Feature extraction method and device for trophoblast cells, electronic equipment and storage medium

Abstract

The invention provides a trophoblast cell feature extraction method and device, electronic equipment and a storage medium, and relates to the technical field of assisted reproduction. The method comprises the following steps: determining a mask image of an inner cell mass in a target blastocyst, a first trophoblast cell mask image and a second trophoblast cell mask image; according to the first trophoblast cell mask image, the second trophoblast cell mask image and the mask image of the inner cell cluster, determining marking information of trophoblast cells in the target blastocyst; performing curved surface mapping on the mask image of the inner cell cluster, the mask image of the first trophoblast cell and the mask image of the second trophoblast cell according to marking information of the trophoblast cells to generate a curved surface mapping result; and according to a curved surface mapping result, carrying out feature extraction on the trophoblast cell region in the target blastocyst to obtain multiple three-dimensional features of the trophoblast cell region. According to the method, more core three-dimensional characteristics of trophoblast cells can be fully excavated, and the blastocyst development potential evaluation accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of assisted reproductive technology, and more specifically, to a method, apparatus, electronic device, and storage medium for extracting features of trophoblast cells. Background Technology

[0002] In assisted reproductive technology, assessing the developmental potential of an embryo is a crucial step in determining the success of embryo implantation. The trophoblast cells in the blastocyst are an important factor in determining the success of blastocyst implantation, especially the morphological characteristics of the polar trophoblast cells, which are closely related to the development and implantation potential of the blastocyst.

[0003] In traditional methods, the assessment of embryonic developmental potential is based on morphological evaluation, which relies on embryologists' subjective observation of the blastocyst in a two-dimensional plane. The main indicators include the degree of expansion of the blastocyst cavity and the morphological grading of the inner cell mass and trophoblast cells.

[0004] The above methods are highly subjective and have limited ability to extract characteristic information from trophoblast cells, resulting in low accuracy in assessing embryonic developmental potential. Summary of the Invention

[0005] The purpose of this application is to address the shortcomings of the prior art by providing a method, apparatus, electronic device, and storage medium for extracting features of trophoblast cells, so as to facilitate the extraction of core three-dimensional features of trophoblast cells and improve the accuracy of embryonic development potential assessment.

[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a method for feature extraction of trophoblast cells, including: Based on the multifocal target blastocyst image corresponding to the target blastocyst, determine the mask image of the inner cell mass, the mask image of the first trophoblast cell, and the mask image of the second trophoblast cell in the target blastocyst. Based on the mask image of the first trophoblast cell, the mask image of the second trophoblast cell, and the mask image of the inner cell mass, the marking information of the trophoblast cells in the target blastocyst is determined; Based on the labeling information of the trophoblast cells, surface mapping is performed on the mask image of the inner cell mass, the mask image of the first trophoblast cells, and the mask image of the second trophoblast cells to generate a surface mapping result. The surface mapping result is used to indicate the region on the surface model where the inner cell mass and trophoblast cells are located after being mapped to the surface model of the target blastocyst. Based on the surface mapping results, feature extraction is performed on the trophoblast cell region in the target blastocyst to obtain multiple three-dimensional features of the trophoblast cell region.

[0007] Optionally, determining the mask image of the inner cell mass, the first trophoblast cell mask image, and the second trophoblast cell mask image in the target blastoblast based on the multifocal target blastoblast image corresponding to the target blastoblast includes: Image fusion processing is performed on the multifocal target blastocyst image corresponding to the target blastocyst to obtain a first fused image and a second fused image; Image sharpness analysis is performed on each focal target blastocyst image to determine the target focal blastocyst image, and based on the target focal blastocyst image, the mask image of the inner cell mass is segmented to obtain the mask image. The first fused image and the second fused image are segmented into trophoblast cells respectively to obtain the first trophoblast cell mask image and the second trophoblast cell mask image.

[0008] Optionally, the step of performing image fusion processing on the multifocal target blastocyst images corresponding to the target blastocyst to obtain a first fused image and a second fused image includes: Multifocal primitive blastocyst images corresponding to the target blastocyst are acquired, and the blastocyst body is segmented in each primitive blastocyst image to obtain images of each blastocyst. Image enhancement processing was performed on each focal blastocyst image to obtain the target blastocyst image for each focal point; Based on the focal plane segmentation strategy, the images of each focal target blastocyst are grouped to obtain the first group of images and the second group of images. The focal target blastocyst images in the first group of images are fused to obtain the first fused image, and the focal target blastocyst images in the second group of images are fused to obtain the second fused image.

[0009] Optionally, the step of performing image sharpness analysis on each focal target blastocyst image to determine the target focal blastocyst image includes: Each focal target blastocyst image is divided into multiple sub-image blocks, and the image sharpness of each sub-image block corresponding to each focal target blastocyst image is determined; The image sharpness of each focal target blastocyst image is determined based on the image sharpness of each sub-image block corresponding to each focal target blastocyst image. The target focal blastocyst image is determined from the target focal blastocyst images based on the image sharpness of each focal target blastocyst image.

[0010] Optionally, determining the labeling information of trophoblast cells in the target blastocyst based on the first trophoblast cell mask image, the second trophoblast cell mask image, and the mask image of the inner cell mass includes: Based on the mask image of the first trophoblast cells and the mask image of the inner cell mass, the marking information of each trophoblast cell in the first trophoblast cell mask image is determined, and the marking information is used to indicate whether the trophoblast cell is a polar trophoblast cell or a parietal trophoblast cell. Based on the mask image of the second trophoblast cells and the mask image of the inner cell mass, the labeling information of each trophoblast cell in the mask image of the second trophoblast cells is determined.

[0011] Optionally, determining the labeling information of each trophoblast cell in the first trophoblast cell mask image based on the first trophoblast cell mask image and the inner cell mass mask image includes: Traverse each trophoblast cell in the first trophoblast cell mask image. For the current trophoblast cell, determine the overlap index between the current trophoblast cell and the inner cell mass based on the area of ​​the mask region of the current trophoblast cell and the area of ​​the mask image of the inner cell mass. The labeling information of the current trophoblast cells is determined based on the overlap index between the current trophoblast cells and the inner cell mass.

[0012] Optionally, based on the labeling information of the trophoblast cells, the step of performing surface mapping on the mask image of the inner cell mass, the first trophoblast cell mask image, and the second trophoblast cell mask image to generate a surface mapping result includes: Based on the coordinates of each pixel in the mask image of the inner cell mass, determine the coordinate information of each pixel in the mask image of the inner cell mass after mapping to the surface model of the target blastocyst; Based on the labeling information of the trophoblast cells, determine each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image; and determine each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell in the second trophoblast cell mask image. Based on each pixel corresponding to each pole trophoblast cell and each pixel corresponding to each wall trophoblast cell in the first trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. Based on each pixel corresponding to each pole trophoblast cell and each pixel corresponding to each wall trophoblast cell in the second trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. The surface mapping result is generated based on the coordinate information of each pixel in the inner cell mass mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image mapped to the surface model of the target blastocyst, and the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst.

[0013] Optionally, determining the coordinate information of each pixel in the mask image of the inner cell mass after mapping to the surface model of the target blastocyst, based on the coordinates of each pixel in the mask image of the inner cell mass, includes: Based on the coordinates of the target pixel in the mask image of the inner cell mass and the scaling factor, a first angle and a second angle are determined; the first angle is used to determine the horizontal position of the target pixel when it is mapped onto the curved surface model of the target blastocyst; the second angle is used to determine the vertical height information of the target pixel when it is mapped onto the curved surface model of the target blastocyst. Based on the first angle and the second angle, determine the coordinate information of the target pixel after it is mapped to the surface model of the target blastocyst.

[0014] Optionally, based on the surface mapping result, feature extraction is performed on the trophoblast cell region in the target blastocyst to obtain multiple three-dimensional features of the trophoblast cell region, including: Based on the surface mapping results, the surface regions corresponding to each polar trophoblast cell and each wall trophoblast cell in the surface model of the target blastocyst are determined. Based on the surface regions corresponding to each trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the trophoblast cell region: area of ​​the trophoblast cell region, perimeter of the trophoblast cell region, roundness of the trophoblast cell region, width of the trophoblast cell region, length of the trophoblast cell region, and ratio of the major and minor axes of the trophoblast cell region. Based on the surface region corresponding to each parietal trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the parietal trophoblast cell region: area of ​​the parietal trophoblast cell region, perimeter of the parietal trophoblast cell region, roundness of the parietal trophoblast cell region, width of the parietal trophoblast cell region, length of the parietal trophoblast cell region, and ratio of the minor axis to the major axis of the parietal trophoblast cell region. Based on the characteristics of the polar trophoblast cell region and the wall trophoblast cell region, multiple three-dimensional features of the trophoblast cell region were determined.

[0015] Secondly, embodiments of this application also provide a feature extraction device for trophoblast cells, including: a determination module, a generation module, and a feature extraction module; The determining module is used to determine the mask image of the inner cell mass, the mask image of the first trophoblast cell, and the mask image of the second trophoblast cell in the target blastoblast based on the multifocal target blastoblast image corresponding to the target blastoblast. The determining module is used to determine the marking information of trophoblast cells in the target blastocyst based on the first trophoblast cell mask image, the second trophoblast cell mask image, and the mask image of the inner cell mass. The generation module is used to perform surface mapping on the mask image of the inner cell mass, the first trophoblast cell mask image, and the second trophoblast cell mask image based on the labeling information of the trophoblast cells, and generate a surface mapping result. The surface mapping result is used to indicate the region on the surface model where the inner cell mass and trophoblast cells are located after being mapped to the surface model of the target blastocyst. The feature extraction module is used to extract features from the trophoblast cell region in the target blastocyst based on the surface mapping result, thereby obtaining multiple three-dimensional features of the trophoblast cell region.

[0016] Optionally, the determining module is specifically used to perform image fusion processing on the multifocal target blastocyst image corresponding to the target blastocyst to obtain a first fused image and a second fused image; Image sharpness analysis is performed on each focal target blastocyst image to determine the target focal blastocyst image, and based on the target focal blastocyst image, the mask image of the inner cell mass is segmented to obtain the mask image. The first fused image and the second fused image are segmented into trophoblast cells respectively to obtain the first trophoblast cell mask image and the second trophoblast cell mask image.

[0017] Optionally, the determining module is specifically used to acquire multifocal primitive blastocyst images corresponding to the target blastocyst, and to perform blastocyst body segmentation on each focal primitive blastocyst image to obtain each focal blastocyst image; Image enhancement processing was performed on each focal blastocyst image to obtain the target blastocyst image for each focal point; Based on the focal plane segmentation strategy, the images of each focal target blastocyst are grouped to obtain the first group of images and the second group of images. The focal target blastocyst images in the first group of images are fused to obtain the first fused image, and the focal target blastocyst images in the second group of images are fused to obtain the second fused image.

[0018] Optionally, the determining module is specifically used to divide each focal target blastocyst image into multiple sub-image blocks and determine the image sharpness of each sub-image block corresponding to each focal target blastocyst image; The image sharpness of each focal target blastocyst image is determined based on the image sharpness of each sub-image block corresponding to each focal target blastocyst image. The target focal blastocyst image is determined from the target focal blastocyst images based on the image sharpness of each focal target blastocyst image.

[0019] Optionally, the determining module is specifically used to determine the marking information of each trophoblast cell in the first trophoblast cell mask image based on the first trophoblast cell mask image and the mask image of the inner cell mass, wherein the marking information is used to indicate whether the trophoblast cell is a polar trophoblast cell or a parietal trophoblast cell. Based on the mask image of the second trophoblast cells and the mask image of the inner cell mass, the labeling information of each trophoblast cell in the mask image of the second trophoblast cells is determined.

[0020] Optionally, the determining module is specifically used to traverse each trophoblast cell in the first trophoblast cell mask image, and for the current trophoblast cell that has been traversed, to determine the overlap index between the current trophoblast cell and the inner cell mass based on the area of ​​the mask region of the current trophoblast cell and the area of ​​the mask image of the inner cell mass. The labeling information of the current trophoblast cells is determined based on the overlap index between the current trophoblast cells and the inner cell mass.

[0021] Optionally, the generation module is specifically used to determine the coordinate information of each pixel in the mask image of the inner cell mass after it is mapped to the surface model of the target blastocyst, based on the coordinates of each pixel in the mask image of the inner cell mass. Based on the labeling information of the trophoblast cells, determine each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image; and determine each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell in the second trophoblast cell mask image. Based on each pixel corresponding to each pole trophoblast cell and each pixel corresponding to each wall trophoblast cell in the first trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. Based on each pixel corresponding to each pole trophoblast cell and each pixel corresponding to each wall trophoblast cell in the second trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. The surface mapping result is generated based on the coordinate information of each pixel in the inner cell mass mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image mapped to the surface model of the target blastocyst, and the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst.

[0022] Optionally, the generation module is specifically used to determine a first angle and a second angle based on the coordinates of the target pixel in the mask image of the inner cell mass and the scaling factor; the first angle is used to determine the horizontal position of the target pixel when it is mapped onto the curved surface model of the target blastocyst; the second angle is used to determine the vertical height information of the target pixel when it is mapped onto the curved surface model of the target blastocyst. Based on the first angle and the second angle, determine the coordinate information of the target pixel after it is mapped to the surface model of the target blastocyst.

[0023] Optionally, the feature extraction module is specifically used to determine, based on the surface mapping results, the surface regions corresponding to each pole trophoblast cell and the surface regions corresponding to each wall trophoblast cell in the surface model of the target blastocyst. Based on the surface regions corresponding to each trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the trophoblast cell region: area of ​​the trophoblast cell region, perimeter of the trophoblast cell region, roundness of the trophoblast cell region, width of the trophoblast cell region, length of the trophoblast cell region, and ratio of the major and minor axes of the trophoblast cell region. Based on the surface region corresponding to each parietal trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the parietal trophoblast cell region: area of ​​the parietal trophoblast cell region, perimeter of the parietal trophoblast cell region, roundness of the parietal trophoblast cell region, width of the parietal trophoblast cell region, length of the parietal trophoblast cell region, and ratio of the minor axis to the major axis of the parietal trophoblast cell region. Based on the characteristics of the polar trophoblast cell region and the wall trophoblast cell region, multiple three-dimensional features of the trophoblast cell region were determined.

[0024] Thirdly, embodiments of this application provide an electronic device, including: a processor, a storage medium, and a bus. The storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to implement the feature extraction method for trophoblast cells as provided in the first aspect.

[0025] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the feature extraction method for trophoblast cells as provided in the first aspect.

[0026] The beneficial effects of this application are: This application provides a method, apparatus, electronic device, and storage medium for feature extraction of trophoblast cells, comprising: determining a mask image of the inner cell mass, a first trophoblast cell mask image, and a second trophoblast cell mask image in the target blastoblast based on a multifocal target blastoblast image corresponding to the target blastoblast; determining labeling information of trophoblast cells in the target blastoblast based on the first trophoblast cell mask image, the second trophoblast cell mask image, and the inner cell mass mask image; performing surface mapping on the inner cell mass mask image, the first trophoblast cell mask image, and the second trophoblast cell mask image based on the labeling information of the trophoblast cells to generate a surface mapping result; and extracting features from the trophoblast cell region in the target blastoblast based on the surface mapping result to obtain multiple three-dimensional features of the trophoblast cell region. This method distinguishes and labels polar trophoblast cells and parietal trophoblast cells within the trophoblast layer by segmenting the inner cell mass mask image and the trophoblast cell mask image. Based on the labeling information, surface mapping is performed on the inner cell mass mask image and the trophoblast cell mask image, mapping the inner cell mass and trophoblast cells onto a curved surface model of the outer surface of the blastocyst. Based on the surface mapping result, the size of each polar trophoblast cell and each parietal trophoblast cell in the three-dimensional space of the blastocyst can be accurately extracted. This allows for the calculation of three-dimensional features of the trophoblast cell region, breaking through the traditional method of feature analysis based on two-dimensional planar images. It can fully explore more core three-dimensional features of the trophoblast cells and improve the accuracy of blastocyst development potential assessment. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 1 ; Figure 2 This is a schematic diagram of the structure of a blastocyst provided in an embodiment of this application; Figure 3 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 2 ; Figure 4 This is a schematic diagram of the segmentation result of trophoblast cells provided in an embodiment of this application; Figure 5 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 3 ; Figure 6 This is a schematic diagram illustrating an image fusion result provided in an embodiment of this application; Figure 7 This is a schematic diagram illustrating another image fusion result provided in an embodiment of this application; Figure 8 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 4 ; Figure 9 This is a schematic diagram of image sharpness filtering provided in an embodiment of this application; Figure 10 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 5 ; Figure 11 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 6 ; Figure 12 This is a schematic diagram illustrating the labeling results of trophoblast cells provided in an embodiment of this application; Figure 13 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 7 ; Figure 14 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 8 ; Figure 15 A schematic diagram illustrating a surface mapping provided in an embodiment of this application; Figure 16 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 9 ; Figure 17 A schematic diagram comparing the recognition capabilities of various models provided in the embodiments of this application; Figure 18 A schematic diagram of a feature extraction device for trophoblast cells provided in an embodiment of this application; Figure 19 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0030] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0031] In assisted reproductive technology (ART), assessing embryonic developmental potential and determining implantation success are crucial aspects of ART treatment cycles. Traditional embryonic developmental potential assessment is based on morphological evaluation, relying on embryologists' subjective observation of the blastocyst in a two-dimensional plane. Its indicators mainly include the degree of blastocoel expansion, and the morphological grading of the inner cell mass (ICM) and trophoblast cells (TE). However, this two-dimensional grading method has the following limitations: 1. High subjectivity: The assessment results are highly dependent on the experience of embryologists, and there are differences among different embryologists, making it difficult to standardize and reproduce the assessment results.

[0032] 2. Incomplete Information: Two-dimensional grading cannot fully reflect the three-dimensional structure of the blastocyst, especially crucial information such as the distribution, morphology, density, and number of trophoblast cells in three-dimensional space. For example, relying solely on a two-dimensional view makes it difficult to accurately determine the integrity and density of trophoblast cells, characteristics closely related to the blastocyst's chromosome ploidy and developmental potential.

[0033] 3. Limited predictive ability: Although the existing grading system is widely used, its predictive ability for embryo implantation potential still needs to be improved, especially in predicting chromosomal aneuploidy in blastocysts. The accuracy of two-dimensional morphological assessment is insufficient, which leads to the incorrect implantation of some morphologically high-quality embryos, thereby increasing the risk of miscarriage or implantation failure.

[0034] Based on this, the trophoblast cell feature extraction method provided in this solution segments the acquired multifocal blastocyst images to obtain mask images of trophoblast cells and inner cell mass. Using the regional overlap index between the inner cell mass mask image and the trophoblast cell mask image, polar trophoblast cells and parietal trophoblast cells are accurately labeled. Based on the labeling information, surface mapping is performed on the trophoblast cell mask image and the inner cell mass mask image to obtain the mapping results of trophoblast cells and inner cell mass on the blastocyst surface model. Based on the mapping results, the core three-dimensional features of the trophoblast cell region, such as topology, spatial arrangement, area, and volume, can be accurately extracted and quantitatively analyzed in three-dimensional space. Based on the extracted three-dimensional features of the trophoblast cell region, the accuracy of blastocyst development potential assessment can be improved, providing a basis for screening high-quality blastocysts.

[0035] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0036] Figure 1 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 1 ;like Figure 1 As shown, the method may include: S101. Based on the multifocal target blastocyst image corresponding to the target blastocyst, determine the mask image of the inner cell mass, the mask image of the first trophoblast cell, and the mask image of the second trophoblast cell in the target blastocyst.

[0037] Multifocal images of the target blastocyst can be acquired using a time-varying incubator, and after blastocyst image preprocessing, multifocal target blastocyst images can be obtained. Based on the multifocal target blastocyst images, the first trophoblast cell mask image and the second trophoblast cell mask image can be segmented and extracted after image fusion; in addition, based on the multifocal target blastocyst images, the mask image of the inner cell mass can be segmented and extracted from the clear focal blastocyst image.

[0038] In the process of image fusion, the multifocal target blastocyst images above and below a certain focal plane (usually the F0 focal plane) can be fused to obtain two fused images. The trophoblast cells of the two fused images can be segmented to obtain the first trophoblast cell mask image and the second trophoblast cell mask image.

[0039] S102. Based on the mask image of the first trophoblast cell, the mask image of the second trophoblast cell, and the mask image of the inner cell mass, determine the labeling information of the trophoblast cells in the target blastocyst.

[0040] Figure 2 This is a schematic diagram of the structure of a blastocyst provided in an embodiment of this application; wherein, ① represents the cavity region of the blastocyst, ② represents the parietal trophoblast cells of the blastocyst, ③ represents the inner cell mass of the blastocyst, and ④ represents the polar trophoblast cells of the blastocyst. Typically, trophoblast cells can be divided into polar trophoblast cells and parietal trophoblast cells; polar trophoblast cells are the trophoblast cells located on the side closest to the inner cell mass; parietal trophoblast cells are the trophoblast cells located on the outer wall of the blastocyst, away from the inner cell mass.

[0041] Since the trophoblast cells and the inner cell mass have a certain positional relationship, the labeling information of the trophoblast cells in the target blastocyst can be generated by overlaying the mask images of the inner cell mass, the first trophoblast cell mask image, and the second trophoblast cell mask image.

[0042] It is worth noting that, since both the first and second trophoblast cell mask images contain multiple independent mask regions of trophoblast cells, the generated labeling information is used to indicate which mask regions in the first and second trophoblast cell mask images are polar trophoblast cells and which mask regions are parietal trophoblast cells, so as to facilitate the independent extraction of features from polar trophoblast cells and parietal trophoblast cells in the subsequent process.

[0043] S103. Based on the labeling information of the trophoblast cells, perform surface mapping on the mask image of the inner cell mass, the mask image of the first trophoblast cells, and the mask image of the second trophoblast cells to generate surface mapping results.

[0044] The surface mapping results are used to indicate the regions on the surface model where the inner cell mass and trophoblast cells are located after being mapped to the surface model of the target blastocyst.

[0045] Based on the mask image of the inner cell mass, the inner cell mass can be mapped onto the outer surface model of the target blastocyst. Based on the labeling information of the trophoblast cells, the regions containing each polar trophoblast cell and each parietal trophoblast cell in the first and second trophoblast cell mask images can be determined respectively. This allows for accurate mapping of each polar trophoblast cell and each parietal trophoblast cell onto the outer surface model of the target blastocyst, generating a surface mapping result. The surface mapping result then shows the region to which the inner cell mass, each polar trophoblast cell, and each parietal trophoblast cell belong on the surface model. This allows for accurate segmentation of the patch regions of each polar trophoblast cell and each parietal trophoblast cell from the surface model.

[0046] S104. Based on the surface mapping results, feature extraction is performed on the trophoblast cells in the target blastocyst to obtain multiple three-dimensional features of the trophoblast cells.

[0047] Based on the surface mapping results, the region sizes of the trophoblast cells at each pole and the wall trophoblast cells of interest can be determined from the surface model. Therefore, based on the region sizes of the trophoblast cells at each pole, the three-dimensional features of the trophoblast cell region can be calculated, and based on the region sizes of the wall trophoblast cells, the three-dimensional features of the wall trophoblast cell region can be calculated. Finally, these features are fused and analyzed to obtain multiple three-dimensional features of the trophoblast cell region. Based on these extracted multiple three-dimensional features of the trophoblast cell region, a pre-constructed blastocyst development potential prediction model can be used to predict the developmental potential of the target blastocyst.

[0048] In summary, the feature extraction method for trophoblast cells provided in this embodiment includes: determining a mask image of the inner cell mass, a first trophoblast cell mask image, and a second trophoblast cell mask image in the target blastoblast based on a multifocal target blastoblast image corresponding to the target blastoblast; determining the labeling information of trophoblast cells in the target blastoblast based on the first trophoblast cell mask image, the second trophoblast cell mask image, and the inner cell mass mask image; performing surface mapping on the inner cell mass mask image, the first trophoblast cell mask image, and the second trophoblast cell mask image based on the labeling information of the trophoblast cells to generate a surface mapping result; and extracting features from the trophoblast cell region in the target blastoblast based on the surface mapping result to obtain multiple three-dimensional features of the trophoblast cell region. This method distinguishes and labels polar trophoblast cells and parietal trophoblast cells within the trophoblast layer by segmenting the inner cell mass mask image and the trophoblast cell mask image. Based on the labeling information, surface mapping is performed on the inner cell mass mask image and the trophoblast cell mask image, mapping the inner cell mass and trophoblast cells onto a curved surface model of the outer surface of the blastocyst. Based on the surface mapping result, the size of each polar trophoblast cell and each parietal trophoblast cell in the three-dimensional space of the blastocyst can be accurately extracted. This allows for the calculation of three-dimensional features of the trophoblast cell region, breaking through the traditional method of feature analysis based on two-dimensional planar images. It can fully explore more core three-dimensional features of the trophoblast cells and improve the accuracy of blastocyst development potential assessment.

[0049] Figure 3 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 2 Optionally, in step S101, based on the multifocal target blastocyst image corresponding to the target blastocyst, the mask image of the inner cell mass, the mask image of the first trophoblast cell layer, and the mask image of the second trophoblast cell layer in the target blastocyst are determined, including: S201. Perform image fusion processing on the multifocal target blastocyst image corresponding to the target blastocyst to obtain a first fused image and a second fused image.

[0050] The core technology of multifocal image fusion lies in identifying and extracting the focal regions in an image to form a new, clear image. In order to improve the usability of the fused image, this solution can divide the multifocal target blastocyst image into two groups, upper and lower, with a certain focal plane as the boundary (as mentioned above, with the F0 focal plane as the boundary), so as to obtain two fused images, namely the first fused image and the second fused image.

[0051] S202. Perform image sharpness analysis on each focal target blastocyst image to determine the target focal blastocyst image, and based on the target focal blastocyst image, segment to obtain the mask image of the inner cell mass.

[0052] For the segmentation of the inner cell mass, the focal image with the best clarity can be determined from each focal target blastocyst image, which is the target focal blastocyst image. The inner cell mass has the highest clarity in the target focal blastocyst image, making it easier to segment the effective inner cell mass.

[0053] Based on the identified target blastocyst image, an inner cell mass segmentation model can be used to segment and obtain a mask image of the inner cell mass. The inner cell mass segmentation model can be implemented using a semantic segmentation network based on an upgraded version of the U-shaped network (Unet++).

[0054] S203. Perform trophoblast cell segmentation on the first fused image and the second fused image respectively to obtain the first trophoblast cell mask image and the second trophoblast cell mask image.

[0055] For the segmentation of trophoblast cells, a trophoblast cell segmentation model can be used to segment the first trophoblast cell mask image in the first fused image and the second trophoblast cell mask image in the second fused image.

[0056] In this embodiment, the trophoblast cell segmentation model can be implemented based on the CellPose algorithm, a segmentation algorithm for cells and nuclei. CellPose consists of generalist models and specialized models. Generalist models can segment a wide range of image types, while specialized models can perform precise segmentation on specific cell types. This allows the cell contour segmentation model to achieve good segmentation results with a relatively small dataset. Furthermore, the cell contour segmentation model possesses topological XY gradient graph inference, which helps determine the boundaries between cells, making it more conducive to obtaining the texture features of trophoblast cells.

[0057] Figure 4 This is a schematic diagram of the segmentation result of trophoblast cells provided in an embodiment of this application, as shown below. Figure 4 As shown in 'a', the image displayed is the first fused image / the second fused image, as shown in 'a'. Figure 4 The image shown in b is the XY gradient map obtained after segmenting the feeder cell layer of the fused image based on CellPose, as shown in the figure. Figure 4 In the diagram, 'c' represents the segmented mask image of the first trophoblast cell layer / the mask image of the second trophoblast cell layer. Figure 4 Each individual small region shown in c corresponds to a trophoblast cell obtained from the segmentation.

[0058] Figure 5 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 3 Optionally, in step S201, image fusion processing is performed on the multifocal target blastocyst image corresponding to the target blastocyst to obtain a first fused image and a second fused image, including: S301. Acquire multifocal primitive blastocyst images corresponding to the target blastocyst, and perform blastocyst body segmentation on each primitive blastocyst image to obtain images of each blastocyst.

[0059] First, the initial acquisition consists of multifocal primitive blastocyst images. Before image fusion, the main blastocyst region can be extracted from each primitive blastocyst image to remove the background from the primitive blastocyst image and retain only the blastocyst region, thus obtaining each primitive blastocyst image.

[0060] S302. Perform image enhancement processing on each focal blastocyst image to obtain the target blastocyst image for each focal blastocyst.

[0061] Subsequently, image enhancement methods, primarily Richardson-Lucy Total Variation (RLTV) and adaptive histogram equalization, are used to enhance cell edges and contrast in the image, aiming to improve the contrast of image contour features. After image enhancement, images of the focal target blastocysts are obtained.

[0062] S303. Based on the focal plane division strategy, the images of each focal target blastocyst are grouped to obtain the first group of images and the second group of images.

[0063] When performing image fusion, the F75 joule to F0 joule images of each target blastocyst can be divided into the first group of images, and the F-75 joule to F0 joule images can be divided into the second group of images.

[0064] S304. The focal target blastocyst images in the first group of images are fused to obtain a first fused image, and the focal target blastocyst images in the second group of images are fused to obtain a second fused image.

[0065] In this embodiment, an unsupervised dual U-shaped encoder fusion network (U2FusionNet) can be used for image fusion. Specifically, for each focal target blastocyst image in the first group of images, a first fused image TextMap_Up can be obtained by fusion, and for each focal target blastocyst image in the second group of images, a second fused image TextMap_Down can be obtained by fusion.

[0066] Figure 6 This is a schematic diagram illustrating an image fusion result provided in an embodiment of this application. The diagram schematically shows the fused image obtained after fusing AF multifocal target blastocyst images, as shown below. Figure 6 As shown in h, and Figure 6 The image g shown in the image represents the blastocyst at the 0 focal plane. It can be seen that the fused image h can reveal clearer internal details.

[0067] Figure 7 This is a schematic diagram illustrating another image fusion result provided in an embodiment of this application, wherein, Figure 7 The image shown in b represents the target blastocyst images at various focal lengths from the first set of images, specifically the target blastocyst images at focal lengths from F75 joules to F0 joules. The first fused image obtained after fusing the first set of images is shown below. Figure 7 As shown in 'a'; Figure 7 The image shown in 'c' represents the target blastocyst images at various focal lengths in the second set of images, specifically the target blastocyst images from F-75 joules to F0 joules. The second fused image obtained after fusing the second set of images is shown below. Figure 7 As shown in d.

[0068] Figure 8Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 4 Optionally, in step S202, image sharpness analysis is performed on each focal target blastocyst image to determine the target focal blastocyst image, including: S401. Divide each focal target blastocyst image into multiple sub-image blocks and determine the image sharpness of each sub-image block corresponding to each focal target blastocyst image.

[0069] In one implementation, each focal target blastocyst image can be divided into 5... For each sub-image block of image 5, a sharpness evaluation operator is used to calculate the sharpness of each sub-image block, thereby obtaining the image sharpness of each sub-image block corresponding to each focal target blastocyst image.

[0070] S402. Determine the image sharpness of each focal target blastocyst image based on the image sharpness of each sub-image block corresponding to each focal target blastocyst image.

[0071] Based on the image sharpness of each sub-image block corresponding to each focal target blastocyst image, the maximum sharpness of the sub-image block can be selected from each focal target blastocyst image as the image sharpness of each focal target blastocyst image.

[0072] S403. Based on the image clarity of each focal target blastocyst image, determine the target focal blastocyst image from each focal target blastocyst image.

[0073] Then, iterate through each focal blastocyst image and select the focal blastocyst image with the highest image clarity as the selected focal blastocyst image with the clearest inner cell mass.

[0074] Figure 9 This is a schematic diagram of image sharpness filtering provided in an embodiment of this application, such as... Figure 9 The image shown in 'a' is a focal target blastocyst image. Figure 9 b in the text is for... Figure 9 The sub-image blocks divided from the focal target blastocyst image of 'a' in the image. Figure 9 The value 'c' in the image shows the clarity heatmap of each sub-image patch. The higher the clarity of the sub-image patch, the brighter the corresponding heatmap area.

[0075] Figure 10 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 5 Optionally, in step S102, based on the mask image of the first trophoblast cell, the mask image of the second trophoblast cell, and the mask image of the inner cell mass, the marking information of the trophoblast cells in the target blastocyst is determined, including: S501. Based on the mask image of the first trophoblast cells and the mask image of the inner cell mass, determine the labeling information of each trophoblast cell in the mask image of the first trophoblast cells.

[0076] The labeling information is used to indicate whether the trophoblast cells are polar trophoblast cells or wall trophoblast cells.

[0077] The labeling here refers to the type labeling of each trophoblast cell in the trophoblast cell mask image, in order to distinguish between polar trophoblast cells and wall trophoblast cells.

[0078] When labeling polar trophoblast cells or parietal trophoblast cells, the masked image of the inner cell mass can be used as a vertically mapped region on the trophoblast layer. The type of trophoblast cell can be determined based on the size of the masked region of a single trophoblast cell in the trophoblast cell masked image and the masked image of the inner cell mass.

[0079] Specifically, by calculating the size of the mask region of each trophoblast cell in the first trophoblast cell mask image and the mask image of the inner cell mass, the labeling information of each trophoblast cell in the first trophoblast cell mask image can be determined.

[0080] S502. Based on the mask image of the second trophoblast cells and the mask image of the inner cell mass, determine the labeling information of each trophoblast cell in the mask image of the second trophoblast cells.

[0081] By calculating the size of the mask region of each trophoblast cell in the second trophoblast cell mask image and the mask image of the inner cell mass, the labeling information of each trophoblast cell in the second trophoblast cell mask image can be determined.

[0082] Figure 11 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 6 Optionally, in step S501, based on the mask image of the first trophoblast cells and the mask image of the inner cell mass, the labeling information of each trophoblast cell in the mask image of the first trophoblast cells is determined, including: S601. Traverse each trophoblast cell in the mask image of the first trophoblast cell. For the current trophoblast cell, determine the overlap index between the current trophoblast cell and the inner cell mass based on the area of ​​the mask region of the current trophoblast cell and the area of ​​the mask image of the inner cell mass.

[0083] like Figure 4As shown in c, each mask region in the trophoblast cell mask image corresponds to one trophoblast cell. Each trophoblast cell in the first trophoblast cell mask image can be traversed. For the current trophoblast cell, the mask region of the current trophoblast cell is extracted. Based on the area of ​​the mask region of the current trophoblast cell and the area of ​​the mask image of the inner cell mass, the intersection over union (IOU) of the mask region of the current trophoblast cell and the mask image of the inner cell mass is calculated to obtain the overlap index between the current trophoblast cell and the inner cell mass.

[0084] S602. Determine the labeling information of the current trophoblast cells based on the overlap index between the current trophoblast cells and the inner cell mass.

[0085] In some embodiments, if the overlap index (IOU) between the current trophoblast cell and the inner cell mass is greater than 0, it indicates that the current trophoblast cell and the inner cell mass overlap, and the current trophoblast cell can be identified as a polar trophoblast cell. If the overlap index (IOU) between the current trophoblast cell and the inner cell mass is equal to 0, it indicates that the current trophoblast cell and the inner cell mass do not overlap, and the current trophoblast cell can be identified as a parietal trophoblast cell.

[0086] By performing the above operations on all trophoblast cells in the trophoblast cell mask image, the specific type of all trophoblast cells can be identified.

[0087] Figure 12 This is a schematic diagram illustrating the labeling results of trophoblast cells provided in an embodiment of this application, as shown below. Figure 12 As shown, marker ① represents cell wall trophoblast cells, and marker ② represents cell polar trophoblast cells. In practical applications, the generated marker information can be stored in tabular form for easy retrieval and use.

[0088] Figure 13 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 7 Optionally, in step S103, based on the labeling information of the trophoblast cells, surface mapping is performed on the mask image of the inner cell mass, the mask image of the first trophoblast cells, and the mask image of the second trophoblast cells to generate surface mapping results, including: S701. Based on the coordinates of each pixel in the mask image of the inner cell mass, determine the coordinate information of each pixel in the mask image of the inner cell mass after mapping to the surface model of the target blastocyst.

[0089] The outer surface of the blastocyst can be approximated as an ellipsoidal surface model. When calculating the three-dimensional surface features of a single polar trophoblast cell, it is necessary to map the segmented mask images of the first trophoblast cell, the second trophoblast cell, and the inner cell mass onto the surface model of the target blastocyst. The surface model of the target blastocyst can be pre-constructed.

[0090] When performing surface mapping, a (top and bottom) hemispherical texture mapping technique can be used. Based on the two-dimensional coordinates of each pixel in the mask image of the inner cell mass, coordinate mapping processing can be performed to obtain the three-dimensional coordinate information of each pixel after mapping to the surface model of the target blastocyst. In other words, the three-dimensional coordinates of pixels in a two-dimensional planar image are found in a three-dimensional surface model.

[0091] S702. Based on the labeling information of the trophoblast cells, determine the pixel points corresponding to each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image; and determine the pixel points corresponding to each pole trophoblast cell and each wall trophoblast cell in the second trophoblast cell mask image.

[0092] Since the labeling information of trophoblast cells indicates whether each independent image region in the trophoblast cell mask image is a polar trophoblast cell or a parietal trophoblast cell, the image region where each polar trophoblast cell and the image region where each parietal trophoblast cell are located can be accurately located in the first trophoblast cell mask image based on the labeling information of trophoblast cells. Thus, based on the image region where each polar trophoblast cell is located, the pixels constituting each polar trophoblast cell can be extracted, and based on the image region where each parietal trophoblast cell is located, the pixels constituting each parietal trophoblast cell can be extracted.

[0093] Similarly, the pixels constituting each polar trophoblast cell and the pixels constituting each wall trophoblast cell can be accurately extracted from the second trophoblast cell mask image.

[0094] S703. Based on the pixels corresponding to each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image after mapping to the surface model of the target blastocyst, and the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst.

[0095] For the first trophoblast cell mask image, coordinate mapping processing can be performed on each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell to obtain the coordinate information of each pixel of each pole trophoblast cell in the first trophoblast cell mask image mapped to the surface model of the target blastocyst, and to obtain the coordinate information of each pixel of each wall trophoblast cell mapped to the surface model of the target blastocyst.

[0096] S704. Based on the pixels corresponding to each pole trophoblast cell and each wall trophoblast cell in the second trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image after mapping to the surface model of the target blastocyst, and the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst.

[0097] Similarly, for the second trophoblast cell mask image, coordinate mapping can be performed on each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell to obtain the coordinate information of each pixel of each pole trophoblast cell in the second trophoblast cell mask image mapped to the surface model of the target blastocyst, and the coordinate information of each pixel of each wall trophoblast cell mapped to the surface model of the target blastocyst.

[0098] S705. Generate a surface mapping result based on the coordinate information of each pixel in the inner cell mass mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image mapped to the surface model of the target blastocyst, and the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst.

[0099] Through the above calculations, the coordinate information of each pixel in the inner cell mass mask image, the pixels of each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image, and the pixels of each pole trophoblast cell and each wall trophoblast cell in the second trophoblast cell mask image, after being mapped to the surface model of the target blastocyst, is known. Therefore, the cell regions of the inner cell mass, each pole trophoblast cell, and each wall trophoblast cell can be directly generated on the surface model based on the mapped coordinate information, and the surface mapping result is obtained. The surface mapping result shows the region where the inner cell mass and trophoblast cells are located on the surface model.

[0100] Figure 14 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 8 Optionally, in step S701, based on the coordinates of each pixel in the mask image of the inner cell mass, the coordinate information of each pixel after mapping to the surface model of the target blastocyst is determined, including: S801. Determine the first angle and the second angle based on the coordinates of the target pixel in the mask image of the inner cell mass and the scaling factor.

[0101] The first angle is used to determine the horizontal position of the target pixel when it is mapped onto the surface model of the target blastocyst; the second angle is used to determine the vertical height information of the target pixel when it is mapped onto the surface model of the target blastocyst.

[0102] This embodiment uses the mapping process of the inner cell mass as an example to illustrate the mapping steps. The mapping steps for trophoblast cells are the same and will not be repeated here.

[0103] It is worth noting that the inner cell mass is usually located in the upper or lower hemisphere of the blastocyst. There are slight differences in the mapping formulas for the upper and lower hemispheres, so it is necessary to use the corresponding mapping formula for calculation based on the specific location of the inner cell mass.

[0104] The mapping formula for the upper hemisphere is as follows:

[0105] The mapping formula for the lower hemisphere is as follows:

[0106] in, , and These are all scaling factors, which are used to prevent stretching of planar images when mapped to ellipsoidal surfaces. = , = , , ; These are the length and width of the first fused image, TextMap_Up, respectively. These are the length and width of the second fused image TextMap_Down, respectively; a, b, and c are the semi-major axis lengths of the three axes of the surface model of the target blastocyst.

[0107] Taking the calculation of any target pixel in the mask image of the inner cell mass as an example, assume the coordinates of the target pixel are ( If v), then the first angle can be calculated based on the coordinates of the target pixel using either the upper hemisphere mapping formula or the lower hemisphere mapping formula. Second angle Among them, the first angle Also known as the horizontal angle, it represents the magnitude of the angle offset along the y-axis in a three-dimensional coordinate system; the second angle. It can be called the pitch angle, which represents the angle of offset along the X-axis in a three-dimensional coordinate system.

[0108] S802. Based on the first angle and the second angle, determine the coordinate information of the target pixel point after it is mapped to the surface model of the target blastocyst.

[0109] Based on the first and second angles and the positional relationship in the spherical coordinate system, the unit spherical coordinate information of the target pixel point mapped to the surface model of the target blastocyst can be calculated. ( , , ):

[0110] Based on the following formula, the unit spherical coordinate information is converted into ellipsoid vertex coordinates to obtain the coordinate information of the target pixel point mapped to the surface model of the target blastocyst. , , ):

[0111] For the mapping processing of the trophoblast cell mask image, since the first trophoblast cell mask image corresponds to the upper hemisphere, the upper hemisphere mapping formula can be used for calculation; since the second trophoblast cell mask image corresponds to the lower hemisphere, the lower hemisphere mapping formula can be used for calculation.

[0112] Figure 15 This is a schematic diagram illustrating a surface mapping provided in an embodiment of this application. Figure 15 The image shown as 'a' represents the masked image of the first trophoblast cell layer / the masked image of the second trophoblast cell layer. Figure 15 The 'c' in the diagram shows the mapping results of the first trophoblast cell mask image / second trophoblast cell mask image onto the surface model of the target blastocyst. Different colors in the mapping results represent different regions that correspond to a trophoblast cell. Figure 15 The diagram shown in Figure 'b' illustrates the mask region of a single trophoblast cell. Figure 15 The d in the figure shows the mapping result of a single trophoblast cell mask region onto the surface model of the target blastocyst. Figure 15 The 'e' in the figure represents the basis of Figure 15 The curved region of a single trophoblast cell extracted from d.

[0113] Figure 16 Flowchart of the feature extraction method for trophoblast cells provided in the embodiments of this application Figure 9Optionally, in step S104, based on the surface mapping results, feature extraction is performed on the trophoblast cell region in the target blastocyst to obtain multiple three-dimensional features of the trophoblast cell region, including: S901. Based on the surface mapping results, determine the surface regions corresponding to each pole trophoblast cell in the surface model of the target blastocyst and the surface regions corresponding to each wall trophoblast cell in the surface model of the target blastocyst.

[0114] In some embodiments, different regions in the surface mapping result can be traversed to segment the surface region of each trophoblast cell from the surface, including segmenting the surface region of each polar trophoblast cell and the surface region of each wall trophoblast cell.

[0115] S902. Based on the surface regions corresponding to each trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the trophoblast cell region: area of ​​the trophoblast cell region, perimeter of the trophoblast cell region, roundness of the trophoblast cell region, width of the trophoblast cell region, length of the trophoblast cell region, and ratio of the major and minor axes of the trophoblast cell region.

[0116] Optionally, the surface area, roundness, perimeter, and ratio of the major and minor axes of the polar trophoblast cell regions can be calculated based on the surface regions of each polar trophoblast cell.

[0117] S903. Based on the surface region corresponding to each parietal trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the parietal trophoblast cell region: area of ​​the parietal trophoblast cell region, perimeter of the parietal trophoblast cell region, roundness of the parietal trophoblast cell region, width of the parietal trophoblast cell region, length of the parietal trophoblast cell region, and ratio of the minor axis to the major axis of the parietal trophoblast cell region.

[0118] Similarly, based on the curved surface regions of each cell in the cell wall, the three-dimensional features of the cell wall region, such as surface area, roundness, perimeter, and ratio of major to minor axis, can be calculated.

[0119] S904. Based on the characteristics of the polar trophoblast cell region and the characteristics of the wall trophoblast cell region, determine multiple three-dimensional features of the trophoblast cell region.

[0120] The overall three-dimensional features of all polar trophoblast cells can be calculated by combining the various three-dimensional features of each polar trophoblast cell region, and the overall three-dimensional features of all wall trophoblast cells can be calculated by combining the various three-dimensional features of each wall trophoblast cell region; it is also possible to combine the overall three-dimensional features of all polar trophoblast cells and the overall three-dimensional features of all wall trophoblast cells to calculate multiple three-dimensional features of the trophoblast cell region, where the trophoblast cell region includes all polar trophoblast cells and all wall trophoblast cells.

[0121] For example, the extracted three-dimensional features include, but are not limited to: the number of polar trophoblast cells, the density of polar trophoblast cells within the inner cell cluster projection area, the mean area of ​​the curved polar trophoblast cell region, the standard deviation of the mean area of ​​the curved polar trophoblast cell region, the mean area of ​​the curved wall trophoblast cell region, the standard deviation of the mean area of ​​the curved wall trophoblast cell region, the mean area of ​​the curved trophoblast cell region, and the standard deviation of the mean area of ​​the curved trophoblast cell region.

[0122] In addition, some two-dimensional features of trophoblast cells can be extracted from the trophoblast cell mask image, such as: the average area of ​​the extreme trophoblast cell region, the average perimeter of the extreme trophoblast cell region, the average roundness of the extreme trophoblast cell region, the average length of the extreme trophoblast cell region, the average ratio of the minor axis to the major axis of the extreme trophoblast cell region, the average area of ​​the wall trophoblast cell region, the average perimeter of the wall trophoblast cell region, the average area of ​​the trophoblast cell region, and the average perimeter of the trophoblast cell region.

[0123] By utilizing the three-dimensional features of the trophoblast cell region extracted in this scheme, the developmental potential of the target blastocyst can be predicted. Experiments have shown that the three-dimensional features extracted in this scheme can provide a good predictive effect for the developmental potential of the embryo.

[0124] The following section explains the selection and construction of the prediction model used in this scheme: When constructing the prediction model, the method described above in this scheme can be used to extract the three-dimensional features of the trophoblast cells of the sample blastocysts to obtain the training sample data for constructing the model.

[0125] To ensure the convergence, stability, and accuracy of model training, the three-dimensional feature data extracted using the above method can be preprocessed. Preprocessing steps include, but are not limited to: (1) Imputation and processing of missing values. The missing values ​​that may occur during the calculation process are systematically identified, and the median imputation technique is used to fill the missing features.

[0126] (2) Outlier detection and correction. For numerical features, outlier detection is performed using a method based on the interquartile range (IQR) and the standard score (Z-Score). For detected extreme outliers, a truncation method is used to limit them to a reasonable range.

[0127] (3) Data Standardization. Numerical features are scaled to eliminate the impact of features with different dimensions on model performance. Z-score standardization is used to standardize each feature... Transform into new features with a mean of 0 and a standard deviation of 1. :

[0128] in, It is the sample mean of the feature. It is the sample standard deviation of the feature. This method can accelerate the convergence speed of the gradient descent process used for decision tree construction in the model, and ensure that features contribute equally during training.

[0129] Based on the preprocessed training sample data, seven mainstream algorithms in the field of clinical prediction can be used to construct a clinical pregnancy prediction model, which is then compared and validated. The main algorithms involved are: Logistic Regression (LR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), Stochastic Gradient Descent (SGD) classifier, and ensemble models.

[0130] Figure 17 This diagram illustrates the comparison of the recognition capabilities of various models provided in the embodiments of this application. The method uses the area under the ROC curve (AUC) as the main performance evaluation index to objectively measure the recognition capability of the model under different classification thresholds.

[0131] like Figure 17As shown, the ROC curves of each model are plotted for intuitive comparison and quantitative evaluation. The results show that the lightweight gradient booster model achieved the highest AUC value on the test set. The lightweight gradient booster model has significant advantages in handling large-scale, high-dimensional, and nonlinear data. This model employs a histogram-based tree learning method, greatly improving the speed of training and inference. Furthermore, through gradient-based one-side sampling (GOSS) and exclusive feature bundling (EFB) techniques, it effectively overcomes the computational efficiency bottleneck of traditional boosting tree algorithms when processing large-scale clinical data, thus providing a high-precision, high-efficiency, and robust optimal solution for clinical pregnancy rate prediction. Therefore, this scheme ultimately selects the lightweight gradient booster model as the core algorithm for clinical pregnancy rate prediction. By inputting the extracted three-dimensional features of the trophoblast cells of the target blastocyst into the clinical pregnancy rate prediction, the success rate of pregnancy after implantation of the target blastocyst can be predicted, thereby assessing the developmental potential of the target blastocyst.

[0132] In summary, the feature extraction method for trophoblast cells provided in this embodiment includes: determining a mask image of the inner cell mass, a first trophoblast cell mask image, and a second trophoblast cell mask image in the target blastoblast based on a multifocal target blastoblast image corresponding to the target blastoblast; determining the labeling information of the trophoblast cells in the target blastoblast based on the first trophoblast cell mask image, the second trophoblast cell mask image, and the inner cell mass mask image; performing surface mapping on the inner cell mass mask image, the first trophoblast cell mask image, and the second trophoblast cell mask image based on the labeling information of the trophoblast cells to generate a surface mapping result; and extracting features from the trophoblast cells in the target blastoblast based on the surface mapping result to obtain multiple three-dimensional features of the trophoblast cell region. This method distinguishes and labels polar trophoblast cells and parietal trophoblast cells within the trophoblast layer by segmenting the inner cell mass mask image and the trophoblast cell mask image. Based on the labeling information, surface mapping is performed on the inner cell mass mask image and the trophoblast cell mask image, mapping the inner cell mass and trophoblast cells onto a curved surface model of the outer surface of the blastocyst. Based on the surface mapping result, the size of each polar trophoblast cell and each parietal trophoblast cell in the three-dimensional space of the blastocyst can be accurately extracted. This allows for the calculation of three-dimensional features of the trophoblast cell region, breaking through the traditional method of feature analysis based on two-dimensional planar images. It can fully explore more core three-dimensional features of the trophoblast cells and improve the accuracy of blastocyst development potential assessment.

[0133] The following describes the apparatus, equipment, and storage medium used to perform the feature extraction method for trophoblast cells provided in this application. The specific implementation process and technical effects are described above and will not be repeated below.

[0134] Figure 18 This is a schematic diagram of a feature extraction device for trophoblast cells provided in an embodiment of this application. The function of this feature extraction device corresponds to the steps performed by the method described above. This device can be understood as a server, or a server processor, or it can be understood as a component that implements the functions of this application under the control of a server, independent of the aforementioned server or processor. Figure 18 As shown, the device may include: a determining module 100, a generating module 200, and a feature extraction module 300; The determination module 100 is used to determine the mask image of the inner cell mass, the mask image of the first trophoblast cell, and the mask image of the second trophoblast cell in the target blastoblast based on the multifocal target blastoblast image corresponding to the target blastoblast. The determination module 100 is used to determine the labeling information of trophoblast cells in the target blastocyst based on the mask image of the first trophoblast cells, the mask image of the second trophoblast cells, and the mask image of the inner cell mass. The generation module 200 is used to perform surface mapping on the mask image of the inner cell mass, the mask image of the first trophoblast cell, and the mask image of the second trophoblast cell based on the labeling information of the trophoblast cells, and generate surface mapping results. The surface mapping results are used to indicate the regions on the surface model where the inner cell mass and trophoblast cells are located after being mapped to the surface model of the target blastocyst. The feature extraction module 300 is used to extract features from the trophoblast cell region in the target blastocyst based on the surface mapping results, and obtain multiple three-dimensional features of the trophoblast cell region.

[0135] Optionally, the determining module 100 is specifically used to perform image fusion processing on the multifocal target blastocyst image corresponding to the target blastocyst to obtain a first fused image and a second fused image; Image sharpness analysis was performed on each focal target blastocyst image to determine the target focal blastocyst image, and based on the target focal blastocyst image, the mask image of the inner cell mass was segmented. The first fused image and the second fused image are segmented into trophoblast cells respectively to obtain the first trophoblast cell mask image and the second trophoblast cell mask image.

[0136] Optionally, the determination module 100 is specifically used to acquire multifocal primitive blastocyst images corresponding to the target blastocyst, and to perform blastocyst body segmentation on each focal primitive blastocyst image to obtain each focal blastocyst image. Image enhancement processing was performed on each focal blastocyst image to obtain the target blastocyst image for each focal point; Based on the focal plane segmentation strategy, the images of each focal target blastocyst are grouped to obtain the first group of images and the second group of images. The focal target blastocyst images in the first group of images are fused to obtain a first fused image, and the focal target blastocyst images in the second group of images are fused to obtain a second fused image.

[0137] Optionally, the determining module 100 is specifically used to divide each focal target blastocyst image into multiple sub-image blocks and determine the image sharpness of each sub-image block corresponding to each focal target blastocyst image; The image sharpness of each focal target blastocyst image is determined based on the image sharpness of each sub-image block corresponding to each focal target blastocyst image. The target focal blastocyst image is determined from the target focal blastocyst images based on the image sharpness of each focal target blastocyst image.

[0138] Optionally, the determining module 100 is specifically used to determine the marking information of each trophoblast cell in the first trophoblast cell mask image based on the first trophoblast cell mask image and the inner cell mass mask image. The marking information is used to indicate whether the trophoblast cell is a polar trophoblast cell or a parietal trophoblast cell. Based on the mask image of the second trophoblast cells and the mask image of the inner cell mass, the labeling information of each trophoblast cell in the mask image of the second trophoblast cells is determined.

[0139] Optionally, the determining module 100 is specifically used to traverse each trophoblast cell in the first trophoblast cell mask image, and for the current trophoblast cell that has been traversed, to determine the overlap index between the current trophoblast cell and the inner cell mass based on the area of ​​the mask region of the current trophoblast cell and the area of ​​the mask image of the inner cell mass. The labeling information of the current trophoblast cells is determined based on the overlap index between the current trophoblast cells and the inner cell mass.

[0140] Optionally, the generation module 200 is specifically used to determine the coordinate information of each pixel in the mask image of the inner cell mass after it is mapped to the surface model of the target blastocyst, based on the coordinates of each pixel in the mask image of the inner cell mass. Based on the labeling information of the trophoblast cells, determine the pixel points corresponding to each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image; and determine the pixel points corresponding to each pole trophoblast cell and each wall trophoblast cell in the second trophoblast cell mask image. Based on the pixels corresponding to each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. Based on each pixel corresponding to each pole trophoblast cell and each pixel corresponding to each wall trophoblast cell in the second trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. Based on the coordinate information of each pixel in the inner cell mass mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image mapped to the surface model of the target blastocyst, and the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst, a surface mapping result is generated.

[0141] Optionally, the generation module 200 is specifically used to determine a first angle and a second angle based on the coordinates of the target pixel in the mask image of the inner cell mass and the scaling factor; the first angle is used to determine the horizontal position of the target pixel when it is mapped onto the curved surface model of the target blastocyst; the second angle is used to determine the vertical height information of the target pixel when it is mapped onto the curved surface model of the target blastocyst. Based on the first angle and the second angle, determine the coordinate information of the target pixel after it is mapped to the surface model of the target blastocyst.

[0142] Optionally, the feature extraction module 300 is specifically used to determine the surface regions corresponding to each pole trophoblast cell in the surface model of the target blastocyst and the surface regions corresponding to each wall trophoblast cell in the surface model of the target blastocyst based on the surface mapping results. Based on the surface regions corresponding to each trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the trophoblast cell region: area of ​​the trophoblast cell region, perimeter of the trophoblast cell region, roundness of the trophoblast cell region, width of the trophoblast cell region, length of the trophoblast cell region, and ratio of the major and minor axes of the trophoblast cell region. Based on the surface region corresponding to each parietal trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the parietal trophoblast cell region: area of ​​the parietal trophoblast cell region, perimeter of the parietal trophoblast cell region, roundness of the parietal trophoblast cell region, width of the parietal trophoblast cell region, length of the parietal trophoblast cell region, and ratio of the minor axis to the major axis of the parietal trophoblast cell region. Based on the characteristics of the polar trophoblast cell region and the wall trophoblast cell region, multiple three-dimensional features of the trophoblast cell region were determined.

[0143] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0144] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more digital signal processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0145] The modules described above can be connected or communicate with each other via wired or wireless connections. Wired connections can include metal cables, optical fibers, hybrid cables, or any combination thereof. Wireless connections can include connections via LAN, WAN, Bluetooth, ZigBee, or NFC, or any combination thereof. Two or more modules can be combined into a single module, and any module can be divided into two or more units. Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here.

[0146] Figure 19 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The device may be a computing device with data processing capabilities.

[0147] The device includes a processor 801, a storage medium 802, and a bus 803. The storage medium 802 stores program instructions that can be executed by the processor 801. When the electronic device is running, the processor 801 and the storage medium 802 communicate through the bus 803. The processor 801 executes the program instructions to implement the feature extraction method of trophoblast cells as described in the embodiment.

[0148] The storage medium 802 stores program code, which, when executed by the processor 801, causes the processor 801 to perform various steps in the feature extraction method for trophoblast cells according to various exemplary embodiments of this application as described in the "Exemplary Methods" section above.

[0149] The processor 801 can be a general-purpose processor, such as a central processing unit (CPU), digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0150] Storage medium 802, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. The storage medium can include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type storage medium, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage medium, magnetic disk, optical disk, etc. The storage medium is any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. In the embodiments of this application, storage medium 802 can also be a circuit or any other device capable of implementing storage functions for storing program instructions and / or data.

[0151] Optionally, this application also provides a program product, such as a computer-readable storage medium, including a program that, when executed by a processor, performs the above-described method embodiments.

[0152] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0153] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0154] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.

[0155] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for feature extraction of trophoblast cells, characterized in that, include: Based on the multifocal target blastocyst image corresponding to the target blastocyst, determine the mask image of the inner cell mass, the mask image of the first trophoblast cell, and the mask image of the second trophoblast cell in the target blastocyst. Based on the mask image of the first trophoblast cell, the mask image of the second trophoblast cell, and the mask image of the inner cell mass, the marking information of the trophoblast cells in the target blastocyst is determined; Based on the labeling information of the trophoblast cells, surface mapping is performed on the mask image of the inner cell mass, the mask image of the first trophoblast cells, and the mask image of the second trophoblast cells to generate a surface mapping result. The surface mapping result is used to indicate the region on the surface model where the inner cell mass and trophoblast cells are located after being mapped to the surface model of the target blastocyst. Based on the surface mapping results, feature extraction is performed on the trophoblast cell region in the target blastocyst to obtain multiple three-dimensional features of the trophoblast cell region.

2. The method according to claim 1, characterized in that, The step of determining the mask image of the inner cell mass, the mask image of the first trophoblast cell layer, and the mask image of the second trophoblast cell layer in the target blastoblast based on the multifocal target blastoblast image corresponding to the target blastoblast includes: Image fusion processing is performed on the multifocal target blastocyst image corresponding to the target blastocyst to obtain a first fused image and a second fused image; Image sharpness analysis is performed on each focal target blastocyst image to determine the target focal blastocyst image, and based on the target focal blastocyst image, the mask image of the inner cell mass is segmented to obtain the mask image. The first fused image and the second fused image are segmented into trophoblast cells respectively to obtain the first trophoblast cell mask image and the second trophoblast cell mask image.

3. The method according to claim 2, characterized in that, The step of performing image sharpness analysis on each focal target blastocyst image to determine the target focal blastocyst image includes: Each focal target blastocyst image is divided into multiple sub-image blocks, and the image sharpness of each sub-image block corresponding to each focal target blastocyst image is determined; The image sharpness of each focal target blastocyst image is determined based on the image sharpness of each sub-image block corresponding to each focal target blastocyst image. The target focal blastocyst image is determined from the target focal blastocyst images based on the image sharpness of each focal target blastocyst image.

4. The method according to claim 1, characterized in that, The step of determining the marker information of trophoblast cells in the target blastocyst based on the first trophoblast cell mask image, the second trophoblast cell mask image, and the mask image of the inner cell mass includes: Based on the mask image of the first trophoblast cells and the mask image of the inner cell mass, the marking information of each trophoblast cell in the first trophoblast cell mask image is determined, and the marking information is used to indicate whether the trophoblast cell is a polar trophoblast cell or a parietal trophoblast cell. Based on the mask image of the second trophoblast cells and the mask image of the inner cell mass, the labeling information of each trophoblast cell in the mask image of the second trophoblast cells is determined.

5. The method according to claim 1, characterized in that, Based on the labeling information of the trophoblast cells, the process of performing surface mapping on the mask image of the inner cell mass, the first trophoblast cell mask image, and the second trophoblast cell mask image to generate a surface mapping result includes: Based on the coordinates of each pixel in the mask image of the inner cell mass, determine the coordinate information of each pixel in the mask image of the inner cell mass after mapping to the surface model of the target blastocyst; Based on the labeling information of the trophoblast cells, determine each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell in the first trophoblast cell mask image; and determine each pixel corresponding to each pole trophoblast cell and each wall trophoblast cell in the second trophoblast cell mask image. Based on each pixel corresponding to each pole trophoblast cell and each pixel corresponding to each wall trophoblast cell in the first trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. Based on each pixel corresponding to each pole trophoblast cell and each pixel corresponding to each wall trophoblast cell in the second trophoblast cell mask image, determine the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image after mapping to the surface model of the target blastocyst, and determine the coordinate information of each pixel corresponding to each wall trophoblast cell after mapping to the surface model of the target blastocyst. The surface mapping result is generated based on the coordinate information of each pixel in the inner cell mass mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the first trophoblast cell mask image mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst, the coordinate information of each pixel corresponding to each pole trophoblast cell in the second trophoblast cell mask image mapped to the surface model of the target blastocyst, and the coordinate information of each pixel corresponding to each wall trophoblast cell mapped to the surface model of the target blastocyst.

6. The method according to claim 5, characterized in that, The step of determining the coordinate information of each pixel in the mask image of the inner cell mass after mapping it to the surface model of the target blastocyst, based on the coordinates of each pixel in the mask image of the inner cell mass, includes: Based on the coordinates of the target pixel in the mask image of the inner cell mass and the scaling factor, a first angle and a second angle are determined; the first angle is used to determine the horizontal position of the target pixel when it is mapped onto the curved surface model of the target blastocyst; the second angle is used to determine the vertical height information of the target pixel when it is mapped onto the curved surface model of the target blastocyst. Based on the first angle and the second angle, determine the coordinate information of the target pixel after it is mapped to the surface model of the target blastocyst.

7. The method according to claim 1, characterized in that, Based on the surface mapping result, feature extraction is performed on the trophoblast cell region in the target blastocyst to obtain multiple three-dimensional features of the trophoblast cell region, including: Based on the surface mapping results, the surface regions corresponding to each polar trophoblast cell and each wall trophoblast cell in the surface model of the target blastocyst are determined. Based on the surface regions corresponding to each trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the trophoblast cell region: area of ​​the trophoblast cell region, perimeter of the trophoblast cell region, roundness of the trophoblast cell region, width of the trophoblast cell region, length of the trophoblast cell region, and ratio of the major and minor axes of the trophoblast cell region. Based on the surface region corresponding to each parietal trophoblast cell in the surface model of the target blastocyst, determine at least one of the following characteristics of the parietal trophoblast cell region: area of ​​the parietal trophoblast cell region, perimeter of the parietal trophoblast cell region, roundness of the parietal trophoblast cell region, width of the parietal trophoblast cell region, length of the parietal trophoblast cell region, and ratio of the minor axis to the major axis of the parietal trophoblast cell region. Based on the characteristics of the polar trophoblast cell region and the wall trophoblast cell region, multiple three-dimensional features of the trophoblast cell region were determined.

8. A device for extracting the characteristics of trophoblast cells, characterized in that, include: The module consists of a determination module, a generation module, and a feature extraction module; The determining module is used to determine the mask image of the inner cell mass, the mask image of the first trophoblast cell, and the mask image of the second trophoblast cell in the target blastoblast based on the multifocal target blastoblast image corresponding to the target blastoblast. The determining module is used to determine the marking information of trophoblast cells in the target blastocyst based on the first trophoblast cell mask image, the second trophoblast cell mask image, and the mask image of the inner cell mass. The generation module is used to perform surface mapping on the mask image of the inner cell mass, the first trophoblast cell mask image, and the second trophoblast cell mask image based on the labeling information of the trophoblast cells, and generate a surface mapping result. The surface mapping result is used to indicate the region on the surface model where the inner cell mass and trophoblast cells are located after being mapped to the surface model of the target blastocyst. The feature extraction module is used to extract features from the trophoblast cell region in the target blastocyst based on the surface mapping result, thereby obtaining multiple three-dimensional features of the trophoblast cell region.

9. An electronic device, characterized in that, include: The device includes a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to implement the feature extraction method for trophoblast cells as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is executed by a processor to implement the feature extraction method for trophoblast cells as described in any one of claims 1 to 7.

Images