Biological strip identification system and method for Marker strip labeling

Abstract

The invention discloses a biological strip recognition system and method for Marker strip labeling. The method comprises the steps that firstly, an electrophoretogram comprising at least two Marker strips is obtained according to a collected gel electrophoretogram; then, a preset strip segmentation model is applied to segment the image area of each Marker strip in the electrophoretogram to obtain a strip segmentation image; inputting the strip segmentation image into a preset strip identification neural network and obtaining a strip identification graph; and finally, marking the Marker strip of which the strip type is a true strip in the strip identification graph to obtain and output a strip marking graph. Due to the fact that the segmentation model and the recognition neural network are used for sequentially conducting strip segmentation and pseudo-strip recognition on the gel electrophoretogram and then marking the Marker strip fragment values, the Marker strip marking efficiency and precision are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of gel electrophoresis image processing, and in particular to a biological band recognition system and method for marker band annotation. Background Art

[0002] Marker bands are used to calibrate electrophoresis results, helping to determine the size and purity of DNA fragments in the sample being tested, verifying the electrophoresis results, and ensuring experimental accuracy. Currently, the most common development method involves capturing images using an imaging system and then manually annotating the marker bands in the image with fragment values ​​to indicate the range of the extracted sample fragments. This manual annotation process can lead to technical issues such as low efficiency and high error rates. Summary of the Invention

[0003] The main technical problem solved by the present invention is how to improve the efficiency and accuracy of Marker strip labeling.

[0004] According to a first aspect, an embodiment provides a biological band identification system for marker band annotation, comprising: Image acquisition module, used to collect gel electrophoresis images in gel electrophoresis experiments; A pattern recognition module is used to obtain an electrophoresis pattern based on the gel electrophoresis pattern; the electrophoresis pattern is an image including at least two marker bands; a band detection module, configured to segment the image area of ​​each Marker band in the electrophoresis pattern using a preset band segmentation model to obtain a band segmentation image; the band segmentation image is used to distinguish and identify the image area of ​​each Marker band in the electrophoresis pattern; a stripe recognition module, configured to input the stripe segmentation image into a preset stripe recognition neural network and obtain a stripe recognition map output by the stripe recognition neural network; the stripe recognition map is used to identify the stripe type of each marker stripe in the stripe segmentation image; the stripe types include true stripes and pseudo stripes; The stripe identification module is used to mark the marker stripe whose stripe type is the true stripe in the stripe identification map to obtain and output a stripe marking map; the stripe marking map is used to mark the segment value of the marker stripe.

[0005] In one embodiment, the biological band identification system further includes a lane construction module for identifying the lane of each marker band in the band segmentation image and outputting the band segmentation image after lane identification to the band identification module; the lane is composed of a plurality of lane blocks with the same geometric shapes arranged in series.

[0006] In one embodiment, the biological strip identification system further includes a lane correction module for matching and calibrating the position and / or shape of each lane block in the lane to correct the tilt and / or displacement deviation of the lane, thereby improving the consistency of the lane.

[0007] In one embodiment, marking the marker stripe whose stripe type is the true stripe in the stripe identification graph includes: Fitting a band position-molecular weight mapping function based on the molecular weight corresponding to the marker band as a reference marker band; Applying the band position-molecular weight mapping function to estimate the molecular weight of the other marker bands and automatically annotating them; The annotated image and the structure data table obtained after automatic annotation are output as the strip annotated graph.

[0008] In one embodiment, the biological band identification system further includes an image preprocessing module for performing image preprocessing on the gel electrophoresis image and outputting the preprocessed gel electrophoresis image to the pattern recognition module; the image preprocessing on the gel electrophoresis image is used to eliminate image noise and / or interference to improve the clarity and / or resolution of the gel electrophoresis image.

[0009] In one embodiment, performing image preprocessing on the gel electrophoresis image includes: Applying Gaussian filtering or median filtering to perform noise reduction on the gel electrophoresis image; Converting the gel electrophoresis image from a color image to a grayscale image; Adjusting the vertical direction of the marker strip in the gel electrophoresis by image rotation; The brightness value of the gel electrophoresis graph is normalized by applying a histogram equalization method.

[0010] In one embodiment, the stripe segmentation model is an instance segmentation model; and the stripe recognition neural network is a graph neural network.

[0011] According to the second aspect, an embodiment provides a biological band identification method for marker band annotation, comprising: Collect gel electrophoresis images in gel electrophoresis experiments; Obtaining an electrophoresis pattern according to the gel electrophoresis pattern; the electrophoresis pattern is an image including at least two marker bands; Applying a preset band segmentation model to segment the image area of ​​each Marker band in the electrophoresis pattern to obtain a band segmentation image; the band segmentation image is used to distinguish and identify the image area of ​​each Marker band in the electrophoresis pattern; Inputting the stripe segmentation image into a preset stripe recognition neural network, and obtaining a stripe recognition map output by the stripe recognition neural network; the stripe recognition map is used to identify the stripe type of each marker stripe in the stripe segmentation image; the stripe types include true stripes and pseudo stripes; The Marker strips whose strip types are the true strips in the strip identification map are marked to obtain and output a strip marking map; the strip marking map is used to mark the segment values ​​of the Marker strips.

[0012] According to a third aspect, an embodiment provides a computer-readable storage medium having a computer program stored thereon. The computer program can be executed by a processor to implement the biological strip identification method as described in the second aspect.

[0013] According to a fourth aspect, an embodiment provides a computer program product, comprising a computer program and / or instructions, which, when executed by a processor, implement the biological band identification method as described in the second aspect.

[0014] According to the biological band recognition system of the above embodiment, the segmentation model and recognition neural network are used to perform band segmentation and pseudo-band recognition on the gel electrophoresis image in sequence and then label the marker band fragment values, so that the efficiency and accuracy of marker band labeling are greatly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a marker strip marking process in an embodiment; Figure 2 is a gel electrophoresis diagram in one embodiment; Figure 3 A stripe marking diagram in one embodiment; Figure 4 A schematic diagram of a process for identifying biological strips in one embodiment; Figure 5 is a structural block diagram of a biological strip identification system in one embodiment; Figure 6 is a structural block diagram of a biological strip identification system in another embodiment; Figure 7This is a strip marking diagram in another embodiment. DETAILED DESCRIPTION

[0016] The present invention will be further described in detail below by means of specific embodiments in conjunction with the accompanying drawings. Similar elements in different embodiments are numbered with associated similar elements. In the following embodiments, many detailed descriptions are provided to enable the present application to be better understood. However, those skilled in the art will readily appreciate that some of the features may be omitted in different circumstances, or may be replaced by other elements, materials, or methods. In some cases, some operations related to the present application are not shown or described in the specification. This is to avoid the core portion of the present application being overwhelmed by excessive descriptions, and for those skilled in the art, it is not necessary to describe these related operations in detail. They will fully understand the related operations based on the description in the specification and the general technical knowledge in the art.

[0017] In addition, the features, operations, or characteristics described in the specification may be combined in any appropriate manner to form various embodiments. Furthermore, the steps or actions in the method description may be reordered or adjusted in a manner readily apparent to those skilled in the art. Therefore, the various sequences in the specification and drawings are provided solely for the purpose of clearly describing a particular embodiment and are not intended to be mandatory, unless otherwise specified.

[0018] Component numbers used herein, such as "first" and "second," are used solely to distinguish the components being described and do not convey any sequential or technical meaning. References to "connection" and "coupling" herein, unless otherwise specified, include both direct and indirect connections (couplings).

[0019] Please refer to Figure 1 , is a schematic diagram of a marker band annotation process in one embodiment, wherein the gel electrophoresis pattern obtained in the gel electrophoresis experiment is automatically identified and annotated by computer software, and the specific process includes: a. Respond to the photo taking command to obtain the gel electrophoresis image, and prompt "photo taking abnormality" if the photo is not taken; b. After executing the photo command, the photo (gel electrophoresis image) is detected. If the photo is not detected or the photo quality is too poor (for example, overexposure), a prompt will be displayed indicating that the photo display is abnormal; c. Identify the marker bands in the gel electrophoresis diagram, specifically identifying the lane where one or more marker bands are located and the bands of each concentration of each marker band. If no marker band is identified, a message will be displayed indicating that the marker band cannot be found. d. Identify the true and false marker bands. If the marker band type cannot be identified, a marker band identification error message will be displayed. e. Identify the band fragment and molecular weight of each marker band. If no identification is possible, it will indicate that the marker band analysis is abnormal. f. Identify the range of the sample marker strips. If the range cannot be identified, it will prompt that the sample is abnormal. g. According to the type and model of the marker band, mark the fragment and molecular weight of the marker band. If the marking cannot be normal, it will prompt the marking abnormality; h. Determine whether the recognition of each marker strip is successful or not, and indicate recognition abnormality if it is judged to be a failure; i. Mark the lane where the sample is located (manually confirm by left-clicking the mouse).

[0020] j. Automatically label other marker bands and further verify the labeling results.

[0021] Please refer to Figure 2 and Figure 3 , respectively, are the gel electrophoresis image and the band annotation image. In the band annotation image, numbers 1 to 13 are lanes containing sample wells. The marker bands for wells 1 and 13 have marker ranges of 50bp-600bp and 100bp-5000bp, respectively. Sample fragments in wells 2, 3, 11, and 12 are dispersed between 50-300bp, with the main fragments concentrated between 50-100bp. Sample fragments in wells 4 and 5 are dispersed between 100-5000bp, with the main fragments concentrated between 100-500bp. Samples in wells 7, 8, and 9 are dispersed, with no concentrated area.

[0022] In the above-mentioned marking process, the most core step is to identify the Marker band category and mark the Marker band fragment size function, which can be automatically identified by software and then marked, or manually input marker parameter name, quantity and size etc. to realize. In one embodiment, multiple groups of parameter information can be preset to identify the marker category and mark the marker band fragment size. In the actual marking process, there may be only a part of the Marker band separation, and the situation that other bands are adhered together, now just need only mark the separated band. The accuracy rate of the adhesion situation judgment at this stage is the main interference factor affecting the Marker band marking efficiency and precision.

[0023] The biological band recognition system disclosed in the embodiments of the present application provides a biological band image recognition method and system based on graph structure modeling and multi-scale fusion, which is used to efficiently and accurately extract band structures from gel electrophoresis or similar biological experimental images, eliminate pseudo-band interference, and complete lane consistency correction and accurate molecular weight estimation.

[0024] Example 1: Please refer to Figure 4 , is a schematic flow chart of a biological band identification method in an embodiment, comprising: Step 101: Obtain a gel electrophoresis image.

[0025] Collecting a gel electrophoresis pattern in a gel electrophoresis experiment: In one embodiment, the gel electrophoresis pattern is obtained by importing a gel electrophoresis experiment image.

[0026] In one embodiment, image preprocessing is performed on the gel electrophoresis image to eliminate image noise and / or interference, so as to improve the clarity and / or resolution of the gel electrophoresis image.

[0027] Step 102: Obtain an electrophoresis pattern.

[0028] An electrophoresis pattern is obtained based on a gel electrophoresis pattern, where the electrophoresis pattern is an image containing at least two marker bands. Since the gel electrophoresis pattern is a raw image directly captured during a gel electrophoresis experiment, either by photographing or importing it, to improve the efficiency of subsequent image processing, the gel electrophoresis pattern is subjected to operations such as denoising, rotation correction, and brightness normalization during image preprocessing to obtain the electrophoresis pattern.

[0029] In one embodiment, image preprocessing includes: 1) Applying Gaussian filtering or median filtering to reduce noise on the gel electrophoresis image; 2) converting the gel electrophoresis image from a color image to a grayscale image; 3) Adjusting the vertical direction of the marker band in the gel electrophoresis by image rotation; 4) Normalizing the brightness values ​​of the gel electrophoresis graph using a histogram equalization method.

[0030] Step 103: stripe segmentation.

[0031] A preset band segmentation model is applied to segment the image area of ​​each marker band in the electrophoresis pattern to obtain a band segmentation image. The band segmentation image is used to distinguish and identify the image area of ​​each marker band in the electrophoresis pattern. In one embodiment, the lane of each marker band in the band segmentation image is identified, and the lane is composed of a plurality of lane blocks with the same geometric shape arranged in series. In one embodiment, during the band segmentation process, the position and / or shape of each lane block in the lane is matched and calibrated to correct the tilt and / or displacement deviation of the lane, thereby improving the consistency of the lane.

[0032] In one embodiment, the stripe segmentation model is an instance segmentation model such as YoloV8-seg, which directly outputs multiple target boxes (differentiated by selecting multiple target boxes) to perform preliminary detection of the marker stripe image area. In one embodiment, based on the target boxes described above, the lane is divided into a number of lane blocks from top to bottom with a fixed step size, such as 3 pixels. Structural information such as the position center, width, height, color mean, and ssim of each lane block is calculated and labeled as V = {v1, v2, …, vn}, where v1 is the sum of the squared coordinates of the position center, v2 is the width, and so on.

[0033] In one embodiment, a graph structure G is constructed as a strip segmentation image, which is expressed as: G=(V,E); Among them, V is the strip node and E is the adjacent edge (adjacent to the same lane or aligned across lanes).

[0034] Step 104: Identify the stripe type.

[0035] The strip segmentation image is input into a preset strip recognition neural network, and a strip recognition map is obtained as the output of the strip recognition neural network. The strip recognition map is used to identify the strip type of each marker strip in the strip segmentation image, where strip types include true stripes and pseudo stripes. In one embodiment, the strip recognition neural network is a graph neural network (GNN model). In one embodiment, the graph structure G serving as the strip segmentation image is input into the GNN model, and the type prediction p of each node is output to determine whether it is a pseudo stripe. The GNN model can have various forms. The following uses a propagation formula as an example, specifically including: ; In the above formula, the left side of the equal sign is the representation of node i in the first layer, the left side of the equal sign is the degree of node d, W is the weight of the layer, and σ is the activation function.

[0036] In one embodiment, the last layer output of the GNN model is connected to a sigmoid layer as an abnormality probability prediction, which is expressed as follows: ; Among them, y is the abnormal probability value; Then remove all nodes predicted to be pseudo-bands to obtain the cleaned band set M'.

[0037] Step 105: Mark the strips.

[0038] Mark the true band marker bands in the band identification map to obtain and output a band annotation map. The band annotation map is used to mark the fragment values ​​of the marker bands.

[0039] In one embodiment, a lane alignment reference line is first constructed based on the position of the internal reference band, and the tilt / displacement deviation of all lanes is corrected to obtain a lane-consistent image I''; then, for each lane, a band position-molecular weight mapping function is fitted based on the molecular weight corresponding to the reference marker band; then, molecular weight estimation and automatic annotation are performed on all bands; finally, the final annotated image and structure data table are output.

[0040] This biological band recognition method combines structured cognition (GNN), multi-scale image modeling, and a lane alignment mechanism driven by biological experimental knowledge. It consistently outputs high-quality analysis results even in complex image backgrounds and irregular lane structures, achieving true automatic band image recognition, correction, and quantitative analysis. This method is widely applicable to intelligent image processing in life science experiments such as gel electrophoresis, Western blot, and nucleic acid analysis.

[0041] Please refer to Figure 5, is a block diagram of the structure of a biological band identification system in one embodiment. In one embodiment of the present application, a biological band identification system is also disclosed, comprising an image acquisition module 10, a pattern recognition module 20, a band detection module 30, a band identification module 40, and a band identification module 50. The image acquisition module 10 is used to capture a gel electrophoresis pattern in a gel electrophoresis experiment. The pattern recognition module 20 is used to obtain an electrophoresis spectrum based on the gel electrophoresis pattern, which is an image including at least two marker bands. The band detection module 30 is used to apply a preset band segmentation model to segment the image area of ​​each marker band in the electrophoresis spectrum to obtain a band segmentation image. The band segmentation image is used to distinguish and identify the image area of ​​each marker band in the electrophoresis spectrum. The band identification module 40 is used to input the band segmentation image into a preset band identification neural network and obtain a band identification map output by the band identification neural network. The band identification map is used to identify the band type of each marker band in the band segmentation image. The stripe types include true stripe and pseudo stripe. The stripe identification module 40 is used to mark the marker stripe whose stripe type is true stripe in the stripe identification map to obtain and output the stripe marking map. The stripe marking map is used to mark the segment value of the marker stripe.

[0042] Please refer to Figure 6 , is a structural block diagram of a biological band identification system in another embodiment. In one embodiment of the present application, the biological band identification system also includes a lane construction module 60, a lane correction module 70 and an image preprocessing module 80. The lane construction module 60 is used to identify the lanes of each marker band in the band segmentation image, and output the band segmentation image after the identified lanes to the band identification module. The lane is composed of a plurality of lane blocks with the same geometric shapes arranged in series. The lane correction module 70 is used to match and calibrate the position and / or shape of each lane block in the lane to correct the tilt and / or displacement deviation of the lane, thereby improving the consistency of the lane. The image preprocessing module 08 is used to perform image preprocessing on the gel electrophoresis image, and output the gel electrophoresis image after image preprocessing to the pattern recognition module. Image preprocessing of the gel electrophoresis image is used to eliminate image noise and / or interference to improve the clarity and / or resolution of the gel electrophoresis image. In one embodiment, the gel electrophoresis image is preprocessed, that is, the original gel electrophoresis image is normalized to reduce background interference, correct tilt, and improve the accuracy of subsequent band detection and identification. In one embodiment, the gel electrophoresis image is preprocessed including: (1) Use Gaussian Blur or Median Filter to reduce the noise of the image; (2) Convert the color image into a grayscale image so that only the band structure can be focused on later. Some gel imagers use black and white cameras to obtain gel electrophoresis images. When the image is colored, it is mostly pseudo-color processed. In order to obtain the original gel electrophoresis image, pseudo-color removal is required.

[0043] (3) Estimate the main direction of the lane or strip through edge detection + Hough line transform; then rotate the main direction until it is upward and perpendicular to the horizontal.

[0044] (4) Normalize the image brightness value, such as histogram equalization.

[0045] In one embodiment of the present application, the biological band recognition system includes six main stages: image preprocessing, band detection, structure graph construction, anomaly discrimination, lane alignment and feature annotation. It applies methods such as graph structure modeling and multi-scale fusion biological band image recognition to efficiently and accurately extract band structures from gel electrophoresis or similar biological experimental images, eliminate pseudo-band interference, and complete lane consistency correction and accurate molecular weight estimation.

[0046] 1. The process of modeling the strip image structure diagram includes: First, each detected biological band is regarded as a node in the graph. Then, a graph structure at the lane level or the entire graph level is constructed, and the lane sequence relationship and the structural similarity between adjacent lanes are modeled as graph edges. Finally, a graph neural network (GNN) is used for structural learning to extract context-aware features and effectively distinguish real bands from pseudo bands and interference bands.

[0047] 2. The multi-scale strip fusion and pseudo-band identification process includes: First, high- and low-resolution images or feature pyramids are used to construct strip regions at different scales, and each strip is comprehensively scored for local clarity, texture characteristics, and spatial consistency. Then, based on the learned strip map structure, GNN aggregates contextual information inside and outside the lane to automatically identify and filter out pseudo or abnormal bands.

[0048] 3. Lane consistency correction and internal reference alignment process includes: First, a weakly supervised method based on internal reference bands is used to estimate the positional offset between lanes. Then, a lane alignment network or a correction strategy based on geometric coding is constructed to ensure that all lane structures are consistent in scale and position. Finally, the accuracy of the band position information in calculating molecular weight is improved.

[0049] 4. Considering simultaneous enhancement and robustness training strategies include: First, a synchronous occlusion enhancement strategy (such as synchronous Random Erasing) is designed during the training phase. Then, interference conditions such as noise pollution, strip blur, and edge loss are simulated to improve the robustness of the model in actual complex environments. In addition, the model's adaptability to complex backgrounds and low-quality images during the training and inference phases is supported.

[0050] Please refer to Figure 7 , is a strip marking diagram in an embodiment, wherein symbol S1 (1), symbol S1 (2), symbol S2 (1), symbol IP-1, symbol IP-1, symbol IP-2, symbol IP-3, symbol S3 and symbol S4 are lanes of sample wells, the green line frame is the image segmentation area of ​​the Marker strip, and the green grid 1, green grid 2 and green grid 3 in the green line frame are lane blocks.

[0051] The biological band identification method disclosed in the embodiments of the present application first obtains an electrophoresis pattern including at least two marker bands based on a collected gel electrophoresis pattern; then applies a preset band segmentation model to segment the image area of ​​each marker band in the electrophoresis pattern to obtain a band segmentation image; then inputs the band segmentation image into a preset band recognition neural network to obtain a band recognition map; finally, annotates the marker bands whose band types are true bands in the band recognition map to obtain and output a band annotation map. Since the segmentation model and the recognition neural network are applied to the gel electrophoresis pattern to perform band segmentation and pseudo-band recognition in sequence before annotating the marker band fragment values, the efficiency and accuracy of marker band annotation are greatly improved.

[0052] Those skilled in the art will appreciate that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer program. When all or part of the functions in the above embodiments are implemented by computer program, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to implement the above functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above functions can be implemented. In addition, when all or part of the functions in the above embodiments are implemented by computer program, the program can also be stored in a storage medium such as a server, another computer, disk, optical disk, flash disk or mobile hard disk, and saved in the memory of the local device by downloading or copying, or the system of the local device is updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be implemented.

[0053] The above examples are used to illustrate the present invention, which are only used to help understand the present invention and are not intended to limit the present invention. Those skilled in the art can make several simple deductions, modifications or substitutions based on the concept of the present invention.

Claims

1. A biological band identification system for marker band annotation, characterized in that: include: Image acquisition module, used to collect gel electrophoresis images in gel electrophoresis experiments; A pattern recognition module is used to obtain an electrophoresis pattern based on the gel electrophoresis pattern; the electrophoresis pattern is an image including at least two marker bands; a band detection module, configured to segment the image area of ​​each Marker band in the electrophoresis pattern using a preset band segmentation model to obtain a band segmentation image; the band segmentation image is used to distinguish and identify the image area of ​​each Marker band in the electrophoresis pattern; a stripe recognition module, configured to input the stripe segmentation image into a preset stripe recognition neural network and obtain a stripe recognition map output by the stripe recognition neural network; the stripe recognition map is used to identify the stripe type of each marker stripe in the stripe segmentation image; the stripe types include true stripes and pseudo stripes; The stripe identification module is used to mark the marker stripe whose stripe type is the true stripe in the stripe identification map to obtain and output a stripe marking map; the stripe marking map is used to mark the segment value of the marker stripe.

2. The biological strip recognition system according to claim 1, wherein: It also includes a lane construction module for marking the lane of each marker strip in the strip segmentation image and outputting the strip segmentation image after marking the lane to the strip identification module; the lane is composed of a plurality of lane blocks with the same geometric shapes arranged in series.

3. The biological strip recognition system according to claim 2, wherein: It also includes a lane correction module for matching and calibrating the position and / or shape of each lane block in the lane to correct the tilt and / or displacement deviation of the lane, thereby improving the consistency of the lane.

4. The biological strip recognition system according to claim 2, wherein: Marking the marker stripe whose stripe type in the stripe identification diagram is the true stripe includes: Fitting a band position-molecular weight mapping function based on the molecular weight corresponding to the marker band as a reference marker band; Applying the band position-molecular weight mapping function to estimate the molecular weight of the other marker bands and automatically annotating them; The annotated image and the structure data table obtained after automatic annotation are output as the strip annotated graph.

5. The biological strip recognition system according to claim 1, wherein: It also includes an image preprocessing module for performing image preprocessing on the gel electrophoresis image and outputting the preprocessed gel electrophoresis image to the pattern recognition module; the image preprocessing on the gel electrophoresis image is used to eliminate image noise and / or interference to improve the clarity and / or resolution of the gel electrophoresis image.

6. The biological strip recognition system according to claim 5, wherein: Performing image preprocessing on the gel electrophoresis image includes: Applying Gaussian filtering or median filtering to perform noise reduction on the gel electrophoresis image; Converting the gel electrophoresis image from a color image to a grayscale image; Adjusting the vertical direction of the marker strip in the gel electrophoresis by image rotation; The brightness value of the gel electrophoresis graph is normalized by applying a histogram equalization method.

7. The biological strip identification system according to claim 1, wherein: The strip segmentation model is an instance segmentation model; the strip recognition neural network is a graph neural network.

8. A biological band identification method for marker band annotation, characterized in that: include: Collect gel electrophoresis images in gel electrophoresis experiments; Obtaining an electrophoresis pattern according to the gel electrophoresis pattern; the electrophoresis pattern is an image including at least two marker bands; Applying a preset band segmentation model to segment the image area of ​​each Marker band in the electrophoresis pattern to obtain a band segmentation image; the band segmentation image is used to distinguish and identify the image area of ​​each Marker band in the electrophoresis pattern; Inputting the stripe segmentation image into a preset stripe recognition neural network, and obtaining a stripe recognition map output by the stripe recognition neural network; the stripe recognition map is used to identify the stripe type of each marker stripe in the stripe segmentation image; the stripe types include true stripes and pseudo stripes; The Marker strips whose strip types are the true strips in the strip identification map are marked to obtain and output a strip marking map; the strip marking map is used to mark the segment values ​​of the Marker strips.

9. A computer-readable storage medium, characterized in that The medium stores a computer program, which can be executed by a processor to implement the biological strip identification method according to claim 8.

10. A computer program product comprising a computer program and / or instructions, characterized in that When the computer program and / or instructions are executed by a processor, the biological strip identification method according to claim 8 is implemented.

Images