Sperm identification method, system and device, computer equipment and readable storage medium

Abstract

The invention relates to a sperm identification method, system and device, computer equipment and a readable storage medium. The method comprises the following steps: in response to a sperm identification instruction, controlling a three-axis electric control objective table to move a culture dish placed on the objective table in spaces in X, Y and Z directions of an inverted microscope according to a preset scanning path, so that the inverted microscope scans sperm observation liquid in the culture dish, obtaining a microscopic image of the sperm observation liquid; taking the microscopic image as input, calling a trained sperm identification model, identifying sperms in the microscopic image, and obtaining a sperm identification result, the trained sperm identification model being obtained by training based on a historical microscopic image of category labels and position information of a real frame carrying sperms; and determining the spatial position of each identified sperm based on the sperm identification result. By adopting the method, the sperm identification efficiency can be improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a sperm identification method, system, device, computer equipment, computer-readable storage medium and computer program product. Background Art

[0002] NOA (Non-Obstructive Azoospermia, NOA) is a type of male infertility characterized by the lack of sperm in the semen due to the inability of the testicles to produce enough sperm normally. NOA can be caused by a variety of factors, including genetic, environmental, inflammatory and drug factors. In order to help these couples conceive through in vitro fertilization, the surgeon needs to obtain a testicular specimen through microsurgical testicular sperm extraction (microTESE), and after confirming the presence of sperm under a microscope, the sperm is frozen and preserved or extracted from the testicle at an optional time, and then the sperm is injected through intracytoplasmic sperm injection to achieve fertilization.

[0003] However, in current clinical practice, there is a problem of low sperm recognition efficiency after microTESE sperm retrieval. Summary of the invention

[0004] Based on this, it is necessary to provide a sperm identification method, device, computer equipment, computer-readable storage medium and computer program product that can improve the efficiency of sperm identification in order to address the above technical problems.

[0005] In a first aspect, the present application provides a sperm identification method, which is applied to a sperm identification system. The sperm identification system includes an inverted microscope equipped with a three-axis electric-controlled stage, including:

[0006] In response to the sperm recognition instruction, the three-axis electric-controlled stage is controlled to move the culture dish placed on the stage in the space of the X, Y, and Z directions of the inverted microscope according to a preset scanning path, so that the inverted microscope scans the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid;

[0007] Taking the microscopic image as input, calling the trained sperm recognition model, identifying the sperm in the microscopic image, and obtaining the sperm recognition result. The trained sperm recognition model is trained based on the historical microscopic images carrying the category label and position information of the real frame of the sperm;

[0008] Based on the sperm identification results, the spatial position of each identified sperm is determined.

[0009] In one embodiment, the sperm identification result includes the position, length, width and head shape of the sperm, and the number of microscopic images is multiple. After obtaining the sperm identification result, the method further includes:

[0010] For each sperm, the movement trajectory of the sperm is determined according to the position of the sperm in the sperm identification results of multiple microscopic images;

[0011] The sperm morphological characteristics score is determined based on the length, width and head shape of the sperm;

[0012] According to the sperm movement trajectory, the sperm motility score is determined;

[0013] The sperm quality score is determined based on the morphological characteristics score and the motility score.

[0014] In one embodiment, after determining the sperm quality score, the method further comprises:

[0015] A sperm quality assessment report is generated based on the morphological characteristic score, motility score and quality score of each sperm.

[0016] In one embodiment, before identifying sperm in a microscopic image based on a trained sperm identification model to obtain a sperm identification result, the method further includes:

[0017] Obtain historical microscopic images of sperm observation fluid, annotate the category and position information of the real frame of sperm in the historical microscopic images, and obtain a training set;

[0018] Constructing an initial sperm recognition model;

[0019] Using the training set as input, the initial sperm recognition model is called to determine the category, location information and confidence of the prediction box, which is used to represent the area where the sperm is located in the historical microscopic image;

[0020] Based on the category and location information of the true box, as well as the category, location information and confidence of the predicted box, determine the error between the predicted box and the true box;

[0021] Based on the error, the parameters of the initial sperm recognition model are iteratively updated until the preset training end condition is reached to obtain a trained sperm recognition model.

[0022] In one embodiment, the method includes determining an error between a predicted box and a true box, including:

[0023] Determine the classification loss, positioning loss, and confidence loss between the predicted box and the true box;

[0024] Based on classification loss, localization loss, and confidence loss, the error between the predicted box and the true box is determined.

[0025] In one embodiment, after obtaining the trained sperm recognition model, the method further includes:

[0026] Construct a test set based on historical microscopic images annotated with the category and position information of the sperm ground truth box;

[0027] Using the test set as input, the trained sperm recognition model is called to identify sperm in historical microscopic images to obtain historical sperm recognition results;

[0028] Based on the historical sperm identification results, the sperm detection accuracy of each historical microscopic image in the test set is determined;

[0029] According to the sperm detection accuracy of each historical microscopic image, the target detection accuracy of the trained sperm recognition model is determined;

[0030] When the target detection accuracy is less than a preset target accuracy threshold, the structure of the trained sperm model is adjusted.

[0031] In a second aspect, the present application further provides a sperm identification system, comprising a three-axis electric-controlled stage, a controller, an inverted microscope and a data processing module, wherein the three-axis electric-controlled stage is connected to the controller and is disposed on the inverted microscope, and the inverted microscope is connected to the data processing module;

[0032] The controller, in response to the sperm recognition instruction of the data processing module, controls the three-axis electric-controlled stage to move the culture dish placed on the stage in the space of the inverted microscope in the X, Y and Z directions according to a preset scanning path;

[0033] An inverted microscope is used to scan the sperm observation fluid in the culture dish to obtain a microscopic image of the sperm observation fluid, and send the microscopic image to a data processing module;

[0034] The data processing module is used to send a sperm recognition instruction to the controller, use the sperm recognition method to recognize sperm in the microscopic image, and determine the spatial position of each recognized sperm.

[0035] In one embodiment, the three-axis electric-controlled stage includes a connecting plate, a Z-axis motion mechanism, a Y-axis motion mechanism, an X-axis motion mechanism, a loading platform, and a Z-axis motion fine-tuning knob;

[0036] The three-axis electric-controlled stage is connected to the base of the inverted microscope through a connecting plate. The loading platform, X-axis motion mechanism and Y-axis motion mechanism are stacked in sequence. The Z-axis motion mechanism is arranged between the Y-axis motion mechanism and the connecting plate, and is respectively connected to the Y-axis motion mechanism and the connecting plate. The Z-axis motion fine-tuning knob is arranged on the Z-axis motion mechanism.

[0037] In a third aspect, the present application also provides a sperm identification device, comprising:

[0038] The sperm observation liquid scanning module is used to respond to the sperm recognition instruction and control the three-axis electric control stage to move the culture dish placed on the stage in the space of the X, Y and Z directions of the inverted microscope according to the preset scanning path, so that the inverted microscope scans the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid;

[0039] The sperm recognition module is used to use the microscopic image as input, call the trained sperm recognition model, identify the sperm in the microscopic image, and obtain the sperm recognition result. The trained sperm recognition model is trained based on the historical microscopic images carrying the category label and position information of the real frame of the sperm;

[0040] The spatial position determination module is used to determine the spatial position of each identified sperm based on the sperm identification result.

[0041] In a fourth aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in any one of the above-mentioned sperm identification method embodiments when executing the computer program.

[0042] In a fifth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in any one of the above-mentioned sperm identification method embodiments are implemented.

[0043] In a sixth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps in any one of the above-mentioned sperm identification method embodiments.

[0044] The sperm identification system is connected to the controller by mounting a three-axis electric-controlled stage on an inverted microscope. When the controller responds to the sperm identification instruction, the three-axis electric-controlled stage is controlled to move the culture dish placed on the stage in the space of the X, Y, and Z directions of the inverted microscope according to the preset scanning path, so that the inverted microscope scans the sperm observation liquid in the culture dish. By adjusting the position of the culture dish, the focal length and scanning area of ​​the microscope are changed accordingly, and there is no need to perform tedious microscope adjustment work, thereby realizing the automatic scanning process of the sperm observation liquid. Different from the existing method of manually adjusting the focal length and coordinates of the microscope, the adjustment accuracy and smoothness are improved, and the possibility of inaccurate scanning due to fatigue caused by long-term manual operation during manual scanning is reduced, thereby improving the scanning efficiency, scanning accuracy and precision of the sperm observation liquid.

[0045] The sperm identification method, system, device, computer equipment, computer-readable storage medium and computer program product control the movement of the three-axis electric-controlled stage in the sperm identification system by electric-controlled automation, so as to realize automatic and precise adjustment of the position of the culture dish placed on the stage in the three directions of X, Y and Z, so that the scanning range of the inverted microscope can cover the entire area of ​​the sperm observation fluid in the culture dish, and obtain multiple microscopic images of the sperm observation fluid, which is conducive to improving the accuracy of sperm identification. In addition, in combination with target detection to identify sperm, after scanning and obtaining the microscopic image of the sperm observation fluid, the historical microscopic image based on the category label and position information of the real frame carrying the sperm is used. The sperm recognition model is trained to identify sperm in the microscopic image and obtain the recognition result. The spatial position of each identified sperm is determined based on the sperm recognition result. On the one hand, the sperm in the microscopic image is automatically identified by the sperm recognition model, which greatly reduces the detection time, reduces the workload of the operator, reduces the possibility of missed detection due to fatigue during manual detection, and improves the efficiency and accuracy of sperm recognition. On the other hand, the spatial position of the sperm can be quickly and accurately located through the sperm recognition result, which is beneficial to evaluate the sperm quality according to the identified spatial position of the sperm to extract sperm samples, assist in medical procedures such as in vitro fertilization, and thus help improve the success rate of medical procedures such as in vitro fertilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A diagram showing an application environment of a sperm identification method in one embodiment;

[0048] Figure 2 A schematic diagram of a sperm identification method according to an embodiment;

[0049] Figure 3 A schematic diagram of a sperm identification method according to another embodiment;

[0050] Figure 4 A schematic diagram of a sperm identification method in another embodiment;

[0051] Figure 5 is a structural block diagram of a sperm identification device in one embodiment;

[0052] Figure 6 is a structural block diagram of a sperm identification system in one embodiment;

[0053] Figure 7 A schematic diagram of the assembly structure of a sperm recognition system in one embodiment;

[0054] Figure 8 It is a structural schematic diagram of an installation platform in one embodiment;

[0055] Fig. 9 is a schematic structural diagram of an inverted microscope in one embodiment;

[0056] Fig.10 A schematic diagram of the structure of a three-axis electrically controlled stage in one embodiment;

[0057] Fig.11 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0059] The sperm identification method provided in the embodiments of the present application can be applied to Figure 1 In the sperm identification system shown in the figure, the control terminal 102 communicates with the three-axis electric-controlled stage 104 and the inverted microscope 106 respectively, and the three-axis electric-controlled stage 104 is mounted on the inverted microscope 106.

[0060] Specifically, the control terminal 102 may respond to a sperm recognition instruction initiated by an operator, control the three-axis electric-controlled stage to move the culture dish placed on the stage in the X, Y, and Z directions of the inverted microscope according to a preset scanning path, so that the inverted microscope scans the sperm observation fluid in the culture dish to obtain a microscopic image of the sperm observation fluid. The inverted microscope sends the microscopic image to the control terminal 102. The control terminal 102 uses the microscopic image as input, calls a trained sperm recognition model, recognizes sperm in the microscopic image, and obtains a sperm recognition result. The trained sperm recognition model is trained based on historical microscopic images that carry category labels and position information of real frames of sperm. Finally, based on the sperm recognition result, the spatial position of each identified sperm is determined.

[0061] The control terminal 102 may be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, etc.

[0062] In an exemplary embodiment, Figure 2 As shown, a sperm identification method is provided, which is applied to Figure 1The control terminal 102 in FIG. 1 is used as an example to illustrate the method, which includes the following S100 to S300. Among them:

[0063] S100, in response to the sperm recognition instruction, controlling the three-axis electric-controlled stage to move the culture dish placed on the stage in the space of the inverted microscope in the X, Y, and Z directions according to a preset scanning path, so that the inverted microscope scans the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid;

[0064] The three-axis electric stage is used to precisely control the spatial position of the culture dish. The inverted microscope is a microscope in which the objective lens and the illumination system are inverted. The sperm observation fluid is obtained by configuring the semen extract of NOA patients. The microscopic image is an electron scanning microscope image.

[0065] In practical applications, the movement distance and movement sequence of the motion mechanism for controlling the movement of the stage in the three directions of X, Y, and Z in the three-axis electric-controlled stage can be predetermined to obtain a set scanning path. The sperm observation liquid is placed in a culture dish, and the culture dish is placed on the stage. When the controller responds to the sperm recognition instruction, the three-axis electric-controlled stage is controlled to move the position of the stage according to the set scanning path, so that the culture dish moves in the space of the three directions of X, Y, and Z on the objective lens of the inverted microscope, so that the inverted microscope scans the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid.

[0066] S200, taking the microscopic image as input, calling the trained sperm recognition model, identifying the sperm in the microscopic image, and obtaining the sperm recognition result. The trained sperm recognition model is trained based on the historical microscopic images carrying the category label and position information of the real frame of the sperm.

[0067] The real frame is used to identify the location area of ​​sperm in the image. The category label of the real frame may include sperm and background. The location information of the real frame may include the center point coordinates of the real frame and the width and height of the real frame. The sperm recognition result may include the location information of the area where each sperm is located in the microscopic image.

[0068] In practical applications, historical microscopic images obtained by scanning sperm observation fluid in a historical time period can be obtained first, and the real frame of sperm, the category label and position information of the real frame can be annotated on the historical microscopic images. The pre-built target detection model can be trained using the historical microscopic images to identify sperm in the image and obtain a trained sperm recognition model. After obtaining the microscopic image of the sperm observation fluid by configuring and scanning the semen extract extracted by microTESE, the trained sperm recognition model is called to obtain the sperm recognition result.

[0069] S300, based on the sperm identification result, determine the spatial position of each identified sperm.

[0070] In practical applications, the coordinates (x, y) of a fixed point in the microscope image and the relative distance between the X value and Y value of the X and Y axis motion mechanism of the three-axis electric-controlled stage during scanning can be determined in advance. After the sperm recognition result is obtained, the two-dimensional coordinates of the sperm are determined based on the position information of the identified sperm in the image and the relative distance, and the spatial position of the sperm is determined in combination with the Z value of the Z axis motion mechanism of the three-axis electric-controlled stage.

[0071] In other embodiments, after the spatial position of the sperm is obtained, the three-axis electric-controlled stage can be controlled to quickly locate to the spatial position of the sperm based on the spatial position of the sperm to facilitate subsequent sperm extraction.

[0072] In the above-mentioned sperm identification method, the movement of the three-axis electric-controlled stage in the system is controlled by electric-controlled automation to realize automatic and precise adjustment of the position of the culture dish placed on the stage in the three directions of X, Y, and Z, so that the scanning range of the inverted microscope can cover the entire area of ​​the sperm observation fluid in the culture dish, and multiple microscopic images of the sperm observation fluid are obtained, which is conducive to improving the accuracy of sperm identification. In addition, combined with target detection to identify sperm, after scanning to obtain the microscopic image of the sperm observation fluid, a sperm recognition model is obtained according to historical microscopic image training based on the category label and position information of the real frame carrying the sperm, and the sperm in the microscopic image is identified. The sperm is identified and the identification result is obtained. The spatial position of each identified sperm is determined based on the sperm identification result. On the one hand, the sperm in the microscopic image is automatically identified by the sperm identification model, which greatly reduces the detection time, reduces the workload of the operator, reduces the possibility of fatigue and missed detection during manual detection, and improves the efficiency and accuracy of sperm identification. On the other hand, the spatial position of the sperm is quickly and accurately located through the sperm identification result, which is conducive to evaluating the sperm quality according to the identified spatial position of the sperm to extract sperm samples, assist in in vitro fertilization and other medical procedures, thereby helping to improve the success rate of in vitro fertilization and other medical procedures.

[0073] After obtaining the sperm identification result, the quality of the identified sperm can be evaluated. In an exemplary embodiment, the sperm identification result includes the position, length, width and head shape of the sperm. The number of microscopic images is multiple, such as Figure 3 As shown, it also includes S420 to S480. Among them:

[0074] S420, for each sperm, determining the movement trajectory of the sperm according to the position of the sperm in the sperm identification results of the multiple microscopic images.

[0075] The movement trajectory of sperm may include movement form and movement amplitude. The movement form may include linear movement, rotational movement, and no movement, etc., and the movement amplitude may include movement speed.

[0076] In actual applications, the inverted microscope continuously scans while the culture dish moves to obtain multiple continuous microscopic images. During the scanning process, the sperm position information in the microscopic image is identified in sequence by the sperm recognition model. The change in the sperm position information in multiple continuous microscopic images can be analyzed through differential images to determine the movement form and speed of the sperm.

[0077] S440, determining the sperm morphological characteristic score based on the length, width and head shape of the sperm.

[0078] Among them, the length and width of the sperm may include the length and width of the sperm head, the length and width of the midsection, and the length and width of the tail.

[0079] In practical applications, the sperm head shape standard, head length standard, head width standard, midsection length standard, midsection width standard, tail length standard and tail width standard can be set according to the strict morphological standard of Kruger, and corresponding scoring conditions can be set for each standard, and then the sperm morphological score can be determined by comprehensive scoring. Specifically, the head shape is 1 point for an oval shape, and 0 points for other shapes. The head length is 1 point between 4.0 and 5.5 microns, otherwise 0 points; the head width is 1 point between 2.5 and 3.5 microns, otherwise 0 points. The midsection length is 1 point between 1.1 and 1.5 microns, otherwise 0 points; the midsection width is 1 point between 0.5 and 0.8 microns, otherwise 0 points. The tail length is 1 point between 45 and 55 microns, otherwise 0 points; the tail width is 1 point between 0.2 and 0.3 microns, otherwise 0 points. The comprehensive score is obtained by summing up the scores of various criteria of the morphological characteristics of sperm in the sperm identification result, and the morphological characteristic score of sperm is obtained according to the preset morphological characteristic scoring standard. The morphological characteristic scoring standard can be that the higher the comprehensive score, the higher the morphological characteristic score.

[0080] S460, determining the motility score of the sperm according to the movement trajectory of the sperm.

[0081] In practical applications, the movement form scoring condition and the movement amplitude scoring condition may be set separately, and the movement form score and the movement amplitude score may be determined according to the movement form and the movement amplitude in the movement trajectory of the sperm. The sperm motility score is determined according to the movement form score and the movement amplitude score. Exemplarily, the movement form scoring condition may include: 2 points for linear movement, 1 point for rotational movement, and 0 points for no movement; the movement amplitude scoring condition may include: 2 points for linear movement speed greater than 20 microns per second, 1 point for linear movement speed between 10 microns per second and 20 microns per second, and 0 points for linear movement speed less than 10 microns per second; 2 points for the curve speed of rotational movement greater than 30 microns per second, 1 point for the curve speed between 20 microns per second and 30 microns per second, and 0 points for the curve speed less than 20 microns per second. According to the sum of the scores of the movement trajectory of the sperm in the sperm identification result, a comprehensive score is obtained, and according to the preset motility scoring standard, the sperm motility score is obtained. The motility scoring standard may be that the higher the comprehensive score, the higher the motility score.

[0082] S480, determining a sperm quality score based on the morphological characteristic score and the motility score.

[0083] In practical applications, weights corresponding to the morphological characteristic scores and motility scores can be pre-assigned. After the morphological characteristic scores and motility scores of sperm are obtained, the sperm quality score is obtained by weighted summation based on the assigned weights.

[0084] In this embodiment, by obtaining the movement trajectory of the sperm based on the sperm identification results in multiple microscopic images, the quality of the sperm is evaluated from two aspects, namely, the morphological characteristics and motility of the sperm according to the movement trajectory of the sperm, the length, width and head shape of the sperm, which is beneficial to improve the accuracy of the evaluation, and further, it is beneficial to improve the accuracy of sperm sample extraction based on the evaluation results.

[0085] After obtaining the sperm quality score, in an exemplary embodiment, step S500 is further included.

[0086] S500, generating a sperm quality assessment report based on the morphological characteristic score, motility score and quality score of each sperm.

[0087] In practical applications, after obtaining the quality score of each identified sperm, the morphological characteristic score and detailed parameters of the morphological characteristics (such as the length and width of the sperm head, the length and width of the midsection, and the length and width of the tail), the motility score and detailed parameters of the motility (such as the form of movement and the amplitude of movement) and the corresponding quality score of each sperm can be integrated to generate a sperm quality assessment report, so that medical staff can quickly understand the quality of the sperm, which is conducive to the extraction of high-quality sperm according to the quality assessment report.

[0088] In this embodiment, a sperm quality assessment report is generated based on the morphological characteristic score, motility score and quality score of each sperm, which is helpful for medical staff to quickly understand the quality of sperm based on the quality assessment report.

[0089] In order to improve the efficiency of sperm recognition, in an exemplary embodiment, before sperm in a microscopic image is recognized based on a trained sperm recognition model and sperm recognition results are obtained, Figure 4 As shown, it also includes S610 to S650. Among them:

[0090] S610, obtaining historical microscopic images of sperm observation fluid, annotating the category and position information of the real frame of sperm in the historical microscopic images, and obtaining a training set.

[0091] The category label of the true frame may include sperm and background. The location information of the true frame may include the coordinates of the center point of the true frame and the width and height of the true frame. The training set is used to train the object detection model to identify sperm in the image.

[0092] In practical applications, historical microscopic images of sperm observation fluid scanned within a historical time period may be obtained, the category labels of the real frames in the historical microscopic images, the coordinates of the center points of the real frames, and the width and height of the real frames may be annotated, and the width and height of the real frames may be normalized and preprocessed to obtain a training set.

[0093] S620, constructing an initial sperm recognition model.

[0094] In practical applications, it can be constructed by selecting a deep learning model, such as the R-CNN model (Regions with CNN features), the YOLO model or the SSD (Single Shot MultiBox Detector) model. For example, the YOLOv5 (You Only Look Once version 5) model is selected to construct the initial sperm recognition model. It can also be designed based on the characteristics of the sperm training set, and the convolutional neural network is selected as the network architecture of the model to determine the number of network layers, the number of convolution kernels per layer, the size, the pooling method and the activation function to construct the initial sperm recognition model. After the initial sperm recognition model is constructed, the hyperparameters of the model can also be set, such as the learning rate, the maximum training batch and the batch size.

[0095] S630, using the training set as input, calling the initial sperm recognition model, determining the category, location information and confidence of the prediction box, where the prediction box is used to represent the area where the sperm in the historical microscopic image is located.

[0096] Among them, the confidence of the prediction box represents the probability estimation of the prediction accuracy of the category and location information of the prediction box.

[0097] In practical applications, the training set is input into the initial sperm recognition model. The initial sperm recognition model generates sperm prediction boxes in the historical microscopic images by extracting and fusing features of the historical microscopic images in the training set. For each prediction box, its category is predicted to obtain the category, location information and confidence of the prediction box.

[0098] S640, based on the category and position information of the real box, and the category, position information and confidence of the predicted box, determine the error between the predicted box and the real box.

[0099] In practical applications, the error between the predicted box and the true box can be determined by calculating the difference between the position information of the predicted box and the position information of the true box, the difference between the category of the predicted box and the true category, and the intersection ratio (confidence) of the predicted box and the true box. The error can be calculated by a loss function, and the loss function includes but is not limited to a mean square error loss function, a cross entropy loss function, and a smoothed L1 loss function.

[0100] S650, iteratively updating the parameters of the initial sperm recognition model based on the error until a preset training end condition is reached to obtain a trained sperm recognition model.

[0101] In practical applications, after the prediction box category and position information are obtained, the loss function, the category, position information and confidence of the real box are used to calculate the loss function value (error), and then the back propagation can be started from the output layer of the initial sperm recognition model according to the loss function value, and the parameter gradient value of each layer is calculated, and the parameters of the initial sperm recognition model are updated to minimize the loss function value. The preset training end condition can be that the loss function value is continuously less than the preset loss threshold within the preset number of iterations, and the training of the sperm recognition model is stopped to obtain the trained sperm recognition model.

[0102] In this embodiment, by training the sperm recognition model, it is helpful to realize automated sperm recognition, thereby, compared with the manual recognition method, it is helpful to greatly improve the efficiency of sperm recognition, reduce the workload of operators, reduce the possibility of fatigue and missed detection during manual detection, and improve the accuracy of sperm recognition.

[0103] In order to quantify the recognition effect of the sperm recognition model, the error is determined by the classification loss, positioning loss and confidence loss of the predicted box and the true box. In an exemplary embodiment, S650 also includes S652 to S654. Among them:

[0104] S652, determining the classification loss, positioning loss, and confidence loss between the predicted box and the true box.

[0105] S654: Determine the error between the predicted box and the true box based on the classification loss, the positioning loss, and the confidence loss.

[0106] In practical applications, based on the intersection-over-union ratio of the predicted box and the true box, the intersection-over-union ratio loss is calculated to obtain the positioning loss. The confidence loss is determined by the binary cross entropy loss function. The classification loss is obtained by the following formula:

[0107]

[0108] Where B represents the number of historical microscopic images in the training set, Represents the category label (1 or 0) of the ground-truth box of historical microscopic image i, 1 represents sperm and 0 represents background. It represents the probability that the model predicts that the category label of the prediction box of the historical microscopic image i is correct.

[0109] After obtaining the classification loss, positioning loss and confidence loss between the predicted box and the true box, the error between the predicted box and the true box is obtained by weighted summing the classification loss, positioning loss and confidence loss.

[0110] In this embodiment, the error between the predicted box and the true box is determined by determining the classification loss, positioning loss and confidence loss of the predicted box and the true box, which is conducive to adjusting the model parameters according to the error to improve the accuracy of sperm recognition.

[0111] In order to improve the recognition accuracy of the sperm recognition model, in an exemplary embodiment, after obtaining the trained sperm recognition model, S710 to S750 are further included. Among them:

[0112] S710, construct a test set based on historical microscopic images annotated with the category and position information of the true sperm box.

[0113] Among them, the test set is used to evaluate the performance of the sperm recognition model.

[0114] In practical applications, after obtaining historical microscopic images with the category and position information of the real sperm box marked, the order of these images can be shuffled and divided into training sets and test sets according to a certain ratio. For example, 20% of the historical microscopic images are obtained to construct a test set.

[0115] S720, taking the test set as input, calling the trained sperm recognition model, identifying sperm in the historical microscopic image, and obtaining the historical sperm recognition result.

[0116] Among them, the historical sperm recognition results include the category, position information and confidence of the predicted box of sperm in the historical microscopic images.

[0117] In practical applications, the test set is input into the trained sperm recognition model. The model recognizes sperm in historical microscopic images in the training set and outputs the category, location information and confidence of the sperm prediction box.

[0118] S730, based on the historical sperm identification results, determine the sperm detection accuracy of each historical microscopic image in the test set.

[0119] Among them, the sperm detection accuracy of historical microscopic images indicates the percentage of sperm counts correctly predicted by the sperm recognition model to the total number of sperm in the image, specifically:

[0120]

[0121] in, Indicates the total number of sperm detected correctly. Indicates the total number of sperm contained in the image.

[0122] In practical applications, the category and position information of the real frame of sperm in the historical microscopic images in the training set are compared with the category and position information of the predicted frame of sperm in the corresponding historical microscopic images to determine the accuracy of sperm detection in each historical microscopic image.

[0123] S740, determining the target detection accuracy of the trained sperm recognition model according to the sperm detection accuracy of each historical microscopic image.

[0124] In practical applications, after obtaining the sperm recognition accuracy of each historical microscopic image, the target detection accuracy can be obtained by summing the sperm detection accuracy of each historical microscopic image and dividing the sum by the total number of historical microscopic images in the test set.

[0125] S750: When the target detection accuracy is less than a preset target accuracy threshold, adjust the structure of the trained sperm recognition model.

[0126] In practical applications, when the target detection rate is less than a preset target accuracy threshold (such as 0.9), the network design of the sperm recognition model can be adjusted, and the model training and model evaluation can be re-performed until the target table detection accuracy is higher than the preset target accuracy threshold.

[0127] In this embodiment, the detection accuracy of the sperm recognition model is determined through a test set, and the detection performance of the model is evaluated based on the detection accuracy. According to the evaluation results, it is decided whether to adjust the structure of the sperm recognition model, so as to ensure that the detection performance of the sperm recognition model meets the requirements.

[0128] In order to make a clearer description of the sperm identification method provided by the present application, a specific embodiment is described below, and the specific embodiment includes the following steps:

[0129] S1, in response to the sperm recognition instruction, controls the three-axis electric-controlled stage to move the culture dish placed on the stage in the space of the X, Y, and Z directions of the inverted microscope according to a preset scanning path, so that the inverted microscope scans the sperm observation fluid in the culture dish to obtain a microscopic image of the sperm observation fluid.

[0130] S2, taking the microscopic image as input, calls the trained sperm recognition model to identify the sperm in the microscopic image and obtain the sperm recognition result. The trained sperm recognition model is trained based on the historical microscopic images carrying the category label and position information of the real box of the sperm.

[0131] S3, based on the sperm identification result, determining the spatial position of each identified sperm.

[0132] S4, for each sperm, the movement trajectory of the sperm is determined according to the position of the sperm in the sperm identification results of multiple microscopic images, and the morphological characteristic score of the sperm is determined according to the length, width and head shape of the sperm.

[0133] S5, determining the motility score of the sperm according to the movement trajectory of the sperm, and determining the quality score of the sperm based on the morphological characteristic score and the motility score.

[0134] S6, generating a sperm quality assessment report based on the morphological characteristic score, motility score and quality score of each sperm.

[0135] In this embodiment, before identifying sperm in a microscopic image based on a trained sperm identification model and obtaining a sperm identification result, a sperm identification model is constructed and trained. The construction and training process of the sperm identification model is as follows:

[0136] S101, obtaining historical microscopic images of sperm observation fluid, annotating the category and position information of the real frame of sperm in the historical microscopic images, and obtaining a training set.

[0137] S102, constructing an initial sperm recognition model.

[0138] S103, using the training set as input, calling the initial sperm recognition model, determining the category, location information and confidence of the prediction box, where the prediction box is used to represent the area where the sperm in the historical microscopic image is located.

[0139] S104, based on the category and position information of the real box, as well as the category, position information and confidence of the predicted box, determine the classification loss, positioning loss and confidence loss between the predicted box and the real box, and based on the classification loss, positioning loss and confidence loss, determine the error between the predicted box and the real box.

[0140] S105, iteratively updating the parameters of the initial sperm recognition model based on the error until a preset training end condition is reached to obtain a trained sperm recognition model.

[0141] S106, constructing a test set based on historical microscopic images annotated with the category and position information of the true sperm box.

[0142] S107, taking the test set as input, calling the trained sperm recognition model, identifying sperm in the historical microscopic image, and obtaining the historical sperm recognition result.

[0143] S108, based on the historical sperm recognition results, determine the sperm detection accuracy of each historical microscopic image in the test set, and determine the target detection accuracy of the trained sperm recognition model according to the sperm detection accuracy of each historical microscopic image.

[0144] S109, when the target detection accuracy is less than a preset target accuracy threshold, adjusting the structure of the trained sperm model.

[0145] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0146] In an exemplary embodiment, Figure 5 As shown, a sperm identification device 600 is provided, comprising: a sperm observation liquid scanning module 610, a sperm identification module 620 and a spatial position determination module 630, wherein:

[0147] The sperm observation fluid scanning module 610 is used to respond to the sperm recognition instruction and control the three-axis electric-controlled stage to move the culture dish placed on the stage in the X, Y, and Z directions of the inverted microscope according to a preset scanning path, so that the inverted microscope scans the sperm observation fluid in the culture dish to obtain a microscopic image of the sperm observation fluid.

[0148] The sperm recognition module 620 is used to take the microscopic image as input, call the trained sperm recognition model, identify the sperm in the microscopic image, and obtain the sperm recognition result. The trained sperm recognition model is trained based on the historical microscopic images carrying the category label and position information of the real frame of the sperm.

[0149] The spatial position determination module 630 is used to determine the spatial position of each identified sperm based on the sperm identification result.

[0150] In an exemplary embodiment, the sperm identification device 600 further includes a sperm quality assessment module 640 for determining, for each sperm, the movement trajectory of the sperm according to the position of the sperm in the sperm identification results of the plurality of microscopic images; and determining the morphological characteristic score of the sperm according to the length, width and head shape of the sperm;

[0151] The sperm motility score is determined based on the sperm movement trajectory; the sperm quality score is determined based on the morphological characteristic score and the motility score.

[0152] In an exemplary embodiment, the sperm identification device 600 further includes a sperm quality assessment report generating module 650, which is further used to generate a sperm quality assessment report based on the morphological feature score, motility score and quality score of each sperm.

[0153] In an exemplary embodiment, the sperm identification device 600 further includes an image acquisition module 660, a model building module 670 and a model training module 680, wherein:

[0154] The image acquisition module 660 is used to acquire historical microscopic images of sperm observation fluid, annotate the category and position information of the real frame of sperm in the historical microscopic images, and obtain a training set.

[0155] The model building module 670 is used to build an initial sperm recognition model.

[0156] The model training module 680 is used to call the initial sperm recognition model with the training set as input, determine the category, position information and confidence of the prediction box, the prediction box is used to represent the area where the sperm in the historical microscopic image is located; based on the category and position information of the real box, as well as the category, position information and confidence of the prediction box, determine the error between the prediction box and the real box; based on the error, iteratively update the parameters of the initial sperm recognition model until the preset training end condition is reached, thereby obtaining a trained sperm recognition model.

[0157] In an exemplary embodiment, the model training module 680 is also used to determine the classification loss, positioning loss and confidence loss between the predicted box and the true box; based on the classification loss, positioning loss and confidence loss, determine the error between the predicted box and the true box.

[0158] In an exemplary embodiment, the sperm identification device 600 further includes a model evaluation module 690. In which:

[0159] The image acquisition module 660 is further used to construct a test set based on the historical microscopic images annotated with the category and position information of the real sperm box.

[0160] The model evaluation module 690 is used to take the test set as input, call the trained sperm recognition model, identify the sperm in the historical microscopic images, and obtain the historical sperm recognition results; based on the historical sperm recognition results, determine the sperm detection accuracy of each historical microscopic image in the test set; determine the target detection accuracy of the trained sperm recognition model according to the sperm detection accuracy of each historical microscopic image; when the target detection accuracy is less than the preset target accuracy threshold, adjust the structure of the trained sperm model.

[0161] Each module in the sperm identification device 600 can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0162] In an exemplary embodiment, Figure 6 As shown, a sperm identification system is provided, comprising an inverted microscope 110, a three-axis electric-controlled stage 120 and its controller 130, and a data processing module 140. The three-axis electric-controlled stage 120 is connected to the controller 130 and is arranged on the inverted microscope 110. The inverted microscope 110 and the controller 130 are respectively connected to the data processing module 140. Among them:

[0163] The controller 130, in response to the sperm identification instruction of the data processing module 140, controls the three-axis electric-controlled stage 120 to move the culture dish placed on the stage in the space of the inverted microscope in the X, Y, and Z directions according to a preset scanning path;

[0164] The inverted microscope 110 is used to scan the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid, and send the microscopic image to the data processing module 134;

[0165] The data processing module 140 is used to send a sperm identification instruction to the controller 130, use the above-mentioned sperm identification method to identify the sperm in the microscopic image, and determine the spatial position of each identified sperm.

[0166] In practical applications, the assembly structure of the sperm recognition system is as follows Figure 7As shown, the inverted microscope 2, the three-axis electric-controlled stage 3 and its controller 4, and the data processing module (not numbered in the figure) of the sperm identification system are placed on the installation platform 1. Among them, the data processing module is the data processing terminal of the sperm identification system, including computer display 6, computer host 7, keyboard 8, mouse 9 and other components, which are used for data processing and human-computer interaction.

[0167] Wherein, the installation platform 1, such as Figure 8 As shown, it includes a mounting platform body 101 and four bottom casters 102. Installing the casters 102 can increase the flexibility of application scenarios of the sperm identification system.

[0168] The inverted microscope 2 is the core component of the system. The prepared sperm observation solution is placed in the culture dish 5, and the culture dish 5 is placed on the three-axis electric-controlled stage 3. The three-axis electric-controlled stage 3 is controlled to move according to the specified scanning path. The inverted microscope 2 captures the observation image (microscopic image) of the culture dish in real time and uploads the image data to the computer. Fig. 9 As shown, the inverted microscope 2 mainly includes an eyepiece 201, an objective lens, a condenser 202, a focusing knob 203, etc. In this embodiment, the focusing knob 203 is only used for calibration of the initial observation state before the sperm observation fluid is automatically scanned and identified.

[0169] Among them, the three-axis electric-controlled stage 3 is loaded with a culture dish 5, and the controller 4 of the three-axis electric-controlled stage 3 is used for accurately adjusting the spatial position of the culture dish 5, so as to facilitate a detailed search for sperm in the sperm observation fluid at the microscopic level.

[0170] In an exemplary embodiment, Fig.10 As shown, the three-axis electric-controlled stage 3 includes a connecting plate 301, a Z-axis motion mechanism 302, a Y-axis motion mechanism 303, an X-axis motion mechanism 304, a loading platform 305, and a Z-axis motion fine-tuning knob 306;

[0171] The three-axis electric-controlled stage 3 is connected to the base of the inverted microscope 2 via a connecting plate 301; the loading platform 305, the X-axis motion mechanism 304 and the Y-axis motion mechanism 303 are stacked in sequence; the Z-axis motion mechanism 302 is disposed between the Y-axis motion mechanism 303 and the connecting plate 301, and is respectively connected to the Y-axis motion mechanism 303 and the connecting plate 301; and the Z-axis motion fine-tuning knob 306 is disposed on the Z-axis motion mechanism 302.

[0172] In practical applications, the connecting plate 301 is used to fix the three-axis electric-controlled stage 3 on the base of the inverted microscope 2. The Z-axis motion mechanism 302 is used to adjust the position of the culture dish 5 in the vertical direction, the Y-axis motion mechanism 303 is used to adjust the position of the culture dish 5 along the Y-axis direction (front and back) at a certain focal length, and the X-axis motion mechanism 304 is used to adjust the position of the culture dish 5 along the X-axis direction (left and right) at a certain focal length. The loading platform 305 is fixed above the X-axis motion mechanism 304. The Z-axis motion fine-tuning knob 306 is used for system calibration, and the culture dish 5 is placed on the loading platform 305. In this embodiment, the culture dish 5 is specially used for the observation of the sperm observation fluid, which can realize a wide range of rapid sperm identification without damaging the sperm, thereby facilitating subsequent operations.

[0173] In one embodiment, the automated scanning and identification process of the sperm identification system can be:

[0174] The data processing terminal stores a sperm recognition model and a sperm quality assessment program for analyzing the image data collected by the inverted microscope. Since the sperm extracted from the testicles of NOA patients by microTESE are mostly non-motile or low-motile sperm, the sperm recognition model can quickly identify and locate the three-dimensional coordinates of the sperm in the culture dish, and evaluate the motility and morphological characteristics of the captured sperm.

[0175] Place the pre-configured sperm operation solution in the culture dish, place the culture dish at zero position, start the sperm automatic identification process through the sperm identification command initiated by the data processing terminal, and the controller responds to the sperm identification command to control the three-axis electric-controlled stage to move according to the preset scanning path to change the spatial position of the culture dish placed on the stage accordingly:

[0176] The Z-axis motion mechanism automatically rises / falls a specified displacement Lz and then maintains the current position, the X-axis motion mechanism moves along the X-axis direction from the front limit position to the rear limit position, the Y-axis motion mechanism moves along the Y-axis direction a specified displacement Ly and then maintains the current position, the X-axis motion mechanism moves along the X-axis direction from the right limit position to the left limit position, until the Y-axis motion mechanism reaches the limit position in the current motion direction, and the single-layer scanning on the microscopic level is completed.

[0177] Since the sperm in the sperm observation solution may settle to the bottom of the culture dish or suspend in the observation solution, resulting in the sperm being distributed in different layers of the sperm observation solution, it is necessary to adjust the position of the Z axis to adjust the focal length of the culture dish and the objective lens of the inverted microscope to focus on the sperm in different layers. Therefore, the aforementioned moving process is repeated until the scanning path is completed to ensure that the scanning range of the microscope covers the entire area of ​​the sperm observation solution.

[0178] During the movement of the loading platform, the inverted microscope captures the microscopic image of the sperm observation fluid in real time and uploads it to the data processing terminal. The data processing terminal automatically identifies and locates the sperm in the microscopic image according to the sperm recognition model and determines the spatial position of the sperm. The sperm quality assessment report is generated through the sperm quality assessment program. The spatial positions of all sperm scanned and identified are recorded and stored. A complete sperm automatic recognition operation is completed, and the located sperm spatial position is used for subsequent sperm automatic extraction operations.

[0179] In this embodiment, by combining a three-axis electric-controlled stage equipped with an inverted microscope and a sperm recognition model, automated sperm recognition and positioning are achieved. On the one hand, by controlling the X, Y, and Z axis motion mechanisms of the three-axis electric-controlled stage to move according to a preset scanning path, the focal length and coordinates of the microscope are automatically adjusted without manual adjustment, thereby improving the efficiency and accuracy of scanning the sperm observation fluid. On the other hand, by scanning and acquiring the microscopic image of the sperm observation fluid in real time, the automatic recognition and positioning of the sperm in the microscopic image greatly improves the efficiency of sperm recognition, reduces the possibility of fatigue miss detection during manual detection, and improves the accuracy of sperm recognition. On the other hand, by determining the spatial position of the sperm through the recognition results, combined with the sperm quality assessment report, it is conducive to quickly locating the sperm that needs to be extracted to assist in medical procedures such as in vitro fertilization.

[0180] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.11As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a method for determining the focus height is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0181] Those skilled in the art will understand that Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0182] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps in any one of the above-mentioned sperm identification method embodiments are implemented.

[0183] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in any one of the above sperm identification method embodiments are implemented.

[0184] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps in any one of the above sperm identification method embodiments.

[0185] It should be noted that 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 this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0186] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0187] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0188] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A sperm identification method, characterized in that: Applied to a sperm identification system, the sperm identification system includes an inverted microscope equipped with a three-axis electrically controlled stage, and the method includes: In response to the sperm recognition instruction, the three-axis electric-controlled stage is controlled to move the culture dish placed on the stage in the space of the inverted microscope in the X, Y, and Z directions according to a preset scanning path, so that the inverted microscope scans the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid; Taking the microscopic image as input, calling a trained sperm recognition model to recognize sperm in the microscopic image to obtain a sperm recognition result, wherein the trained sperm recognition model is trained based on historical microscopic images carrying category labels and position information of real frames of sperm; Based on the sperm identification results, the spatial position of each identified sperm is determined.

2. The method according to claim 1, characterized in that The sperm identification result includes the position, length, width and head shape of the sperm, the number of the microscopic images is multiple, and after obtaining the sperm identification result, the method further includes: For each sperm, determining the movement trajectory of the sperm according to the position of the sperm in the sperm identification results of the multiple microscopic images; Determining the morphological characteristic score of the sperm according to the length, width and head shape of the sperm; Determining a motility score of the sperm according to the motion trajectory of the sperm; The sperm quality score is determined based on the morphological characteristic score and the motility score.

3. The method according to claim 2, characterized in that After determining the quality score of the sperm, the method further comprises: A sperm quality assessment report is generated based on the morphological characteristic score, the motility score and the quality score of each sperm.

4. The method according to any one of claims 1 to 3, characterized in that: Before identifying the sperm in the microscopic image based on the trained sperm identification model to obtain the sperm identification result, the method further includes: Obtaining historical microscopic images of sperm observation fluid, marking the category and position information of the real frame of sperm in the historical microscopic images, and obtaining a training set; Constructing an initial sperm recognition model; Taking the training set as input, calling the initial sperm recognition model, and determining the category, position information, and confidence of a prediction box, wherein the prediction box is used to represent the area where the sperm in the historical microscopic image is located; Determine an error between the predicted frame and the true frame based on the category and position information of the true frame, and the category, position information, and confidence of the predicted frame; Based on the error, the parameters of the initial sperm recognition model are iteratively updated until a preset training end condition is reached to obtain a trained sperm recognition model.

5. The method according to claim 4, characterized in that Determining the error between the predicted frame and the true frame includes: Determine the classification loss, positioning loss and confidence loss between the predicted box and the true box; Based on the classification loss, the positioning loss and the confidence loss, an error between the predicted box and the true box is determined.

6. The method according to claim 4, characterized in that After obtaining the trained target detection model and the trained sperm recognition model, the method further includes: Construct a test set based on historical microscopic images annotated with the category and position information of the sperm ground truth box; Taking the test set as input, calling the trained sperm recognition model, identifying sperm in the historical microscopic image, and obtaining a historical sperm recognition result; Based on the historical sperm identification results, determining the sperm detection accuracy of each of the historical microscopic images in the test set; Determining the target detection accuracy of the trained sperm recognition model according to the sperm detection accuracy of each of the historical microscopic images; When the target detection accuracy is less than a preset target accuracy threshold, the structure of the trained sperm recognition model is adjusted.

7. A sperm identification system, characterized in that: The system comprises: a three-axis electric-controlled stage, a controller, an inverted microscope and a data processing module, wherein the three-axis electric-controlled stage is connected to the controller and arranged on the inverted microscope, and the inverted microscope and the controller are respectively connected to the data processing module; A controller, in response to the sperm recognition instruction of the data processing module, controls the three-axis electric-controlled stage to move the culture dish placed on the stage in the space of the inverted microscope in three directions of X, Y and Z according to a preset scanning path; The inverted microscope is used to scan the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid, and send the microscopic image to the data processing module; The data processing module is used to identify the sperm in the microscopic image using the sperm identification method as described in any one of claims 1 to 6, and determine the spatial position of each identified sperm.

8. The system according to claim 7, characterized in that The three-axis electric-controlled stage includes a connecting plate, a Z-axis motion mechanism, a Y-axis motion mechanism, an X-axis motion mechanism, a loading platform, and a Z-axis motion fine-tuning knob; The three-axis electrically controlled stage is connected to the base of the inverted microscope via the connecting plate; the loading platform, the X-axis motion mechanism and the Y-axis motion mechanism are stacked in sequence; the Z-axis motion mechanism is disposed between the Y-axis motion mechanism and the connecting plate, and is respectively connected to the Y-axis motion mechanism and the connecting plate; and the Z-axis motion fine-tuning knob is disposed on the Z-axis motion mechanism.

9. A sperm identification device, characterized in that: The device comprises: A sperm observation liquid scanning module, which is used to control the three-axis electric-controlled stage to move the culture dish placed on the stage in the space of the inverted microscope in the X, Y and Z directions according to a preset scanning path in response to the sperm recognition instruction, so that the inverted microscope scans the sperm observation liquid in the culture dish to obtain a microscopic image of the sperm observation liquid; A sperm recognition module, used to use the microscopic image as input, call a trained sperm recognition model, identify sperm in the microscopic image, and obtain a sperm recognition result, wherein the trained sperm recognition model is trained based on historical microscopic images carrying category labels and position information of real frames of sperm; The spatial position determination module is used to determine the spatial position of each identified sperm based on the sperm identification result.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

Images