WO2020211403A1 - 电泳图的识别方法、装置、设备及存储介质 - Google Patents

电泳图的识别方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2020211403A1
WO2020211403A1 PCT/CN2019/123944 CN2019123944W WO2020211403A1 WO 2020211403 A1 WO2020211403 A1 WO 2020211403A1 CN 2019123944 W CN2019123944 W CN 2019123944W WO 2020211403 A1 WO2020211403 A1 WO 2020211403A1
Authority
WO
WIPO (PCT)
Prior art keywords
column
target
image
maximum pixel
pixel values
Prior art date
Application number
PCT/CN2019/123944
Other languages
English (en)
French (fr)
Inventor
赵文妍
丁砚书
段广有
闵文波
葛毅
廖国娟
Original Assignee
苏州金唯智生物科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 苏州金唯智生物科技有限公司 filed Critical 苏州金唯智生物科技有限公司
Publication of WO2020211403A1 publication Critical patent/WO2020211403A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning

Definitions

  • the present disclosure relates to the field of electrophoresis technology, for example, to a method, device, equipment, and storage medium for identifying electrophoresis diagrams.
  • Electrophoresis refers to the phenomenon that colloidal particles move towards the cathode or anode in the dispersion medium under the action of an external DC power supply.
  • the technique that uses electrophoresis to separate substances is called electrophoresis.
  • Electrophoresis technology has been widely used in the field of biochemistry.
  • electrophoresis technology a mixture sample is divided into strips on a support medium, and the recorded spectrum obtained by scanning on a densitometer is called an electrophoresis. According to the shape and color of the bands in the electrophoresis graph, the gel running situation of the mixture sample can be determined, and then the sample amount of the mixture sample in the subsequent test can be determined.
  • the identification and judgment of the gel image in the electrophoresis diagram is achieved manually, that is, the staff judges the gel image in the electrophoresis diagram based on usual experience. Because manual judgment and recognition are subject to a certain degree, As a result, there is a big difference in the recognition results of the gel image in the electrophoresis diagram.
  • the present disclosure provides a method, device, equipment and storage medium for identifying electrophoresis diagrams, so as to solve the problem of discrepancies in the recognition results of the electrophoretic diagrams.
  • the embodiment of the present invention provides a method for recognizing an electropherogram, including:
  • Each target pixel matrix is input into a pre-trained prediction model for prediction, so as to obtain the target sample addition amount of the sample.
  • the embodiment of the present invention also provides an electropherogram recognition device, including:
  • An identification module configured to obtain an electropherogram of the sample, and identify a frame of the gel image in the electrophoresis image, wherein the gel image includes at least one band image;
  • An extraction module configured to extract a preset number of pixel values in each of the striped images to form a target pixel matrix of each of the striped images
  • the input module is configured to input the target pixel matrix into a pre-trained prediction model for prediction, so as to obtain the target sample addition amount of the sample.
  • the embodiment of the present invention also provides a device, which includes:
  • One or more processors are One or more processors;
  • Memory used to store one or more programs
  • the one or more processors When the one or more programs are executed by the one or more processors, the one or more processors implement the electropherogram identification method described above.
  • the embodiment of the present invention also provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the electropherogram identification method described above is realized.
  • the technical solution of this embodiment uses image recognition technology to identify the electrophoresis pattern of the sample, and forms the target pixel matrix of the striped image, which is input to the pre-trained prediction model, and the prediction model outputs the target addition of the sample in subsequent experiments. Sample size. This can make the recognition result of the electropherogram have a unified standard, and avoid the problem of the difference in the recognition result of the gel image in the electropherogram due to the certain subjectivity of manual judgment and recognition.
  • FIG. 1 is a flowchart of a method for identifying electrophoresis diagrams according to Embodiment 1 of the present invention
  • Embodiment 2 is a flowchart of a method for identifying electrophoresis diagrams provided by Embodiment 2 of the present invention
  • FIG. 3 is a schematic structural diagram of an electrophoresis diagram identification device provided by Embodiment 3 of the present invention.
  • FIG. 4 is a schematic structural diagram of a device according to Embodiment 4 of the present invention.
  • Figure 1 is a flow chart of the method for identifying electrophoresis diagrams provided in the first embodiment of the present invention. This embodiment can be applied to identify electrophoresis diagrams and determine the target sample amount in subsequent tests of samples. Identify the device to perform.
  • electrophoresis is a common analysis method. Charged molecules and particles migrate in the separation medium.
  • the separation medium is subjected to an electric field between two electrodes. Isoelectric point (Isoelectric point, pI), molecular weight, charge, or a combination of these factors can be used to separate proteins.
  • the separation medium is usually a gel.
  • the separation gel is usually placed on a support, and the two opposite ends of the separation gel are in contact with the electrode buffer in solution or rigid form.
  • the electrode can be inserted into a container containing the electrode buffer.
  • the buffer solution from the electrolyte and ion storage keeps the pH (pH) value and other parameters constant.
  • the molecules are detected and identified in different ways. For example, by visually detecting and identifying the color of the gel, or using optical means, such as scanning the colored gel or labeled samples with a laser scanner, or imaging them.
  • the identification method of the electrophoresis diagram in this embodiment is to identify and judge the gel image scanned by the laser scanner.
  • the method for identifying an electropherogram includes the following steps:
  • gel electrophoresis is usually used to separate biomolecules, such as proteins, peptides, nucleic acids, and so on.
  • the sample can be a protein sample, peptide or nucleic acid sample, etc.
  • the sample can be a human body, mammalian tissue, cell lysis or bacteria, insect or yeast cell system, etc. It should be noted that in this embodiment, only samples are described, but not limited.
  • the electropherogram of the sample can be understood as the spectrum obtained by electrophoresis after the current sample is added to the gel.
  • the frame of the rubber image can be understood as the frame of the rubber image including all the strip images to be recognized.
  • the electropherogram of the current sample can be obtained from the laser scanner.
  • the electropherogram of the current sample can be acquired from the laser scanner.
  • the method for receiving the acquisition instruction can be designed according to the actual situation. For example, after detecting that the electrophoresis is scanned in the laser scanner, the acquisition instruction is generated and received. For another example, after detecting a click operation on the identification device for the electrophoresis graph, an acquisition instruction is generated and received.
  • a combination of the canny edge detection method and the Laplacian edge detection algorithm is used to accurately locate the border of the gel image.
  • the position of the target position in the whole electropherogram will be accurately changed to white, and the remaining positions are defaulted to background pixels, and the background pigments are changed to black.
  • the entire electrophoresis image has only two colors, black and white, and then the target location is expanded and corroded by morphological methods, which can eliminate the black spots inside the target location, accurately locate the object, and locate the border of the gel image.
  • the target position in this embodiment refers to the striped image contained within the frame of the glue image.
  • the striped image is composed of multiple pixels, and each pixel has its corresponding pixel value.
  • each striped image is converted to their corresponding pixel values.
  • each striped image can obtain a matrix composed of pixel values. This matrix composed of pixel values, It is called the original pixel matrix.
  • the maximum value of each column is searched. Assume that the original pixel matrix is a matrix with N rows and M columns, where M and N are both positive integers. Correspondingly, in each original pixel matrix, by looking up the maximum pixel value of each column, the maximum pixel value of M columns can be found. Wherein, the column maximum pixel value refers to the maximum pixel value in each column of the original pixel matrix. Then the maximum pixel values of the M columns are compared, and the maximum value of the maximum pixel values of the M columns is obtained as the target maximum pixel value. Determine the target pixel matrix according to the target maximum pixel value and preset rules.
  • each striped image corresponds to an original pixel matrix and a target pixel matrix respectively. That is, in the electrophoretic image, the number of target pixel matrixes is equal to the number of striped images.
  • the target sample addition amount of the sample can be understood as the sample weight that the current sample needs to increase in the gel in the subsequent test.
  • the strip images are arranged in the electropherogram in a certain order, and multiple target pixel matrices are sequentially input into the pre-trained prediction model according to the sequence of the strip images in the electrophoresis diagram, and the prediction model outputs the current sample Target sample volume.
  • the electropherogram of a sample is obtained, and the frame of the gel image in the electropherogram is recognized, wherein the gel image includes at least one band image, and then each image is extracted.
  • the preset number of pixel values in the striped image form the target pixel matrix of each striped image.
  • each target pixel matrix is input to a pre-trained prediction model for prediction to obtain the target sample amount of the sample.
  • the technical solution of this embodiment uses image recognition technology to identify the electrophoresis pattern of the sample, and forms the target pixel matrix of the striped image, which is input to the pre-trained prediction model, and the prediction model outputs the target of the sample in subsequent experiments Add sample volume. In this way, the recognition result of the electropherogram can have a unified standard, and the problem of the discrepancy of the recognition result of the gel image in the electropherogram due to the certain subjectivity of manual judgment and recognition is avoided.
  • Fig. 2 is a flowchart of the electropherogram identification method provided in the second embodiment of the present invention.
  • the embodiment of the present invention optimizes the electropherogram identification method.
  • the optimized electropherogram The identification method mainly includes the following steps:
  • S210 Identify the gel image area of the electrophoretic image through an edge detection algorithm.
  • a combination of the canny edge detection method and the Laplacian edge detection algorithm is used to accurately locate the border of the gel image.
  • the edge detection algorithm is not limited, and an appropriate edge detection algorithm can be designed and selected according to actual conditions to identify the glue image area.
  • S220 Eliminate the background area in the electrophoresis image and the interfering pixels in the gel image area to obtain at least one primary gel image.
  • the background area in the electrophoresis graph refers to other areas in the electrophoresis graph except for the strip area, and interfering pixels in the strip area can be understood as non-white pixels in the strip area.
  • the positions of the striped regions in the entire electrophoretic graph will be changed to white, and the remaining positions are defaulted to the background area, and the pixels in the background area are changed to black. In this way, there are only two colors of black and white in the whole electrophoresis graph. Then use morphological methods to dilate and corrode the target position, which can eliminate the black spots inside the target position, accurately locate the object, and locate the frame of the glue image.
  • S230 Perform preprocessing on each primary strip image to obtain a primary pixel matrix corresponding to each primary strip image.
  • the electropherogram is adjusted according to the coordinate system to obtain a normalized electropherogram.
  • preprocessing each primary gel image can be understood as de-drying and normalizing each strip image to obtain the primary pixel matrix corresponding to each primary gel image.
  • the primary pixel matrix refers to a pixel matrix composed of pixel values corresponding to pixels in the primary gel image.
  • the mode is the number that appears most frequently in a set of data.
  • the row mode refers to the value that appears most frequently in each row of data in the primary pixel matrix
  • the column mode refers to the value that appears most frequently in each column of data in the primary pixel matrix.
  • the row mode of each row and the column mode of each column of the primary pixel matrix are calculated. It should be noted that, in this embodiment, only the row mode and the column mode are described without limitation, and suitable methods can be adopted for calculation according to actual conditions.
  • S250 Perform correction on each primary gel image according to the comparison result of the row mode and the background mode and the comparison result of the column mode and the background mode to obtain a frame of the gel image.
  • the background mode refers to the mode of the pixel values of the pixels in the background area.
  • the calculation method of the background mode and the calculation method of the row mode and the column mode can be the same or different, as long as the background mode can be calculated.
  • the morphological method is used to dilate and corrode the target position in the process of determining the primary gel image. Therefore, the frame near the target position will also be determined as the target position. At this time, the target position Make further corrections.
  • the row pixels corresponding to the row mode are eliminated.
  • the column mode is less than or equal to the background mode, it means that the column to which the column mode belongs contains the pixel values of the pixels in the background area, and the column pixels corresponding to the column mode are eliminated. That is, the pixels in the column corresponding to the mode of the column become the same color as the pixels in the background area.
  • the column mode is less than or equal to the background mode, the column pixels corresponding to the column mode are eliminated.
  • the row mode is less than or equal to the background mode, it means that the row to which the row mode belongs contains the pixel values of the pixels in the background area, and the row pixels corresponding to the row mode are eliminated. That is, the pixels in the row corresponding to the mode of the row become the same color as the pixels in the background area.
  • each striped image is converted to their corresponding pixel values.
  • each striped image can obtain a matrix composed of pixel values. This matrix composed of pixel values, It is called the original pixel matrix.
  • S270 Calculate the maximum pixel value of each column of the pixel value in the original pixel matrix to obtain the maximum pixel value of M columns, where M is the number of columns of the original pixel matrix.
  • the maximum pixel value of a column can be understood as the maximum pixel value in each column determined in the original pixel matrix with a column as a unit.
  • find the maximum value of each column Assume that the original pixel matrix is a matrix with N rows and M columns, where M and N are both positive integers.
  • the maximum pixel value of M columns can be found.
  • the maximum pixel values of the M columns are compared, and the maximum value of the maximum pixel values of the M columns is determined as the target maximum pixel value; if the original pixel matrix has a target maximum pixel value, the target maximum The column to which the pixel value belongs and the two adjacent columns to the column to which the target maximum pixel value belongs form a target pixel matrix.
  • the maximum pixel value of the first column is compared with the maximum pixel value of the second column; if the maximum pixel value of the first column is greater than that of the second column Maximum pixel value, the maximum pixel value of the first column is taken as the maximum pixel value of the adjacent column; if the maximum pixel value of the first column is less than the maximum pixel value of the second column, the maximum pixel value of the second column As the maximum pixel value of the adjacent column; the column to which the two same target maximum pixel values belong and the column to which the maximum pixel value of the adjacent column belongs to form a target pixel matrix.
  • the maximum pixel value of the first column can be understood as the maximum pixel value of the column to the left of the first target maximum pixel value
  • the maximum pixel value of the second column can be understood as the maximum pixel value of the column to the right of the second target maximum pixel value.
  • the first target maximum pixel value is located to the left of the second target maximum pixel value.
  • the maximum pixel value of the adjacent column can be understood as the pixel value with the larger value among the maximum pixel value in the first column and the maximum pixel value in the second column.
  • the column to which the first target maximum pixel value belongs is The number of columns between the columns to which the second target maximum pixel value belongs; if the number of columns between the columns to which the first target maximum pixel value belongs and the column to which the second target maximum pixel value belongs is 1, then the first The column to which the target maximum pixel value belongs, the column to which the second target maximum pixel value belongs, and the interval column between the columns to which two identical target maximum pixel values belong form a target pixel matrix.
  • the column continuity length is the continuous existence of the column to which the target maximum pixel value belongs. The number of columns; in each column with the largest continuous length of the column, take any 3 consecutive columns to which the target maximum pixel value belongs to form a target pixel matrix.
  • the electropherogram identification method obtained by the embodiment of the present invention obtains the electropherogram of the sample and recognizes the border of the gel image in the electropherogram, and then extracts a preset number of pixel values in each strip image to form each The target pixel matrix of the striped image, and finally each target pixel matrix is input into a pre-trained prediction model in a preset order for prediction, so as to obtain the target sample amount of the sample.
  • the technical solution of this embodiment uses the image recognition technology to recognize the electrophoresis map, and forms the target pixel matrix of the strip image, which is input into the pre-trained prediction model, and the output sample of the prediction model is targeted for sample addition in subsequent experiments the amount. In this way, the recognition result of the electropherogram can have a unified standard, and the problem of the discrepancy of the recognition result of the gel image in the electropherogram due to the certain subjectivity of manual judgment and recognition is avoided.
  • this embodiment provides a method for recognizing a polymerase chain reaction (Polymerase Chain Reaction, PCR) electrophoresis pattern.
  • PCR Polymerase Chain Reaction
  • the method further includes: extracting the median of each column of the strip pixel matrix corresponding to each strip image in the gel image, wherein the gel image includes standard strip images and non-standard strip images.
  • Strip image determine each non-standard strip according to the median of each column of the strip pixel matrix corresponding to the standard strip image and the median of each column of the strip pixel matrix corresponding to the non-standard strip image The position of the image in the gel image.
  • Fig. 3 is a schematic structural diagram of an electropherogram identification device provided in the third embodiment of the present invention. This embodiment can be applied to identify the electropherogram and determine the target sample amount of the sample in the subsequent experiment, as shown in Fig. 3,
  • the electropherogram recognition device provided in this embodiment mainly includes: a recognition module 310, an extraction module 320, and an input module 330.
  • the identification module 310 is configured to obtain the electrophoresis image of the sample, and identify the frame of the gel image in the electrophoresis image, wherein the gel image includes at least one strip image;
  • the extraction module 320 is configured to extract a preset number of pixel values in each striped image to form a target pixel matrix of each striped image;
  • the input module 330 is configured to input each target pixel matrix into a pre-trained prediction model for prediction, so as to obtain the target sample addition amount of the sample.
  • the electropherogram recognition device recognizes the frame of the gel image in the electrophoresis image by acquiring the electrophoresis image of the sample, wherein the gel image includes at least one band image, and then extracts each band
  • the preset number of pixel values in the image constitute the target pixel matrix of each striped image, and finally each target pixel matrix is input into a pre-trained prediction model for prediction to obtain the target sample amount of the current sample.
  • the technical solution of this embodiment uses image recognition technology to identify the electrophoresis pattern of the sample, and forms the target pixel matrix of the striped image, which is input to the pre-trained prediction model, and the prediction model outputs the target of the sample in subsequent experiments Add sample volume. In this way, the recognition result of the electropherogram can have a unified standard, and the problem of the discrepancy of the recognition result of the gel image in the electropherogram due to the certain subjectivity of manual judgment and recognition is avoided.
  • the identification module 310 includes:
  • a recognition unit configured to recognize the gel image area of the electrophoresis image through an edge detection algorithm
  • the elimination unit is configured to eliminate the background area in the electrophoresis image and the interfering pixels in the gel image area to obtain at least one primary gel image.
  • the identification module 310 further includes:
  • a preprocessing unit configured to preprocess each of the primary strip images to obtain a primary pixel matrix corresponding to each of the primary strip images
  • the first calculation unit is configured to calculate the row mode and the column mode in each of the primary pixel matrixes
  • a correction unit configured to correct each of the primary strip images according to the comparison result of the row mode and the background mode and the comparison result of the column mode and the background mode to obtain the gum image image Border.
  • the correction unit is configured to eliminate the row pixels corresponding to the row mode if the row mode is less than or equal to the background mode; if the column mode is less than or equal to the background mode; If the background mode is described, the column pixels corresponding to the column mode are eliminated.
  • the device further includes:
  • the median has an extraction module, which is configured to extract the median of each column of the strip pixel matrix corresponding to each strip image in the gel image, wherein the gel image includes a standard strip image and a non-standard strip image.
  • the strip image determination module is configured to determine the median of each column of the strip pixel matrix corresponding to the standard strip image and the median of each column of the strip pixel matrix corresponding to the non-standard strip image The position of each non-standard strip image in the gel image.
  • the extraction module 320 includes:
  • An extraction unit configured to extract, for each of the striped images, pixel values in the striped images to form an original pixel matrix
  • the second calculation unit is configured to calculate the maximum pixel value of each column of pixel values in the original pixel matrix to obtain the maximum pixel values of M columns, where M is the number of columns of the original pixel matrix;
  • the acquiring unit is configured to acquire a preset number of pixel values to form a target pixel matrix of the striped image according to the comparison result of the maximum pixel values of the M columns.
  • the acquiring unit is configured to compare the maximum pixel values of the M columns, and determine the maximum value of the maximum pixel values of the M columns as the target maximum pixel value; if the original pixel matrix If there is a target maximum pixel value, the column to which the target maximum pixel value belongs and two columns adjacent to the column to which the target maximum pixel value belongs are formed into a target pixel matrix.
  • the acquiring unit is configured to: if the original pixel matrix only has two identical target maximum pixel values and the two identical target maximum pixel values belong to adjacent columns, or the original pixel matrix If the number of target maximum pixel values is greater than or equal to 3 and at least two identical target maximum pixel values belong to adjacent columns, the maximum pixel value in the first column is compared with the maximum pixel value in the second column; if the first column is The maximum pixel value of one column is greater than the maximum pixel value of the second column, then the maximum pixel value of the first column is taken as the maximum pixel value of the adjacent column; if the maximum pixel value of the first column is less than the maximum pixel value of the second column Value, the maximum pixel value of the second column is taken as the maximum pixel value of the adjacent column; the column to which two identical target maximum pixel values belong and the column to which the maximum pixel value of the adjacent column belongs to form a target pixel matrix.
  • the acquiring unit is configured to determine that if the original pixel matrix only has two identical target maximum pixel values and the columns to which the two identical target maximum pixel values belong are not adjacent, then determine the first target maximum pixel value.
  • the acquiring unit is configured to determine the continuous length of the column if the number of target maximum pixel values in the original pixel matrix is greater than or equal to 3 and at least 3 target maximum pixel values belong to the continuous column, wherein
  • the column continuity length is the number of columns continuously existing in the column to which the target maximum pixel value belongs; in each column with the largest continuous length of the column, any three consecutive columns to which the target maximum pixel value belongs are taken to form the target pixel matrix.
  • the electropherogram identification device provided by the embodiment of the present invention can execute the electropherogram identification method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for the execution method.
  • FIG. 4 is a schematic structural diagram of a device provided by Embodiment 4 of the present invention.
  • the device includes a processor 410, a memory 420, an input device 430, and an output device 440; the number of processors 410 in the device may be One or more, one processor 410 is taken as an example in FIG. 4; the processor 410, the memory 420, the input device 430, and the output device 440 in the device may be connected by a bus or other means.
  • a bus connection is taken as an example .
  • the memory 420 can be used to store software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the electropherogram identification method in the embodiment of the present invention (for example, the electropherogram identification device
  • the processor 410 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 420, that is, realizes the above-mentioned electropherogram identification method.
  • the memory 420 may mainly include a program storage area and a data storage area.
  • the program storage area may store an operating system and an application program required by at least one function; the data storage area may store data created according to the use of the terminal, etc.
  • the memory 420 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices.
  • the memory 420 may further include a memory remotely provided with respect to the processor 410, and these remote memories may be connected to the device through a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
  • the input device 430 may be used to receive inputted numeric or character information, and generate key signal input related to user settings and function control of the device.
  • the output device 440 may include a display device such as a display screen.
  • the fifth embodiment of the present invention also provides a storage medium containing computer-executable instructions, when the computer-executable instructions are executed by a computer processor, are used to perform an electropherogram identification method, the method including:
  • Each target pixel matrix is input into a pre-trained prediction model for prediction, so as to obtain the target sample addition amount of the sample.
  • a storage medium containing computer-executable instructions provided by an embodiment of the present invention is not limited to the method operations described above, and can also execute the electropherogram identification method provided by any embodiment of the present invention. Related operations in.
  • the present invention can be realized by software and necessary general-purpose hardware, of course, it can also be realized by hardware, but in many cases the former is a better embodiment.
  • the technical solution of the present invention essentially or the part that contributes to the prior art can be embodied in the form of a software product.
  • the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk.
  • ROM Read-Only Memory
  • RAM Random Access Memory
  • FLASH Flash memory
  • hard disk or optical disk etc., including several instructions to make a computer device (which can be a personal computer) , A server, or a network device, etc.) execute the method described in each embodiment of the present disclosure.

Abstract

一种电泳图的识别方法、装置、设备及存储介质，通过获取样本的电泳图，识别电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像（S110），然后提取每个条带图像中预设数量的像素值，构成每个条带图像的目标像素矩阵（S120），最后将每个目标像素矩阵输入预先训练的预测模型进行预测，以得到样本的目标加样量（S130）。

Description

电泳图的识别方法、装置、设备及存储介质
本申请要求在2019年4月15日提交中国专利局、申请号为201910299112.4的中国专利申请的优先权，以上申请的全部内容通过引用结合在本申请中。
技术领域
本公开涉及电泳技术领域，例如涉及一种电泳图的识别方法、装置、设备及存储介质。
背景技术
电泳是指在外加直流电源的作用下，胶体微粒在分散介质里向阴极或阳极作定向移动的现象。利用电泳现象使物质分离的技术叫做电泳技术。
电泳技术在生物化学领域得到了广泛的应用。通过电泳技术将一个混合物样品在支持介质上分成条带，并在光密度计上进行扫描而获得的记录图谱称为电泳图。根据电泳图中条带的形状及颜色可以确定混合物样品的跑胶情况，进而确定该混合物样品在后续试验中的加样量。
然而，目前阶段，对于电泳图中胶图的识别及判断都是通过人工来实现的，即工作人员根据往常的经验判断电泳图中的胶图，由于人工判断与识别带有一定的主观性，导致电泳图中胶图的识别结果存在较大的差异。
发明内容
本公开提供一种电泳图的识别方法、装置、设备及存储介质，以解决电泳图中胶图的识别结果存在差异的问题。
本发明实施例提供了一种电泳图的识别方法，包括：
获取样本的电泳图，识别所述电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像；
提取每个所述条带图像中预设数量的像素值，构成每个所述条带图像的目标像素矩阵；
将每个所述目标像素矩阵输入预先训练好的预测模型进行预测，以得到样本的目标加样量。
本发明实施例还提供了一种电泳图的识别装置，包括：
识别模块，设置为获取样本的电泳图，识别所述电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像；
提取模块，设置为提取每个所述条带图像中预设数量的像素值，构成每个所述条带图像的目标像素矩阵；
输入模块，设置为将所述目标像素矩阵输入预先训练好的预测模型进行预测，以得到所述样本的目标加样量。
本发明实施例还提供了一种设备，所述设备包括：
一个或多个处理器；
存储器，用于存储一个或多个程序；
当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如上所述的电泳图的识别方法。
本发明实施例还提供了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现如上所述的电泳图的识别方法。
本实施例的技术方案，采用图像识别的技术对样本的电泳图进行识别，并形成条带图像的目标像素矩阵，输入到预先训练好的预测模型中，预测模型输出样本后续试验时的目标加样量。这样可以使得电泳图的识别结果有一个统一 的规范，避免了由于人工判断与识别带有一定的主观性而导致电泳图中胶图图像的识别结果存在差异的问题。
附图说明
图1为本发明实施例一提供的电泳图的识别方法的流程图；
图2为本发明实施例二提供的电泳图的识别方法的流程图；
图3为本发明实施例三提供的电泳图的识别装置的结构示意图；
图4为本发明实施例四提供的一种设备的结构示意图。
具体实施方式
下面结合附图和实施例对本公开作详细说明。
实施例一
图1为本发明实施例一提供的电泳图的识别方法的流程图，本实施例可适用于识别电泳图，并确定样本后续试验时的目标加样量的情况，该方法可以由电泳图的识别装置来执行。
首先，需要说明的是，电泳是一种常见的分析方法。带电荷的分子和微粒在分离介质中迁移，分离介质在两个电极之间经受电场、可用等电点(Isoelectric point，pI)、分子量、电荷或这些因素的组合来使蛋白质分离。在一实施方式中，所述分离介质通常为凝胶。
分离凝胶通常置于支撑件上，并且分离凝胶的两个相对的端部接触呈溶液形式或者刚性形式的电极缓冲剂。电极可插入包含电极缓冲剂的容器中。来自电解质和离子存储器的缓冲溶液使pH(酸碱度)值和其他参数保持恒定。在分离之后，用不同的方式检测和识别分子。例如：通过对凝胶着色直观地进行检 测和识别，或者，利用光学手段，诸如：用激光扫描仪等扫描经着色的凝胶或带标记样本，或者对它们成像。
例如，本实施例中的电泳图的识别方法就是对激光扫描仪扫描得到的凝胶图像进行识别和判断。
如图1所示，本发明实施例提供的电泳图的识别方法包括如下步骤：
S110、获取样本的电泳图，识别电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像。
在本实施例中，凝胶电泳通常用来分离生物分子，诸如：蛋白质、缩氨酸、核酸等。样本可以是蛋白质样本、缩氨酸或者核酸样本等。例如，样本可以是人体、哺乳动物组织、细胞溶菌或细菌、昆虫或酵母细胞系统等。需要说明的是，本实施例中仅对样本进行说明，而非限定。
样本的电泳图可以理解为当前样本加入凝胶中，经过电泳得到的图谱。胶图图像的边框可以理解为包含所有待识别的条带图像构成的胶图图像的边框。
在一实施方式中，可以从激光扫描仪中获取当前样本的电泳图。例如，可以在接收到获取指令之后，从激光扫描仪中获取当前样本的电泳图。在本实施例中，可以根据实际情况设计获取指令的接收方式，例如：在检测到激光扫描仪中扫描到电泳图之后，则生成并接收到获取指令。又如，在检测到用于对电泳图的识别装置的点击操作之后，则生成并接收到获取指令。
在本实施例中，采用了canny边缘检测法和Laplacian边缘检测算法相结合的方式准确的定位出胶图图像的边框。采用canny边缘检测法和Laplacian边缘检测算法相结合的方式，准确地将整个电泳图中有目标位置的位置将变为白色，其余位置均默认为背景像素，并将背景色素变为黑色。这样整张电泳图中就只有黑白两种颜色，然后运用形态学的方法将目标位置进行膨胀腐蚀，这样可以 消除目标位置内部的黑点，准确定位出物体，定位出胶图图像的边框。需要说明的是，本实施例中的目标位置是指胶图图像的边框之内所包含的条带图像。
S120、提取每个条带图像中预设数量的像素值，构成每个条带图像的目标像素矩阵。
在本实施例中，条带图像是由多个像素点构成的，每个像素点都有其对应的像素值。
在本实施例中，将每个条带图像中的像素点转换为其对应的像素值，这样，每个条带图像都可以得到的由像素值构成的矩阵，这个由像素值构成的矩阵，称为原始像素矩阵。
在一实施方式中，在每个原始像素矩阵中，查找每一列的最大值。假设原始像素矩阵为N行M列的矩阵，其中，M和N均为正整数。相应地，在每个原始像素矩阵中，查找每一列的像素最大值，可以找到M个列最大像素值。其中，所述列最大像素值是指原始像素矩阵中每一列中的最大像素值。然后将M个列最大像素值进行比较，获取M个列最大像素值中的最大值作为目标最大像素值。根据目标最大像素值以及预设的规则，确定目标像素矩阵。
在本实施例中，每个条带图像分别对应一个原始像素矩阵和一个目标像素矩阵。即在电泳图中，目标像素矩阵的数量和条带图像的数量相等。
S130、将每个目标像素矩阵输入预先训练好的预测模型进行预测，以得到样本的目标加样量。
在本实施例中，样本的目标加样量可以理解为在后续的试验中，当前样本需要在凝胶中增加的样本重量。
在本实施例中，条带图像按照一定的顺序在电泳图中排列，将多个目标像素矩阵按照条带图像在电泳图中的排列顺序依次输入预先训练的预测模型，预 测模型输出当前样本的目标加样量。
本发明实施例提供的电泳图的识别方法，通过获取样本的电泳图，并识别出电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像，然后提取每个条带图像中预设数量的像素值，构成每个条带图像的目标像素矩阵，最后将每个目标像素矩阵输入预先训练好的预测模型进行预测，以得到样本的目标加样量。本实施例的技术方案，采用图像识别的技术对样本的电泳图进行识别，并形成条带图像的目标像素矩阵，输入到预先训练好的预测模型中，预测模型输出该样本后续试验时的目标加样量。这样可以使得电泳图的识别结果有一个统一的规范，避免了由于人工判断与识别带有一定的主观性而导致电泳图中胶图图像的识别结果存在差异的问题。
实施例二
图2为本发明实施例二提供的电泳图的识别方法的流程图，在上述实施例的基础上，本发明实施例优化了电泳图的识别方法，如图2所示，优化后的电泳图的识别方法主要包括如下步骤：
S210、通过边缘检测算法识别出所述电泳图的胶图图像区域。
在本实施例中，采用canny边缘检测法和Laplacian边缘检测算法相结合的方式准确地定位出胶图图像的边框。需要说明的是，本实施例中，对边缘检测算法不做限定，可以根据实际情况设计、选择合适的边缘检测算法识别出胶图图像区域。
S220、消除电泳图中的背景区域和胶图图像区域中的干扰像素点，得到至少一个初级胶图图像。
在本实施例中，电泳图中的背景区域是指电泳图中除去条带区域之外的其 他区域，条带区域中的干扰像素点可以理解为条带区域中的非白色的像素点。
在本实施例中，将整个电泳图中有条带区域的位置将变为白色，其余位置均默认为背景区域，并将背景区域的像素点变为黑色。这样整张电泳图中就只有黑白两种颜色。然后运用形态学的方法将目标位置进行膨胀腐蚀，这样可以消除目标位置内部的黑点，准确定位出物体，定位出胶图图像的边框。
S230、对每个初级条带图像进行预处理，得到每个初级条带图像对应的初级像素矩阵。
在一实施方式中，如果电泳图存在倾斜的情况，则根据坐标系对电泳图进行调整，得到放正的电泳图。
在本实施例中，对每个初级胶图图像进行预处理可以理解为对每个条带图像进行除燥、归一化处理，得到每个初级胶图图像对应的初级像素矩阵。初级像素矩阵是指初级胶图图像中的像素点对应的像素值构成的像素矩阵。
S240、计算初级像素矩阵中的行众数和列众数。
在本实施例中，众数是一组数据中出现次数最多的数值。行众数是指初级像素矩阵中每行数据中出现次数最多的值，列众数是指初级像素矩阵中每列数据中出现次数最多的值。
在本实施例中，计算初级像素矩阵每行的行众数和每列的列众数。需要说明的是，本实施例中，仅对行众数以及列众数进行说明而非限定，可以根据实际情况采取合适的方法进行计算。
S250、根据行众数与背景众数的比较结果以及列众数与背景众数的比较结果，对每个初级胶图图像进行校正，得到胶图图像的边框。
在本实施例中，背景众数是指背景区域中像素点的像素值中的众数。背景众数的计算方式和行众数以及列众数的计算方式可以相同，也可以不同，能够 计算出背景众数即可。
需要说明的是，在确定初级胶图图像的过程中应用了形态学的方法将目标位置进行膨胀腐蚀，因此，在目标位置附近的边框也会被确定为目标位置，这时，需要对目标位置进行进一步地校正。
在一实施方式中，如果行众数小于或等于背景众数，则将行众数对应的行像素进行消除。在本实施例中，如果列众数小于或等于背景众数，则说明该列众数所属的列中包含了背景区域中像素点的像素值，则将列众数对应的列像素进行消除，即将该列众数对应的列中的像素点变成与背景区域中像素点相同的颜色。
如果所述列众数小于或等于所述背景众数，则将所述列众数对应的列像素进行消除。在本实施例中，如果行众数小于或等于背景众数，则说明该行众数所属的行中包含了背景区域中像素点的像素值，则将行众数对应的行像素进行消除，即将该行众数对应的行中的像素点变成与背景区域中像素点相同的颜色。
S260、对每个条带图像，提取该条带图像中像素值构成原始像素矩阵。
在本实施例中，将每个条带图像中的像素点转换为其对应的像素值，这样，每个条带图像都可以得到的由像素值构成的矩阵，这个由像素值构成的矩阵，称为原始像素矩阵。
S270、计算原始像素矩阵中每列像素值的列最大像素值，得到M个列最大像素值，其中，M为原始像素矩阵的列数。
在本实施例中，列最大像素值可以理解为原始像素矩阵中，以列为单位，确定的每一列中的最大像素值。在每个原始像素矩阵中，查找每一列的最大值。假设原始像素矩阵为N行M列的矩阵，其中，M和N均为正整数。相应地，在每个原始像素矩阵中，查找每一列的像素最大值，可以找到M个列最大像素 值。
S280、根据M个列最大像素值的比较结果，获取预设数量的像素值构成条带图像的目标像素矩阵。
在一实施方式中，将M个列最大像素值进行比较，并将M个列最大像素值中的最大值确定为目标最大像素值；如果原始像素矩阵存在一个目标最大像素值，则将目标最大像素值所属列以及与目标最大像素值所属列相邻的两列，构成目标像素矩阵。
如果所述原始像素矩阵仅存在两个相同的目标最大像素值且两个相同的目标最大像素值分别所属的列相邻，或者，所述原始像素矩阵中目标最大像素值的个数大于等于3且至少两个相同的目标最大像素值分别所属的列相邻，则将第一列最大像素值与第二列最大像素值进行比较；如果所述第一列最大像素值大于所述第二列最大像素值，则将所述第一列最大像素值作为相邻列最大像素值；如果所述第一列最大像素值小于所述第二列最大像素值，将所述第二列最大像素值作为相邻列最大像素值；将两个相同的目标最大像素值所属列以及相邻列最大像素值所属列，构成目标像素矩阵。
第一列最大像素值可以理解为第一目标最大像素值左侧的列最大像素值，第二列最大像素值可以理解为第二目标最大像素值右侧的列最大像素值。其中，在所述原始像素矩阵中，第一目标最大像素值位于所述第二目标最大像素值的左侧。相邻列最大像素值可以理解为第一列最大像素值和第二列最大像素值中数值比较大的像素值。
在一实施方式中，如果所述原始像素矩阵仅存在两个相同的目标最大像素值且两个相同的目标最大像素值分别所属的列不相邻，则确定第一目标最大像素值所属列与第二目标最大像素值所属列之间间隔的列数；如果所述第一目标 最大像素值所属列与第二目标最大像素值所属列之间间隔的列数为1，则将所述第一目标最大像素值所属列、第二目标最大像素值所属列以及两个相同的目标最大像素值所属列之间的间隔列，构成目标像素矩阵。
如果原始像素矩阵中目标最大像素值的个数大于等于3且至少3个目标最大像素值所属列连续，则确定列连续长度，其中，所述列连续长度是目标最大像素值所属列连续存在的列个数；在所述列连续长度最大的每个列中，取任意3个连续的目标最大像素值所属列构成目标像素矩阵。
S290、将每个目标像素矩阵按照预设顺序输入预先训练好的预测模型进行预测，以得到样本的目标加样量。
本发明实施例提供的电泳图的识别方法，通过获取样本的电泳图，并识别出电泳图中的胶图图像的边框，然后提取每个条带图像中预设数量的像素值，构成每个条带图像的目标像素矩阵，最后将每个目标像素矩阵按照预设顺序输入预先训练好的预测模型进行预测，以得到样本的目标加样量。本实施例的技术方案，采用图像识别的技术对电泳图进行识别，并形成条带图像的目标像素矩阵，输入到预先训练好的预测模型中，预测模型输出样本在后续试验中的目标加样量。这样可以使得电泳图的识别结果有一个统一的规范，避免了由于人工判断与识别带有一定的主观性而导致电泳图中胶图图像的识别结果存在差异的问题。
在上述实施例的基础上，本实施例提供一种聚合酶链式反应(Polymerase Chain Reaction，PCR)电泳图的识别方法。在PCR电泳图的识别过程中，有一个比较重要的部分为条带图像大小的识别。需要把每个条带图像与对应的标准条带图像进行比较，因为标准条带图像中有9条条带，每条条带都表示不同的片段大小。
在一实施方式中，所述方法还包括：提取胶图图像中每个条带图像对应的条带像素矩阵的每一列中位数，其中，所述胶图图像包括标准条带图像和非标准条带图像；根据所述标准条带图像对应的条带像素矩阵的每一列中位数，以及非标准条带图像对应的条带像素矩阵的每一列中位数，确定每个非标准条带图像在胶图图像中的位置。
从标准条带图像对应的条带像素矩阵中的每一列取出其中位数，根据每一列的中位数可以得到一个中位数的分布情况，并取所述中位数中的最大值，每一个最大值则表示为一条条带，将每条条带对应的中位数中的最大值与标准条带图像中像素的最大值进行比较，可以定位每个固定片段的条带在电泳图中的位置，同理对每个条带图像进行上述操作，可以得出每个条带图像大小的范围。
实施例三
图3为本发明实施例三提供的电泳图的识别装置的结构示意图，本实施例可适用于识别电泳图，并确定样本在后续试验中的目标加样量的情况，如图3所示，本实施例提供的电泳图的识别装置主要包括：识别模块310、提取模块320和输入模块330。
其中，识别模块310，设置为获取样本的电泳图，识别所述电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像；
提取模块320，设置为提取每个条带图像中预设数量的像素值，构成每个所述条带图像的目标像素矩阵；
输入模块330，设置为将每个所述目标像素矩阵输入预先训练好的预测模型进行预测，以得到样本的目标加样量。
本发明实施例提供的电泳图的识别装置，通过获取样本的电泳图，识别电 泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像，然后提取每个条带图像中预设数量的像素值，构成每个条带图像的目标像素矩阵，最后将每个目标像素矩阵输入预先训练好的预测模型进行预测，以得到当前样本的目标加样量。本实施例的技术方案，采用图像识别的技术对样本的电泳图进行识别，并形成条带图像的目标像素矩阵，输入到预先训练好的预测模型中，预测模型输出该样本后续试验时的目标加样量。这样可以使得电泳图的识别结果有一个统一的规范，避免了由于人工判断与识别带有一定的主观性而导致电泳图中胶图图像的识别结果存在差异的问题。
在一实施方式中，所述识别模块310包括：
识别单元，设置为通过边缘检测算法识别出所述电泳图的胶图图像区域；
消除单元，设置为消除所述电泳图中的背景区域以及所述胶图图像区域中的干扰像素点，得到至少一个初级胶图图像。
在一实施方式中，所述识别模块310还包括：
预处理单元，设置为对每个所述初级条带图像进行预处理，得到每个所述初级条带图像对应的初级像素矩阵；
第一计算单元，设置为计算每个所述初级像素矩阵中的行众数和列众数；
校正单元，设置为根据所述行众数与背景众数的比较结果以及所述列众数与背景众数的比较结果，对每个所述初级条带图像进行校正，得到所述胶图图像的边框。
在一实施方式中，校正单元是设置为如果所述行众数小于或等于所述背景众数，则将所述行众数对应的行像素进行消除；如果所述列众数小于或等于所述背景众数，则将所述列众数对应的列像素进行消除。
在一实施方式中，所述装置还包括：
中位数有提取模块，设置为提取所述胶图图像中每个条带图像对应的条带像素矩阵的每一列中位数，其中，所述胶图图像包括标准条带图像和非标准条带图像；
条带图像确定模块，设置为根据所述标准条带图像对应的条带像素矩阵的每一列中位数，以及所述非标准条带图像对应的条带像素矩阵的每一列中位数，确定每个非标准条带图像在所述胶图图像中的位置。
在一实施方式中，提取模块320包括：
提取单元，设置为对每个所述条带图像，提取所述条带图像中像素值构成原始像素矩阵；
第二计算单元，设置为计算所述原始像素矩阵中每列像素值的列最大像素值，得到M个列最大像素值，其中，M为所述原始像素矩阵的列数；
获取单元，设置为根据所述M个列最大像素值的比较结果，获取预设数量的像素值构成所述条带图像的目标像素矩阵。
在一实施方式中，获取单元是设置为将所述M个列最大像素值进行比较，并将所述M个列最大像素值中的最大值确定为目标最大像素值；如果所述原始像素矩阵存在一个目标最大像素值，则将所述目标最大像素值所属列以及与所述目标最大像素值所属列相邻的两列，构成目标像素矩阵。
在一实施方式中，获取单元是设置为如果所述原始像素矩阵仅存在两个相同的目标最大像素值且两个相同的目标最大像素值分别所属的列相邻，或者，所述原始像素矩阵中目标最大像素值的个数大于等于3且至少两个相同的目标最大像素值分别所属的列相邻，则将第一列最大像素值与第二列最大像素值进行比较；如果所述第一列最大像素值大于所述第二列最大像素值，则将所述第一列最大像素值作为相邻列最大像素值；如果所述第一列最大像素值小于所述 第二列最大像素值，将所述第二列最大像素值作为相邻列最大像素值；将两个相同的目标最大像素值所属列以及相邻列最大像素值所属列，构成目标像素矩阵。
在一实施方式中，获取单元是设置为如果所述原始像素矩阵仅存在两个相同的目标最大像素值且两个相同的目标最大像素值分别所属的列不相邻，则确定第一目标最大像素值所属列与第二目标最大像素值所属列之间间隔的列数；如果所述第一目标最大像素值所属列与所述第二目标最大像素值所属列之间间隔的列数为1，则将所述第一目标最大像素值所属列、所述第二目标最大像素值所属列以及两个相同的目标最大像素值所属列之间的间隔列，构成目标像素矩阵。
在一实施方式中，获取单元是设置为如果所述原始像素矩阵中目标最大像素值的个数大于等于3且至少3个目标最大像素值所属列连续，则确定列连续长度，其中，所述列连续长度是目标最大像素值所属列连续存在的列个数；在所述列连续长度最大的每个列中，取任意3个连续的目标最大像素值所属列构成目标像素矩阵。
本发明实施例所提供的电泳图的识别装置可执行本发明任意实施例所提供的电泳图的识别方法，具备执行方法相应的功能模块和有益效果。
实施例四
图4为本发明实施例四提供的一种设备的结构示意图，如图4所示，该设备包括处理器410、存储器420、输入装置430和输出装置440；设备中处理器410的数量可以是一个或多个，图4中以一个处理器410为例；设备中的处理器410、存储器420、输入装置430和输出装置440可以通过总线或其他方式连接， 图4中以通过总线连接为例。
存储器420作为一种计算机可读存储介质，可用于存储软件程序、计算机可执行程序以及模块，如本发明实施例中的电泳图的识别方法对应的程序指令/模块(例如，电泳图的识别装置中的识别模块310、提取模块320和输入模块330)。处理器410通过运行存储在存储器420中的软件程序、指令以及模块，从而执行设备的各种功能应用以及数据处理，即实现上述的电泳图的识别方法。
存储器420可主要包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需的应用程序；存储数据区可存储依据终端的使用所创建的数据等。此外，存储器420可以包括高速随机存取存储器，还可以包括非易失性存储器，例如至少一个磁盘存储器件、闪存器件、或其他非易失性固态存储器件。在一些实例中，存储器420可进一步包括相对于处理器410远程设置的存储器，这些远程存储器可以通过网络连接至设备。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
输入装置430可用于接收输入的数字或字符信息，以及产生与设备的用户设置以及功能控制有关的键信号输入。输出装置440可包括显示屏等显示设备。
实施例五
本发明实施例五还提供一种包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行一种电泳图的识别方法，该方法包括：
获取样本的电泳图，识别所述电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像；
提取每个所述条带图像中预设数量的像素值，构成每个所述条带图像的目 标像素矩阵；
将每个所述目标像素矩阵输入预先训练好的预测模型进行预测，以得到样本的目标加样量。
当然，本发明实施例所提供的一种包含计算机可执行指令的存储介质,其计算机可执行指令不限于如上所述的方法操作，还可以执行本发明任意实施例所提供的电泳图的识别方法中的相关操作。
通过以上关于实施方式的描述，所属领域的技术人员可以清楚地了解到，本发明可借助软件及必需的通用硬件来实现，当然也可以通过硬件实现，但很多情况下前者是更佳的实施方式。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品可以存储在计算机可读存储介质中，如计算机的软盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、闪存(FLASH)、硬盘或光盘等，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本公开各个实施例所述的方法。
值得注意的是，上述电泳图的识别装置的实施例中，所包括的各个单元和模块只是按照功能逻辑进行划分的，但并不局限于上述的划分，只要能够实现相应的功能即可；另外，各功能单元的具体名称也只是为了便于相互区分，并不用于限制本公开的保护范围。

Claims (13)

  1. 一种电泳图的识别方法，包括：
    获取样本的电泳图，识别所述电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像；
    提取每个所述条带图像中预设数量的像素值，构成每个所述条带图像的目标像素矩阵；
    将每个所述目标像素矩阵输入预先训练好的预测模型进行预测，以得到所述样本的目标加样量。
  2. 根据权利要求1所述的方法，其中，识别所述电泳图中的胶图图像的边框，包括：
    通过边缘检测算法识别出所述电泳图的胶图图像区域；
    消除所述电泳图中的背景区域和所述胶图图像区域中的干扰像素点，得到至少一个初级胶图图像。
  3. 根据权利要求2所述的方法，其中，识别所述电泳图中的胶图图像的边框，还包括：
    对每个所述初级胶图图像进行预处理，得到每个所述初级胶图图像对应的初级像素矩阵；
    计算每个所述初级像素矩阵中的行众数和列众数；
    根据所述行众数与背景众数的比较结果以及所述列众数与背景众数的比较结果，对每个所述初级条带图像进行校正，得到所述胶图图像的边框。
  4. 根据权利要求3所述的方法，其中，所述根据所述行众数与背景众数的比较结果以及所述列众数与背景众数的比较结果，对每个所述初级胶图图像进行校正，包括：
    如果所述行众数小于或等于所述背景众数，则将所述行众数对应的行像素进行消除；
    如果所述列众数小于或等于所述背景众数，则将所述列众数对应的列像素进行消除。
  5. 根据权利要求1所述的方法，还包括：
    提取所述胶图图像中每个条带图像对应的条带像素矩阵的每一列中位数，其中，所述胶图图像包括标准条带图像和非标准条带图像；
    根据所述标准条带图像对应的条带像素矩阵的每一列中位数，以及所述非标准条带图像对应的条带像素矩阵的每一列中位数，确定每个非标准条带图像在所述胶图图像中的位置。
  6. 根据权利要求1所述的方法，其中，所述提取每个所述条带图像中预设数量的像素值，构成每个所述条带图像的目标像素矩阵，包括：
    对每个所述条带图像，提取所述条带图像中像素值构成原始像素矩阵；
    计算所述原始像素矩阵中每列像素值的列最大像素值，得到M个列最大像素值，其中，M为所述原始像素矩阵的列数且M为正整数；
    根据所述M个列最大像素值的比较结果，获取预设数量的像素值构成所述条带图像的目标像素矩阵。
  7. 根据权利要求6所述的方法，其中，所述根据所述M个列最大像素值的比较结果，获取预设数量的像素值构成目标像素矩阵，包括：
    将所述M个列最大像素值进行比较，并将所述M个列最大像素值中的最大值确定为目标最大像素值；
    如果所述原始像素矩阵存在一个目标最大像素值，则将所述目标最大像素值所属列以及与所述目标最大像素值所属列相邻的两列，构成目标像素矩阵。
  8. 根据权利要求7所述的方法，其中，所述根据M个列最大像素值的比较结果，获取预设数量的像素值构成目标像素矩阵，还包括：
    如果所述原始像素矩阵仅存在两个相同的目标最大像素值且两个相同的目标最大像素值分别所属的列相邻，或者，所述原始像素矩阵中目标最大像素值的个数大于等于3且至少两个相同的目标最大像素值分别所属的列相邻，则将第一列最大像素值与第二列最大像素值进行比较；
    如果所述第一列最大像素值大于所述第二列最大像素值，则将所述第一列最大像素值作为相邻列最大像素值；
    如果所述第一列最大像素值小于所述第二列最大像素值，将所述第二列最大像素值作为相邻列最大像素值；
    将两个相同的目标最大像素值所属列以及相邻列最大像素值所属列，构成目标像素矩阵。
  9. 根据权利要求7所述的方法，其中，所述根据M个列最大像素值的比较结果，获取预设数量的像素值构成目标像素矩阵，还包括：
    如果所述原始像素矩阵仅存在两个相同的目标最大像素值且两个相同的目标最大像素值分别所属的列不相邻，则确定第一目标最大像素值所属列与第二目标最大像素值所属列之间间隔的列数；
    如果所述第一目标最大像素值所属列与所述第二目标最大像素值所属列之间间隔的列数为1，则将所述第一目标最大像素值所属列、所述第二目标最大像素值所属列以及两个相同的目标最大像素值所属列之间的间隔列，构成目标像素矩阵。
  10. 根据权利要求7所述的方法，其中，所述根据M个列最大像素值的比较结果，获取预设数量的像素值构成目标像素矩阵，还包括：
    如果所述原始像素矩阵中目标最大像素值的个数大于等于3且至少3个目标最大像素值所属列连续，则确定列连续长度，其中，所述列连续长度是目标最大像素值所属列连续存在的列个数；
    在所述列连续长度最大的每个列中，取任意3个连续的目标最大像素值所属列构成目标像素矩阵。
  11. 一种电泳图的识别装置，包括：
    识别模块，设置为获取样本的电泳图，识别所述电泳图中的胶图图像的边框，其中，所述胶图图像包括至少一个条带图像；
    提取模块，设置为提取每个所述条带图像中预设数量的像素值，构成每个所述条带图像的目标像素矩阵；
    输入模块，设置为将每个所述目标像素矩阵输入预先训练好的预测模型，以得到所述样本的目标加样量。
  12. 一种设备，包括：
    一个或多个处理器；
    存储器，用于存储一个或多个程序；
    当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如权利要求1-10任一所述的电泳图的识别方法。
  13. 一种计算机可读存储介质，其上存储有计算机程序，其中，该程序被处理器执行时实现如权利要求1-10任一所述的电泳图的识别方法。
PCT/CN2019/123944 2019-04-15 2019-12-09 电泳图的识别方法、装置、设备及存储介质 WO2020211403A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910299112.4 2019-04-15
CN201910299112.4A CN110033449B (zh) 2019-04-15 2019-04-15 电泳图的识别方法、装置、设备及存储介质

Publications (1)

Publication Number Publication Date
WO2020211403A1 true WO2020211403A1 (zh) 2020-10-22

Family

ID=67238427

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/123944 WO2020211403A1 (zh) 2019-04-15 2019-12-09 电泳图的识别方法、装置、设备及存储介质

Country Status (2)

Country Link
CN (1) CN110033449B (zh)
WO (1) WO2020211403A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113177548A (zh) * 2021-05-08 2021-07-27 四川大学 一种针对免疫固定电泳的重点区域识别方法

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110033449B (zh) * 2019-04-15 2023-06-27 苏州金唯智生物科技有限公司 电泳图的识别方法、装置、设备及存储介质
CN111640131B (zh) * 2020-05-29 2021-05-18 广东利元亨智能装备股份有限公司 电芯涂胶检测方法、装置和电子设备
CN111723730B (zh) * 2020-06-18 2023-08-22 中国银行股份有限公司 对目标对象的数量进行预测的方法及相关设备
WO2022188696A1 (zh) * 2021-03-08 2022-09-15 南京金斯瑞生物科技有限公司 自动识别和分析胶图的方法
CN113112432A (zh) * 2021-05-13 2021-07-13 广州道一科学技术有限公司 自动识别图像条带的方法
CN114219752B (zh) * 2021-09-23 2023-07-25 四川大学 一种针对血清蛋白电泳的异常区域检测方法
CN117392679B (zh) * 2023-12-11 2024-03-08 清华大学 一种自动标记pcr胶图的方法和系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106485698A (zh) * 2016-09-21 2017-03-08 上海理工大学 从凝胶电泳条带图像中获取dna色谱的方法
CN106558041A (zh) * 2015-09-29 2017-04-05 中国疾病预防控制中心传染病预防控制所 凝胶电泳数字图像中基于局部抑制的条带探测算法
US20190026890A1 (en) * 2017-07-21 2019-01-24 Panasonic Intellectual Property Management Co., Ltd. Display control apparatus, display control method, and recording medium
CN110033449A (zh) * 2019-04-15 2019-07-19 苏州金唯智生物科技有限公司 电泳图的识别方法、装置、设备及存储介质

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4810348A (en) * 1987-03-16 1989-03-07 Helena Laboratories Corporation Automatic electrophoresis apparatus and method
ES2366380T3 (es) * 2005-06-14 2011-10-19 Protox Therapeutics Incorporated Método para el tratamiento o prevención de hiperplasia prostática benigna utilizando proteínas formadoras de poros modificadas.
CN101532055A (zh) * 2009-02-04 2009-09-16 中国农业科学院北京畜牧兽医研究所 辅助鉴定五指山小型猪近交系的方法及其专用试剂盒
WO2013163451A1 (en) * 2012-04-27 2013-10-31 Bio-Rad Laboratories, Inc. Stain-free protein quantification and normalization
CN105302553A (zh) * 2015-10-21 2016-02-03 北京工业大学 一种基于vc++的凝胶图像识别和定位的方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106558041A (zh) * 2015-09-29 2017-04-05 中国疾病预防控制中心传染病预防控制所 凝胶电泳数字图像中基于局部抑制的条带探测算法
CN106485698A (zh) * 2016-09-21 2017-03-08 上海理工大学 从凝胶电泳条带图像中获取dna色谱的方法
US20190026890A1 (en) * 2017-07-21 2019-01-24 Panasonic Intellectual Property Management Co., Ltd. Display control apparatus, display control method, and recording medium
CN110033449A (zh) * 2019-04-15 2019-07-19 苏州金唯智生物科技有限公司 电泳图的识别方法、装置、设备及存储介质

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113177548A (zh) * 2021-05-08 2021-07-27 四川大学 一种针对免疫固定电泳的重点区域识别方法
CN113177548B (zh) * 2021-05-08 2022-07-08 四川大学 一种针对免疫固定电泳的重点区域识别方法

Also Published As

Publication number Publication date
CN110033449B (zh) 2023-06-27
CN110033449A (zh) 2019-07-19

Similar Documents

Publication Publication Date Title
WO2020211403A1 (zh) 电泳图的识别方法、装置、设备及存储介质
AU743867B2 (en) Method and apparatus for analyzing images
US11860094B2 (en) Analysis system and analysis method
CN110866932A (zh) 多通道舌体边缘检测装置、方法及存储介质
CN110599552A (zh) 一种基于计算机视觉的pH试纸检测方法
CN112381081A (zh) 公章文字自动识别方法、装置、计算机设备及存储介质
EP3353740B1 (en) Image processing method&apparatus for normalisaton and artefact correction
CN112861845A (zh) 车牌识别方法、装置、计算机设备和存储介质
CN114049330A (zh) 一种荧光原位杂交图像中荧光特征融合方法及系统
CN105760869A (zh) 一种基于图像灰度识别及处理的阅卷方法
CN115131784B (zh) 一种图像处理方法、装置、电子设备及存储介质
WO2022188696A1 (zh) 自动识别和分析胶图的方法
Wu et al. A fast spot segmentation algorithm for two‐dimensional gel electrophoresis analysis
JP7050122B2 (ja) キャピラリ電気泳動装置
Abeykoon et al. An automated system for analyzing agarose and polyacrylamide gel images
CN112070678B (zh) 一种批量蛋白质印迹膜条倾斜矫正和分割方法及系统
CN112734767B (zh) 基于病理图像组织区域的提取方法、装置、设备及介质
CN111127434A (zh) 染色分离方法及装置
Gutierrez On the use of distance maps in the analysis of 1D DNA gel images
Faroq et al. 2-D electrophoresis gel matching using proximity graphs
KR101373060B1 (ko) 전기영동 및 젤 다큐멘테이션 기능이 통합된 통합장치의 이미지 분석 방법, 이미지 분석 시스템 및 바이오 진단 방법
Chan et al. Automatic band detection of 1D-gel images
Labyed et al. Automatic segmentation and band detection of protein images based on the standard deviation profile and its derivative
CN112836587A (zh) 跑道识别方法、装置、计算机设备及存储介质
Ruz-Suarez et al. Fully-convolutional neural networks ensemble for comet segmentation in single cell gel electrophoresis assay images

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19924792

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19924792

Country of ref document: EP

Kind code of ref document: A1