Embryo three-dimensional reconstruction method and device, electronic equipment and storage medium

Abstract

The invention provides an embryo three-dimensional reconstruction method and device, electronic equipment and a storage medium, and relates to the technical field of embryo quality evaluation. Comprising the following steps: acquiring a multi-focal original plane image of a blastocyst to be reconstructed based on a time difference incubator; adopting a pre-trained image frame interpolation model to perform iterative frame interpolation processing on the multi-focal original plane image to generate a multi-focal extended plane image corresponding to each iteration; according to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, adopting an auxiliary circle marking algorithm to mark a blastocyst main body area in each focus original plane image and each focus extended plane image; respectively extracting mask images of a plurality of inner cell clusters and mask images of a plurality of trophoblast cells on the basis of marking results; and performing three-dimensional reconstruction according to the extracted mask images of the internal cell clusters and the mask images of the trophoblast cells to obtain curved surface models of the internal cell clusters and curved surface models of the trophoblast cells. The method improves the reconstruction precision.

Description

Technical Field

[0001] The present application relates to the technical field of embryo quality assessment, and in particular to a method, device, electronic device and storage medium for three-dimensional reconstruction of an embryo. Background Art

[0002] Embryo reconstruction usually refers to the use of modern biotechnology to artificially intervene and reconstruct embryos or embryo-like structures. Embryo reconstruction can restore the structure and internal details of the embryo based on the extracted embryo features. Based on the reconstructed embryo model, it can assist experts in more accurately evaluating the development of the embryo and selecting high-quality embryos.

[0003] Existing embryo reconstruction methods mainly rely on fluorescent staining. However, the uniformity of staining is difficult to control, and the potential toxicity of the stain will seriously affect the normal development and growth of the embryo. Summary of the invention

[0004] The purpose of the present application is to provide a method, device, electronic device and storage medium for three-dimensional reconstruction of embryos in order to improve the effect of embryo reconstruction and the efficiency of reconstruction.

[0005] To achieve the above purpose, the technical solution adopted in the embodiment of the present application is as follows:

[0006] In a first aspect, the present application provides a method for three-dimensional reconstruction of an embryo, comprising:

[0007] Acquire multi-focus original plane images of the blastocyst to be reconstructed based on the time difference incubator, wherein the acquisition interval between each focus original plane image is a preset interval;

[0008] The pre-trained image interpolation model is used to iteratively interpolate the multi-focal original plane image to generate the multi-focal extended plane image corresponding to each round of iteration;

[0009] According to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, an auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focus original plane image and each focus extended plane image, and based on the marking results, mask images of multiple inner cell masses and mask images of multiple trophoblast cells are respectively extracted;

[0010] Three-dimensional reconstruction is performed based on the extracted mask images of each inner cell mass and the extracted mask images of each trophoblast cell to obtain a surface model of the inner cell mass and a surface model of the trophoblast cell.

[0011] Optionally, the method of using a pre-trained image interpolation model to perform iterative interpolation processing on the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration includes:

[0012] The multi-focus original plane image is input into the image interpolation model, and the image interpolation model performs iterative interpolation processing according to the original plane images of each focus and iterative control conditions to generate multi-focus extended plane images corresponding to each round of iteration; the iterative control conditions include: the number of iterations and the amount of interpolation data in each round of iteration.

[0013] Optionally, the inputting the multi-focus original plane image into the image interpolation model, and the image interpolation model performing iterative interpolation processing according to each focus original plane image and an iteration control condition to generate a multi-focus extended plane image corresponding to each round of iteration, comprises:

[0014] Inputting the multi-focus original plane images into the image interpolation model, and extracting the image features of the original plane images of each focus by the image interpolation model;

[0015] According to the image features of the extracted original plane images of each focus, according to the amount of interpolation data of the current round of iteration, the current round of iterative interpolation processing is performed to generate a multi-focus extended plane image corresponding to the current round of iteration; and according to the multi-focus extended plane image and the multi-focus original plane image generated by the current round of iteration, the current multi-focus original plane image is updated;

[0016] The number of current total plane images is determined based on the multifocal extended plane image corresponding to the current round of iteration and the multifocal original plane image; if the number of total plane images does not reach the preset required number, the next round of iterative interpolation processing is iteratively performed until the number of total plane images reaches the preset required number, then the iteration is stopped, and the currently generated multifocal extended plane image is used as the multifocal extended plane image.

[0017] Optionally, the training process of the image interpolation model includes:

[0018] Collect multi-focal sample plane images corresponding to multiple blastocysts;

[0019] Inputting each focal sample plane image into an initial image interpolation model, extracting image features of each focal sample plane image by the initial image interpolation model, and predicting and generating a multi-focus new plane image according to the image features of each focal sample plane image;

[0020] Determine the model loss value of the current round according to the image features of each focal new plane image generated in the current round and the image features of each focal sample plane image;

[0021] According to the model loss value of the current round, the parameters of the current initial image interpolation model are optimized until the model loss value of the current round meets the preset loss threshold, and the current initial image interpolation model is used as the image interpolation model.

[0022] Optionally, the method uses an auxiliary circle marking algorithm to mark the blastocyst main body area in each focal original plane image and each focal extended plane image according to the multifocal original plane image and the multifocal extended plane image corresponding to each round of iteration, and extracts mask images of multiple inner cell masses and mask images of multiple trophoblast cells based on the marking results, including:

[0023] Generate first target images of each focus corresponding to the inner cell mass according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the first finger wheel iteration;

[0024] Generate a second target image of each focus corresponding to the trophoblast cells according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the second finger wheel iteration;

[0025] Using an auxiliary circle marking algorithm, marking the blastocyst main body region in each focal first target image corresponding to the inner cell mass, and extracting the mask image of the inner cell mass in each focal first target image from the marked blastocyst main body region;

[0026] The auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focal second target image corresponding to the trophoblast cell, and the mask image of the trophoblast cell in each focal second target image is extracted from the marked blastocyst main body area.

[0027] Optionally, the using of an auxiliary circle marking algorithm to mark the blastocyst main body region in each focused first target image corresponding to the inner cell mass includes:

[0028] According to the radius of the central focal plane image and the length of the blastocyst in the vertical direction, an ellipse fitting is performed on the vertical section of the blastocyst to obtain an ellipse fitting function;

[0029] Determining the radius of the auxiliary circle corresponding to each focused first target image according to the focal plane depth where each focused first target image is located and the ellipse fitting function;

[0030] Taking the center point of each focused first target image as the origin, an auxiliary circle is drawn according to the auxiliary circle radius corresponding to each focused first target image, the main area of ​​the blastocyst in each focused first target image is determined, and the main area of ​​the blastocyst in each focused first target image is marked to obtain each first marked image.

[0031] Optionally, extracting the mask image of the inner cell mass in each focused first target image from the marked blastocyst main body region comprises:

[0032] Performing deconvolution enhancement processing and histogram equalization processing on each first labeled image to obtain each enhanced first labeled image;

[0033] The inner cell clusters in each enhanced first labeled image are segmented using a pre-trained inner cell cluster segmentation model to obtain mask images of multiple inner cell clusters.

[0034] Optionally, before the iterative interpolation process is performed on the multi-focus original plane image using the pre-trained image interpolation model, the method further includes:

[0035] Segmenting the blastocyst region in the central focal plane image in the acquired multi-focus original plane image to obtain a blastocyst reference mask image;

[0036] According to the blastocyst reference mask image, the remaining focal original plane images are preprocessed to obtain preprocessed focal original plane images.

[0037] Optionally, preprocessing the remaining focal original plane images according to the blastocyst reference mask image to obtain the preprocessed focal original plane images includes:

[0038] Taking the blastocyst reference mask image as a reference, translating the center points of the remaining focal original plane images to obtain translated focal original plane images;

[0039] The boundaries of the translated original plane images of each focus are filled to obtain the preprocessed original plane images of each focus.

[0040] Optionally, after obtaining the surface model of the inner cell mass, the method further comprises:

[0041] According to the surface model of the inner cell mass, extracting at least one feature data of the inner cell mass under the surface model;

[0042] According to the inner cell mass fluorescence model obtained by fluorescence labeling, extracting various corresponding characteristic data of the inner cell mass under the fluorescence model;

[0043] The surface model of the inner cell mass is evaluated according to at least one feature data of the inner cell mass under the surface model and each corresponding feature data of the inner cell mass under the fluorescence model.

[0044] In a second aspect, the embodiment of the present application further provides an embryo three-dimensional reconstruction device, comprising: an acquisition module, a processing module, and a reconstruction module;

[0045] The acquisition module is used to acquire multi-focus original plane images of the blastocyst to be reconstructed based on the time difference incubator, and the acquisition interval between each focus original plane image is a preset interval;

[0046] The processing module is used to use a pre-trained image interpolation model to perform iterative interpolation processing on the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration;

[0047] The processing module is used to mark the blastocyst main body area in each focal original plane image and each focal extended plane image according to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, and extract the mask images of multiple inner cell masses and the mask images of multiple trophoblast cells respectively based on the marking results;

[0048] The reconstruction module is used to perform three-dimensional reconstruction according to the extracted mask images of each inner cell mass and the mask images of each trophoblast cell, so as to obtain the surface model of the inner cell mass and the surface model of the trophoblast cell.

[0049] Optionally, the processing module is specifically used to input the multi-focus original plane image into the image interpolation model, and the image interpolation model performs iterative interpolation processing according to the original plane images of each focus and iterative control conditions to generate multi-focus extended plane images corresponding to each round of iteration; the iterative control conditions include: the number of iterations and the amount of interpolation data in each round of iteration.

[0050] Optionally, the processing module is specifically used to input the multi-focus original plane image into the image interpolation model, and the image interpolation model extracts the image features of the original plane image of each focus;

[0051] According to the image features of the extracted original plane images of each focus, according to the amount of interpolation data of the current round of iteration, the current round of iterative interpolation processing is performed to generate a multi-focus extended plane image corresponding to the current round of iteration; and according to the multi-focus extended plane image and the multi-focus original plane image generated by the current round of iteration, the current multi-focus original plane image is updated;

[0052] The number of current total plane images is determined based on the multifocal extended plane image corresponding to the current round of iteration and the multifocal original plane image; if the number of total plane images does not reach the preset required number, the next round of iterative interpolation processing is iteratively performed until the number of total plane images reaches the preset required number, then the iteration is stopped, and the currently generated multifocal extended plane image is used as the multifocal extended plane image.

[0053] Optionally, a training module is also included;

[0054] The training module is used to collect multi-focus sample plane images corresponding to multiple blastocysts; input each focus sample plane image into an initial image interpolation model, extract the image features of each focus sample plane image by the initial image interpolation model, and predict and generate multi-focus new plane images based on the image features of each focus sample plane image; determine the model loss value of the current round based on the image features of each focus new plane image generated in the current round and the image features of each focus sample plane image; according to the model loss value of the current round, optimize the parameters of the current initial image interpolation model until the model loss value of the current round meets the preset loss threshold, and then use the current initial image interpolation model as the image interpolation model.

[0055] Optionally, the processing module is specifically configured to generate each focal first target image corresponding to the inner cell mass according to the multifocal original plane image and the multifocal extended plane image corresponding to the first fingerwheel iteration;

[0056] Generate a second target image of each focus corresponding to the trophoblast cells according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the second finger wheel iteration;

[0057] Using an auxiliary circle marking algorithm, marking the blastocyst main body region in each focal first target image corresponding to the inner cell mass, and extracting the mask image of the inner cell mass in each focal first target image from the marked blastocyst main body region;

[0058] The auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focal second target image corresponding to the trophoblast cell, and the mask image of the trophoblast cell in each focal second target image is extracted from the marked blastocyst main body area.

[0059] Optionally, the processing module is specifically configured to perform ellipse fitting on a vertical section of the blastocyst according to a radius of the central focal plane image and a length of the blastocyst in a vertical direction to obtain an ellipse fitting function;

[0060] Determining the radius of the auxiliary circle corresponding to each focused first target image according to the focal plane depth where each focused first target image is located and the ellipse fitting function;

[0061] Taking the center point of each focused first target image as the origin, an auxiliary circle is drawn according to the auxiliary circle radius corresponding to each focused first target image, the main area of ​​the blastocyst in each focused first target image is determined, and the main area of ​​the blastocyst in each focused first target image is marked to obtain each first marked image.

[0062] Optionally, the processing module is specifically used to perform deconvolution enhancement processing and histogram equalization processing on each first marked image to obtain each enhanced first marked image;

[0063] The inner cell clusters in each enhanced first labeled image are segmented using a pre-trained inner cell cluster segmentation model to obtain mask images of multiple inner cell clusters.

[0064] Optionally, the processing module is further used to segment the blastocyst region in the central focal plane image in the acquired multi-focus original plane image to obtain a blastocyst reference mask image;

[0065] According to the blastocyst reference mask image, the remaining focal original plane images are preprocessed to obtain preprocessed focal original plane images.

[0066] Optionally, the processing module is specifically used to translate the center points of the remaining focal original plane images based on the blastocyst reference mask image to obtain translated focal original plane images;

[0067] The boundaries of the translated original plane images of each focus are filled to obtain the preprocessed original plane images of each focus.

[0068] Optionally, it also includes: an analysis module;

[0069] The analysis module is used to extract at least one feature data of the inner cell mass under the surface model according to the surface model of the inner cell mass; extract each corresponding feature data of the inner cell mass under the fluorescence model according to the fluorescence model of the inner cell mass obtained by fluorescent labeling; and evaluate the surface model of the inner cell mass according to at least one feature data of the inner cell mass under the surface model and each corresponding feature data of the inner cell mass under the fluorescence model.

[0070] In a third aspect, an embodiment of the present application provides an electronic device comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the storage medium communicate via the bus, and the processor executes the machine-readable instructions to implement the embryo three-dimensional reconstruction method provided in the first aspect.

[0071] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for three-dimensional reconstruction of an embryo as provided in the first aspect is executed.

[0072] The beneficial effects of this application are:

[0073] The present application provides a method, device, electronic device and storage medium for three-dimensional reconstruction of an embryo, comprising: obtaining multi-focus original plane images of a blastocyst to be reconstructed based on a time-difference incubator, wherein the acquisition interval between each focus original plane image is a preset interval; using a pre-trained image interpolation model to iteratively interpolate the multi-focus original plane images to generate multi-focus extended plane images corresponding to each round of iteration; using an auxiliary circle marking algorithm based on the multi-focus original plane images and the multi-focus extended plane images corresponding to each round of iteration, marking the main area of ​​the blastocyst in each focus original plane image and each focus extended plane image, and extracting mask images of multiple inner cell masses and mask images of multiple trophoblast cells based on the marking results; performing three-dimensional reconstruction based on the extracted mask images of each inner cell mass and the extracted mask images of each trophoblast cell to obtain a surface model of the inner cell mass and a surface model of the trophoblast cell. This method obtains multi-focus original plane images by using an HMC imaging system under a time-difference culture system, and then simulates and generates multi-focus extended plane images based on the multi-focus original plane images based on the image interpolation algorithm, so as to perform three-dimensional reconstruction of the embryo based on the multi-focus original plane images and the multi-focus extended plane images. Among them, the method of obtaining images using a time-difference culture system is more convenient and safer; and through image interpolation processing, frame plane filling can be performed between multi-focus original plane images, so that information connection and transition of adjacent focal original plane images can be performed through extended plane images, so that the data used to reconstruct the model is more comprehensive and rich; and the extraction of the image region of interest by the auxiliary circle marking algorithm can eliminate the interference of the artifact image, and through two image enhancement processes, the clarity of the extracted image of the region of interest can be increased, so that the details in the image are more obvious, and the accuracy of the subsequent segmentation of the mask image is improved, thereby improving the accuracy of the embryo three-dimensional reconstruction. Compared with traditional two-dimensional reconstruction, the three-dimensional reconstruction effect can better display the stereoscopic structure and internal details of the embryo, and compared with fluorescent staining or other optical imaging techniques, it will not have adverse effects on the development and growth of the embryo. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0075] Figure 1 A schematic diagram of a process for three-dimensional reconstruction of an embryo provided in an embodiment of the present application;

[0076] Figure 2 A schematic diagram of another process for three-dimensional reconstruction of embryos provided in an embodiment of the present application;

[0077] Figure 3 A schematic diagram of a process flow of another embryo three-dimensional reconstruction method provided in an embodiment of the present application;

[0078] Figure 4 A schematic diagram of another process for three-dimensional reconstruction of embryos provided in an embodiment of the present application;

[0079] Figure 5 A schematic diagram of another process for three-dimensional reconstruction of embryos provided in an embodiment of the present application;

[0080] Figure 6 A schematic diagram of a process of extracting a mask image of a trophoblast cell provided in an embodiment of the present application;

[0081] Figure 7 A schematic diagram of a process flow of another embryo three-dimensional reconstruction method provided in an embodiment of the present application;

[0082] Figure 8 A schematic diagram of a comparison before and after image enhancement processing provided in an embodiment of the present application;

[0083] Fig. 9 A schematic diagram of a mask image provided in an embodiment of the present application;

[0084] Fig.10 A schematic diagram of another mask image provided in an embodiment of the present application;

[0085] Fig.11 A schematic diagram of another process for three-dimensional reconstruction of embryos provided in an embodiment of the present application;

[0086] Fig.12 A schematic diagram of a process flow of another embryo three-dimensional reconstruction method provided in an embodiment of the present application;

[0087] Fig.13 A schematic diagram of image pre-processing comparison provided in an embodiment of the present application;

[0088] Fig.14 A schematic diagram of a process flow of another embryo three-dimensional reconstruction method provided in an embodiment of the present application;

[0089] Fig.15 A schematic diagram of an embryo reconstruction result provided in an embodiment of the present application;

[0090] Fig.16 A schematic diagram of an embryo three-dimensional reconstruction device provided in an embodiment of the present application;

[0091] Fig.17 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0092] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0093] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0094] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0095] Embryo selection plays a vital role in embryo transplantation. By observing the blastomere morphology, growth rate, cell division, blastocyst size, trophoblast cell and inner cell mass morphology of embryos in the cleavage or blastocyst stage, embryologists can evaluate the developmental potential of the embryo through consensus based on these observed characteristics and select high-quality embryos to improve the embryo implantation rate and pregnancy success rate. The current mainstream method of obtaining embryo image data is to obtain multi-focal plane data of the embryo through a time-difference culture system and an imaging system, and embryologists evaluate through two-dimensional images.

[0096] Embryo reconstruction technology uses modern biotechnology to artificially intervene in and reconstruct embryos or embryo-like structures. Embryo reconstruction can restore the structure and internal details of the embryo based on the extracted embryonic characteristics, thereby assisting in the study of the embryo.

[0097] Existing methods for embryo reconstruction mainly rely on fluorescent staining for reconstruction through confocal microscopy. Fluorescent staining reconstruction technology has obvious shortcomings in the three-dimensional reconstruction of embryos. The potential toxicity of the dye will seriously affect the normal development and growth of the embryo, and the uniformity of the staining is difficult to control. The imaging equipment is complex to operate and the cost is high. Even from a scientific research perspective, obtaining a clear and complete three-dimensional embryo model is still a complex and difficult process.

[0098] In addition to fluorescent staining methods, there are also technologies that use other optical imaging systems, such as OCT (Optical Coherence Tomography). However, OCT technology relies on the propagation of light and has limited imaging depth. In embryo reconstruction, this may not be enough to capture detailed information about deep embryonic structures. OCT imaging mainly relies on the sample's absorption and scattering characteristics of light, which may lead to insufficient contrast in some embryonic tissues and make them difficult to distinguish. In addition, the phototoxicity of the embryo must be considered. Although OCT is a non-invasive technology, long-term exposure will cause certain interference to the embryo, which can easily have an adverse effect on the development and growth of the embryo.

[0099] Based on this, this scheme provides a method for three-dimensional reconstruction of embryos, in which a multi-focus original plane image is obtained through an HMC (Hoffman Modulation Contrast) imaging system under a time-difference culture system, and then based on an image interpolation algorithm, a multi-focus extended plane image is simulated and generated on the basis of the multi-focus original plane image, thereby performing three-dimensional reconstruction of the embryo based on the multi-focus original plane image and the multi-focus extended plane image. Through image interpolation processing, frame plane filling can be performed between multi-focus original plane images, so that information connection and transition of adjacent focal original plane images can be performed through extended plane images, thereby improving the accuracy of three-dimensional reconstruction of embryos. Compared with traditional two-dimensional reconstruction, the three-dimensional reconstruction effect can better display the stereoscopic structure and internal details of the embryo, and compared with fluorescent staining or other optical imaging techniques, it will not adversely affect the development and growth of the embryo.

[0100] Figure 1 A schematic diagram of a flow chart of a method for three-dimensional reconstruction of an embryo provided in an embodiment of the present application; the execution subject of the method may be a computer device, such as Figure 1 As shown, the method may include:

[0101] S101. Acquire multi-focus original plane images of the blastocyst to be reconstructed based on a time difference incubator, wherein the acquisition interval between each focus original plane image is a preset interval.

[0102] Optionally, a time-difference incubator can be used to obtain multi-focal original plane images of the blastocyst that needs to be reconstructed using HMC bright field imaging technology. Compared with confocal microscopy or other microscopy imaging, the method of obtaining images under a time-difference system is simpler and safer.

[0103] The interval between each two adjacent focal planes in the collected multi-focus original plane image may be 15 um. The F0 focal plane may be used as the central focal plane, and the multi-focus original plane image may be collected above the F0 focal plane, and the multi-focus original plane image may be collected below the F0 focal plane.

[0104] In one achievable manner, 5 original plane images may be collected above the F0 focal plane, and 5 original plane images may also be collected below the F0 focal plane, that is, 11 focal plane original plane images may be collected. However, in actual applications, the number of collected images is not limited to 11 focal plane original images, and the number of collected images may be adaptively adjusted.

[0105] According to the existing consensus: the blastocyst can be divided into 6 periods according to the degree of expansion and hatching of the blastocyst. In this plan, the blastocyst in the third or fourth stage will be given priority for reconstruction, that is, the stage from when the cyst cavity expands more than 1 / 2 of the total volume to when the blastocyst is fully expanded. This stage is the critical time period for morphological evaluation of the blastocyst.

[0106] S102: using a pre-trained image interpolation model, performing iterative interpolation processing on the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration.

[0107] In some embodiments, the multi-focus original plane image can be used as input data and input into a pre-trained image interpolation model, so that the image interpolation model can generate an intermediate focal plane image of adjacent focal original plane images based on image feature simulation between upper and lower adjacent focal original plane images, thereby obtaining a multi-focus extended plane image by interpolating the multi-focus original plane image.

[0108] Optionally, considering the different image requirements for reconstruction of the inner cell mass and trophoblast cells, the image interpolation model can perform multiple rounds of interpolation processing, and in each round of iteration, a corresponding number of multifocal extended plane images can be generated. Therefore, when reconstructing, all plane images after any round of iteration can be used as the data basis for reconstruction according to the needs.

[0109] S103. According to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, an auxiliary circle marking algorithm is used to mark the main area of ​​the blastocyst in each focal original plane image and each focal extended plane image, and based on the marking results, mask images of multiple inner cell masses and mask images of multiple trophoblast cells are respectively extracted.

[0110] Optionally, based on the obtained multi-focal original plane images and multi-focal extended plane images, image region of interest extraction can be performed to mark and extract the blastocyst main body area in each focal original plane image and each focal extended plane image. The blastocyst main body area may refer to the clear area inside the blastocyst.

[0111] Due to the particularity of HMC bright field imaging, in order to eliminate the interference of artifact images between adjacent layers, an auxiliary circle marking algorithm can be used in this embodiment to mark the main area of ​​the blastocyst in each focal original plane image and each focal expanded plane image, so as to extract multiple mask images of the inner cell mass and multiple mask images of the trophoblast cells based on the marking results.

[0112] S104 , performing three-dimensional reconstruction based on the extracted mask images of each inner cell mass and the mask images of each trophoblast cell to obtain a surface model of the inner cell mass and a surface model of the trophoblast cell.

[0113] Optionally, the contour of the inner cell mass can be extracted based on the extracted mask images of each inner cell mass, thereby fitting a point cloud model of the inner cell mass based on the contour data, and finally obtaining a surface model of the inner cell mass based on the point cloud model of the inner cell mass. The surface model of the inner cell mass is also the inner cell mass model after three-dimensional reconstruction.

[0114] Similarly, the contours of the trophoblast cells can be extracted based on the extracted mask images of each trophoblast cell, thereby simulating a point cloud model of the trophoblast cells based on the contour data, and finally obtaining a surface model of the trophoblast cells based on the point cloud model of the trophoblast cells. The surface model of the trophoblast cells is also the trophoblast cell model after three-dimensional reconstruction.

[0115] In summary, the embryo three-dimensional reconstruction method provided in this embodiment includes: obtaining a multi-focus original plane image of the blastocyst to be reconstructed based on a time difference incubator, and the acquisition interval between each focus original plane image is a preset interval; using a pre-trained image interpolation model, iterative interpolation processing is performed on the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration; according to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, an auxiliary circle marking algorithm is used to mark the main area of ​​the blastocyst in each focus original plane image and each focus extended plane image, and based on the marking result, a plurality of mask images of inner cell masses and a plurality of mask images of trophoblast cells are respectively extracted; according to the extracted mask images of each inner cell mass and the mask images of each trophoblast cell, a three-dimensional reconstruction is performed to obtain a surface model of the inner cell mass and a surface model of the trophoblast cell. This method obtains a multi-focus original plane image with an HMC imaging system under a time difference incubation system, and then simulates and generates a multi-focus extended plane image based on the multi-focus original plane image based on the image interpolation algorithm, thereby performing three-dimensional reconstruction of the embryo based on the multi-focus original plane image and the multi-focus extended plane image. Among them, the method of acquiring images using a time-difference culture system is more convenient and safer; and through image interpolation processing, frame plane filling can be performed between multi-focus original plane images, so that information connection and transition of adjacent focus original plane images can be performed through extended plane images, making the data used to reconstruct the model more comprehensive and rich; and through the auxiliary circle marking algorithm to extract the image area of ​​interest, the interference of artifact images can be eliminated, and the accuracy of subsequent mask image segmentation can be improved, thereby improving the accuracy of embryo three-dimensional reconstruction. Compared with traditional two-dimensional reconstruction, the three-dimensional reconstruction effect can better display the three-dimensional structure and internal details of the embryo, and compared with fluorescent staining or other optical imaging techniques, it will not have adverse effects on the development and growth of the embryo.

[0116] Optionally, in step S102, a pre-trained image interpolation model is used to iteratively interpolate the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration, which may include: inputting the multi-focus original plane image into the image interpolation model, and the image interpolation model performs iterative interpolation processing according to the original plane images of each focus and an iterative control condition to generate a multi-focus extended plane image corresponding to each round of iteration; the iterative control condition includes: the number of iterations and the amount of interpolation data in each round of iteration.

[0117] It should be noted that the image interpolation model used in this embodiment can be constructed using an anti-generation network model. The network consists of a generator (G) and a discriminator (D). The generator uses a U-net codec network as the backbone network. The U-net network structure is a fully convolutional symmetric network structure. The convolution layer on the left side of the structure is composed of a downsampling layer, and the right side of the structure is composed of an upsampling layer composed of deconvolution. Compared with the general U-Net network structure, the generator will perform batch normalization after each convolution layer. The discriminator (D) uses ResNet-50 as the backbone network model. The discriminator is used to distinguish between the simulated plane image generated by G and the multi-focus original plane image. In the training of the adversarial generation network, corresponding loss functions will be set for G and D. During the network training process, G and D are optimized simultaneously to ensure that the final generated simulated plane image is close to the original plane image.

[0118] In one implementation, the multi-focus original plane image may be input into an image interpolation model, and the image interpolation model performs iterative interpolation processing according to a set iterative control condition.

[0119] Optionally, the iteration control condition may include: the number of iterations and the amount of interpolation frame data per iteration, wherein the number of iterations is used to control the number of iterations, and the amount of interpolation frame data per iteration is used to indicate the number of extended plane images to be generated in each iteration.

[0120] For example, if the amount of interpolated frame data in the first round of iteration is 10, then after the first round of iteration, 10 focal length expansion plane images can be generated.

[0121] Optionally, the image interpolation model may learn image features of adjacent focal plane images based on image features of the multi-focus original plane image, so as to generate an intermediate layer extended plane image of the adjacent focal plane images.

[0122] Figure 2 A schematic flow chart of another embryo 3D reconstruction method provided in an embodiment of the present application; Optionally, in the above steps, the multi-focal original plane image is input into the image interpolation model, and the image interpolation model performs iterative interpolation processing according to each focal original plane image and iterative control conditions to generate a multi-focal extended plane image corresponding to each round of iteration, which may include:

[0123] S201, inputting multi-focus original plane images into an image interpolation model, and extracting image features of the original plane images of each focus by the image interpolation model.

[0124] Optionally, after the multi-focus original plane images are input into the image interpolation model, the image interpolation model may first learn the original plane images of each focus to extract image features of the original plane images of each focus.

[0125] S202. Based on the image features of the extracted original plane images of each focus, and according to the amount of interpolation data of the current round of iteration, perform the interpolation processing of the current round of iteration to generate a multi-focus extended plane image corresponding to the current round of iteration; and update the current multi-focus original plane image based on the multi-focus extended plane image and the multi-focus original plane image generated by the current round of iteration.

[0126] Taking any round of iteration as an example, the number of extended plane images that need to be generated in the current round of iteration can be determined based on the amount of interpolated frame data in the current round of iteration, so as to generate multi-focus extended plane images corresponding to the current round of iteration based on the image features of the extracted original plane images of each focus and the number of extended plane images to be generated.

[0127] Optionally, for any adjacent two-focus original plane images, at least one focal extended plane image can be generated according to the image features of the adjacent two-focus original plane images, and the generated extended plane image can be used as the intermediate layer focal plane image of the adjacent two-focus original plane images to fill the intermediate layer of the adjacent two-focus original plane images. The purpose of doing so is that the number of multi-focus original plane images is small and the spacing is large, and such spacing is not enough to fit the contour change trend of the inner cell mass or trophoblast cells.

[0128] Through image interpolation processing, image expansion can be performed on the basis of the original plane image, and the intermediate layer image of any two adjacent focal original plane images can be simulated and filled to enrich the number of images. The generated extended plane image can play a role of transition and information connection between adjacent original plane images, which is beneficial to subsequent contour fitting and surface reconstruction.

[0129] In some embodiments, after the current round of iterations ends, the extended plane image generated by the current round and all existing plane images before the current round can be used as a whole to update the current multi-focus original plane image. In this way, in the next round of iterations, the updated multi-focus original plane image can be used as input to continue iterative interpolation and generate a multi-focus extended plane image corresponding to the next round of iterations.

[0130] S203. Determine the current number of total plane images based on the multi-focus extended plane image corresponding to the current round of iteration and the multi-focus original plane image; if the number of total plane images does not reach the preset required number, iteratively perform the next round of iterative frame insertion processing until the number of total plane images reaches the preset required number, then stop the iteration and use the currently generated multi-focus extended plane image as the multi-focus extended plane image.

[0131] Optionally, after the current round of iterations ends, the sum of the current round of multifocal extended plane images and the existing multifocal original plane images can be calculated. If the sum reaches a preset required number, that is, the sum of the current number of existing plane images reaches the required number of images, the iteration can be stopped.

[0132] On the contrary, if the sum does not reach the preset required quantity, the next round of iteration will continue until the sum reaches the preset required quantity.

[0133] Exemplarily, it is assumed that the multi-focus original plane image includes an 11-focus image, the number of iterations is 3, the amount of interpolation data corresponding to the first round of iterations is 10, the amount of interpolation data corresponding to the second round of iterations is 20, and the amount of interpolation data corresponding to the third round of iterations is 40. Then, in the first round of interpolation, the 11-focus original plane image is used as the input of the image interpolation model, and the first round of interpolation generates a 10-focus extended plane image from the 11-focus original plane image. After the first round of interpolation, the total number of images is 21; in the second round of interpolation, the 21-focus image is used as the input of the image interpolation model, and the second round of interpolation generates a 20-focus extended plane image from the 21-focus image. After the second round of interpolation, the total number of images is 41; in the third round of interpolation, the 41-focus image is used as the input of the image interpolation model, and the third round of interpolation generates a 40-focus extended plane image from the 41-focus image. After the third round of interpolation, the total number of images is 81.

[0134] In practical applications, the number of iterations and the amount of interpolation data corresponding to each round of iteration can be adaptively adjusted according to demand.

[0135] Figure 3 A schematic flow chart of another embryo 3D reconstruction method provided in an embodiment of the present application; Optionally, the training process of the image interpolation model may include:

[0136] S301, collecting multi-focus sample plane images corresponding to a plurality of blastocysts.

[0137] During training, multiple blastocysts may be selected, and multi-focal sample plane images may be collected for each blastocyst. Alternatively, 200 blastocysts may be selected, and 11-focal sample plane images may be collected for each blastocyst, so that a total of 2200 focal sample plane images may be obtained. Of course, this is only one possible data selection method.

[0138] S302: Input each focus sample plane image into an initial image interpolation model, extract image features of each focus sample plane image by the initial image interpolation model, and predict and generate a new multi-focus plane image based on the image features of each focus sample plane image.

[0139] Optionally, the above-collected sample plane images of each focus can be input into the initial image interpolation model, and the initial image interpolation model extracts the image features of the sample plane images of each focus, and based on the extracted image features, predicts and simulates the generation of a multi-focus new plane image. The new plane image is an intermediate layer image between adjacent sample plane images.

[0140] Among them, the initial image interpolation model and the image interpolation model finally trained have the same model architecture, but the network parameters of the initial image interpolation model have not been trained to be used for executing the image interpolation processing method in this scheme. The network parameters of the initial image interpolation model can be considered as the default initial parameters. After training with sample plane images, the network parameters can be confirmed and the final image interpolation model can be obtained.

[0141] S303 , determining the model loss value of the current round according to the image features of each new plane image of each focus generated in the current round and the image features of each sample plane image of each focus.

[0142] Optionally, for any round of model training, after the training is completed, the model loss value of the current round can be calculated based on the image features of each focal new plane image generated in the current round and the image features of each focal sample plane image.

[0143] S304. Optimize the parameters of the current initial image interpolation model according to the model loss value of the current round, until the model loss value of the current round meets the preset loss threshold, and then use the current initial image interpolation model as the image interpolation model.

[0144] Therefore, according to the model loss value of the current round, the parameters of the current initial image interpolation model can be optimized. When the model loss value of the current round meets the preset loss threshold, it means that the new plane image predicted and generated by the current initial image interpolation model is close to the sample plane image, and the prediction effect of the model has reached a relatively satisfactory level. At this time, the current initial image interpolation model can be used as the final image interpolation model.

[0145] Figure 4 A flow chart of another embryo three-dimensional reconstruction method provided in an embodiment of the present application; Optionally, in step S103, based on the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, an auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focal original plane image and each focal extended plane image, including:

[0146] S401, generating first target images of each focus corresponding to the inner cell mass according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the first thumbwheel iteration.

[0147] In some embodiments, a multifocal image for reconstructing trophoblast cells and a multifocal image for reconstructing inner cell mass may be constructed from the multifocal original plane images and the multifocal extended plane images generated in at least one round of iterations.

[0148] Alternatively, since the trophoblast cells are regularly arranged on the surface of the blastocyst and compact in spatial distribution, a smaller vertical resolution (i.e., more focal plane data) is required for detail completion. Therefore, more images are needed to capture and record detailed information of the trophoblast cells during the reconstruction process. In addition, the trophoblast cells are arranged along the inner wall of the zona pellucida and wrap around the inner cell mass. This distribution makes the morphology and position of the trophoblast cells in three-dimensional space more complex, so more images are needed to fully reflect their structural characteristics.

[0149] Therefore, the number of images required to reconstruct the inner cell mass may be less than the number of images required to reconstruct the trophoblast cells.

[0150] In one embodiment, referring to the example in step S203, the original 11-focal plane image and the 10-focal extended plane image generated after the first round of iteration can be used to form the first target image of each focus corresponding to the inner cell mass. That is, the number of target images required to reconstruct the inner cell mass is 21-focal.

[0151] S402, generating second target images of each focus corresponding to trophoblast cells according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the second finger wheel iteration.

[0152] The total image after the third round of iteration can be used as the second target image of each focus corresponding to the trophoblast cells, that is, the number of target images required to reconstruct the trophoblast cells is 81 focuses.

[0153] Of course, this is only a specific example of value selection. In actual applications, the number of images required for trophoblast cells may be more than the number of images required for inner cell mass, but the specific number is not limited.

[0154] S403, using an auxiliary circle marking algorithm to mark the blastocyst main body region in each focused first target image corresponding to the inner cell mass, and extracting the mask image of the inner cell mass in each focused first target image from the marked blastocyst main body region.

[0155] Optionally, for each focus first target image corresponding to the inner cell mass, an auxiliary circle marking algorithm is used to mark the main area of ​​the blastocyst, that is, to mark the clear area inside the blastocyst. Based on the marking results, the mask image of the inner cell mass can be extracted from the main area of ​​the blastocyst, wherein a mask image of the inner cell mass can be extracted from each focus first target image, thereby obtaining a plurality of mask images of the inner cell masses.

[0156] S404, using an auxiliary circle marking algorithm to mark the blastocyst main body region in each focal second target image corresponding to the trophoblast cells, and extracting the mask image of the trophoblast cells in each focal second target image from the marked blastocyst main body region.

[0157] Similarly, for each focal second target image corresponding to the trophoblast cells, an auxiliary circle marking algorithm is used to mark the blastocyst main body region therein, and the mask image of the trophoblast cells is extracted from the blastocyst main body region according to the same method as above.

[0158] Thus, the number of mask images extracted of the inner cell mass may be 21, and the number of mask images extracted of the trophoblast cells may be 81.

[0159] Figure 5 A schematic flow chart of another embryo three-dimensional reconstruction method provided in an embodiment of the present application; Optionally, in step S403, an auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focal first target image corresponding to the inner cell mass, which may include:

[0160] S501 . Perform ellipse fitting on the vertical section of the blastocyst according to the radius of the central focal plane image and the length of the blastocyst in the vertical direction to obtain an ellipse fitting function.

[0161] It should be noted that the principle of the auxiliary circle marking algorithm is to approximate the multi-layer focal plane data as the horizontal "cut" of the spherical surface by the structural characteristics of the blastocyst spherical shape.

[0162] Optionally, the radius of the central focal plane image can be manually measured and the length of the blastocyst in the vertical direction can be estimated, and an ellipse fitting function can be obtained by performing ellipse fitting on the "vertical section" of the blastocyst.

[0163] The central focal plane image may refer to the F0 focal plane blastocyst image; and the length of the blastocyst in the vertical direction may refer to the z-axis length of the blastocyst.

[0164] S502 : Determine the radius of the auxiliary circle corresponding to each focused first target image according to the focal plane depth where each focused first target image is located and the ellipse fitting function.

[0165] For an arbitrary first target image with a focus, the radius of the auxiliary circle corresponding to the first target image can be calculated according to the depth of the focal plane where the first target image is located and the ellipse fitting function.

[0166] The depth of the focal plane refers to the vertical distance from the focal plane to the F0 focal plane.

[0167] Then, based on the above method, the auxiliary circle radius corresponding to each focused first target image can be calculated respectively.

[0168] S503, taking the center point of each focused first target image as the origin, drawing an auxiliary circle according to the auxiliary circle radius corresponding to each focused first target image, determining the main area of ​​the blastocyst in each focused first target image, and marking the main area of ​​the blastocyst in each focused first target image to obtain each first marked image.

[0169] Then, taking the center point of each focused first target image as the origin, an auxiliary circle is drawn in each focused first target image according to the calculated auxiliary circle radius corresponding to each focused first target image, and the area surrounded by the auxiliary circle is the main body area of ​​the blastocyst, in which the cell structure is clearer.

[0170] Thus, the main area of ​​the blastocyst marked in each focal first target image is cropped to obtain the first marked image corresponding to each focal first target image.

[0171] Likewise, according to the same method as above, the second marking image corresponding to each focused second target image can be obtained by cutting out each focused second target image.

[0172] Figure 6 A schematic diagram of a mask image extraction process of a trophoblast cell provided in an embodiment of the present application. Figure 6 The figure a shows a slice model of an 11-focus image, with the boundary line of each "section" as the auxiliary circle marking range of the trophoblast cells, that is, the location of the trophoblast cells in this layer should be within the auxiliary circle marking range. Figure 6 The outline of the trophoblast cells obtained by segmentation after auxiliary circle marking is shown in b. Figure 6 Figure c shows the marking result of the auxiliary circle on the multi-focus image.

[0173] Figure 7 A schematic flow chart of another embryo three-dimensional reconstruction method provided in an embodiment of the present application; Optionally, in step S403, extracting the mask image of the inner cell mass in each focal first target image from the marked blastocyst main body area may include:

[0174] S701 : Perform deconvolution enhancement processing and histogram equalization processing on each first labeled image to obtain each enhanced first labeled image.

[0175] In some embodiments, image enhancement processing may be performed on the extracted first marker images to improve the clarity and display effect of the extracted first marker images.

[0176] Optionally, the present solution performs two image enhancement processes, the first of which is performed by the RLTV (Richardson-lucy with tolar variation, deconvolution enhancement) algorithm. RLTV is a deconvolution algorithm that uses iterative deblurring. RL is an iterative deblurring algorithm based on Bayesian theory independently proposed by Richardson and Lucy, which can effectively restore images. The basic principle is to compare the differences between the reconstructed image and the target image, and update the reconstructed image based on these differences. Tolar Variation is a metric for evaluating image quality, which is used to control the convergence of the iterative process in the RL algorithm. Tolar Variation is a standard deviation used to measure the magnitude of the change in intensity of adjacent pixels in an image. By using Tolar Variation as a convergence indicator, it can be ensured that the iterative process will not stop before reaching a certain threshold, thereby avoiding falling into a local minimum.

[0177] After RLTV processing, the second image enhancement is performed. The second enhancement is achieved through Equalize Adaotive Hitogram (adaptive histogram equalization), which is a method used to improve image contrast, especially when the histogram of the image presents a skewed distribution. Due to the particularity of HMC imaging, the image will appear bright on the left and dark on the right, and the image brightness distribution is uneven. Therefore, it is necessary to use adaptive histogram equalization to improve the overall visual effect of the image and make the details in the image clearer.

[0178] Figure 8 This is a schematic diagram of a comparison of an image before and after enhancement processing provided in an embodiment of the present application. Figure 8 a in the figure is the original image extracted by the auxiliary circle marking algorithm; Figure 8 b in the figure is the image enhanced by the RLTV algorithm; Figure 8 The c in the figure is the image after histogram equalization. Compared with the original image, it can be seen that after two image enhancement processes, the image details are clearer and more conducive to subsequent reconstruction.

[0179] S702 , using a pre-trained inner cell cluster segmentation model, segmenting the inner cell clusters in each enhanced first labeled image to obtain mask images of multiple inner cell clusters.

[0180] Next, a pre-trained segmentation model is used to segment the image to obtain a mask image. The corresponding segmentation models are trained for the inner cell mass and trophoblast cells respectively.

[0181] Among them, for the inner cell clusters, a pre-trained inner cell cluster segmentation model can be used to identify and segment the inner cell clusters in each enhanced first labeled image to obtain mask images of multiple inner cell clusters, and according to the focal plane where the mask image of each inner cell cluster is located, the mask images of each inner cell cluster can be sorted to generate a mask sequence of the inner cell clusters.

[0182] Similarly, a pre-trained trophoblast cell segmentation model can be used to segment the trophoblast cells in each enhanced second labeled image to obtain multiple trophoblast cell mask images, and the mask images of each trophoblast cell can be sorted according to the focal plane where the mask image of each trophoblast cell is located to generate a trophoblast cell mask sequence.

[0183] Then, based on the mask image of each inner cell mass, the contour information of the inner cell mass can be extracted, and based on the contour information, an initial point cloud model of the inner cell mass is generated. The normal line of the initial point cloud model is estimated and the surface is reconstructed to obtain the surface model of the inner cell mass.

[0184] Similarly, based on the mask images of each trophoblast cell, the contour information of the trophoblast cell can be extracted, and based on the contour information, an initial point cloud model of the trophoblast cell can be generated. The normal line estimation and surface reconstruction of the initial point cloud model can be performed to obtain a surface model of the trophoblast cell.

[0185] It is worth noting that the inner cell mass segmentation model and the trophoblast cell segmentation model can be obtained by training the models after correspondingly annotating a large number of sample blastocyst images collected.

[0186] In some embodiments, in addition to extracting masks from the inner cell mass and trophoblast cells, the blastocyst segmentation model may also be used to extract a mask image of the blastocyst.

[0187] Fig. 9 A schematic diagram of a mask image provided in an embodiment of the present application. Fig. 9 1 in the figure shows an image sequence of interest of the blastocyst region at 81 focal lengths; Fig. 9 The image b in the figure shows a mask image sequence of 81 focal trophoblast cells; Fig. 9 The c in the figure shows the trophoblast cells obtained after segmentation by the segmentation model, wherein the grayscale surrounding the circle in c is different from other areas and is the trophoblast cells.

[0188] Fig.10 Another mask image schematic diagram provided in the embodiment of the present application. Fig.10 The image a in the figure shows the image sequence of interest of the blastocyst at 21 focal lengths; Fig.10 The image b in the figure shows a 21-focus mask image sequence of the inner cell mass; Fig.10 The image c in the figure shows the inner cell mass obtained after segmentation by the segmentation model, wherein the small block in the upper left part of c that is different from other areas is the inner cell mass.

[0189] Fig.11 A schematic flow chart of another embryo 3D reconstruction method provided in an embodiment of the present application; Optionally, in step S102, before the multi-focus original plane image is iteratively interpolated using a pre-trained image interpolation model, the method further includes:

[0190] S1101 . Segment the blastocyst region in the central focal plane image in the acquired multi-focus original plane image to obtain a blastocyst reference mask image.

[0191] In some embodiments, based on the multi-focus original plane images collected in step S101, each original plane image may be preprocessed first. Optionally, the multi-focus original plane images include an F0 focus original plane image, and the blastocyst region in the F0 focus original plane image may be segmented first to obtain a blastocyst F0 focus mask image as a reference mask image.

[0192] S1102 , preprocessing the remaining focal original plane images according to the blastocyst reference mask image to obtain preprocessed focal original plane images.

[0193] Optionally, the blastocyst reference mask image can be used as a reference to move and fill the boundaries of the remaining focal original plane images to obtain preprocessed focal original plane images.

[0194] Fig.12 A flow chart of another embryo three-dimensional reconstruction method provided in an embodiment of the present application; Optionally, in step S1102, preprocessing the remaining focal original plane images according to the blastocyst reference mask image to obtain the preprocessed focal original plane images may include:

[0195] S1201, taking the blastocyst reference mask image as a reference, translating the center points of the remaining focal original plane images to obtain translated focal original plane images.

[0196] The positions of the remaining focal original plane images are moved using the blastocyst reference mask image to move the center point of each focal original plane image to the center of the field of view to ensure the display effect of each focal original plane image.

[0197] S1202 : Fill the boundaries of the translated original plane images of each focus to obtain the preprocessed original plane images of each focus.

[0198] The borders of each focal original plane image are filled until the image resolution is consistent with the original image resolution. The purpose of border filling is to set the reference point of the spatial coordinate system of the reconstructed model to the center point of the blastocyst.

[0199] Fig.13 A schematic diagram of comparison before and after image preprocessing provided in an embodiment of the present application. Fig.13 a in the equation is the original plane image obtained. Fig.13 The preprocessed image obtained by segmenting the blastocyst region in the original plane image and moving the blastocyst image to the center of the field of view and then filling the border is shown in b.

[0200] Fig.14 A schematic flow chart of another embryo 3D reconstruction method provided in an embodiment of the present application; Optionally, in step S104, after obtaining the surface model of the inner cell mass, the method may further include:

[0201] S1401. According to the surface model of the inner cell mass, extract at least one feature data of the inner cell mass under the surface model.

[0202] In some embodiments, based on the reconstructed surface model of the inner cell mass, at least one feature data of the inner cell mass can be extracted, for example, key morphological feature data such as the volume and surface area of ​​the inner cell mass can be extracted.

[0203] S1402. Extracting corresponding characteristic data of the inner cell mass under the fluorescence model according to the inner cell mass fluorescence model obtained by fluorescence labeling.

[0204] In order to evaluate the accuracy of this method for embryo reconstruction, a waste embryo staining experiment was performed to verify it. After obtaining the multifocal image of the blastocyst, the waste embryo was fluorescently stained, and the trophoblast cells, inner cell mass, blastocyst cell membrane and other cell structures of the blastocyst were fluorescently labeled. Then, the confocal microscope was used to take pictures and images. The confocal imaging data were reconstructed in three dimensions with the help of Imaris software, and the key morphological characteristic parameters such as the volume and surface area of ​​the blastocyst, the number of trophoblast cells of the blastocyst, the volume and surface area of ​​the inner cell mass and other key morphological characteristic parameters were extracted.

[0205] S1403, evaluating the surface model of the inner cell mass according to at least one feature data of the inner cell mass under the surface model and each corresponding feature data of the inner cell mass under the fluorescence model.

[0206] Based on the various characteristic data of the inner cell mass extracted under the above-mentioned fluorescence model experiment, the corresponding characteristic data of the inner cell mass extracted from the inner cell mass surface model reconstructed based on this scheme can be compared and analyzed to perform model evaluation on the surface model of the inner cell mass reconstructed by this scheme.

[0207] Of course, the surface model of trophoblast cells can also be evaluated using this method.

[0208] In addition, based on the comparison of fluorescence models, parameters such as key parameters in the fitting function in the inner cell mass reconstruction and the coefficient of surface smoothing in the point cloud surface reconstruction can also be tuned.

[0209] Fig.15 A schematic diagram of an embryo reconstruction result provided in an embodiment of the present application. Fig.15 1 in a shows the focal plane image of blastocyst 0; 2 in a shows the result of fluorescent staining of blastocyst; 3 in a and 4 in a are the results of reconstruction of the outer surface of blastocyst. Fig.15 1 in b shows the clearest focal plane image of the inner cell mass in the blastocyst; 2 in b shows the result of fluorescent staining of the inner cell mass; 3 in b and 4 in b show the results of inner cell mass reconstruction.

[0210] In summary, the embryo three-dimensional reconstruction method provided in this embodiment includes: obtaining a multi-focus original plane image of the blastocyst to be reconstructed based on a time difference incubator, and the acquisition interval between each focus original plane image is a preset interval; using a pre-trained image interpolation model, iterative interpolation processing is performed on the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration; according to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, an auxiliary circle marking algorithm is used to mark the main area of ​​the blastocyst in each focus original plane image and each focus extended plane image, and based on the marking result, a plurality of mask images of inner cell masses and a plurality of mask images of trophoblast cells are respectively extracted; according to the extracted mask images of each inner cell mass and the mask images of each trophoblast cell, a three-dimensional reconstruction is performed to obtain a surface model of the inner cell mass and a surface model of the trophoblast cell. This method obtains a multi-focus original plane image with an HMC imaging system under a time difference incubation system, and then simulates and generates a multi-focus extended plane image based on the multi-focus original plane image based on the image interpolation algorithm, thereby performing three-dimensional reconstruction of the embryo based on the multi-focus original plane image and the multi-focus extended plane image. Among them, the method of acquiring images using a time-difference culture system is more convenient and safer; and through image interpolation processing, frame plane filling can be performed between multi-focus original plane images, so that information connection and transition of adjacent focus original plane images can be performed through extended plane images, making the data used to reconstruct the model more comprehensive and rich; and through the auxiliary circle marking algorithm to extract the image region of interest, the interference of artifact images can be eliminated, and through two image enhancement processes, the clarity of the extracted image of the region of interest can be increased, making the details in the image more obvious, and improving the accuracy of subsequent mask image segmentation, thereby improving the accuracy of embryo three-dimensional reconstruction. Compared with traditional two-dimensional reconstruction, the three-dimensional reconstruction effect can better display the three-dimensional structure and internal details of the embryo, and compared with fluorescent staining or other optical imaging techniques, it will not have adverse effects on the development and growth of the embryo.

[0211] The following is a description of the devices, equipment, storage media, etc. used to execute the embryo three-dimensional reconstruction method provided in this application. The specific implementation process and technical effects are mentioned above and will not be repeated below.

[0212] Fig.16 A schematic diagram of an embryo 3D reconstruction device provided in an embodiment of the present application, the functions implemented by the embryo 3D reconstruction device correspond to the steps performed by the above method. The device can be understood as the above server, or the processor of the server, or can be understood as a component independent of the above server or processor that implements the functions of the present application under the control of the server, such as Fig.16 As shown, the device may include: an acquisition module 160, a processing module 161, and a reconstruction module 162;

[0213] An acquisition module 160 is used to acquire multi-focus original plane images of the blastocyst to be reconstructed based on a time difference incubator, wherein the acquisition interval between each focus original plane image is a preset interval;

[0214] The processing module 161 is used to use a pre-trained image interpolation model to perform iterative interpolation processing on the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration;

[0215] The processing module 161 is used to mark the blastocyst main body area in each focal original plane image and each focal extended plane image according to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration by using an auxiliary circle marking algorithm, and extract a plurality of inner cell mass mask images and a plurality of trophoblast cell mask images respectively based on the marking results;

[0216] The reconstruction module 162 is used to perform three-dimensional reconstruction based on the extracted mask images of each inner cell mass and the mask images of each trophoblast cell to obtain the surface model of the inner cell mass and the surface model of the trophoblast cell.

[0217] Optionally, the processing module 161 is specifically used to input the multi-focus original plane image into the image interpolation model, and the image interpolation model performs iterative interpolation processing according to the original plane images of each focus and iterative control conditions to generate multi-focus extended plane images corresponding to each round of iteration; the iterative control conditions include: the number of iterations and the amount of interpolation data in each round of iteration.

[0218] Optionally, the processing module 161 is specifically configured to input the multi-focus original plane image into the image interpolation model, and extract image features of each focus original plane image by the image interpolation model;

[0219] According to the image features of the extracted original plane images of each focus, according to the amount of interpolation data of the current round of iteration, the current round of iterative interpolation processing is performed to generate a multi-focus extended plane image corresponding to the current round of iteration; and according to the multi-focus extended plane image and the multi-focus original plane image generated by the current round of iteration, the current multi-focus original plane image is updated;

[0220] The current number of total plane images is determined based on the multifocal extended plane image corresponding to the current round of iteration and the multifocal original plane image; if the number of total plane images does not reach the preset required number, the next round of iterative interpolation processing is iteratively performed until the number of total plane images reaches the preset required number, then the iteration is stopped, and the currently generated multifocal extended plane image is used as the multifocal extended plane image.

[0221] Optionally, a training module is also included;

[0222] The training module is used to collect multi-focus sample plane images corresponding to multiple blastocysts; input each focus sample plane image into the initial image interpolation model, extract the image features of each focus sample plane image by the initial image interpolation model, and predict and generate multi-focus new plane images based on the image features of each focus sample plane image; determine the model loss value of the current round based on the image features of each focus new plane image generated in the current round and the image features of each focus sample plane image; according to the model loss value of the current round, optimize the parameters of the current initial image interpolation model until the model loss value of the current round meets the preset loss threshold, and then use the current initial image interpolation model as the image interpolation model.

[0223] Optionally, the processing module 161 is specifically configured to generate a first target image of each focus corresponding to the inner cell mass according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the first finger wheel iteration;

[0224] Generate a second target image of each focus corresponding to the trophoblast cells according to the multi-focal original plane image and the multi-focal extended plane image corresponding to the second finger wheel iteration;

[0225] Using an auxiliary circle marking algorithm, the blastocyst main body area in each focal first target image corresponding to the inner cell mass is marked, and the mask image of the inner cell mass in each focal first target image is extracted from the marked blastocyst main body area;

[0226] The auxiliary circle marking algorithm is used to mark the blastocyst main body region in each focal second target image corresponding to the trophoblast cells, and the mask image of the trophoblast cells in each focal second target image is extracted from the marked blastocyst main body region.

[0227] Optionally, the processing module 161 is specifically configured to perform ellipse fitting on a vertical section of the blastocyst according to a radius of the central focal plane image and a length of the blastocyst in a vertical direction to obtain an ellipse fitting function;

[0228] Determine the radius of the auxiliary circle corresponding to each of the first-focus target images according to the focal plane depth where each of the first-focus target images is located and the ellipse fitting function;

[0229] Taking the center point of each focused first target image as the origin, an auxiliary circle is drawn according to the auxiliary circle radius corresponding to each focused first target image, the main area of ​​the blastocyst in each focused first target image is determined, and the main area of ​​the blastocyst in each focused first target image is marked to obtain each first marked image.

[0230] Optionally, the processing module 161 is specifically configured to perform deconvolution enhancement processing and histogram equalization processing on each first labeled image to obtain each enhanced first labeled image;

[0231] The inner cell clusters in each enhanced first labeled image are segmented using a pre-trained inner cell cluster segmentation model to obtain mask images of multiple inner cell clusters.

[0232] Optionally, the processing module 161 is further configured to segment the blastocyst region in the central focal plane image in the acquired multi-focus original plane image to obtain a blastocyst reference mask image;

[0233] According to the blastocyst reference mask image, the remaining focal original plane images are preprocessed to obtain the preprocessed focal original plane images.

[0234] Optionally, the processing module 161 is specifically configured to translate the center points of the remaining focal original plane images based on the blastocyst reference mask image to obtain translated focal original plane images;

[0235] The boundaries of the translated original plane images of each focus are filled to obtain the preprocessed original plane images of each focus.

[0236] Optionally, it also includes: an analysis module;

[0237] The analysis module is used to extract at least one feature data of the inner cell mass under the surface model according to the surface model of the inner cell mass; extract each corresponding feature data of the inner cell mass under the fluorescence model according to the fluorescence model of the inner cell mass obtained by fluorescent labeling; and evaluate the surface model of the inner cell mass according to at least one feature data of the inner cell mass under the surface model and each corresponding feature data of the inner cell mass under the fluorescence model.

[0238] The above-mentioned device is used to execute the method provided by the aforementioned embodiment, and its implementation principle and technical effect are similar, which will not be repeated here.

[0239] The above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASIC), or one or more digital singnal processors (DSP), or one or more field programmable gate arrays (FPGA). For another example, when a module is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0240] The above modules can be connected or communicated with each other via a wired connection or a wireless connection. The wired connection may include a metal cable, an optical cable, a hybrid cable, etc., or any combination thereof. The wireless connection may include a connection in the form of a LAN, a WAN, Bluetooth, a ZigBee, or NFC, or any combination thereof. Two or more modules may be combined into a single module, and any one module may be divided into two or more units. Those skilled in the art will clearly understand that for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application.

[0241] Fig.17 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application, which may be a computing device with data processing capabilities.

[0242] The device may include: a processor 801 and a storage medium 802 .

[0243] The storage medium 802 is used to store programs, and the processor 801 calls the programs stored in the storage medium 802 to execute the above method embodiment. The specific implementation method and technical effect are similar and will not be repeated here.

[0244] Among them, the storage medium 802 stores program code, and when the program code is executed by the processor 801, the processor 801 executes various steps in the embryo three-dimensional reconstruction method according to various exemplary embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0245] Processor 801 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor (DigitalSignal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, and can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly embodied as a hardware processor to be executed, or a combination of hardware and software modules in the processor can be executed.

[0246] 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. Storage medium may include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type storage medium, random access storage medium (Random Access Memory, RAM), static random access storage medium (Static Random Access Memory, SRAM), programmable read-only storage medium (Programmable Read Only Memory, PROM), read-only storage medium (Read Only Memory, ROM), electrically erasable programmable read-only storage medium (Electrically Erasable Programmable Read-Only Memory, EEPROM), magnetic storage medium, disk, optical disk, etc. Storage medium is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The storage medium 802 in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.

[0247] Optionally, the present application also provides a program product, such as a computer-readable storage medium, including a program, which is used to execute the above method embodiment when executed by a processor.

[0248] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0249] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0250] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0251] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (English: processor) to perform some steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only storage medium (English: Read-Only Memory, referred to as: ROM), random access storage medium (English: Random Access Memory, referred to as: RAM), disk or optical disk and other media that can store program code.

Claims

1. A method for three-dimensional reconstruction of an embryo, characterized in that: include: Acquire multi-focus original plane images of the blastocyst to be reconstructed based on the time difference incubator, wherein the acquisition interval between each focus original plane image is a preset interval; The pre-trained image interpolation model is used to iteratively interpolate the multi-focal original plane image to generate the multi-focal extended plane image corresponding to each round of iteration; According to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, an auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focus original plane image and each focus extended plane image, and based on the marking results, mask images of multiple inner cell masses and mask images of multiple trophoblast cells are respectively extracted; Three-dimensional reconstruction is performed based on the extracted mask images of each inner cell mass and the extracted mask images of each trophoblast cell to obtain a surface model of the inner cell mass and a surface model of the trophoblast cell.

2. The method according to claim 1, characterized in that The method uses a pre-trained image interpolation model to iteratively interpolate the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration, including: The multi-focus original plane image is input into the image interpolation model, and the image interpolation model performs iterative interpolation processing according to the original plane images of each focus and iterative control conditions to generate multi-focus extended plane images corresponding to each round of iteration; the iterative control conditions include: the number of iterations and the amount of interpolation data in each round of iteration.

3. The method according to claim 2, characterized in that The multi-focus original plane image is input into the image interpolation model, and the image interpolation model performs iterative interpolation processing according to each focus original plane image and an iterative control condition to generate a multi-focus extended plane image corresponding to each round of iteration, including: Inputting the multi-focus original plane images into the image interpolation model, and extracting the image features of the original plane images of each focus by the image interpolation model; According to the image features of the extracted original plane images of each focus, according to the amount of interpolation data of the current round of iteration, the current round of iterative interpolation processing is performed to generate a multi-focus extended plane image corresponding to the current round of iteration; and according to the multi-focus extended plane image and the multi-focus original plane image generated by the current round of iteration, the current multi-focus original plane image is updated; The number of current total plane images is determined based on the multifocal extended plane image corresponding to the current round of iteration and the multifocal original plane image; if the number of total plane images does not reach the preset required number, the next round of iterative interpolation processing is iteratively performed until the number of total plane images reaches the preset required number, then the iteration is stopped, and the currently generated multifocal extended plane image is used as the multifocal extended plane image.

4. The method according to claim 1, characterized in that: The training process of the image interpolation model includes: Collect multi-focal sample plane images corresponding to multiple blastocysts; Inputting each focal sample plane image into an initial image interpolation model, extracting image features of each focal sample plane image by the initial image interpolation model, and predicting and generating a multi-focus new plane image according to the image features of each focal sample plane image; Determine the model loss value of the current round according to the image features of each focal new plane image generated in the current round and the image features of each focal sample plane image; According to the model loss value of the current round, the parameters of the current initial image interpolation model are optimized until the model loss value of the current round meets the preset loss threshold, and the current initial image interpolation model is used as the image interpolation model.

5. The method according to claim 1, characterized in that The method comprises: marking the main area of ​​the blastocyst in each focal original plane image and each focal extended plane image according to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration by using an auxiliary circle marking algorithm, and extracting mask images of multiple inner cell masses and mask images of multiple trophoblast cells based on the marking results, including: Generate first target images of each focus corresponding to the inner cell mass according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the first round of iteration; Generate a second target image of each focus corresponding to the trophoblast cells according to the multi-focus original plane image and the multi-focus extended plane image corresponding to the second round of iteration; Using an auxiliary circle marking algorithm, marking the blastocyst main body region in each focal first target image corresponding to the inner cell mass, and extracting the mask image of the inner cell mass in each focal first target image from the marked blastocyst main body region; The auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focal second target image corresponding to the trophoblast cell, and the mask image of the trophoblast cell in each focal second target image is extracted from the marked blastocyst main body area.

6. The method according to claim 5, characterized in that The auxiliary circle marking algorithm is used to mark the blastocyst main body area in each focal first target image corresponding to the inner cell mass, including: According to the radius of the central focal plane image and the length of the blastocyst in the vertical direction, an ellipse fitting is performed on the vertical section of the blastocyst to obtain an ellipse fitting function; Determining the radius of the auxiliary circle corresponding to each focused first target image according to the focal plane depth where each focused first target image is located and the ellipse fitting function; Taking the center point of each focused first target image as the origin, an auxiliary circle is drawn according to the auxiliary circle radius corresponding to each focused first target image, the main area of ​​the blastocyst in each focused first target image is determined, and the main area of ​​the blastocyst in each focused first target image is marked to obtain each first marked image.

7. The method according to claim 6, characterized in that The step of extracting the mask image of the inner cell mass in each focal first target image from the marked blastocyst main body region comprises: Performing deconvolution enhancement processing and histogram equalization processing on each first labeled image to obtain each enhanced first labeled image; The inner cell clusters in each enhanced first labeled image are segmented using a pre-trained inner cell cluster segmentation model to obtain mask images of multiple inner cell clusters.

8. The method according to claim 1, characterized in that Before the pre-trained image interpolation model is used to iteratively interpolate the multi-focus original plane image, the method further includes: Segmenting the blastocyst region in the central focal plane image in the acquired multi-focus original plane image to obtain a blastocyst reference mask image; According to the blastocyst reference mask image, the remaining focal original plane images are preprocessed to obtain preprocessed focal original plane images.

9. The method according to claim 8, characterized in that The method of preprocessing the remaining focal original plane images according to the blastocyst reference mask image to obtain the preprocessed focal original plane images comprises: Taking the blastocyst reference mask image as a reference, translating the center points of the remaining focal original plane images to obtain translated focal original plane images; The boundaries of the translated original plane images of each focus are filled to obtain the preprocessed original plane images of each focus.

10. The method according to claim 1, characterized in that After obtaining the surface model of the inner cell mass, the method further includes: According to the surface model of the inner cell mass, extracting at least one feature data of the inner cell mass under the surface model; According to the inner cell mass fluorescence model obtained by fluorescence labeling, extracting various corresponding characteristic data of the inner cell mass under the fluorescence model; The surface model of the inner cell mass is evaluated according to at least one feature data of the inner cell mass under the surface model and each corresponding feature data of the inner cell mass under the fluorescence model.

11. An embryo three-dimensional reconstruction device, characterized in that: include: Acquisition module, processing module, reconstruction module; The acquisition module is used to acquire multi-focus original plane images of the blastocyst to be reconstructed based on the time difference incubator, and the acquisition interval between each focus original plane image is a preset interval; The processing module is used to use a pre-trained image interpolation model to perform iterative interpolation processing on the multi-focus original plane image to generate a multi-focus extended plane image corresponding to each round of iteration; The processing module is used to mark the blastocyst main body area in each focal original plane image and each focal extended plane image according to the multi-focus original plane image and the multi-focus extended plane image corresponding to each round of iteration, and extract the mask images of multiple inner cell masses and the mask images of multiple trophoblast cells respectively based on the marking results; The reconstruction module is used to perform three-dimensional reconstruction according to the extracted mask images of each inner cell mass and the mask images of each trophoblast cell, so as to obtain the surface model of the inner cell mass and the surface model of the trophoblast cell.

12. An electronic device, characterized in that: include: A processor, a storage medium and a bus, wherein 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 embryo three-dimensional reconstruction method as described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the method for three-dimensional reconstruction of embryos as claimed in any one of claims 1 to 10 is implemented.

Images