Methods, systems, and electronic devices for morphological evaluation of te cells
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUZHOU BOUNDLESS MEDICAL TECH CO LTD
- Filing Date
- 2026-05-09
- Publication Date
- 2026-08-07
AI Technical Summary
[0006]本发明创造实施例提供的一种TE细胞的形态学评估方法、系统及电子设备，至少解决相关技术中对TE细胞进行形态学评估的准确性不足的问题
Smart Images
Figure CN122530102A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical image processing technology, and in particular to a method, system and electronic device for morphological evaluation of TE cells. Background Technology
[0002] The trophoblast cells (TE) are an important component of the blastocyst, and their morphological parameters directly reflect the developmental potential of the blastocyst.
[0003] Morphological assessment of TE cells mainly relies on embryologists observing the number and morphological uniformity of TE cells in blastocysts under a microscope, which serves as an important reference for embryo selection and transfer decisions. This assessment process depends on human experience and subjective judgment, resulting in insufficient repeatability and consistency.
[0004] To address this, related technologies use image erosion and dilation to segment and identify TE cell contours, and then calculate the number of TE cells, area standard deviation, and area coefficient of variation for evaluation. However, image erosion and dilation can easily cause the contours of small TE cells to disappear, making it difficult to segment overlapping TE cell contours, resulting in insufficient accuracy in morphological evaluation.
[0005] There is currently no effective solution to the problem of insufficient accuracy in morphological assessment of TE cells in related technologies. Summary of the Invention
[0006] The present invention provides a method, system, and electronic device for morphological evaluation of TE cells, which at least solves the problem of insufficient accuracy in morphological evaluation of TE cells in related technologies.
[0007] This invention provides a method for morphological evaluation of TE cells, comprising: inputting a 2D microscopic image of the target blastocyst surface into a segmentation model to obtain multiple TE cell contour masks; wherein the segmentation model is based on a Mask R-CNN convolutional neural network model for the mask region, with a multi-scale attention fusion module added to the output of the feature pyramid network and a boundary loss function set in the mask prediction branch, the weight coefficients of the boundary loss function being set to be related to the actual area of the TE cells; color-coding each TE cell contour mask to obtain a labeled image, wherein each TE cell contour mask corresponds to a unique color; statistically processing the labeled image to obtain the number of color types and the TE cell contour integral corresponding to each color, wherein the number of color types is used to characterize the number of TE cells; calculating the actual area of each TE cell based on the TE cell contour integral; and calculating the area standard deviation and area coefficient of variation based on the number of color types and the actual area to evaluate the morphological uniformity of the TE cells.
[0008] Preferably, before inputting the 2D micrograph of the target blastocyst surface into the segmentation model, the above method further includes: constructing an initial model of the segmentation model based on the Mask R-CNN model, a multi-scale attention fusion module, and a boundary loss function; training the initial model to obtain the training result; and using the corresponding initial model as the segmentation model if the training result meets the preset performance requirements.
[0009] Preferably, before constructing the initial model of the segmentation model based on the Mask R-CNN model, the multi-scale attention fusion module, and the boundary loss function, the above method further includes: determining the multi-scale attention fusion module based on joint modeling of channel attention and spatial attention; and constructing the boundary loss function based on the weight coefficients and the mask prediction loss of each TE cell.
[0010] Preferably, the boundary loss function is expressed as follows: ; In the formula, Represents the boundary loss function. Indicates the number of colors. Indicates the first The weighting coefficient of each TE cell Indicates the first Masking prediction loss for TE cells; Weighting coefficient of each TE cell The formula is expressed as: ; In the formula, Indicates the first The actual area of each TE cell This represents the preset smoothing constant. This indicates the preset adjustment factor.
[0011] Preferably, training the initial model to obtain training results includes: providing multiple 2D microscopic images of blastocysts; expanding the 2D microscopic images to obtain expanded samples; sequentially adjusting the size and labeling the expanded samples to obtain training samples; dividing the training samples into training set, validation set, and test set based on a preset ratio; and training the initial model based on the training set, validation set, and test set to obtain training results; wherein the preset performance requirements include that the specificity, sensitivity, and similarity coefficient of the initial model all reach above 90%.
[0012] Preferably, statistical processing is performed on the marker image to obtain the number of color types and the TE cell contour integral corresponding to each color, including: performing color statistics on the marker image to obtain the number of color types; performing contour point statistics on the marker image according to each color to determine the TE cell contour integral corresponding to each color; wherein, the formula for the TE cell contour integral is expressed as: ; In the formula, Indicates the first Contour integral of TE cells Indicates the first Number of contour points per TE cell Indicates the first On the first TE cell The x-coordinates of the contour points Indicates the first On the first TE cell The ordinates of the contour points Indicates the first On the first TE cell The x-coordinates of the contour points Indicates the first On the first TE cell The ordinates of the contour points.
[0013] Preferably, the formulas for calculating the area standard deviation and area coefficient of variation based on the number of color types and the actual area are expressed as follows: ; ; In the formula, Indicates the standard deviation of area. Indicates the number of colors. Indicates the first The actual area of each TE cell This represents the coefficient of variation of area.
[0014] This invention provides a morphological evaluation system for TE cells, comprising: a segmentation module that inputs a 2D microscopic image of the target blastocyst surface into a segmentation model to obtain multiple TE cell contour masks. The segmentation model is based on a Mask R-CNN convolutional neural network model for the mask region, with a multi-scale attention fusion module added to the output of the feature pyramid network and a boundary loss function set in the mask prediction branch. The weight coefficients of the boundary loss function are set to be related to the actual area of the TE cells. An evaluation module that colors each TE cell contour mask to obtain a labeled image, wherein each TE cell contour mask corresponds to a unique color; statistically processes the labeled image to obtain the number of color types and the TE cell contour integral corresponding to each color, wherein the number of color types is used to characterize the number of TE cells; calculates the actual area of each TE cell based on the TE cell contour integral; and calculates the area standard deviation and area coefficient of variation based on the number of color types and the actual area to evaluate the morphological uniformity of the TE cells.
[0015] Preferably, it also includes an acquisition module; the acquisition module rotates the target blastocyst and acquires initial images of the target blastocyst from different orientations; the initial images are stitched together to obtain a 2D micrograph of the target blastocyst.
[0016] An embodiment of the present invention provides an electronic device comprising: a processor, and a memory storing a program, the program including instructions which, when executed by the processor, cause the processor to perform any of the methods described above.
[0017] This invention provides a morphological evaluation method for TE cells. A 2D microscopic image of the target blastocyst surface is input into a segmentation model to obtain multiple TE cell contour masks. The segmentation model is based on a Mask R-CNN convolutional neural network model for the mask region, with a multi-scale attention fusion module added to the output of the feature pyramid network and a boundary loss function set in the mask prediction branch. The weight coefficients of the boundary loss function are set to be related to the actual area of the TE cells. Each TE cell contour mask is color-labeled to obtain a labeled image, where each TE cell contour mask corresponds to a unique color. The labeled image is statistically processed to obtain the number of color types and the TE cell contour integral corresponding to each color, where the number of color types represents the number of TE cells. The actual area of each TE cell is calculated based on the TE cell contour integral. The area standard deviation and area coefficient of variation are calculated based on the number of color types and the actual area to evaluate the morphological uniformity of the TE cells.
[0018] This technology can automatically acquire the number of TE cells, area standard deviation, and area coefficient of variation, eliminating the need for direct microscopic observation of TE cell morphology and overcoming the subjectivity of embryologists. Even when TE cells have overlapping outlines or significant differences in size, it can learn useful features, reducing the impact of image segmentation instability on the accuracy of TE cell morphological assessment. This addresses the insufficient accuracy of TE cell morphological assessment in related technologies. Attached Figure Description
[0019] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other embodiments based on these drawings without creative effort.
[0020] Figure 1 This is a flowchart of the steps of a morphological evaluation method for TE cells in an embodiment of the present invention.
[0021] Figure 2 This is a schematic diagram of the network architecture of the segmentation model in an embodiment of the present invention.
[0022] Figure 3 These are 2D microscopic images of the blastocyst surface in an embodiment of the present invention.
[0023] Figure 4 This is a schematic diagram illustrating the execution of a morphological evaluation method for TE cells in an embodiment of the present invention.
[0024] Figure 5 This is a comparative schematic diagram of TE cell contour images in an embodiment of the present invention.
[0025] Figure 6 This is a schematic diagram of a morphological evaluation system for TE cells in an embodiment of the present invention.
[0026] Figure 7 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation
[0027] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.
[0028] The trophoblast cells (TE) are an important component of the blastocyst, and their morphological parameters directly reflect the developmental potential of the blastocyst.
[0029] Morphological assessment of TE cells mainly relies on embryologists observing the number and morphological uniformity of TE cells in blastocysts under a microscope, which serves as an important reference for embryo selection and transfer decisions. This assessment process depends on human experience and subjective judgment, resulting in insufficient repeatability and consistency.
[0030] To address this, related technologies use image erosion and dilation to segment and identify TE cell contours, and then calculate the number of TE cells, area standard deviation, and area coefficient of variation for evaluation. However, image erosion and dilation can easily cause the contours of small TE cells to disappear, making it difficult to segment overlapping TE cell contours, resulting in insufficient accuracy in morphological evaluation.
[0031] Therefore, please refer to Figure 1 As shown, the present invention provides a method for morphological evaluation of TE cells, including steps S101 to S105.
[0032] Step S101: Input the 2D microscopic image of the target blastocyst surface into the segmentation model to obtain multiple TE cell contour masks. The segmentation model is obtained by adding a multi-scale attention fusion module to the output of the feature pyramid network and setting a boundary loss function in the mask prediction branch, based on the Mask R-CNN model of the mask region convolutional neural network. The weight coefficient of the boundary loss function is set to be related to the actual area of the TE cells.
[0033] Step S102: Color-mark each TE cell contour mask to obtain a marker map, wherein each TE cell contour mask corresponds to a unique color.
[0034] Step S103: Perform statistical processing on the labeled image to obtain the number of color types and the TE cell contour integral corresponding to each color, wherein the number of color types is used to characterize the number of TE cells.
[0035] Step S104: Calculate the actual area of each TE cell based on the TE cell contour integral.
[0036] Step S105: Calculate the area standard deviation and area coefficient of variation based on the number of color types and the actual area to assess the morphological uniformity of TE cells.
[0037] The segmentation model is used to automatically segment and identify the TE cell contours in 2D microscopic images of the blastocyst surface and obtain a mask for each TE cell contour. The Mask R-CNN model includes, but is not limited to, the backbone network (containing a feature pyramid network FPN), the region proposal network (RPN), classification and regression branches, and mask prediction branches.
[0038] Please refer to Figure 2 As shown, P3, P4, and P5 are three feature maps of different scales output by the feature pyramid network. The segmentation model provided in this embodiment is based on the Mask R-CNN model architecture. A boundary prediction branch (adding boundary loss function constraints) is added to the mask prediction branch to obtain the boundary enhancement mask branch. This allows the model to pay more attention to target contour information during segmentation, thereby improving the segmentation accuracy of adjacent TE cells. Furthermore, unlike the class-weighted loss function in related technologies, this embodiment introduces an instance-level adaptive weight mechanism based on TE cell area into the boundary loss function. By constructing a weight function related to the actual area of the TE cells, smaller TE cells occupy a higher proportion in the loss function, effectively improving the model's ability to segment small targets.
[0039] In other words, the segmentation model provided in this embodiment is a structurally improved model that introduces attention mechanism, boundary modeling, and small target weighting on the basis of Mask R-CNN model, and can still achieve accurate segmentation even when TE cells are densely distributed and have overlapping contours.
[0040] Color labeling of each TE cell contour mask means mapping each TE cell contour mask to a unique color value to obtain a labeled image containing multiple colors, where different colors correspond to different TE cells.
[0041] The formula for calculating the actual area of each TE cell based on the TE cell contour integral is expressed as follows: ; In the formula, Indicates the first The actual area of each TE cell Indicates the first Contour integral of TE cells This represents the pixel area conversion factor, used to convert pixel space to actual physical space. Pixel area conversion factor The specific value can be determined by those skilled in the art based on a limited number of experiments.
[0042] Unlike methods employing image erosion and dilation, the method described in this embodiment performs color statistical analysis on the TE cell contour mask obtained from the segmentation model. This avoids segmentation errors caused by contour adhesion and loss of small targets, improving the accuracy and stability of TE cell counting and area assessment. Furthermore, unlike related technologies that use image pixel counts to characterize TE cell area, the method in this embodiment calculates the actual area based on TE cell contour integrals, improving the accuracy of TE cell morphological parameter calculation to sub-pixel precision.
[0043] by Figure 3 Taking the 2D micrograph of the blastocyst surface shown as an example, the execution process of the above method provided in this embodiment can be referred to... Figure 4 As shown, the smaller the area standard deviation and area coefficient of variation, the better the morphological uniformity of TE cells.
[0044] In blastocyst quality assessment, the number and morphological homogeneity of TE cells are important indicators, closely related to blastocyst implantation rate and sustained pregnancy rate. During preimplantation genetic testing (PGT), the number and morphological homogeneity of TE cells not only affect the safety of the biopsy procedure but also the reliability of subsequent genetic testing results. Although TE cell morphological assessment cannot directly determine chromosomal abnormalities, blastocysts with poor TE cell morphological indicators often have a higher incidence of aneuploidy or mosaicism. Therefore, using the method described in this embodiment for TE cell morphological assessment helps to link embryo morphological observation with embryo selection and genetic testing, thereby optimizing embryo transfer strategies and improving pregnancy outcomes.
[0045] The method provided in this invention automatically acquires the number, area standard deviation, and area variation coefficient of TE cells through 2D microscopic images of the blastocyst surface, eliminating the need for direct microscopic observation of TE cell morphology and overcoming the subjectivity of embryologists. It can learn useful features even when TE cells have overlapping outlines and significant differences in size, reducing the impact of image segmentation instability on the accuracy of TE cell morphological assessment. Therefore, the method provided in this embodiment can achieve objective, accurate, and automated assessment of TE cell morphology, solving the problem of insufficient accuracy in morphological assessment of TE cells in related technologies, and helping to improve the success rate of embryo transfer and PGT detection.
[0046] Preferably, before inputting the 2D micrograph of the target blastocyst surface into the segmentation model in step S101, the above method further includes: rotating the target blastocyst and acquiring initial images of the target blastocyst from different orientations; stitching the initial images together to obtain a 2D micrograph of the target blastocyst.
[0047] Rotating the target blastocyst can be achieved through microtubule rotation blastocyst manipulation, including but not limited to biaxial orthogonal rotation and a combination of uniaxial and 90° rotation.
[0048] The algorithms for stitching the initial images can be, but are not limited to, Sift, Surf, and Harris Corner.
[0049] Preferably, initial images of the target blastocyst are acquired from six positions: front, back, left, right, top, and bottom.
[0050] Preferably, before inputting the 2D micrograph of the target blastocyst surface into the segmentation model in step S101, the above method further includes: constructing an initial model of the segmentation model based on the Mask R-CNN model, a multi-scale attention fusion module, and a boundary loss function; training the initial model to obtain the training result; and using the corresponding initial model as the segmentation model if the training result meets the preset performance requirements.
[0051] Preferably, before constructing the initial model of the segmentation model based on the Mask R-CNN model, the multi-scale attention fusion module, and the boundary loss function, the above method further includes: determining the multi-scale attention fusion module based on joint modeling of channel attention and spatial attention; and constructing the boundary loss function based on the weight coefficients and the mask prediction loss of each TE cell.
[0052] The boundary loss function is expressed as follows: ; In the formula, Represents the boundary loss function. Indicates the number of colors. Indicates the first The weighting coefficient of each TE cell Indicates the first Masking prediction loss for TE cells.
[0053] No. Weighting coefficient of each TE cell The formula is expressed as: ; In the formula, Indicates the first The actual area of each TE cell This represents a preset smoothing constant used to prevent the denominator from being zero; This represents a preset regulatory factor used to control the sensitivity of the weights to the area of TE cells.
[0054] The multi-scale attention fusion module is determined by joint modeling of channel attention and spatial attention, which is an existing technology. However, in this embodiment, the multi-scale attention fusion module is set at the output of the feature pyramid network of the Mask R-CNN model, which can improve the model's ability to express features of small-scale TE cells and overlapping regions.
[0055] Preferably, training the initial model to obtain training results includes: providing multiple 2D microscopic images of blastocysts; expanding the 2D microscopic images to obtain expanded samples; sequentially adjusting the size and labeling the expanded samples to obtain training samples; dividing the training samples into training set, validation set, and test set based on a preset ratio; and training the initial model based on the training set, validation set, and test set to obtain training results; wherein the preset performance requirements include that the specificity, sensitivity, and similarity coefficient of the initial model all reach above 90%.
[0056] For example, by rotating the blastocyst through a microtube under a microscope, one 2D micrograph of the blastocyst surface can be obtained for each blastocyst. By preparing 400 blastocysts in this way, a total of 400 2D micrograph samples of the blastocyst surface can be obtained. In order to increase the amount of data, rotation, horizontal flipping, vertical flipping and contrast adjustment are performed to expand the 400 2D micrograph samples to 4000 2D micrograph samples.
[0057] The image size of 4000 2D micrographs was adjusted to 1300 pixels × 800 pixels to fit the input size of the segmentation model. Simultaneously, the TE cell contours in these 4000 images were labeled using the LabelImg annotation software to generate corresponding label data, which was in *.json format.
[0058] The 4000 images and their corresponding labels were divided into training, validation, and test sets for the model based on an 8:1:1 ratio.
[0059] The performance of the trained model is shown in Table 1.
[0060] Table 1 Performance of the segmentation model
[0061] Segmentation Model 98.3 94.4 95.0
[0062] The specificity reached 98.3%, the sensitivity reached 94.4%, and the similarity coefficient reached 95.0%, indicating that the trained segmentation model has significant inference performance.
[0063] Preferably, the marker image is statistically processed to obtain the number of color types and the TE cell contour integral corresponding to each color, including: performing color statistics on the marker image to obtain the number of color types; and performing contour point statistics on the marker image according to each color to determine the TE cell contour integral corresponding to each color.
[0064] The formula for the TE cell contour integral is expressed as: ; In the formula, Indicates the first Contour integral of TE cells Indicates the first Number of contour points per TE cell Indicates the first On the first TE cell The x-coordinates of the contour points Indicates the first On the first TE cell The ordinates of the contour points Indicates the first On the first TE cell The x-coordinates of the contour points Indicates the first On the first TE cell The ordinates of the contour points.
[0065] The formulas for calculating the standard deviation of area and the coefficient of variation of area based on the number of color types and the actual area are expressed as follows: ; ; In the formula, Indicates the standard deviation of area. Indicates the number of colors. Indicates the first The actual area of each TE cell This represents the coefficient of variation of area.
[0066] For example, to further illustrate the accuracy of the method provided in this embodiment, 360 2D micrographs of blastocyst surfaces were divided into three groups: a manual observation group, an image erosion and dilation group, and the method group of this embodiment, corresponding to the manual observation method, the image erosion and dilation method, and the method provided in this embodiment, respectively.
[0067] In the morphological assessment of TE cells, morphological parameters of TE cells were obtained using these three groups. Each group of blastocyst TE cells was stained, and the true values of the morphological parameters for each group were obtained using confocal microscopy. The accuracy results of the three assessment methods are shown in Table 2. In addition to confocal microscopy, two-photon microscopy or laser scanning microscopy can also be used to obtain the true values of TE cell morphological parameters.
[0068] Table 2. Accuracy of different TE cell morphology assessment methods
[0069] Manual observation method 78.3% Image erosion and dilation method 87.5% Method of this embodiment 95.8%
[0070] The method group in this embodiment has the highest accuracy of TE cell morphology parameters, reaching 95.8%, which is significantly higher than that of manual observation and image erosion dilation method.
[0071] For example, to further illustrate the evaluation effect of the above-mentioned method provided in this embodiment, 100 blastocyst surface 2D micrographs were divided into two groups: an image erosion and dilation group and a method group of this embodiment, corresponding to the image erosion and dilation method and the above-mentioned method provided in this embodiment, respectively.
[0072] For easier comparison, the TE cell contour images obtained by the two methods are stitched together. Please refer to [link / reference]. Figure 5 As shown. Among them, located in Figure 5 The lower half of the black-and-white image is a TE cell contour image obtained by image erosion and dilation. It suffers from issues such as the disappearance of contours of some small TE cells and the failure to separate overlapping TE cell contours, resulting in poor image quality. Figure 5 The upper half of the color image is a TE cell outline image obtained by the method described above in this embodiment. It can preserve the outline of small TE cells, separate overlapping TE cell outlines, and has a better image effect, which helps to improve the accuracy of TE cell morphological assessment.
[0073] Please refer to Figure 6 As shown in the embodiment of the present invention, a morphological evaluation system for TE cells is also provided, including a segmentation module 62 and an evaluation module 63.
[0074] The segmentation module 62 inputs a 2D microscopic image of the target blastocyst surface into the segmentation model to obtain multiple TE cell contour masks. The segmentation model is based on the Mask R-CNN model of the mask region convolutional neural network, with a multi-scale attention fusion module added to the output of the feature pyramid network and a boundary loss function set in the mask prediction branch. The weight coefficients of the boundary loss function are set to be related to the actual area of the TE cells.
[0075] Evaluation module 63 colors-labels each TE cell contour mask to obtain a labeled map, where each TE cell contour mask corresponds to a unique color; statistically processes the labeled map to obtain the number of color types and the TE cell contour integral corresponding to each color, where the number of color types is used to characterize the number of TE cells; calculates the actual area of each TE cell based on the TE cell contour integral; and calculates the area standard deviation and area coefficient of variation based on the number of color types and the actual area to evaluate the morphological uniformity of TE cells.
[0076] Preferably, please refer to Figure 6 As shown, the system also includes a data acquisition module 61.
[0077] The acquisition module 61 rotates the target blastocyst and acquires initial images of the target blastocyst from different orientations; the initial images are then stitched together to obtain a 2D micrograph of the target blastocyst.
[0078] In summary, the technical solutions provided by the embodiments of the present invention can achieve objective, accurate and automatic evaluation of TE cell morphology, solve the problem of insufficient accuracy in morphological evaluation of TE cells in related technologies, and provide more assistance for embryo quality assessment and PGT detection.
[0079] The present invention also provides a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a method according to an embodiment of the present invention.
[0080] The present invention also provides a computer program product, including a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform the method of the embodiments of the present invention.
[0081] This invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, which, when executed by the at least one processor, causes the electronic device to perform the method of this invention.
[0082] refer to Figure 7This is a structural block diagram of an electronic device, either a server or a client, according to an embodiment of the present invention. It is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.
[0083] like Figure 7 As shown, the electronic device includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 702 or a computer program loaded into a random access memory (RAM) 703 from a storage unit 708. The RAM 703 may also store various programs and data required for the operation of the electronic device. The computing unit 701, the ROM 702, and the RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.
[0084] Multiple components in the electronic device are connected to I / O interface 705, including: input unit 706, output unit 707, storage unit 708, and communication unit 709. Input unit 706 can be any type of device capable of inputting information into the electronic device. Input unit 706 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of the electronic device. Output unit 707 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 708 may include, but is not limited to, disks and optical discs. Communication unit 709 allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, and / or wireless communication transceivers, such as Bluetooth devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.
[0085] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, CPUs, graphics processing units (GPUs), various special-purpose artificial intelligence (AI) computing units, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above. For example, in some embodiments, the method embodiments of the present invention can be implemented as computer programs tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on an electronic device via ROM 702 and / or communication unit 709. In some embodiments, the computing unit 701 can be configured to perform the methods described above by any other suitable means (e.g., by means of firmware).
[0086] Computer programs for implementing the methods of embodiments of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0087] In the context of embodiments of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable signal medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, or infrared systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0088] It should be noted that the term "comprising" and its variations used in the embodiments of this invention are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "a plurality" mentioned in the embodiments of this invention are illustrative and not restrictive, and those skilled in the art should understand that unless explicitly indicated otherwise in the context, they should be understood as "one or more". The descriptions of terms such as "first", "second", etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of indicated technical features.
[0089] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of this invention are subject to strict compliance with relevant laws, regulations, and regulatory requirements in their collection, storage, use, processing, transmission, provision, and disclosure, and adhere to the principles of legality, legitimacy, necessity, and good faith. The acquisition of relevant information and data is premised on the user's explicit consent or other legitimate reasons, and a clear and convenient authorization management approach is provided to the user, allowing the user to independently choose to consent, withdraw consent, or refuse to provide relevant information. For functions that rely on user information, if the user does not authorize or withdraws authorization, the corresponding technical function cannot be implemented, and the technical solution of this invention is not applicable in this scenario.
[0090] The steps described in the method embodiments provided by the present invention can be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of protection of the present invention is not limited in this respect.
[0091] The term "embodiment" in this specification refers to a specific feature, structure, or characteristic described in connection with an embodiment that may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily imply the same embodiment, nor does it imply independence or alternativeity from other embodiments. The various embodiments in this specification are described in a related manner, with reference to each other for similar or identical parts. In particular, for apparatus, device, and system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, and relevant details are referred to in the description of the method embodiments.
[0092] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of protection. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.
Claims
1. A method for morphological evaluation of TE cells, characterized in that, include: The 2D microscopic image of the target blastocyst surface is input into the segmentation model to obtain multiple TE cell contour masks. The segmentation model is based on the Mask R-CNN model of the mask region convolutional neural network, with the addition of a multi-scale attention fusion module at the output of the feature pyramid network and the setting of a boundary loss function in the mask prediction branch. The weight coefficients of the boundary loss function are set to be related to the actual area of the TE cells. Each TE cell contour mask is color-labeled to obtain a label map, where each TE cell contour mask corresponds to a unique color; The labeled image is statistically processed to obtain the number of color types and the TE cell contour integral corresponding to each color, wherein the number of color types is used to characterize the number of TE cells; The actual area of each TE cell is calculated based on the TE cell contour integral. The area standard deviation and area coefficient of variation are calculated based on the number of color types and the actual area to assess the morphological uniformity of TE cells.
2. The method according to claim 1, characterized in that, Before inputting a 2D micrograph of the target blastocyst surface into the segmentation model, the method further includes: The initial model of the segmentation model is constructed based on the Mask R-CNN model, the multi-scale attention fusion module, and the boundary loss function; The initial model is trained to obtain the training results; If the training results meet the preset performance requirements, the corresponding initial model will be used as the segmentation model.
3. The method according to claim 2, characterized in that, Before constructing the initial model of the segmentation model based on the Mask R-CNN model, the multi-scale attention fusion module, and the boundary loss function, the method further includes: The multi-scale attention fusion module is determined based on joint modeling of channel attention and spatial attention; The boundary loss function is constructed based on the weighting coefficients and the mask prediction loss of each TE cell.
4. The method according to claim 3, characterized in that, The formula for the boundary loss function is expressed as follows: ； In the formula, Represents the boundary loss function. Indicates the number of colors. Indicates the first The weighting coefficient of each TE cell Indicates the first Masking prediction loss for individual TE cells; No. Weighting coefficient of each TE cell The formula is expressed as: ； In the formula, Indicates the first The actual area of each TE cell This represents the preset smoothing constant. This indicates the preset adjustment factor.
5. The method according to claim 2, characterized in that, The initial model is trained to obtain training results, including: Provides 2D micrographs of multiple blastocysts; The 2D micrograph sample is augmented to obtain an augmented sample; The augmented samples are then subjected to size adjustment and labeling processes in sequence to obtain training samples; The training samples are divided into a training set, a validation set, and a test set based on a preset ratio. The initial model is trained based on the training set, the validation set, and the test set to obtain the training result; The preset performance requirements include that the specificity, sensitivity, and similarity coefficient of the initial model all reach 90% or higher.
6. The method according to claim 1, characterized in that, Statistical processing of the labeled image yields the number of color types and the TE cell contour integral corresponding to each color, including: Perform color statistics on the marked image to obtain the number of color types; The outline points of the marked map are counted according to each color, and the TE cell outline integral corresponding to each color is determined. The formula for the TE cell contour integral is expressed as follows: ； In the formula, Indicates the first Contour integral of TE cells Indicates the first Number of contour points per TE cell Indicates the first On the first TE cell The x-coordinates of the contour points Indicates the first On the first TE cell The ordinates of the contour points Indicates the first On the first TE cell The x-coordinates of the contour points Indicates the first On the first TE cell The ordinates of the contour points.
7. The method according to claim 1, characterized in that, The formulas for calculating the area standard deviation and area coefficient of variation based on the number of color types and the actual area are expressed as follows: ； ； In the formula, Indicates the standard deviation of area. Indicates the number of colors. Indicates the first The actual area of each TE cell This represents the coefficient of variation of area.
8. A morphological evaluation system for TE cells, characterized in that, include: The segmentation module inputs a 2D microscopic image of the target blastocyst surface into the segmentation model to obtain multiple TE cell contour masks. The segmentation model is based on the Mask R-CNN model of the mask region convolutional neural network, with a multi-scale attention fusion module added to the output of the feature pyramid network and a boundary loss function set in the mask prediction branch. The weight coefficients of the boundary loss function are set to be related to the actual area of the TE cells. The evaluation module colors-labels each TE cell contour mask to obtain a labeled map, where each TE cell contour mask corresponds to a unique color; it performs statistical processing on the labeled map to obtain the number of color types and the TE cell contour integral corresponding to each color, where the number of color types is used to characterize the number of TE cells; it calculates the actual area of each TE cell based on the TE cell contour integral; and it calculates the area standard deviation and area coefficient of variation based on the number of color types and the actual area to evaluate the morphological uniformity of TE cells.
9. The system according to claim 8, characterized in that, It also includes a data acquisition module; The acquisition module rotates the target blastocyst and acquires initial images of the target blastocyst from different orientations; the initial images are then stitched together to obtain a 2D micrograph of the target blastocyst.
10. An electronic device, comprising: A processor and a memory storing a program, characterized in that the program includes instructions that, when executed by the processor, cause the processor to perform the method according to any one of claims 1 to 7.