In this embodiment, the virtual matrix mapping step includes: Initialize the virtual matrix and fill it with elements whose sum of row and column numbers is an even number. .

[0091] Remember the standard rectangle The coordinates of the four vertices are , assuming that the microbeads are The center coordinates in the matrix area are ( ), whose brightness and darkness are classified as , the gray value is , then this center point is mapped to the standard rectangle The coordinates of the coordinate system are: in, For microbeads The center coordinates in the matrix region, For microbeads The transpose of the center coordinates in the matrix region, is the target affine matrix.

[0092] The calculation method corresponding to the row and column coordinates of the virtual matrix includes: in, is the row number, is the column number, and A standard rectangle The row and column coordinates in and are the vertex coordinates in the region, is the column value of the virtual matrix IM, is the row value of the virtual matrix IM.

[0093] The virtual matrix IM Line The columns are filled with information such as .

[0094] The virtual matrix IM with some element information updated is shown in Table 1.

[0095] Table 1 Traverse all the elements that have not been updated, that is, traverse each element in the virtual matrix IM according to the row and column numbers, and find the coordinates of the center point elements, that is, elements that have not been updated, and add microbead related parameters for each element that has not been updated.

[0096] In one embodiment, if the second element (microbead center coordinate) of the i-th row and j-th column of the virtual matrix IM is is (-1,-1), then the center point coordinates ( ) is updated to , and then update the i-th row and j-th column of the virtual matrix IM as [0, ( ), ( )].

[0097] In summary, this embodiment provides a gene chip microbead detection method, which realizes image acquisition of the rectangular area image of the target microbead through an array camera, which can improve the image acquisition accuracy while reducing the equipment cost and use cost of image acquisition. The target detection model using deep learning can learn the various geometric forms of microbeads, accurately locate them, and directly judge the brightness of the microbeads based on the pixel value of the microbead position and the pixels in the adjacent area, which has a higher accuracy rate than extracting the pixel value features of the microbeads to judge the brightness. It can also provide more intuitive detection results, which is convenient for the subsequent use and detection display of the gene chip microbead detection results.

[0098] Based on the same inventive concept, the present application embodiment also provides a gene chip microbead detection device for implementing the gene chip microbead detection method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more gene chip microbead detection device embodiments provided below can refer to the limitations of the gene chip microbead detection method above, and will not be repeated here.

[0099] In one embodiment, reference Fig.13 This embodiment also provides a gene chip microbead detection device 1300, including: an acquisition module 1310, a detection module 1320 and an output module 1330, wherein: An acquisition module 1310 is used to acquire a target microbead rectangular region image, wherein the target microbead rectangular region image is formed by splicing a plurality of sub-images, and the sub-images cover a portion of the target microbead rectangular region; A detection module 1320 is used to process the target microbead rectangular area image using a target detection model to obtain microbead related parameters, wherein the target detection model is a model obtained based on deep learning training, and the microbead related parameters include microbead position, microbead light and dark classification results, and microbead area grayscale mean; The output module 1330 is used to construct a target matrix according to the microbead related parameters and the target microbead rectangular area to obtain the microbead detection results, wherein the target matrix stores the microbead related parameters according to the arrangement of the microbeads in the target microbead rectangular area.

[0100] In one embodiment, the acquisition module 1310 is specifically used to divide the target microbead rectangular area into regions to obtain multiple sub-regions to be photographed, wherein adjacent sub-regions to be photographed include preset overlapping regions, the target microbead rectangular area includes a first side and a second side, and the length of the first side is greater than the length of the second side; acquire images of the sub-regions to be photographed to obtain multiple sub-images; and perform preset splicing processing on the multiple sub-images to obtain images of the target microbead rectangular area.

[0101] In one embodiment, the acquisition module 1310 is specifically used to determine, for two adjacent sub-images, a first adjacent area of ​​one sub-image as a template area, and a second adjacent area of ​​the other sub-image as a matched area; search for a position that matches the template area in the matched area until a target matching position is obtained, wherein the target matching position is a position where the matched area has the greatest similarity with the template area; move the other sub-image according to the target matching position, and discard redundant parts of the image until the pixel points of the template area and the matched area are consistent; when any two adjacent sub-images are spliced, determine the spliced ​​image as the target microbead rectangular area image.

[0102] In one embodiment, the detection module 1320 is specifically used to obtain different gene chip images, wherein the gene chip images include microbeads; annotate the gene chip images with microbead-related parameters; and perform model training using a target detection network or a segmentation network based on the annotated gene chip images to obtain a target detection model.

[0103] In one embodiment, the output module 1330 is specifically used to obtain a target affine matrix based on the position coordinates of the target microbead rectangular area, wherein the distribution state of the target affine matrix is ​​a horizontal and vertical state; construct a virtual matrix according to the microbead distribution situation in the target microbead rectangular area, wherein the number of rows and columns of the virtual matrix corresponds to the number of microbeads in the row direction and column direction in the microbead distribution situation; combine the target affine matrix and the virtual matrix, map the microbead related parameters to each element of the virtual matrix, and obtain the microbead detection results, wherein one element corresponds to one microbead.

[0104] In one embodiment, the output module 1330 is specifically used to extract four boundary lines of the target microbead rectangular area according to a preset iterative fitting algorithm; connect the four boundary lines to obtain an edge matrix; and perform an affine transformation on the edge matrix to obtain a target affine matrix.

[0105] In one embodiment, the output module 1330 is specifically used to initialize the virtual matrix, fill the preset content in the target element, wherein the target element is the element in the virtual matrix corresponding to the microbead position in the target affine matrix; traverse the unupdated elements in the virtual matrix, and fill the corresponding microbead-related parameters for each unupdated element mapping, wherein the unupdated elements are filled with the preset content.

[0106] In summary, this embodiment provides a gene chip microbead detection device, which realizes image acquisition of the rectangular area image of the target microbead through an array camera, which can improve the image acquisition accuracy while reducing the equipment cost and use cost of image acquisition. The target detection model using deep learning can learn the various geometric forms of microbeads, accurately locate them, and directly judge the brightness of the microbeads based on the pixel value of the microbead position and the pixels in the adjacent area, which has a higher accuracy rate than extracting the pixel value features of the microbeads to judge the brightness. It can also provide more intuitive detection results, which is convenient for the subsequent use and detection display of the gene chip microbead detection results.

[0107] Fig.13 The specific implementation of the device shown can refer to the specific implementation of the aforementioned method embodiment, which will not be repeated here.

[0108] In addition, an embodiment of the present application further provides an electronic device, the electronic device comprising: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the gene chip microbead detection method in the aforementioned method embodiment.

[0109] The embodiment of the present application also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the gene chip microbead detection method in the aforementioned method embodiment.

[0110] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. Examples of computer-readable storage media may be, but are not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0111] In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0112] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0113] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains a target microbead rectangular area image, wherein the target microbead rectangular area image is formed by splicing multiple sub-images, and the sub-images cover a partial area of ​​the target microbead rectangular area; uses a target detection model to process the target microbead rectangular area image to obtain microbead related parameters, wherein the target detection model is a model obtained based on deep learning training, and the microbead related parameters include microbead positions, microbead light and dark classification results, and microbead area grayscale mean; constructs a target matrix based on the microbead related parameters and the target microbead rectangular area to obtain microbead detection results, wherein the target matrix stores the microbead related parameters according to the arrangement of the microbeads in the target microbead rectangular area.

[0114] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0115] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0116] The units involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of a unit does not, in some cases, constitute a limitation on the unit itself.

[0117] It should be understood that various parts of the present disclosure may be implemented in hardware, software, firmware, or a combination thereof.

[0118] The above is only a specific implementation of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present disclosure should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A gene chip microbead detection method, characterized in that: include: Acquire a target microbead rectangular region image, wherein the target microbead rectangular region image is formed by splicing a plurality of sub-images, and the sub-images cover a portion of the target microbead rectangular region; Processing the target microbead rectangular area image using a target detection model to obtain microbead related parameters, wherein the target detection model is a model obtained based on deep learning training, and the microbead related parameters include microbead position, microbead light and dark classification results, and microbead area grayscale mean; A target matrix is ​​constructed according to the microbead-related parameters and the target microbead rectangular area to obtain a microbead detection result, wherein the target matrix stores the microbead-related parameters according to the arrangement of the microbeads in the target microbead rectangular area.

2. The method according to claim 1, characterized in that The step of obtaining a rectangular area image of a target microbead comprises: Dividing the target microbead rectangular area into multiple sub-areas to be photographed, wherein adjacent sub-areas to be photographed include a preset overlapping area, and the target microbead rectangular area includes a first side and a second side, and the length of the first side is greater than the length of the second side; Acquire an image of the sub-area to be photographed to obtain a plurality of sub-images; The plurality of sub-images are subjected to a preset stitching process to obtain the rectangular region image of the target microbeads.

3. The method according to claim 2, characterized in that The step of performing a preset stitching process on the plurality of sub-images to obtain the target microbead rectangular area image comprises: For two adjacent sub-images, a first adjacent region of one sub-image is determined as a template region, and a second adjacent region of the other sub-image is determined as a matched region; Searching for a position matching the template region in the matched region until a target matching position is obtained, wherein the target matching position is a position where the similarity between the matched region and the template region is the greatest; Moving the other sub-image according to the target matching position and discarding redundant partial images until the pixel points of the template area are consistent with those of the matched area; When any two adjacent sub-images are stitched together, the stitched image is determined as the target microbead rectangular area image.

4. The method according to claim 1, characterized in that Before using the target detection model to process the target microbead rectangular area image to obtain microbead related parameters, the method further includes: Acquiring different gene chip images, wherein the gene chip images include microbeads; Annotating the gene chip image with microbead-related parameters; According to the annotated gene chip image, a target detection network or a segmentation network is used to perform model training to obtain the target detection model.

5. The method according to claim 1, characterized in that The step of constructing a target matrix according to the microbead related parameters and the target microbead rectangular area to obtain microbead detection results includes: Acquire a target affine matrix according to the position coordinates of the target microbead rectangular area, wherein the distribution state of the target affine matrix is ​​a horizontal and vertical state; Constructing a virtual matrix according to the microbead distribution in the target microbead rectangular area, wherein the number of rows and columns of the virtual matrix corresponds to the number of microbeads in the microbead distribution in the row direction and the column direction; The target affine matrix and the virtual matrix are combined to map the microbead related parameters to each element of the virtual matrix to obtain the microbead detection result, wherein one element corresponds to one microbead.

6. The method according to claim 5, characterized in that The step of obtaining a target affine matrix according to the position coordinates of the target microbead rectangular area includes: Extracting four boundary lines of the target microbead rectangular area according to a preset iterative fitting algorithm; Connecting the four boundary lines to obtain an edge matrix; Performing an affine transformation on the edge matrix to obtain the target affine matrix.

7. The method according to claim 5, characterized in that Mapping the microbead related parameters to each element of the virtual matrix to obtain the microbead detection result includes: Initializing the virtual matrix and filling the target element with preset content, wherein the target element is the element in the virtual matrix corresponding to the position of the microbead in the target affine matrix; The unupdated elements in the virtual matrix are traversed, and the corresponding micro-bead related parameters are filled in for each unupdated element mapping, wherein the unupdated elements are filled with the preset content.

8. A gene chip microbead detection device, characterized in that: include: An acquisition module, used for acquiring a rectangular area image of a target microbead, wherein the rectangular area image of the target microbead is formed by splicing a plurality of sub-images, and the sub-images cover a part of the rectangular area of ​​the target microbead; A detection module, used to process the target microbead rectangular area image using a target detection model to obtain microbead related parameters, wherein the target detection model is a model obtained based on deep learning training, and the microbead related parameters include microbead position, microbead light and dark classification results, and microbead area grayscale mean; The output module is used to construct a target matrix according to the microbead related parameters and the target microbead rectangular area to obtain the microbead detection results, wherein the target matrix stores the microbead related parameters according to the arrangement of the microbeads in the target microbead rectangular area.

9. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the gene chip microbead detection method according to any one of the preceding claims 1-7.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the gene chip microbead detection method described in any one of claims 1 to 7.

Images