Internal cell mass analysis method and device, electronic equipment and storage medium

Abstract

The invention provides an internal cell mass analysis method and device, electronic equipment and a storage medium, and relates to the technical field of morphology. The method comprises the following steps: collecting a plurality of blastocyst image sets; determining a target blastocyst image set according to the number of the inner cell clusters contained in each group of blastocyst image sets; performing three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the internal cell cluster; according to the point cloud model, determining multiple three-dimensional features of the internal cell cluster; according to the three-dimensional features of the inner cell mass, determining a corresponding relation between the three-dimensional features and an embryo development result, and performing embryo scoring modeling; the method comprises the following steps: performing three-dimensional modeling on an internal cell cluster to obtain a point cloud model of the internal cell cluster, thereby extracting three-dimensional characteristics of the internal cell cluster, and analyzing a corresponding relationship between the extracted three-dimensional characteristics and an embryonic development result. According to the method, the research of embryonic development is deeply performed, and the three-dimensional features are added into a traditional scoring system for embryo modeling analysis, so that the accuracy of embryo clinical outcome prediction is improved.

Description

Technical Field

[0001] The present application relates to the field of morphological technology, and in particular to an inner cell mass analysis method, device, electronic device and storage medium. Background Art

[0002] The inner cell mass is a key part of the early embryo and has a vital influence on embryonic development. It not only determines the formation of various organs and tissues of the embryo, but also regulates the overall development process of the embryo through cell differentiation and the transmission of genetic information. It is particularly important to study the relationship between the morphology of the inner cell mass and the results of embryonic development.

[0003] At present, the morphological study of the inner cell mass generally measures the morphology of the inner cell mass from two-dimensional perspectives such as the size and compactness of the inner cell mass. This method makes the evaluation of embryonic development results less accurate. Summary of the invention

[0004] The purpose of the present application is to provide an inner cell mass analysis method, device, electronic device and storage medium to address the deficiencies in the above-mentioned prior art, so as to improve the predictive performance of the embryo scoring model by mining the three-dimensional characteristics of the inner cell mass.

[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, an embodiment of the present application provides a method for analyzing an inner cell mass, comprising:

[0007] Acquiring a plurality of sets of blastocyst image sets, each set of blastocyst image sets including multi-focal plane blastocyst images;

[0008] Determining a target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets;

[0009] Performing three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes position information of each data point constituting the inner cell mass in three-dimensional space;

[0010] determining a plurality of three-dimensional features of the inner cell mass according to the point cloud model;

[0011] According to the multiple three-dimensional features of the inner cell mass, the correspondence between each three-dimensional feature and the embryo development result is determined, and embryo scoring modeling is performed; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result.

[0012] Optionally, determining multiple three-dimensional features of the inner cell mass based on the point cloud model includes:

[0013] Determining a plurality of basic three-dimensional features of the inner cell mass according to the point cloud model;

[0014] Performing feature mining analysis based on various basic three-dimensional features to determine at least one combined three-dimensional feature of the inner cell mass;

[0015] Based on each of the basic three-dimensional features and each of the combined three-dimensional features, multiple three-dimensional features of the inner cell mass are obtained.

[0016] Optionally, determining a plurality of basic three-dimensional features of the inner cell mass according to the point cloud model includes:

[0017] Determining a first basic three-dimensional feature of the inner cell mass according to the point cloud model, wherein the first basic three-dimensional feature is used to characterize a ratio of a short axis to a long axis of the inner cell mass;

[0018] determining a second basic three-dimensional feature of the inner cell mass according to the point cloud model, wherein the second basic three-dimensional feature is used to characterize a ratio of a surface area to a volume of the inner cell mass;

[0019] A third basic three-dimensional feature of the inner cell mass is determined according to the point cloud model, and the third basic three-dimensional feature is used to characterize the spherical similarity of the inner cell mass.

[0020] Optionally, determining the first basic three-dimensional feature of the inner cell mass according to the point cloud model includes:

[0021] Performing principal component analysis on all data points in the point cloud model to determine three directions in which data point variation meets conditions;

[0022] Determining the first axis and the second axis of the inner cell mass according to the degree of variation of the data points in each direction; the first axis represents the long axis, and the second axis represents the short axis;

[0023] Determine a first value according to the first axis and the second axis;

[0024] The first value is used as a first basic three-dimensional feature of the inner cell mass.

[0025] Optionally, determining the second basic three-dimensional feature of the inner cell mass according to the point cloud model includes:

[0026] Performing triangular mesh processing on the point cloud model to obtain a triangular mesh model;

[0027] Based on the triangular mesh model, determining the surface area of ​​the inner cell mass;

[0028] Based on the point cloud model, graphics calculation is performed to obtain the volume of the inner cell mass;

[0029] determining a second value based on the volume and the surface area;

[0030] The second value is used as a second basic three-dimensional feature of the inner cell mass.

[0031] Optionally, determining the third basic three-dimensional feature of the inner cell mass according to the point cloud model includes:

[0032] Determining the sphericity of the inner cell mass according to the volume of the inner cell mass and the volume of a first equivalent sphere; the surface area of ​​the first equivalent sphere is equal to the surface area of ​​the inner cell mass;

[0033] Determining the surface sphericity of the inner cell mass according to the surface area of ​​the inner cell mass and the surface area of ​​a second equivalent sphere; the volume of the second equivalent sphere is equal to the volume of the inner cell mass;

[0034] Determining a morphological factor of the inner cell mass; the morphological factor is related to the number of tiny spikes on the surface of the inner cell mass;

[0035] The body sphericity of the inner cell mass, the surface sphericity of the inner cell mass and the morphological factor of the inner cell mass are used as the third basic three-dimensional feature of the inner cell mass.

[0036] Optionally, performing feature mining analysis based on various basic three-dimensional features to determine at least one combined three-dimensional feature of the inner cell mass includes:

[0037] determining a first combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the third basic three-dimensional feature;

[0038] A second combined three-dimensional feature of the inner cell mass is determined based on the second basic three-dimensional feature and the first basic three-dimensional feature.

[0039] Optionally, determining a first combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the third basic three-dimensional feature includes:

[0040] determining a third value according to the second basic three-dimensional feature and the body sphericity in the third basic three-dimensional feature;

[0041] The third value is determined as a first combined three-dimensional feature of the inner cell mass.

[0042] Optionally, determining a second combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the first basic three-dimensional feature includes:

[0043] A weighted calculation is performed based on the second basic three-dimensional feature, the first basic three-dimensional feature, the weight corresponding to the second basic three-dimensional feature, and the weight corresponding to the first basic three-dimensional feature to obtain a second combined three-dimensional feature of the inner cell mass.

[0044] Optionally, before determining the correspondence between each three-dimensional feature and the embryonic development result based on the multiple three-dimensional features of the inner cell mass, the method further includes:

[0045] Draw a feature scatter plot based on each basic three-dimensional feature and each combined three-dimensional feature;

[0046] Determine, according to the feature scatter plot, the correlation between any two of the three-dimensional features in the plurality of three-dimensional features;

[0047] According to the correlation between two three-dimensional features, three-dimensional features with nonlinear relationships and those satisfying linear relationships are eliminated from multiple three-dimensional features.

[0048] Optionally, determining the correspondence between each three-dimensional feature and the embryonic development result according to the multiple three-dimensional features of the inner cell mass includes:

[0049] According to the three-dimensional features, statistical analysis is performed to determine the corresponding relationship between the three-dimensional features and the embryo development results;

[0050] According to the analysis results, various three-dimensional features are screened to obtain screened three-dimensional features.

[0051] Optionally, the embryo scoring modeling comprises:

[0052] One or more of the screened three-dimensional features are added as new training parameters to the original training parameters of the original embryo scoring model to establish a new embryo scoring model.

[0053] In a second aspect, an embodiment of the present application further provides an inner cell mass analysis device, including: a collection module, a determination module, a generation module, and an analysis module;

[0054] The acquisition module is used to acquire multiple groups of blastocyst image sets, each group of blastocyst image sets includes multi-focal plane blastocyst images;

[0055] The determination module is used to determine the target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets;

[0056] The generating module is used to perform three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes position information of each data point constituting the inner cell mass in three-dimensional space;

[0057] The determination module is used to determine multiple three-dimensional features of the inner cell mass according to the point cloud model;

[0058] The analysis module is used to determine the correspondence between each three-dimensional feature and the embryo development result based on multiple three-dimensional features of the inner cell mass, and to perform embryo scoring modeling; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result.

[0059] Optionally, the determination module is specifically used to determine a plurality of basic three-dimensional features of the inner cell mass according to the point cloud model;

[0060] Performing feature mining analysis based on various basic three-dimensional features to determine at least one combined three-dimensional feature of the inner cell mass;

[0061] Based on each of the basic three-dimensional features and each of the combined three-dimensional features, multiple three-dimensional features of the inner cell mass are obtained.

[0062] Optionally, the determination module is specifically used to determine a first basic three-dimensional feature of the inner cell mass according to the point cloud model, wherein the first basic three-dimensional feature is used to characterize the ratio of the short axis to the long axis of the inner cell mass;

[0063] determining a second basic three-dimensional feature of the inner cell mass according to the point cloud model, wherein the second basic three-dimensional feature is used to characterize a ratio of a surface area to a volume of the inner cell mass;

[0064] A third basic three-dimensional feature of the inner cell mass is determined according to the point cloud model, and the third basic three-dimensional feature is used to characterize the spherical similarity of the inner cell mass.

[0065] Optionally, the determination module is specifically used to perform principal component analysis on all data points in the point cloud model to determine three directions in which data point variation meets conditions;

[0066] Determining the first axis and the second axis of the inner cell mass according to the degree of variation of the data points in each direction; the first axis represents the long axis, and the second axis represents the short axis;

[0067] determining a first ratio based on the first axis and the second axis;

[0068] The first ratio is used as a first basic three-dimensional feature of the inner cell mass.

[0069] Optionally, the determining module is specifically used to perform triangulation processing on the point cloud model to obtain a triangulated mesh model;

[0070] Based on the triangular mesh model, determining the surface area of ​​the inner cell mass;

[0071] Based on the point cloud model, graphics calculation is performed to obtain the volume of the inner cell mass;

[0072] determining a second ratio based on the volume and the surface area;

[0073] The second ratio is used as the second basic three-dimensional feature of the inner cell mass.

[0074] Optionally, the determination module is specifically used to determine the sphericity of the inner cell mass according to the volume of the inner cell mass and the volume of a first equivalent sphere; the surface area of ​​the first equivalent sphere is equal to the surface area of ​​the inner cell mass;

[0075] Determining the surface sphericity of the inner cell mass according to the surface area of ​​the inner cell mass and the surface area of ​​a second equivalent sphere; the volume of the second equivalent sphere is equal to the volume of the inner cell mass;

[0076] Determining a morphological factor of the inner cell mass; the morphological factor is related to the number of tiny spikes on the surface of the inner cell mass;

[0077] The body sphericity of the inner cell mass, the surface sphericity of the inner cell mass and the morphological factor of the inner cell mass are used as the third basic three-dimensional feature of the inner cell mass.

[0078] Optionally, the determination module is specifically configured to determine a first combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the third basic three-dimensional feature;

[0079] A second combined three-dimensional feature of the inner cell mass is determined based on the second basic three-dimensional feature and the first basic three-dimensional feature.

[0080] Optionally, the determination module is specifically configured to determine a third ratio according to the second basic three-dimensional feature and the body sphericity in the third basic three-dimensional feature;

[0081] The third ratio is determined as a first combined three-dimensional feature of the inner cell mass.

[0082] Optionally, the determination module is specifically used to perform weighted calculation according to the second basic three-dimensional feature, the first basic three-dimensional feature, the weight corresponding to the second basic three-dimensional feature, and the weight corresponding to the first basic three-dimensional feature to obtain the second combined three-dimensional feature of the inner cell mass.

[0083] Optionally, the analysis module is further used to draw a feature scatter plot according to each basic three-dimensional feature and each combined three-dimensional feature;

[0084] Determine, according to the feature scatter plot, the correlation between any two of the three-dimensional features in the plurality of three-dimensional features;

[0085] According to the correlation between two three-dimensional features, three-dimensional features with nonlinear relationships and those satisfying linear relationships are eliminated from multiple three-dimensional features.

[0086] Optionally, the analysis module is specifically used to perform statistical analysis based on various three-dimensional features to determine the correspondence between various three-dimensional features and embryonic development results; based on the analysis results, the various three-dimensional features are screened to obtain screened three-dimensional features.

[0087] Optionally, the analysis module is specifically used to add one or more of the screened three-dimensional features as new training parameters to the original training parameters of the original embryo scoring model to establish a new embryo scoring model.

[0088] 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 through the bus, and the processor executes the machine-readable instructions to implement the inner cell mass analysis method provided in the first aspect.

[0089] 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 inner cell mass analysis method provided in the first aspect is executed.

[0090] The beneficial effects of this application are:

[0091] The present application provides an inner cell mass analysis method, device, electronic device and storage medium, including:

[0092] Collect multiple sets of blastocyst image sets, each set of blastocyst image sets includes multi-focal plane blastocyst images; determine the target blastocyst image set according to the number of inner cell masses contained in each set of blastocyst image sets; perform three-dimensional modeling based on the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes the position information of each data point constituting the inner cell mass in three-dimensional space; determine multiple three-dimensional features of the inner cell mass according to the point cloud model; determine the correspondence between each three-dimensional feature and the embryo development result according to the multiple three-dimensional features of the inner cell mass, and perform embryo scoring modeling; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result. This scheme obtains a point cloud model of the inner cell mass by performing three-dimensional spatial modeling on the inner cell mass, so that based on the analysis of the point cloud model, multiple three-dimensional features of the inner cell mass can be extracted, thereby analyzing the correspondence between the extracted three-dimensional features and the embryo development result, deepening the research on embryo development, and adding three-dimensional features to the traditional scoring system for embryo modeling analysis to form a new embryo scoring model to improve the accuracy of embryo clinical outcome prediction. Among them, by extracting the three-dimensional features of the inner cell mass, we can make up for the problem of the relatively single two-dimensional features extracted in traditional methods, and the research and analysis of embryonic development results based on the three-dimensional features of the inner cell mass can improve the accuracy of the analysis results, thereby providing assistance for clinical research. Adding the three-dimensional features of the inner cell mass to the existing scoring system improves the measurement of the inner cell mass morphology and improves the predictive performance of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] 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.

[0094] Figure 1 A schematic diagram of a process for analyzing an inner cell mass provided in an embodiment of the present application;

[0095] Figure 2 A schematic diagram of another inner cell mass analysis method provided in an embodiment of the present application;

[0096] Figure 3 A schematic diagram of a process flow of another inner cell mass analysis method provided in an embodiment of the present application;

[0097] Figure 4 A schematic diagram of another inner cell mass analysis method provided in an embodiment of the present application;

[0098] Figure 5A schematic diagram of a process flow of another inner cell mass analysis method provided in an embodiment of the present application;

[0099] Figure 6 A schematic diagram of another inner cell mass analysis method provided in an embodiment of the present application;

[0100] Figure 7 A schematic diagram of another inner cell mass analysis method provided in an embodiment of the present application;

[0101] Figure 8 A schematic diagram of the ROC curve of the pregnancy prediction model provided in the embodiment of the present application;

[0102] Fig. 9 A schematic diagram of the ROC curve of the live birth prediction model provided in the embodiment of the present application;

[0103] Fig.10 A schematic diagram of the ROC curve of the miscarriage prediction model provided in the embodiment of the present application;

[0104] Fig.11 A schematic diagram of an inner cell mass analysis device provided in an embodiment of the present application;

[0105] Fig.12 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0106] 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 order, 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.

[0107] 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.

[0108] The inner cell mass is an important component in the early stages of embryonic development. It has developmental omnipotence and can develop into various tissues of the fetus. Research on the morphology of the inner cell mass helps to deeply understand the mechanism and process of embryonic development. The study of the morphology of the inner cell mass is of great significance for improving the success rate of assisted reproductive technology. By optimizing the embryo culture conditions and improving the development state of the inner cell mass, the embryo implantation rate and pregnancy rate can be improved.

[0109] Although significant progress has been made in the study of the morphology of the inner cell mass, many challenges still exist. For example, how to accurately assess the developmental potential and quality of the inner cell mass and how to apply it in clinical practice still require further research and exploration.

[0110] Currently, the studies on the inner cell mass are all based on the evaluation of the size and compactness of the inner cell mass. These are all evaluations of the inner cell mass from a two-dimensional perspective. The evaluation results are not comprehensive enough, thus affecting the accuracy of the evaluation of clinical results.

[0111] 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.

[0112] Figure 1 The following is a flow chart of an inner cell mass analysis method provided in an embodiment of the present application; the execution subject of the method may be a computer device. Figure 1 As shown, the method may include:

[0113] S101. Collect multiple groups of blastocyst image sets, each group of blastocyst image sets includes multi-focal plane blastocyst images.

[0114] Optionally, multiple groups of blastocyst image sets under the time difference culture system (Hoffman imaging) can be collected, and each group of blastocyst image sets can include blastocyst images of 11 focal planes or more.

[0115] Each blastocyst image in each blastocyst image set may contain an inner cell mass or may not contain an inner cell mass. A blastocyst image containing an inner cell mass means that the inner cell mass is within the field of view of the blastocyst image.

[0116] S102: Determine a target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets.

[0117] Optionally, in order to ensure the accuracy and modeling effect of the three-dimensional modeling of the inner cell mass, a set of blastocyst images in which at least three blastocyst images all contain the inner cell mass can be selected from multiple sets of blastocyst image sets as target blastocyst images.

[0118] Exemplarily, assuming that a group of blastocyst image sets contains 11 blastocyst images, at least three of which contain inner cell masses, then the group of blastocyst image sets can be used as the target blastocyst image set.

[0119] Of course, when there are multiple groups of blastocyst image sets that all meet the conditions, the group of blastocyst image sets containing the largest number of inner cell masses can be used as the target blastocyst image set.

[0120] S103. Perform three-dimensional modeling based on the target blastocyst image set to generate a point cloud model of the inner cell mass.

[0121] The point cloud model includes the position information of each data point constituting the inner cell mass in three-dimensional space.

[0122] Optionally, in one method, three-dimensional embryo modeling can be performed based on the blastocyst images of each focal plane in the selected target blastocyst image set to obtain a point cloud model of the embryo, and the point cloud model of the inner cell mass can be extracted based on the point cloud model of the embryo.

[0123] In another way, each blastocyst image containing the inner cell mass may be first determined from the selected target blastocyst image set, and then three-dimensional modeling may be performed based on relevant features of the inner cell mass extracted from each blastocyst image.

[0124] Optionally, a machine learning-based method can be used to analyze and identify the selected blastocyst images using a deep learning algorithm, automatically extract the features of the inner cell mass and construct a three-dimensional model. The constructed three-dimensional model is then converted into point cloud data, which can be achieved by sampling the surface of the three-dimensional model, and each sampling point contains its three-dimensional coordinate information; the generated point cloud is subjected to denoising, smoothing and other processing operations to improve its quality and usability; point cloud processing software can be used for further analysis and editing.

[0125] The generated point cloud model of the inner cell mass may include the position information of each data point constituting the inner cell mass in three-dimensional space, and the position is determined by three coordinate values ​​of X, Y, and Z.

[0126] Of course, attribute information of each data point may also be included, and the attribute information includes but is not limited to: color information, intensity or reflectivity, normal information, etc.

[0127] S104. Determine multiple three-dimensional features of the inner cell mass based on the point cloud model.

[0128] Based on the obtained point cloud model of the inner cell mass, the morphological analysis of the inner cell mass can be performed to determine multiple three-dimensional features of the inner cell mass.

[0129] This embodiment extracts the three-dimensional features of the inner cell mass from a three-dimensional perspective. Based on the analysis of the point cloud model, multiple three-dimensional features of the inner cell mass can be extracted.

[0130] S105. According to the multiple three-dimensional features of the inner cell mass, the corresponding relationship between each three-dimensional feature and the embryo development result is determined, and embryo scoring modeling is performed.

[0131] Among them, the corresponding relationship is used to characterize the influence of three-dimensional features on embryonic development outcomes.

[0132] In some embodiments, based on the multiple three-dimensional features of the extracted inner cell mass, the correspondence between the three-dimensional features and the embryonic development results can be analyzed, and then the three-dimensional features can be added to the original embryo scoring model (traditional scoring system) to perform embryo scoring modeling and form a new embryo scoring model.

[0133] Optionally, the three-dimensional features are used for embryo scoring modeling, that is, the three-dimensional features are added as evaluation parameters to the original embryo scoring model, a new embryo scoring model is constructed, and the embryo development results are predicted based on the new embryo scoring model.

[0134] In summary, the inner cell mass analysis method provided in this embodiment includes: collecting multiple groups of blastocyst image sets, each group of blastocyst image sets includes multi-focal plane blastocyst images; determining the target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets; performing three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes the position information of each data point constituting the inner cell mass in three-dimensional space; according to the point cloud model, determining multiple three-dimensional features of the inner cell mass; according to the multiple three-dimensional features of the inner cell mass, determining the correspondence between each three-dimensional feature and the embryo development result, and performing embryo scoring modeling; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result. This scheme obtains a point cloud model of the inner cell mass by performing three-dimensional space modeling on the inner cell mass, so that based on the analysis of the point cloud model, multiple three-dimensional features of the inner cell mass can be extracted, thereby analyzing the correspondence between the extracted three-dimensional features and the embryo development result. In-depth research on embryonic development, and adding three-dimensional features to the traditional scoring system for embryo modeling analysis to form a new embryo scoring model, the application of the new embryo scoring model can improve the accuracy of embryo clinical outcome prediction. Among them, by extracting the three-dimensional features of the inner cell mass, we can make up for the problem of the relatively single two-dimensional features extracted in traditional methods, and the research and analysis of embryonic development results based on the three-dimensional features of the inner cell mass can improve the accuracy of the analysis results, thereby providing assistance for clinical research. Adding the three-dimensional features of the inner cell mass to the existing scoring system improves the measurement of the inner cell mass morphology and improves the predictive performance of the model.

[0135] Figure 2 A schematic flow chart of another inner cell mass analysis method provided in an embodiment of the present application; Optionally, in step S104, determining multiple three-dimensional features of the inner cell mass according to the point cloud model may include:

[0136] S201. Determine multiple basic three-dimensional features of the inner cell mass based on the point cloud model.

[0137] First, by analyzing the point cloud model, multiple basic three-dimensional features of the inner cell mass can be determined, among which the basic three-dimensional features can also be understood as a single three-dimensional feature.

[0138] S202. Perform feature mining analysis based on various basic three-dimensional features to determine at least one combined three-dimensional feature of the inner cell mass.

[0139] Then, by performing feature mining analysis, feature combination, etc. on each basic three-dimensional feature, at least one combined three-dimensional feature of the inner cell mass can be derived based on the basic three-dimensional features.

[0140] Combined 3D features refer to the combined features obtained by combining basic 3D features through mathematical calculations, etc. This allows the features to be derived and expanded on the basis of the basic 3D features to obtain more 3D features.

[0141] S203. Obtain multiple three-dimensional features of the inner cell mass based on the basic three-dimensional features and the combined three-dimensional features.

[0142] The basic three-dimensional features and combined three-dimensional features obtained above together constitute multiple three-dimensional features of the inner cell mass.

[0143] In some embodiments, for each obtained three-dimensional feature, the continuous features in the three-dimensional feature may be discretized, the purpose of which mainly includes the following aspects:

[0144] Reduce model complexity: Discretizing continuous features can reduce the number of segmentation points that the model needs to consider. Especially for tree-based models such as decision trees and random forests, discretization can significantly reduce the depth and complexity of the tree and improve the training and prediction speed of the model.

[0145] Handling missing values: For some models, especially distance-based models (such as K-means clustering), handling missing values ​​for continuous features can be complicated. By discretizing continuous features, it is easier to handle missing values, for example, by treating missing values ​​as a separate category or filling them based on the data distribution.

[0146] Avoid overfitting: In some cases, small changes in continuous features may have a significant impact on the model's prediction results, which may cause the model to overfit the noise in the training data. Through discretization, this sensitivity can be reduced and the generalization ability of the model can be improved.

[0147] Enhance feature interpretability: After converting continuous features into categorical features, it is easier to interpret the model's prediction results.

[0148] Nonlinear relationship modeling: Sometimes, the relationship between continuous features and target variables may be nonlinear. Through discretization, this nonlinear relationship can be approximated as a piecewise linear relationship, which helps the model capture this relationship.

[0149] Adapt to specific algorithm requirements: Some algorithms may be more suitable for processing categorical features, such as the Naive Bayes classifier. These algorithms can be better applied by discretizing continuous features.

[0150] Data privacy protection: In some cases, continuous features may contain sensitive information. By discretizing, personal privacy can be protected to a certain extent while retaining enough information for modeling.

[0151] In this embodiment, the discretization methods that can be used include but are not limited to: equal-width discretization, equal-frequency discretization, quantile discretization, cluster discretization, custom discretization, discretization based on statistical parameters, and discretization based on decision trees, etc., and decision tree binning is preferred.

[0152] a) Equal width discretization, equal frequency discretization and quantile discretization: divide the data into several intervals according to the range or frequency;

[0153] b) Clustering discretization: Use a clustering algorithm (such as K-means) to cluster the data and treat each cluster as an interval;

[0154] c) Custom discretization: define data boundaries and number of groups based on requirements and expert consensus;

[0155] d) Discretization based on statistical parameters: Divide the data into different groups based on statistical parameters such as chi-square analysis, mutual information, variance, etc.

[0156] e) Use ROC (Receiver Operating Characteristic) to determine the optimal threshold: Commonly used parameters include Youden's J statistic, maximizing F1 score, balancing error rate, and being closest to the ideal point;

[0157] f) Decision tree binning: Use the decision tree algorithm to divide the data, and each leaf node of the tree represents an interval.

[0158] Optionally, in step S201, determining multiple basic three-dimensional features of the inner cell mass according to the point cloud model may include: determining a first basic three-dimensional feature of the inner cell mass according to the point cloud model, the first basic three-dimensional feature being used to characterize the ratio of the short axis to the long axis of the inner cell mass; determining a second basic three-dimensional feature of the inner cell mass according to the point cloud model, the second basic three-dimensional feature being used to characterize the ratio of the surface area to the volume of the inner cell mass; determining a third basic three-dimensional feature of the inner cell mass according to the point cloud model, the third basic three-dimensional feature being used to characterize the spherical similarity of the inner cell mass.

[0159] In some embodiments, the ratio of the short axis to the long axis, the ratio of the surface area to the volume, and the spherical similarity of the inner cell mass can be determined based on the point cloud model, and the ratio of the short axis to the long axis, the ratio of the surface area to the volume, and the spherical similarity are each used as a basic three-dimensional feature. Among them, the ratio of the short axis to the long axis of the inner cell mass can be used as a better indicator to describe the three-dimensional spatial morphology of the inner cell mass.

[0160] The ratio of the short axis to the long axis of the inner cell mass is an important feature for describing the inner cell mass, reflecting the anisotropy and shape characteristics of the inner cell mass, and the value range is 0-1; when the ratio of the short axis to the long axis is closer to 1, it means that the extension of the inner cell mass in all directions is more similar, that is, the stronger the isotropy, and the closer the inner cell mass is to a sphere; when the ratio of the short axis to the long axis is much smaller than 1, it means that the extension of the inner cell mass in one direction is much higher than that in other directions, that is, the anisotropy is strong, and the shape of the inner cell mass is more slender or flat.

[0161] The surface area to volume ratio of the inner cell mass is a key morphological feature of the inner cell mass, and describes the structural stability of the inner cell mass. Compared with inner cell masses with high surface area to volume ratios, inner cell masses with low surface area to volume ratios have more stable structures and are "stronger". In addition, the surface area to volume ratio is also an auxiliary indicator of metabolic demand and heat loss. A lower surface area to volume ratio means that the inner cell mass has low metabolic demand and less heat loss, which helps to maintain heat and energy inside the inner cell mass.

[0162] The spherical similarity of the inner cell mass can be measured from multiple perspectives such as body sphericity, surface sphericity, morphological factor, etc. The closer the morphology of the inner cell mass is to a sphere or the more regular it is, the greater the body sphericity, surface sphericity and morphological factor of the inner cell mass.

[0163] Figure 3 A schematic flow chart of another inner cell mass analysis method provided in an embodiment of the present application; optionally, determining the first basic three-dimensional feature of the inner cell mass according to the point cloud model may include:

[0164] S301, performing principal component analysis on all data points in the point cloud model to determine three directions in which data point variation meets conditions.

[0165] Optionally, this embodiment describes a method for calculating the ratio of the short axis to the long axis of the inner cell mass. A principal component analysis may be performed on all data points in the point cloud model to determine the three directions with the largest variation of the data points.

[0166] Principal component analysis is a commonly used data dimensionality reduction technique. The goal of principal component analysis is to find three new coordinate axes (called principal components), which are the main change directions of the original data, that is, the three directions in which the above data points vary the most.

[0167] S302, determining the first axis and the second axis of the inner cell mass respectively according to the variation degree of the data points in each direction.

[0168] Therein, the first axis represents the major axis, and the second axis represents the minor axis.

[0169] It can be determined from the three directions that the direction with the largest variation is the long axis of the inner cell mass, and the direction with the smallest variation is the short axis of the inner cell mass.

[0170] S303: Determine a first value according to the first axis and the second axis.

[0171] The ratio of the major axis to the minor axis of the inner cell mass is calculated to obtain a first value, which is also the ratio of the major axis to the minor axis.

[0172] S304: Use the first value as a first basic three-dimensional feature of the inner cell mass.

[0173] Therefore, the first basic three-dimensional feature is the ratio of the short axis to the long axis of the inner cell mass.

[0174] Figure 4 A schematic flow chart of another inner cell mass analysis method provided in an embodiment of the present application; optionally, determining the second basic three-dimensional feature of the inner cell mass according to the point cloud model may include:

[0175] S401, performing triangulation processing on the point cloud model to obtain a triangulated mesh model.

[0176] This embodiment describes the calculation method of the ratio of the surface area to the volume of the inner cell mass. Calculating the surface area of ​​the inner cell mass based on the point cloud model usually involves converting the point cloud data into a format that can calculate the surface area, such as a triangular mesh. Open3D is a popular open source library that provides a variety of tools for processing three-dimensional data, including point clouds and triangular meshes, and allows calculation of surface areas.

[0177] S402. Determine the surface area of ​​the inner cell mass based on the triangular mesh model.

[0178] First, you can use Open3D to load the point cloud model. Since the point cloud itself does not contain surface information (that is, it does not define faces or triangles), you need to convert the point cloud data into a triangular mesh. This can be achieved through a variety of algorithms, such as Poisson (Poisson reconstruction, Poisson reconstruction) reconstruction, Ball Pivoting (3D geometry algorithm based on ball model), etc.

[0179] Based on the converted triangular mesh, the surface area of ​​the inner cell mass can be calculated by calling the relevant functions in Open3D.

[0180] S403. Performing graphics operations based on the point cloud model to obtain the volume of the inner cell mass.

[0181] There are many ways to calculate the volume of the inner cell mass based on the point cloud model, all of which can be achieved through graphics calculation.

[0182] A simple calculation method is to calculate the convex hull of the point cloud and estimate the volume of the inner cell cluster based on the volume of the convex hull. The convex hull is the smallest convex polyhedron that contains all the point cloud data points.

[0183] It can also be achieved through voxelization, which is the process of converting point cloud data into a 3D grid, where the space is divided into small cubes (voxels) and the number of point cloud points contained in each voxel is counted. Then, the volume can be estimated based on the occupancy of the voxels.

[0184] When the surface of the point cloud model is closed, the triangular mesh method can also be used to convert the point cloud data into a triangular mesh and calculate the volume of the triangular mesh.

[0185] Of course, there are other methods that can be used to estimate the volume of the inner cell mass, such as the level set method, etc. These methods are usually more complicated and require more computing resources.

[0186] S404: Determine a second value according to the volume and the surface area.

[0187] Optionally, based on the calculated surface area and volume of the inner cell mass, a ratio of the surface area to the volume may be calculated to obtain a second value.

[0188] S405. Use the second value as the second basic three-dimensional feature of the inner cell mass.

[0189] Then, the second basic three-dimensional characteristic is the ratio of the surface area to the volume of the inner cell mass.

[0190] Figure 5 A schematic flow chart of another inner cell mass analysis method provided in an embodiment of the present application; optionally, determining the third basic three-dimensional feature of the inner cell mass according to the point cloud model may include:

[0191] S501. Determine the sphericity of the inner cell mass according to the volume of the inner cell mass and the volume of the first equivalent sphere.

[0192] Among them, the surface area of ​​the first equivalent sphere is equal to the surface area of ​​the inner cell mass.

[0193] The body sphericity of the inner cell mass can be the ratio of the volume of the inner cell mass itself to the volume of the first equivalent sphere having the same surface area as the inner cell mass. That is, the body sphericity of the inner cell mass is obtained by calculating the ratio of the volume of the inner cell mass to the volume of the first equivalent sphere. The first equivalent sphere has the same surface area as the inner cell mass.

[0194] S502. Determine the surface sphericity of the inner cell mass according to the surface area of ​​the inner cell mass and the surface area of ​​the second equivalent sphere.

[0195] The volume of the second equivalent sphere is equal to the volume of the inner cell mass.

[0196] Similarly, the surface sphericity of the inner cell mass can be the ratio of the surface area of ​​the inner cell mass itself to the surface area of ​​the second equivalent sphere with the same volume as the inner cell mass. That is, the surface sphericity of the inner cell mass is obtained by calculating the ratio of the surface area of ​​the inner cell mass to the surface area of ​​the second equivalent sphere. The second equivalent sphere has the same volume as the inner cell mass.

[0197] S503. Determine the morphological factor of the inner cell mass; the morphological factor is related to the number of tiny spikes on the surface of the inner cell mass.

[0198] In some embodiments, the number of tiny peaks on the surface of the inner cell mass can be determined based on the point cloud model of the inner cell mass, and the morphological factor of the inner cell mass increases as the number of tiny peaks on the inner cell mass increases, thereby determining the morphological factor of the inner cell mass.

[0199] S504. Using the body sphericity of the inner cell mass, the surface sphericity of the inner cell mass, and the morphological factor of the inner cell mass as the third basic three-dimensional feature of the inner cell mass.

[0200] Therefore, the body sphericity of the inner cell mass, the surface sphericity of the inner cell mass, and the morphological factors of the inner cell mass can together constitute the third basic three-dimensional feature.

[0201] Optionally, in step S202, feature mining analysis is performed based on various basic three-dimensional features to determine at least one combined three-dimensional feature of the inner cell mass, which may include: determining a first combined three-dimensional feature of the inner cell mass based on the second basic three-dimensional feature and the third basic three-dimensional feature; determining a second combined three-dimensional feature of the inner cell mass based on the second basic three-dimensional feature and the first basic three-dimensional feature.

[0202] Feature mining is mainly carried out from two perspectives: the distribution of the features themselves and the interaction between the features. The combined features are then screened from aspects such as their meaning, their indicative effect on clinical outcomes, and the distribution of feature data, ultimately obtaining combined features with practical and clinical significance.

[0203] The feature mining method is mainly to increase the dimension. In this embodiment, the original basic three-dimensional features can be mathematically operated to generate combined features. Optionally, operations can be performed from methods such as addition, multiplication, division and exponential functions, such as the volume of the inner cell mass*the ratio of the short axis to the long axis of the inner cell mass, log (the sphericity of the surface of the inner cell mass) + log (the ratio of the short axis to the long axis of the inner cell mass). The use of exponential functions can avoid the situation where the value of a smaller number after multiplication is smaller (i.e., numerical underflow).

[0204] When performing feature mining, multiple combined three-dimensional features can be combined. In this embodiment, two relatively typical combined three-dimensional features are used for illustration. The first combined three-dimensional feature can be calculated based on the obtained surface area to volume ratio of the inner cell mass and the sphericity of the inner cell mass. The second combined three-dimensional feature can be calculated based on the obtained surface area to volume ratio of the inner cell mass and the ratio of the short axis to the long axis of the inner cell mass.

[0205] Figure 6 A schematic flow chart of another inner cell mass analysis method provided in an embodiment of the present application; optionally, determining the first combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the third basic three-dimensional feature may include:

[0206] S601. Determine a third value according to the second basic three-dimensional feature and the body sphericity in the third basic three-dimensional feature.

[0207] Optionally, a third value may be obtained by multiplying the ratio of the surface area to the volume of the inner cell mass and the body sphericity of the inner cell mass, that is, the third value is: the ratio of the surface area to the volume of the inner cell mass*the body sphericity of the inner cell mass.

[0208] S602: Determine the third value as the first combined three-dimensional feature of the inner cell mass.

[0209] Thus, the first combined three-dimensional feature is: the ratio of the surface area to the volume of the inner cell mass*the body sphericity of the inner cell mass. The first combined three-dimensional feature combines the structural stability and body sphericity of the inner cell mass.

[0210] Optionally, determining the second combined three-dimensional feature of the inner cell mass based on the second basic three-dimensional feature and the first basic three-dimensional feature may include: performing weighted calculation based on the second basic three-dimensional feature, the first basic three-dimensional feature, the weight corresponding to the second basic three-dimensional feature, and the weight corresponding to the first basic three-dimensional feature to obtain the second combined three-dimensional feature of the inner cell mass.

[0211] According to the ratio of the surface area to the volume of the inner cell mass and the ratio of the short axis to the long axis of the inner cell mass, the second combined three-dimensional feature can be calculated.

[0212] The calculation of the second combined three-dimensional feature can be: the ratio of the surface area to the volume of the inner cell mass*0.9+the ratio of the short axis to the long axis of the inner cell mass*0.1. The second combined three-dimensional feature is obtained by combining the structural stability and shape of the inner cell mass.

[0213] In practical applications, the obtained combined three-dimensional features are not limited to those listed above, and other combined three-dimensional features can also be obtained by combination. This embodiment takes two combined three-dimensional features that have a greater impact on embryonic development research as examples.

[0214] It is worth mentioning that after obtaining the basic three-dimensional features of the inner cell mass, data cleaning can be performed on each basic three-dimensional feature to eliminate abnormal data, and then feature mining and analysis can be performed.

[0215] Data cleaning can be performed from the perspective of data and feature definition. Each type of feature has a corresponding reasonable data range. Data cleaning can be performed based on the standard data range corresponding to each basic 3D feature to remove data that is not within the standard data range.

[0216] Figure 7 A schematic flow chart of another inner cell mass analysis method provided in an embodiment of the present application; optionally, in step S105, before determining the correspondence between each three-dimensional feature and the embryonic development result based on multiple three-dimensional features of the inner cell mass, the following may also be included:

[0217] S701 . Draw a feature scatter plot based on each basic three-dimensional feature and each combined three-dimensional feature.

[0218] In some embodiments, feature correlation analysis may be performed on each of the obtained three-dimensional features to eliminate nonlinear and highly linear features to avoid feature omission.

[0219] Optionally, a feature scatter plot may be drawn based on each basic three-dimensional feature and each combined three-dimensional feature.

[0220] S702: Determine the correlation between any two three-dimensional features among the multiple three-dimensional features according to the feature scatter plot.

[0221] Based on the feature scatter plot, the linear and nonlinear relationships between features can be intuitively determined. Among them, for any two three-dimensional features, the linear and nonlinear relationships between the two three-dimensional features can be analyzed.

[0222] S703: According to the correlation between two three-dimensional features, three-dimensional features having nonlinear relationships and satisfying linear relationships are eliminated from the multiple three-dimensional features.

[0223] According to the analysis results, three-dimensional features with nonlinear relationships can be eliminated and three-dimensional features with strong linear relationships can be eliminated.

[0224] Optionally, in step S105, determining the correspondence between each three-dimensional feature and the result of embryonic development based on multiple three-dimensional features of the inner cell mass may include: performing statistical analysis based on each three-dimensional feature to determine the correspondence between each three-dimensional feature and the result of embryonic development; and screening each three-dimensional feature based on the analysis results to obtain screened three-dimensional features.

[0225] In some embodiments, the correspondence between each three-dimensional feature and the embryo development result can be determined by performing statistical analysis on the experimental data.

[0226] Table 1 shows the impact of the three-dimensional characteristics of the inner cell mass on clinical pregnancy outcomes:

[0227] Table 1

[0228]

[0229] Table 2 shows the impact of the three-dimensional characteristics of the inner cell mass on clinical live birth outcomes:

[0230] Table 2

[0231]

[0232] Table 3 shows the impact of the three-dimensional characteristics of the inner cell mass on clinical abortion outcomes:

[0233] Table 3

[0234]

[0235] It is worth noting that the miscarriage analysis results are inconsistent when the ratio of the surface area and volume of the inner cell mass is used as continuous data and grouped data respectively. This is mainly because the surface area and volume thresholds of the inner cell mass in the miscarriage data are set according to the thresholds of the clinical pregnancy data, and the miscarriage data and the clinical pregnancy data set are not the same data. Therefore, in essence, the thresholds should be different, but considering the correlation between the three clinical outcomes corresponding to this feature, the thresholds should be controlled to be the same; in addition, even in the same data set, different thresholds will correspond to different results. This is because, compared with continuous data, the discretization of the data will affect the data structure and lose some information, so the results may be different.

[0236] Combined with the table analysis, it can be seen that when the short-to-long axis ratio of the inner cell mass is <=0.864, the blastocyst has a higher clinical pregnancy rate (64.49% VS 40.48%, p=0.003) and clinical live birth rate (53.51% VS 28.57% p=0.003).

[0237] The surface area to volume ratio of the inner cell mass is an indicator that describes the structural stability of the inner cell mass. When the surface area to volume ratio of the inner cell mass was between 0.186 and 0.196, it had the highest clinical pregnancy rate (86.68% VS 63.75% VS 45.16%, p = 0.018) and live birth rate (66.67% VS 52.93% VS 32.26%, p = 0.04).

[0238] When the sphericity of the inner cell mass was <=38.14, the live birth rate was the highest (66.67% VS 53.51%, p=0.003).

[0239] Optionally, based on the analysis of the correspondence between three-dimensional features and embryonic development results, some of the extracted three-dimensional features may not have a significant impact on the embryonic development results and have no particularly obvious measurement value, so they can be screened out from the three-dimensional features to obtain the screened three-dimensional features.

[0240] Optionally, in step S105, embryo scoring modeling may include: adding one or more of the screened three-dimensional features as new training parameters to the original training parameters of the original embryo scoring model to establish a new embryo scoring model.

[0241] In order to further measure the performance of the three-dimensional features of the inner cell mass in the clinical prediction model, this embodiment can add the screened three-dimensional features of the inner cell mass to the traditional scoring system, establish a new embryo scoring model, and compare the performance.

[0242] In some embodiments, each three-dimensional feature can be added to a traditional scoring system as an independent training parameter to analyze and study the impact of a single three-dimensional feature on embryonic development outcomes.

[0243] In other embodiments, two or more three-dimensional features may be added together into a traditional scoring system for comprehensive analysis.

[0244] By adding the short-to-long axis ratio of the inner cell mass as a continuous feature to the existing scoring system, it was found that the performance of the established embryo scoring model was improved regardless of pregnancy, live birth or miscarriage.

[0245] In order to further measure the performance of the extracted three-dimensional features of the inner cell mass in the clinical prediction model, this paper uses the Gardner score as the performance baseline, and adds the basic three-dimensional features of the inner cell mass extracted above to the Gardner scoring system to establish a decision tree prediction model and compare the performance. All data are divided into training sets and test sets according to time, and GridSearch (when modeling a decision tree, there are many hyperparameters, such as the depth of the decision tree, evaluation indicators, minimum sample size, etc., which are called hyperparameters) is used to optimize the hyperparameters of each set of feature inputs in the training set to ensure that the modeling of each set of feature inputs is the best model.

[0246] Figure 8 This is a schematic diagram of the ROC curve of the pregnancy prediction model provided in the embodiment of the present application. Figure 8 (a) is a schematic diagram of the ROC curve of the pregnancy prediction model on the training set. Figure 8 (b) is a schematic diagram of the ROC curve of the pregnancy prediction model on the test set. The horizontal axis of the curve diagram represents the false positive rate of the model prediction results; the vertical axis represents the true positive rate of the model prediction results; Figure 8 In (a), curve 1 represents the ROC curve corresponding to the scoring system when the three-dimensional features of the inner cell mass are not added; curve 2 represents the ROC curve corresponding to the scoring system when the ratio of the short axis to the long axis of the inner cell mass is added. The performance comparison of the pregnancy prediction model is shown in Table 4:

[0247] Table 4

[0248] Pregnancy prediction model Auc: training set Auc: test set Rating system 0.604 0.595 Scoring system + short-to-long axis ratio of inner cell mass 0.631 0.611 Scoring system + surface area to volume ratio of inner cell mass 0.617 0.620 Scoring system + Sphericity of inner cell mass 0.633 0.601 Scoring system + surface area to volume ratio of inner cell mass & sphericity of inner cell mass 0.637 0.614 Scoring system + surface area to volume ratio of inner cell mass & short-to-long axis ratio of inner cell mass 0.621 0.600

[0249] Fig. 9 This is a schematic diagram of the ROC curve of the live birth prediction model provided in the embodiment of the present application. Fig. 9 (a) is a schematic diagram of the ROC curve of the live birth prediction model on the training set. Fig. 9(b) is a schematic diagram of the ROC curve of the live birth prediction model on the test set. The horizontal axis of the curve diagram represents the false positive rate of the model prediction results; the vertical axis represents the true positive rate of the model prediction results; Fig. 9 In (a), curve 1 represents the ROC curve corresponding to the scoring system when the three-dimensional features of the inner cell mass are not added; curve 2 represents the ROC curve corresponding to the scoring system when the ratio of the short axis to the long axis of the inner cell mass is added. The performance comparison of the live birth prediction model is shown in Table 5:

[0250] Table 5

[0251] Live birth prediction model Auc: training set Auc: test set Rating system 0.619 0.606 Scoring system + short-to-long axis ratio of inner cell mass 0.629 0.633 Scoring system + surface area to volume ratio of inner cell mass 0.624 0.622 Scoring system + Sphericity of inner cell mass 0.655 0.612 Scoring system + surface area to volume ratio of inner cell mass & sphericity of inner cell mass 0.610 0.627 Scoring system + surface area to volume ratio of inner cell mass & short-to-long axis ratio of inner cell mass 0.627 0.614

[0252] Fig.10 This is a schematic diagram of the ROC curve of the miscarriage prediction model provided in the embodiment of the present application. Fig.10 (a) is a schematic diagram of the ROC curve of the miscarriage prediction model on the training set. Fig.10 (b) is a schematic diagram of the ROC curve of the miscarriage prediction model on the test set. The horizontal axis of the curve diagram represents the false positive rate of the model prediction results; the vertical axis represents the true positive rate of the model prediction results; Fig.10 In (a), curve 1 represents the ROC curve corresponding to the scoring system when the three-dimensional features of the inner cell mass are not added; curve 2 represents the ROC curve corresponding to the scoring system when the ratio of the short axis to the long axis of the inner cell mass is added. The performance comparison of the miscarriage prediction model is shown in Table 6:

[0253] Table 6

[0254] Live birth prediction model Auc: training set Auc: test set Rating system 0.577 0.589 Scoring system + short-to-long axis ratio of inner cell mass 0.584 0.608 Scoring system + surface area to volume ratio of inner cell mass 0.614 0.682 Scoring system + Sphericity of inner cell mass 0.646 0.639 Scoring system + surface area to volume ratio of inner cell mass & sphericity of inner cell mass 0.700 0.712 Scoring system + surface area to volume ratio of inner cell mass & short-to-long axis ratio of inner cell mass 0.605 0.745

[0255] It can be seen from Tables 4, 5 and 6 that, in the test set, adding the three-dimensional features of the inner cell mass improves the prediction model of clinical outcomes to varying degrees. Among them, the surface area and volume ratio of the inner cell mass, that is, the stability of the structure, has the greatest improvement in the test set.

[0256] In summary, the inner cell mass analysis method provided in this embodiment includes: collecting multiple groups of blastocyst image sets, each group of blastocyst image sets includes multi-focal plane blastocyst images; determining the target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets; performing three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes the position information of each data point constituting the inner cell mass in three-dimensional space; according to the point cloud model, determining multiple three-dimensional features of the inner cell mass; according to the multiple three-dimensional features of the inner cell mass, determining the correspondence between each three-dimensional feature and the embryo development result, and performing embryo scoring modeling; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result. This scheme obtains a point cloud model of the inner cell mass by performing three-dimensional space modeling on the inner cell mass, so that based on the analysis of the point cloud model, multiple three-dimensional features of the inner cell mass can be extracted, thereby analyzing the correspondence between the extracted three-dimensional features and the embryo development result. In-depth research on embryonic development, and adding three-dimensional features to the traditional scoring system for embryo modeling analysis to form a new embryo scoring model, the application of the new embryo scoring model can improve the accuracy of embryo clinical outcome prediction. Among them, by extracting the three-dimensional features of the inner cell mass, we can make up for the problem of the relatively single two-dimensional features extracted in traditional methods, and the research and analysis of embryonic development results based on the three-dimensional features of the inner cell mass can improve the accuracy of the analysis results, thereby providing assistance for clinical research. Adding the three-dimensional features of the inner cell mass to the existing scoring system improves the measurement of the inner cell mass morphology and improves the predictive performance of the model.

[0257] The following is a description of the apparatus, device, storage medium, etc. used to perform the inner cell mass analysis method provided in the present application. The specific implementation process and technical effects are described above and will not be repeated below.

[0258] Fig.11 Schematic diagram of an inner cell mass analysis device provided in an embodiment of the present application, the functions implemented by the inner cell mass analysis device correspond to the steps performed by the above method. Fig.11 As shown, the device may include: a collection module 120, a determination module 121, a generation module 122, and an analysis module 123;

[0259] An acquisition module 120 is used to acquire multiple groups of blastocyst image sets, each group of blastocyst image sets includes multi-focal plane blastocyst images;

[0260] A determination module 121 is used to determine a target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets;

[0261] A generating module 122 is used to perform three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes position information of each data point constituting the inner cell mass in three-dimensional space;

[0262] A determination module 121, configured to determine a plurality of three-dimensional features of the inner cell mass according to the point cloud model;

[0263] The analysis module 123 is used to determine the correspondence between each three-dimensional feature and the embryo development result according to the multiple three-dimensional features of the inner cell mass, and to perform embryo scoring modeling; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result.

[0264] Optionally, the determination module 121 is specifically configured to determine a plurality of basic three-dimensional features of the inner cell mass according to the point cloud model;

[0265] Based on various basic three-dimensional features, feature mining analysis is performed to determine at least one combined three-dimensional feature of the inner cell mass;

[0266] According to each basic three-dimensional feature and each combined three-dimensional feature, multiple three-dimensional features of the inner cell mass are obtained.

[0267] Optionally, the determination module 121 is specifically used to determine a first basic three-dimensional feature of the inner cell mass according to the point cloud model, where the first basic three-dimensional feature is used to characterize the ratio of the short axis to the long axis of the inner cell mass;

[0268] determining a second basic three-dimensional feature of the inner cell mass according to the point cloud model, wherein the second basic three-dimensional feature is used to characterize a ratio of a surface area to a volume of the inner cell mass;

[0269] According to the point cloud model, the third basic three-dimensional feature of the inner cell mass is determined, and the third basic three-dimensional feature is used to characterize the spherical similarity of the inner cell mass.

[0270] Optionally, the determination module 121 is specifically used to perform principal component analysis on all data points in the point cloud model to determine three directions in which the data point variation meets the conditions;

[0271] According to the degree of variation of the data points in each direction, the first axis and the second axis of the inner cell mass are determined respectively; the first axis represents the long axis, and the second axis represents the short axis;

[0272] Determining a first ratio based on the first axis and the second axis;

[0273] The first ratio is taken as the first basic three-dimensional feature of the inner cell mass.

[0274] Optionally, the determination module 121 is specifically used to perform triangulation processing on the point cloud model to obtain a triangulated mesh model;

[0275] Based on the triangular mesh model, the surface area of ​​the inner cell mass was determined;

[0276] Based on the point cloud model, graphics operations are performed to obtain the volume of the inner cell mass;

[0277] Determining a second ratio based on the volume and the surface area;

[0278] The second ratio is used as the second basic three-dimensional feature of the inner cell mass.

[0279] Optionally, the determination module 121 is specifically configured to determine the body sphericity of the inner cell mass according to the volume of the inner cell mass and the volume of the first equivalent sphere; the surface area of ​​the first equivalent sphere is equal to the surface area of ​​the inner cell mass;

[0280] The sphericity of the inner cell mass is determined according to the surface area of ​​the inner cell mass and the surface area of ​​the second equivalent sphere; the volume of the second equivalent sphere is equal to the volume of the inner cell mass;

[0281] determining a morphogenetic factor of the inner cell mass; the morphogenetic factor is related to the number of tiny spikes on the surface of the inner cell mass;

[0282] The body sphericity of the inner cell mass, the surface sphericity of the inner cell mass and the morphological factor of the inner cell mass are taken as the third basic three-dimensional characteristics of the inner cell mass.

[0283] Optionally, the determination module 121 is specifically configured to determine a first combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the third basic three-dimensional feature;

[0284] A second combined three-dimensional feature of the inner cell mass is determined according to the second basic three-dimensional feature and the first basic three-dimensional feature.

[0285] Optionally, the determination module 121 is specifically configured to determine a third ratio according to the second basic three-dimensional feature and the body sphericity in the third basic three-dimensional feature;

[0286] The third ratio is determined as a first combined three-dimensional feature of the inner cell mass.

[0287] Optionally, the determination module 121 is specifically used to perform weighted calculation according to the second basic three-dimensional feature, the first basic three-dimensional feature, the weight corresponding to the second basic three-dimensional feature, and the weight corresponding to the first basic three-dimensional feature to obtain the second combined three-dimensional feature of the inner cell mass.

[0288] Optionally, the analysis module 123 is further used to draw a feature scatter plot according to each basic three-dimensional feature and each combined three-dimensional feature;

[0289] According to the feature scatter plot, determine the correlation between two three-dimensional features in multiple three-dimensional features;

[0290] According to the correlation between two three-dimensional features, three-dimensional features with nonlinear relationships and those satisfying linear relationships are eliminated from multiple three-dimensional features.

[0291] Optionally, the analysis module 123 is specifically used to perform statistical analysis based on various three-dimensional features to determine the corresponding relationship between various three-dimensional features and embryonic development results; based on the analysis results, the various three-dimensional features are screened to obtain screened three-dimensional features.

[0292] Optionally, the analysis module 123 is specifically configured to add one or more of the screened three-dimensional features as new training parameters to the original training parameters of the original embryo scoring model to establish a new embryo scoring model.

[0293] 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.

[0294] 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).

[0295] 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.

[0296] Fig.12A 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.

[0297] The device may include: a processor 801 and a storage medium 802 .

[0298] 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.

[0299] The storage medium 802 stores program code, and when the program code is executed by the processor 801, the processor 801 executes various steps of the inner cell mass analysis method according to various exemplary embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0300] 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.

[0301] 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.

[0302] 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.

[0303] 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.

[0304] 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.

[0305] 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.

[0306] 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 analyzing inner cell mass, characterized in that: include: Acquiring a plurality of sets of blastocyst image sets, each set of blastocyst image sets including multi-focal plane blastocyst images; Determining a target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets; Performing three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes position information of each data point constituting the inner cell mass in three-dimensional space; determining a plurality of three-dimensional features of the inner cell mass according to the point cloud model; According to the multiple three-dimensional features of the inner cell mass, the correspondence between each three-dimensional feature and the embryo development result is determined, and embryo scoring modeling is performed; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result.

2. The method according to claim 1, characterized in that Determining multiple three-dimensional features of the inner cell mass based on the point cloud model includes: Determining a plurality of basic three-dimensional features of the inner cell mass according to the point cloud model; Performing feature mining analysis based on various basic three-dimensional features to determine at least one combined three-dimensional feature of the inner cell mass; Based on each of the basic three-dimensional features and each of the combined three-dimensional features, multiple three-dimensional features of the inner cell mass are obtained.

3. The method according to claim 2, characterized in that Determining multiple basic three-dimensional features of the inner cell mass based on the point cloud model includes: Determining a first basic three-dimensional feature of the inner cell mass according to the point cloud model, wherein the first basic three-dimensional feature is used to characterize a ratio of a short axis to a long axis of the inner cell mass; determining a second basic three-dimensional feature of the inner cell mass according to the point cloud model, wherein the second basic three-dimensional feature is used to characterize a ratio of a surface area to a volume of the inner cell mass; A third basic three-dimensional feature of the inner cell mass is determined according to the point cloud model, and the third basic three-dimensional feature is used to characterize the spherical similarity of the inner cell mass.

4. The method according to claim 3, characterized in that Determining the first basic three-dimensional feature of the inner cell mass according to the point cloud model includes: Performing principal component analysis on all data points in the point cloud model to determine three directions in which data point variation meets conditions; Determining the first axis and the second axis of the inner cell mass according to the degree of variation of the data points in each direction; the first axis represents the long axis, and the second axis represents the short axis; Determine a first value according to the first axis and the second axis; The first value is used as a first basic three-dimensional feature of the inner cell mass.

5. The method according to claim 3, characterized in that: Determining the second basic three-dimensional feature of the inner cell mass according to the point cloud model includes: Performing triangular mesh processing on the point cloud model to obtain a triangular mesh model; Based on the triangular mesh model, determining the surface area of ​​the inner cell mass; Based on the point cloud model, graphics calculation is performed to obtain the volume of the inner cell mass; determining a second value based on the volume and the surface area; The second value is used as a second basic three-dimensional feature of the inner cell mass.

6. The method according to claim 3, characterized in that Determining the third basic three-dimensional feature of the inner cell mass according to the point cloud model includes: Determining the sphericity of the inner cell mass according to the volume of the inner cell mass and the volume of a first equivalent sphere; the surface area of ​​the first equivalent sphere is equal to the surface area of ​​the inner cell mass; Determining the surface sphericity of the inner cell mass according to the surface area of ​​the inner cell mass and the surface area of ​​a second equivalent sphere; the volume of the second equivalent sphere is equal to the volume of the inner cell mass; Determining a morphological factor of the inner cell mass; the morphological factor is related to the number of tiny spikes on the surface of the inner cell mass; The body sphericity of the inner cell mass, the surface sphericity of the inner cell mass and the morphological factor of the inner cell mass are used as the third basic three-dimensional feature of the inner cell mass.

7. The method according to claim 3, characterized in that The step of performing feature mining analysis based on various basic three-dimensional features to determine at least one combined three-dimensional feature of the inner cell mass includes: determining a first combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the third basic three-dimensional feature; A second combined three-dimensional feature of the inner cell mass is determined based on the second basic three-dimensional feature and the first basic three-dimensional feature.

8. The method according to claim 7, characterized in that Determining the first combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the third basic three-dimensional feature includes: determining a third value according to the second basic three-dimensional feature and the body sphericity in the third basic three-dimensional feature; The third value is determined as a first combined three-dimensional feature of the inner cell mass.

9. The method according to claim 7, characterized in that: Determining a second combined three-dimensional feature of the inner cell mass according to the second basic three-dimensional feature and the first basic three-dimensional feature includes: A weighted calculation is performed based on the second basic three-dimensional feature, the first basic three-dimensional feature, the weight corresponding to the second basic three-dimensional feature, and the weight corresponding to the first basic three-dimensional feature to obtain a second combined three-dimensional feature of the inner cell mass.

10. The method according to claim 2, characterized in that Before determining the correspondence between each three-dimensional feature and the embryonic development result according to the multiple three-dimensional features of the inner cell mass, the method further includes: Draw a feature scatter plot based on each basic three-dimensional feature and each combined three-dimensional feature; Determine, according to the feature scatter plot, the correlation between any two of the three-dimensional features in the plurality of three-dimensional features; According to the correlation between two three-dimensional features, three-dimensional features with nonlinear relationships and those satisfying linear relationships are eliminated from multiple three-dimensional features.

11. The method according to claim 1, characterized in that: Determining the correspondence between each three-dimensional feature and the embryonic development result according to the multiple three-dimensional features of the inner cell mass includes: According to the three-dimensional features, statistical analysis is performed to determine the corresponding relationship between the three-dimensional features and the embryo development results; According to the analysis results, various three-dimensional features are screened to obtain screened three-dimensional features.

12. The method according to claim 11, characterized in that The embryo scoring modeling comprises: One or more of the screened three-dimensional features are added as new training parameters to the original training parameters of the original embryo scoring model to establish a new embryo scoring model.

13. An inner cell mass analysis device, characterized in that: include: Acquisition module, determination module, generation module, analysis module; The acquisition module is used to acquire multiple groups of blastocyst image sets, each group of blastocyst image sets includes multi-focal plane blastocyst images; The determination module is used to determine the target blastocyst image set according to the number of inner cell masses contained in each group of blastocyst image sets; The generating module is used to perform three-dimensional modeling according to the target blastocyst image set to generate a point cloud model of the inner cell mass; the point cloud model includes position information of each data point constituting the inner cell mass in three-dimensional space; The determination module is used to determine multiple three-dimensional features of the inner cell mass according to the point cloud model; The analysis module is used to determine the correspondence between each three-dimensional feature and the embryo development result based on multiple three-dimensional features of the inner cell mass, and to perform embryo scoring modeling; the correspondence is used to characterize the degree of influence of the three-dimensional feature on the embryo development result.

14. 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, and when the electronic device is running, the processor and the storage medium communicate via the bus, and the processor executes the program instructions to implement the inner cell mass analysis method as described in any one of claims 1 to 12.

15. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the inner cell mass analysis method according to any one of claims 1 to 12 is implemented.

Images