Method for positioning of polar body of oocyte in intracytoplasmic sperm injection based on image recognition

Abstract

The application relates to an image recognition-based method for positioning an oocyte polar body in an intracytoplasmic sperm injection, obtaining a microscopic digital image of an oocyte; determining an inner edge of a zona pellucida, an outer edge of a cytoplasm and a peripheral gap region located between the two according to the microscopic digital image; extracting a candidate region in the peripheral gap region; respectively judging whether each candidate region is located in the peripheral gap region, is not connected with the cytoplasm region and is not connected with the zona pellucida outer side region; determining a valid candidate region which is located in the peripheral gap region, is not connected with the cytoplasm region and is not connected with the zona pellucida outer side region; determining a target polar body region according to the valid candidate region; determining a polar body center position and polar body orientation information according to the target polar body region, and outputting a polar body positioning result. The method can be implemented based on a microscopic digital image, and is favorable for improving the stability, consistency and automation application value of the polar body positioning result.

Description

Technical Field

[0001] This invention relates to the field of assisted reproduction and medical image processing technology, specifically to a method for locating oocyte polar bodies in intracytoplasmic sperm injection based on image recognition. Background Technology

[0002] During intracytoplasmic sperm injection (ICSI), the position of the oocyte polar body is closely related to the selection of subsequent micromanipulation positions. Therefore, polar body identification and localization have always been key aspects of automated micromanipulation. Existing technologies have attempted to improve the accuracy of related operations through image processing, defocus imaging, trajectory modeling, or cell nucleus manipulation based on polar body references. For example, patent application CN116925911A proposes a method for precise extraction of cell polar bodies, which uses the difference between positive and negative defocused images, binarization, and polar coordinate information to achieve polar body localization. Patent application CN112080387A proposes determining the intracellular trajectory of the microneedle approaching the cell nucleus by offline calibration of the three-dimensional distribution of the cell nucleus relative to the polar body and combining it with three-dimensional finite element modeling. These solutions demonstrate that existing technologies recognize the importance of polar bodies or polar body reference information in oocyte micromanipulation, but overall, they still focus more on the implementation of specific operational steps and have not yet formed a universal localization technology solution for ICSI scenarios that can directly and stably output the center position and orientation information of the polar body.

[0003] Specifically, the technical approach of CN116925911A is to first move the cell under a microscope to bring the polar body into focus, then change the positive and negative defocus amounts to obtain positive and negative defocus images, and then binarize the difference images; subsequently, identify the outline segments of the polar body, cytoplasm, and zona pellucida in the binarized images, and take the outline segment with the largest area located between the cytoplasm and zona pellucida as the polar body outline segment. While this approach can improve the contrast between the polar body and the background to some extent, it has several drawbacks. First, the localization process relies on additional defocus imaging and defocus selection, resulting in a long image acquisition chain that is highly dependent on focus state, defocus setting, and image quality. Second, using the "largest contour segment between the cytoplasm and the zona pellucida" as the basis for polar body identification relies heavily on area features and contour segment selection. When there are debris in the peripheral gaps, local reflections, false edges, adhesion shadows, or irregular polar body shapes, candidate region confusion can easily occur, affecting localization stability. Third, the technical focus of this approach is on serving the subsequent microneedle approach and extraction trajectory control, leaning more towards the polar body extraction operation itself. There is still room for improvement in its specificity regarding the polar body orientation output, attitude adjustment basis, and standardized center localization results directly required in ICSI operations.

[0004] For example, while CN112080387A proposes a technical approach that eliminates the need to determine the precise location of each cell nucleus during actual operation, its core still relies on offline calibration of the three-dimensional distribution of the cell nucleus relative to the polar body using "polar bodies as a reference," and further uses finite element modeling, dominant factor analysis, and trajectory fitting to bring the microneedle closer to the cell nucleus. In this scheme, polar body localization still requires first detecting the cytoplasm's contour and calculating the focus measure, then rotating the polar body to the focal plane by cell manipulation, and determining its three-dimensional position using the contour center when the polar body's contour area reaches its maximum value and the height of the focal plane. Therefore, this type of scheme is highly dependent on the initial calibration samples, the offline modeling process, and the fitting of operational parameters, and its main purpose is to solve cell nucleus manipulation problems, rather than directly addressing polar body region identification and orientation output in single-frame or conventional bright-field microscopy images during ICSI. The model's transferability and process simplicity may still be limited when dealing with different batches of oocytes, different imaging conditions, or significant local morphological variations. The existing technology still has the following shortcomings: First, the polar body search region lacks more explicit structural boundary constraints; second, the determination of the authenticity of candidate regions relies more on area, focus or offline models, and the topological connectivity is not fully utilized; third, it is difficult to balance positioning accuracy, process simplicity and direct adaptability to ICSI attitude control. Summary of the Invention

[0005] The purpose of this invention is to provide a method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition, thereby addressing some of the drawbacks and shortcomings pointed out in the background art.

[0006] The present invention addresses the aforementioned technical problems by employing the following technical solution: a method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition, comprising: acquiring a microscopic digital image of the oocyte; and determining, based on the microscopic digital image, the inner edge of the zona pellucida, the outer edge of the cytoplasm, and the peripheral space region located between the two.

[0007] Candidate regions are extracted within the peripheral gap region; it is then determined whether each candidate region is located within the peripheral gap region, whether it is not connected to the cytoplasmic region, and whether it is not connected to the outer region of the zona pellucida.

[0008] Candidate regions located within the peripheral gap region that are not connected to the cytoplasm region and are not connected to the outer region of the zona pellucida are identified as valid candidate regions; the target polar body region is determined based on the valid candidate regions; the polar body center position and polar body orientation information are determined based on the target polar body region, and the polar body positioning result is output.

[0009] Furthermore, each connected region within the peripheral gap region is projected circumferentially and associated with a first corresponding segment on the inner edge of the zona pellucida and a second corresponding segment on the outer edge of the cytoplasm, respectively; the overlapping interval between the first corresponding segment and the second corresponding segment in the circumferential direction is determined; and the connected region with the overlapping interval is determined as a candidate region.

[0010] Furthermore, when there are multiple effective candidate regions, corresponding points are selected at preset circumferential intervals within the overlapping intervals corresponding to each effective candidate region, and corresponding points on the inner edge of the zona pellucida and corresponding points on the outer edge of the cytoplasm are connected to form a radial line segment group; the number of radial line segments intersecting with each effective candidate region is counted; the effective candidate region that is ranked first and unique is determined as the target polar body region according to the descending order of the number of radial line segments; when there are ties for first place, the target polar body region is determined according to the descending order of the length of the corresponding overlapping interval.

[0011] Furthermore, along the circumferential direction of the overlapping intervals corresponding to each of the effective candidate regions, corresponding points at the same circumferential position are selected at a preset circumferential interval on the inner edge of the zona pellucida and the outer edge of the cytoplasm, and the corresponding points at the inner edge of the zona pellucida and the outer edge of the cytoplasm at the same circumferential position are connected respectively to form the radial line segment group.

[0012] Furthermore, the number of radial segments refers to the number of radial segments that intersect with the same valid candidate region; radial segments that intersect with multiple valid candidate regions are not included in the number of radial segments of each valid candidate region.

[0013] Furthermore, when determining the target polar body region, a preset circumferential direction is used as the positive direction, and the opposite direction is used as the negative direction; corresponding points are selected along the positive and negative directions respectively to form a first radial line segment group and a second radial line segment group; a first candidate result and a second candidate result are determined based on the first radial line segment group and the second radial line segment group respectively; when the first candidate result and the second candidate result are the same, they are determined as the target polar body region; when they are different, the target polar body region is determined according to the length of the overlapping interval.

[0014] Further, the midpoint of the overlapping interval corresponding to the target polar body region is determined; the radial line passing through the midpoint and connecting the inner edge of the zona pellucida and the outer edge of the cytoplasm is determined as the polar body azimuth line; the midpoint of the intersection of the polar body azimuth line and the target polar body region is determined as the polar body center position.

[0015] Further, the starting point and ending point of the overlapping interval along the circumferential direction are determined; the position where the arc length distance from the starting point and the ending point along the circumferential path of the overlapping interval is equal is determined as the midpoint of the overlapping interval.

[0016] Furthermore, when the polar body azimuth line and the target polar body region form multiple intersections, the projection range of each intersection in the circumferential direction is determined; the intersection whose circumferential projection range includes the midpoint of the overlapping interval is determined as a candidate intersection, and the polar body center position is determined based on the candidate intersection.

[0017] Furthermore, when there are multiple candidate intersection segments, the sum of the distances from the endpoints of each candidate intersection segment to the inner edge of the zona pellucida and the outer edge of the cytoplasm is calculated; the candidate intersection segments are arranged in ascending order according to the sum of the distances, and the first and only candidate intersection segment is determined as the target intersection segment; when there are ties for the first position, the target intersection segments are determined by arranging them in descending order according to their lengths; the midpoint of the target intersection segment is determined as the polar body center position.

[0018] The beneficial effects of this invention are as follows: Compared with the prior art, this invention first determines the inner edge of the zona pellucida, the outer edge of the cytoplasm, and the peripheral gap region between them based on microscopic digital images. Then, it extracts candidate regions within the peripheral gap region and filters effective candidate regions by combining conditions such as the candidate region being located within the peripheral gap region, not being connected to the cytoplasm region, and not being connected to the outer region of the zona pellucida. This can limit the polar body from both spatial position and topological connectivity perspectives, reducing the possibility of false detection caused by debris, shadows, and local pseudo-contours within the peripheral gap.

[0019] Furthermore, after determining the effective candidate region, this invention determines the target polar body region and accordingly determines the polar body center position and polar body orientation information, enabling direct output of positioning results suitable for microinjection posture adjustment and path planning. This method can be implemented based on microscopic digital images, has clear processing logic, and is beneficial for improving the stability, consistency, and automated application value of polar body positioning results. Attached Figure Description

[0020] Figure 1 This is a flowchart of the oocyte polar body localization method based on image recognition according to the present invention.

[0021] Figure 2 This is a schematic diagram of the oocyte structural boundary and candidate region in Embodiment 1 of the present invention.

[0022] Figure 3 This is a schematic diagram of the circumferential projection and radial line segment group in Embodiment 1 of the present invention.

[0023] Figure 4 This is a graph showing the main ranking and auxiliary comprehensive score evaluation in Embodiment 1 of the present invention.

[0024] Figure 5 This is a schematic diagram of the oocyte boundary and target polar body region in Embodiment 2 of the present invention.

[0025] Figure 6 This is a diagram showing the determination of the midpoint of the overlapping interval and the polar body azimuth line in Embodiment 2 of the present invention.

[0026] Figure 7 This is a schematic diagram of multi-intersection screening and target intersection determination in Embodiment 2 of the present invention.

[0027] Figure 8 This is a diagram showing the coordinate positioning result of the center of the polar body in Embodiment 2 of the present invention. Detailed Implementation

[0028] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0029] Combined with appendix Figure 1 In this embodiment, a digital microscopic image of the oocyte is first acquired. This digital microscopic image can be obtained by using a microscopic imaging device to capture the oocyte in a culture dish and then transmitted to an image processing unit. To improve subsequent recognition accuracy, the digital microscopic image can be preprocessed. Preprocessing includes at least one of denoising, grayscale normalization, contrast enhancement, and edge smoothing to reduce the impact of uneven illumination, background interference, and local blurring generated during microscopic imaging on the recognition results.

[0030] After preprocessing, the structural boundaries of the oocyte are identified, specifically the inner edge of the zona pellucida and the outer edge of the cytoplasm. The inner edge of the zona pellucida characterizes the boundary of the zona pellucida near the peripheral space, while the outer edge of the cytoplasm characterizes the boundary of the cytoplasm near the peripheral space. The inner edge of the zona pellucida and the outer edge of the cytoplasm can be obtained through at least one of the following methods: edge detection, threshold segmentation, morphological processing, contour fitting, active contour model, or a trained image segmentation model. Preferably, the segmentation results of the zona pellucida and cytoplasm regions can be obtained first, and then the inner boundary of the zona pellucida region and the outer boundary of the cytoplasm region can be extracted separately. To facilitate subsequent circumferential position calculation, the oocyte center can be determined based on the fitted center of the outer edge of the cytoplasm, the fitted ellipse center, or the area center of the cytoplasm region, and this oocyte center can be used as the reference center for subsequent circumferential projection and radial connection.

[0031] After obtaining the inner edge of the zona pellucida and the outer edge of the cytoplasm, the annular or near-annular region enclosed between them is defined as the peripheral gap region. This region is the main spatial extent in which polar bodies may appear, and therefore serves as a limiting area for subsequent candidate region extraction and screening. For irregularly shaped oocytes, the peripheral gap region can also be understood as a spatially continuous band-like region between the inner edge of the zona pellucida and the outer edge of the cytoplasm.

[0032] After determining the peripheral gap region, candidate regions are extracted within it. Candidate regions can be obtained through threshold segmentation, edge aggregation, connected component analysis, or region growing. Alternatively, they can be obtained by detecting local grayscale differences, texture differences, and morphological differences within the peripheral gap region. Preferably, the peripheral gap region can be binarized or labeled first, and then each locally connected region within that region can be obtained, with each locally connected region serving as an initial candidate region. Each extracted candidate region is then considered as a potential polar body region.

[0033] Specifically, it is determined whether each candidate region is located within the peripheral gap region, whether it is not connected to the cytoplasmic region, and whether it is not connected to the outer region of the zona pellucida. Being located within the peripheral gap region ensures that the candidate region is in a reasonable position between the inner edge of the zona pellucida and the outer edge of the cytoplasm. Being not connected to the cytoplasmic region excludes false detection regions caused by cytoplasmic edge protrusions, cytoplasmic debris, or local artifacts. Being not connected to the outer region of the zona pellucida excludes false detection regions caused by impurities, background noise, or suspended matter in the culture medium.

[0034] Disconnection from the cytoplasm and from the outer region of the zona pellucida can be determined through connectivity analysis on binary or labeled maps of the candidate region, cytoplasm, and outer zona pellucida. Connectivity analysis can use 4-connectivity or 8-connectivity methods, with 8-connectivity being preferred. A candidate region is considered disconnected from the cytoplasm if there is no connecting path between it and the cytoplasm; similarly, a candidate region is considered disconnected from the outer zona pellucida if there is no connecting path between it and the outer zona pellucida. Candidate regions that simultaneously meet all three conditions are designated as valid candidate regions.

[0035] After obtaining valid candidate regions, the target polar body region is determined based on these regions. When there is only one valid candidate region, it can be directly identified as the target polar body region. When there are multiple valid candidate regions, at least one of the following can be considered: region area, shape regularity, grayscale characteristics, positional relationship with the inner edge of the zona pellucida and the outer edge of the cytoplasm, and circumferential distribution characteristics. The region that best matches the morphological and spatial characteristics of the polar body is then selected as the target polar body region.

[0036] After identifying the target polar body region, the polar body center location and polar body orientation information are further determined based on this region. The polar body center location can be obtained by determining the geometric center, area center, or midpoint of the radial direction between the inner edge of the zona pellucida and the outer edge of the cytoplasm of the target polar body region. In a preferred embodiment, the polar body center location and polar body orientation information can also be specifically determined using the method described later, based on the midpoint of the overlapping interval and the polar body orientation line. The polar body orientation information can be determined based on the circumferential position of the target polar body region relative to the overall outline of the oocyte, or it can be characterized by the direction of the line connecting the center of the target polar body region and the center of the oocyte. Finally, the polar body center location and polar body orientation information are output as the polar body localization result for subsequent intracytoplasmic sperm injection (ICSI) operations.

[0037] After identifying the peripheral gap region, connectivity analysis is performed on the image content within this region to obtain the connected regions located within it. A connected region can be understood as a local region that is spatially continuous and maintains overall coherence in terms of grayscale features, edge features, or segmentation results. Since the polar body is usually located between the inner edge of the zona pellucida and the outer edge of the cytoplasm, it is necessary to further filter the connected regions based on their circumferential positional relationships.

[0038] Specifically, a circumferential coordinate system is established with the oocyte center as the reference center. The circumferential position can be characterized using polar angle parameters, arc length parameters, or normalized circumferential parameters. Then, each connected region is projected along the circumferential direction and associated with a first corresponding segment on the inner edge of the zona pellucida and a second corresponding segment on the outer edge of the cytoplasm. Circumferential projection refers to mapping the coverage area of ​​the connected region in the peripheral direction of the oocyte onto the inner edge of the zona pellucida and the outer edge of the cytoplasm, thereby obtaining the corresponding position range of the connected region on the two boundaries. Based on this mapping relationship, the first corresponding segment on the inner edge of the zona pellucida and the second corresponding segment on the outer edge of the cytoplasm can be determined.

[0039] In one implementation, the circumferential parameter range of each pixel within a connected region relative to the oocyte center can be statistically analyzed. This circumferential parameter range is then mapped to a set of boundary points at the same circumferential position on the inner edge of the zona pellucida and the outer edge of the cytoplasm, thereby forming a first corresponding segment and a second corresponding segment. If a connected region corresponds to multiple separate continuous intervals in the circumferential direction, each continuous interval can be processed separately. Preferably, the continuous interval with the longest length is taken as the segment corresponding to the connected region for subsequent judgment.

[0040] After obtaining the first and second corresponding segments, their positional relationship in the circumferential direction is further determined, and the overlapping interval between the first and second corresponding segments in the circumferential direction is identified. The overlapping interval characterizes the circumferential range of the connected region that has a common covering relationship between the inner edge of the zona pellucida and the outer edge of the cytoplasm; that is, the effective region range of the connected region that simultaneously corresponds to both sides of the boundary in the circumferential direction. The overlapping interval can be obtained by finding the intersection of the first and second corresponding segments under the same circumferential parameter coordinates. If the intersection is empty, it means that the connected region does not simultaneously correspond to both sides of the boundary in the circumferential direction and is not considered a candidate region; if the intersection is one or more continuous intervals, at least one of the continuous intervals can be used as the overlapping interval, preferably the continuous overlapping interval with the longest length as the overlapping interval corresponding to the connected region.

[0041] If the first and second corresponding segments of a connected region overlap, it indicates that the connected region is spatially located between the inner edge of the zona pellucida and the outer edge of the cytoplasm, consistent with the distribution characteristics of polar bodies located in peripheral gaps. Therefore, connected regions with overlapping segments can be identified as candidate regions. This processing method not only allows for preliminary identification based on the morphology of the connected region itself, but also enables more targeted limitation of candidate regions by combining the circumferential correspondence between the inner edge of the zona pellucida and the outer edge of the cytoplasm, thus providing a foundation for subsequent screening of effective candidate regions and determination of target polar body regions.

[0042] When there are multiple valid candidate regions obtained after the aforementioned screening, in order to further determine the target polar body region from the multiple valid candidate regions, a radial line segment group can be constructed based on the overlapping intervals corresponding to each valid candidate region, and the intersection relationship between the radial line segment and each valid candidate region can be used for judgment.

[0043] Specifically, along the circumferential direction of the overlapping intervals corresponding to each effective candidate region, corresponding points are selected at preset circumferential intervals on the inner edge of the zona pellucida and the outer edge of the cytoplasm. The preset circumferential interval can be a fixed angular interval, a fixed arc length interval, or a fixed number of uniform sampling intervals. Corresponding points are points at the same circumferential position; that is, a corresponding point on the inner edge of the zona pellucida corresponds to a corresponding point on the outer edge of the cytoplasm in terms of circumferential parameters. Subsequently, the corresponding points on the inner edge of the zona pellucida and the corresponding points on the outer edge of the cytoplasm at the same circumferential position are connected sequentially to form multiple radial line segments connecting the inner edge of the zona pellucida and the outer edge of the cytoplasm. These multiple radial line segments together constitute a radial line segment group corresponding to the effective candidate region. In this way, the circumferential range of each effective candidate region can be discretized into multiple detection positions, and the actual occupancy of each effective candidate region between the inner edge of the zona pellucida and the outer edge of the cytoplasm can be evaluated accordingly.

[0044] After forming radial segment groups, the number of radial segments intersecting with each valid candidate region is counted. The number of radial segments refers to the number of radial segments that intersect with the same valid candidate region. "Intersecting" can be understood as the radial segment and the valid candidate region having at least one common pixel, boundary contact point, or the segment passing through the interior of the valid candidate region within the image plane. If a radial segment intersects with multiple valid candidate regions simultaneously, this radial segment is not counted in the radial segment count of any valid candidate region to avoid shared segments interfering with the sorting results.

[0045] After statistical analysis, the valid candidate regions are sorted in descending order of the number of radial segments. If the first-ranked valid candidate region is unique, it is identified as the target polar body region. This is because valid candidate regions that intersect with more radial segments typically have more continuous and stable spatial coverage characteristics within their corresponding overlapping intervals, which better matches the morphological characteristics of polar body regions distributed in peripheral gaps. If multiple valid candidate regions are tied for first place in the number of radial segments, they are further sorted in descending order of the length of their corresponding overlapping intervals, and the valid candidate region with the larger overlapping interval length is identified as the target polar body region.

[0046] To reduce the influence of the circumferential sampling direction on the determination result, in another embodiment, when determining the target polar body region, a circumferential direction is preset as the positive direction, and the direction opposite to the preset circumferential direction is preset as the negative direction. The preset circumferential direction can be the direction of increasing circumferential parameters, and the negative direction can be the direction of decreasing circumferential parameters; alternatively, a reference ray in the image can be preset as the starting direction, and sampling can be performed in clockwise and counterclockwise directions respectively.

[0047] Then, corresponding points are selected within the overlapping intervals of each valid candidate region along both the forward and reverse directions, forming a first radial segment group and a second radial segment group, respectively. Preferably, the forward and reverse directions use the same sampling interval and the same overlapping interval endpoints as sampling boundaries, only changing the sampling order. Based on the first radial segment group, the first candidate result is determined according to the aforementioned intersection statistics and sorting rules. Based on the second radial segment group, the second candidate result is determined according to the same rules. When the first candidate result is the same as the second candidate result, it indicates that the valid candidate region has consistent priority under different circumferential sampling orders, and therefore it is determined as the target polar body region. When the first candidate result is different from the second candidate result, it indicates that different sampling directions have affected the results. In this case, the overlapping interval lengths of each candidate result can be further compared, and the valid candidate region with the larger overlapping interval length is determined as the target polar body region. By making judgments separately in both the forward and reverse directions and combining the overlapping interval length for the final decision, the random deviations caused by discrete sampling starting points and sampling orders can be effectively reduced, making the determination result of the target polar body region more stable and reliable.

[0048] After determining the target polar body region, the polar body's azimuth information and center position are further determined based on the overlapping interval corresponding to the target polar body region. Specifically, the start and end points of the overlapping interval along the circumferential direction are first determined, and the arc distance from each position to the start and end points is calculated along the circumferential path of the overlapping interval. The position where the arc distance to the start and end points are equal is determined as the midpoint of the overlapping interval. The midpoint represents the center position of the target polar body region in the circumferential range.

[0049] After obtaining the midpoint, the radial line passing through the midpoint and connecting the inner edge of the zona pellucida and the outer edge of the cytoplasm is determined as the polar body azimuth line. More specifically, boundary points on the inner edge of the zona pellucida and the outer edge of the cytoplasm, respectively, can be determined at the same circumferential position as the midpoint of the overlapping area, and these two boundary points are connected to form the polar body azimuth line. The polar body azimuth line is used to indicate the orientation of the polar body relative to the overall outline of the oocyte. Subsequently, the intersection of the polar body azimuth line and the target polar body region is determined, and the midpoint of this intersection is determined as the center position of the polar body. The midpoint of the intersection can be the geometric midpoint of the line connecting the two ends of the intersection.

[0050] In some cases, the azimuth line of a polar body may intersect with the target polar body region in multiple ways. When this occurs, to accurately determine the center position of the polar body, these intersections can be further filtered. Specifically, the projection range of each intersection in the circumferential direction is determined, and it is judged whether the circumferential projection range of each intersection includes the midpoint of the overlapping interval. Intersections whose circumferential projection range includes the midpoint of the overlapping interval are identified as candidate intersections. This filtering method is used because the midpoint of the overlapping interval reflects the center position of the target polar body region in the circumferential direction, and intersections whose circumferential projection range includes this midpoint are more likely to correspond to the actual main body of the polar body. When determining the center position of the polar body based on candidate intersections, if there is only one candidate intersection, the midpoint of that candidate intersection can be directly determined as the center position of the polar body.

[0051] When there are multiple candidate intersection segments, to further differentiate the priority of each segment, the sum of the distances from the endpoints of each segment to the inner edge of the zona pellucida and the outer edge of the cytoplasm can be calculated. The sum of these distances characterizes the degree of positional matching between the inner edge of the zona pellucida and the outer edge of the cytoplasm. Specifically, the endpoint closer to the inner edge of the zona pellucida within the same candidate intersection segment can be designated as the first endpoint, and the shortest distance from the first endpoint to the inner edge of the zona pellucida can be calculated; the endpoint closer to the outer edge of the cytoplasm can be designated as the second endpoint, and the shortest distance from the second endpoint to the outer edge of the cytoplasm can be calculated; the sum of these two shortest distances is then used as the sum of the distances for the candidate intersection segment.

[0052] After calculation, the candidate intersection segments are sorted in ascending order of total distance, and the first and only one in the sorted list is determined as the target intersection segment. If there are ties for the first position, the candidate intersection segments are further sorted in descending order of length, and the longer candidate intersection segment is determined as the target intersection segment. After determining the target intersection segment, the midpoint of the target intersection segment is determined as the polar body center position. Through the above processing method, even if the polar body azimuth line passes through multiple local areas of the target polar body region, the intersection segment that truly represents the main body of the polar body can be identified by combining the circumferential center position and the spatial characteristics of the intersection segment itself, thereby improving the accuracy and stability of the polar body center position determination. Finally, the azimuth information represented by the polar body azimuth line and the polar body center position can be output as the polar body positioning result for subsequent intracytoplasmic sperm injection (ICSI) positioning guidance.

[0053] Example 1:

[0054] In this embodiment, microscopic digital images of oocytes are first acquired. These images are obtained by a microscopic imaging device from the oocytes in the culture dish and transmitted to the image processing unit. The resolution of the acquired images is 1024×1024 pixels, and the spatial calibration coefficient is 0.32 μm per pixel. To reduce the impact of uneven illumination, background noise, and local defocusing on subsequent recognition results, the microscopic digital images are preprocessed. Specifically, median filtering, grayscale normalization, and local contrast enhancement are sequentially applied to the original images to make the grayscale transitions between the zona pellucida, cytoplasm, and peripheral space clearer and improve the stability of boundary extraction.

[0055] After preprocessing, the structural boundaries of the oocyte are identified. Specifically, edges are first extracted based on grayscale gradients, and then morphological closure and smoothing fitting are combined to obtain the inner edge of the zona pellucida and the outer edge of the cytoplasm. The inner edge of the zona pellucida is used to characterize the boundary of the zona pellucida near the peripheral space, and the outer edge of the cytoplasm is used to characterize the boundary of the cytoplasm near the peripheral space. Further, the center of the area of ​​the outer edge of the cytoplasm is used as the center of the oocyte, and a polar coordinate system is established based on this center, so that the position of each region in the circumferential direction is uniformly represented as an angular parameter. For the sample image in this embodiment, the equivalent radius of the inner edge of the zona pellucida is 58.6 μm, and the equivalent radius of the outer edge of the cytoplasm is 46.2 μm, therefore the band-like region between the two is determined as the peripheral space region. Figure 2 As shown in the figure, the overall structural boundary of the oocyte, the oocyte center, the inner edge of the zona pellucida, the outer edge of the cytoplasm, and the peripheral gap region located between the two are illustrated. The peripheral gap region is distributed in the form of a ring between the inner edge of the zona pellucida and the outer edge of the cytoplasm, which is used to define the extraction range of subsequent candidate regions.

[0056] Since the polar body is typically located between the inner edge of the zona pellucida and the outer edge of the cytoplasm, candidate regions are extracted within the peripheral gap region. Specifically, local thresholding is first performed on the peripheral gap region, followed by connected component labeling of the segmentation results to obtain several locally connected regions. Each locally connected region serves as an initial candidate region for subsequent evaluation. To prevent false detections caused by cytoplasmic edge protrusions, debris from the outer edge of the zona pellucida, and background noise, each candidate region is evaluated to determine whether it completely falls within the peripheral gap region, whether it is disconnected from the cytoplasmic region, and whether it is disconnected from the outer edge of the zona pellucida. The disconnection determination uses an 8-connectivity criterion. When there is no 8-connected path between the candidate region and the cytoplasmic region, it is considered disconnected from the cytoplasmic region; when there is no 8-connected path between the candidate region and the outer edge of the zona pellucida, it is considered disconnected from the outer edge of the zona pellucida. Candidate regions that meet all three conditions are identified as valid candidate regions. Figure 2The image shows three valid candidate regions, region A, region B, and region C, located within the periphery gap region. Each candidate region is located between the inner edge of the zona pellucida and the outer edge of the cytoplasm, and is spatially separated from the main body of the cytoplasm and the outer region of the zona pellucida.

[0057] To further improve the targeting of the screening, circumferential projection was performed on each connected region within the periphery gap. Since the overall outline of the oocyte is approximately circular or nearly circular, the distribution of connected regions within the periphery gap can be characterized by their polar angle range relative to the oocyte center; therefore, angular parameters were used. Describe the circumferential position. For any connected region... The set of pixels is mapped circumferentially to the inner edge of the transparent band to obtain the first corresponding segment. Mapping along the same circumference to the outer edge of the cytoplasm yields the second corresponding segment. .in, and They represent the first The starting and ending angles of each connected region on the inner edge of the transparent zone. and These represent the initial angle and the termination angle at the outer edge of the cytoplasm, respectively. Figure 3 The projection results of regions A, B, and C in the circumferential direction are shown, as well as the overlapping range of each candidate region between the inner edge of the zona pellucida and the outer edge of the cytoplasm, and the sampling relationship of the subsequent radial line segment groups.

[0058] After obtaining the first and second corresponding segments, the angular length of the overlapping interval is calculated first, followed by the arc length. To avoid ambiguity caused by the same terminology in different units, in this embodiment, the result in angular units is referred to as the angular length of the overlapping interval, and the result in length units is referred to as the arc length of the overlapping interval. The angular parameters are uniformly expressed in radians in the formula calculation; when the original data is expressed in degrees, it is first multiplied by... Convert to radians. angular length of overlapping intervals of connected regions Determine using the following formula:

[0059]

[0060] In the formula, Indicates the first The range of common angles of a connected region in the circumferential direction, in radians; min represents taking the smaller of the two ending angles; max represents taking the larger of the two starting angles; when the two segments do not have a circumferential intersection. Set to 0. The above relationship reflects the circumferential extent to which the connected region simultaneously maintains corresponding coverage between the inner edge of the zona pellucida and the outer edge of the cytoplasm. Therefore... The larger the value, the more it indicates that the region better matches the spatial distribution characteristics of the polar body located in the peripheral gap. Further, let the equivalent mid-diameter radius of the peripheral gap region be... Then the arc length of the overlapping interval Calculate using the following formula:

[0061]

[0062] In the formula, Indicates the first The arc length of the overlapping intervals corresponding to each connected region, in μm; This represents the average of the equivalent radius of the inner edge of the zona pellucida and the equivalent radius of the outer edge of the cytoplasm, used to convert the circumferential angle length into a spatial arc length. The derivation of this formula is based on the fact that the circumferential gap can be approximated as an arc unfolding along the median radius within a local circumferential range, so the arc length is equal to the product of the radius and the radian angle.

[0063] In this embodiment, take

[0064]

[0065] After calculation, the three valid candidate regions are denoted as regions A, B, and C. Region A corresponds to a common angle range of 31.2° to 46.8°, region B to a common angle length of 11.3°, and region C to a common angle length of 13.9°. After converting these to radians, we obtain...

[0066]

[0067]

[0068]

[0069] Further, the arc lengths of the overlapping intervals corresponding to the three valid candidate regions were obtained.

[0070]

[0071]

[0072]

[0073] The results above show that region A has the largest common coverage area between the inner edge of the zona pellucida and the outer edge of the cytoplasm, followed by region C, and region B has the smallest. Therefore, regions A and C are more likely to correspond to the true polar body region. Correspondingly, Figure 2 Among the candidate regions shown, region A is located in a reasonable position within the peripheral gap region and has a large coverage area; Figure 3 In the figure, the circumferential overlap interval and its arc length label value corresponding to area A are also greater than those of areas B and C. The spatial coverage advantage of area A can also be seen from the results shown in the figure.

[0074] When there are multiple valid candidate regions, in order to further determine the target polar body region, corresponding points are selected at a preset circumferential interval within the overlapping interval of each valid candidate region, and the inner edge points of the transparent zone and the outer edge points of the cytoplasm at the same circumferential position are connected to form a radial segment group. In this embodiment, the preset circumferential interval is 1.3°. The reason for choosing this interval is that, on the one hand, it can ensure that a sufficient number of detection segments are formed within the overlapping interval of the candidate regions, and on the other hand, it will not cause adjacent segments to highly overlap due to excessive sampling, thereby affecting statistical stability. If a radial segment has a common pixel point with a valid candidate region or passes through the interior of the region, it is considered that the radial segment intersects with the valid candidate region. If the same radial segment intersects with multiple valid candidate regions at the same time, the radial segment is not counted in the number of intersections with any valid candidate region, so as to avoid common segments interfering with the sorting results. Figure 3 The diagram schematically illustrates the radial line segment groups and corresponding sampling intervals arranged along the overlapping intervals of each candidate region. The arc lengths of the overlapping intervals of regions A, B, and C are labeled as 14.27, 10.33, and 12.71, respectively, which can be used to schematically reflect the degree of continuous occupancy of each candidate region within the peripheral gap.

[0075] After forming radial segment groups, a primary sort is performed based on the number of radial segments intersecting with each valid candidate region. When the number of radial segments is tied, a secondary sort is performed based on the arc length of the overlapping interval. Forward and reverse sampling results are used to verify the stability of the above sorting results. That is, in this embodiment, the determination of the target polar body region still primarily relies on the sorting rules of the number of radial segments and the arc length of the overlapping interval; the comprehensive score serves only as an auxiliary verification indicator and does not replace the primary sorting rules. To comprehensively characterize the radial coverage, spatial overlap, and bidirectional sampling stability of each valid candidate region, the following auxiliary scoring function is further constructed:

[0076]

[0077] In the formula, Indicates the first Auxiliary composite score for each valid candidate region; Indicates the relationship with the first The number of radial line segments that intersect with and only intersect with each valid candidate region; This represents the maximum number of radial segments in all valid candidate regions; Indicates the first The arc length of the overlapping interval of each valid candidate region; This represents the maximum value of the overlapping interval arc length among all valid candidate regions; This indicates a two-way consistency index; , , These represent the weighting coefficients of the radial coverage factor, circumferential overlap factor, and bidirectional stability factor, respectively. In this embodiment, we take... , , Since the polar body region typically exhibits a relatively continuous radial occupancy and a relatively stable circumferential correspondence within the peripheral gap, therefore and These are used to characterize continuous coverage and spatial consistency, respectively; while Used to reduce random deviations caused by discrete sampling start point and sampling order.

[0078] In this embodiment, the case where a region is first in both forward and backward sampling is considered to be ranked first, including both single first-place and joint first-place. If a valid candidate region is ranked first in both forward and backward sampling, then... If it is ranked first only in one direction, then If it is not ranked first in either direction, then Statistically, the number of effective radial line segments corresponding to regions A, B, and C are 12, 9, and 12 respectively during forward sampling, and 12, 8, and 11 respectively during reverse sampling. Therefore, region A is tied for first place with region C in forward sampling, and is the sole first place in reverse sampling. Area B was not ranked first in either direction, therefore Region C only ranked first in the forward sampling, therefore . Figure 3 This correspondingly illustrates the relationship between the circumferential projection and the radial line segment arrangement of the candidate region, while Figure 4 The number of positive line segments, the number of negative line segments, the arc length of the overlapping interval, and the auxiliary comprehensive score are then normalized and compared in a bar chart format to show the priority differences between candidate regions from the perspectives of the main ranking indicator and the auxiliary verification indicator.

[0079] Further take

[0080]

[0081]

[0082] Substituting the above data into the auxiliary scoring function, we get...

[0083]

[0084]

[0085]

[0086] The results above show that region A has the highest auxiliary comprehensive score, followed by region C, and region B has the lowest. Further analysis according to the main ranking rule reveals that in forward sampling, regions A and C are tied for first place in the number of radial segments. Comparing the arc length of the overlapping interval, region A's 14.27 μm is greater than region C's 12.71 μm, therefore region A has priority. In reverse sampling, region A again ranks first with 12 effective radial segments. This indicates that region A has the highest priority based on both the main ranking rule and the auxiliary score results; therefore, region A is identified as the target polar body region. Correspondingly, Figure 4 The four sets of normalized bar indicators reflect the relative size of the number of positive line segments, the number of negative line segments, the arc length of the overlapping interval, and the auxiliary comprehensive score. Among them, area A is at the highest or tied for the highest level in all four indicators, which further verifies the rationality of area A as the target polar body region.

[0087] After determining the target polar body region, the circumferential position corresponding to this region can be output as a priori basis for subsequent injection localization. Specifically, the circumferential angle corresponding to the midpoint of the overlapping area in region A can be used as the dominant circumferential position of the target polar body region, and output together with the identification results of the inner edge of the zona pellucida, the outer edge of the cytoplasm, and the peripheral gap region to the subsequent control module. Through the above processing method, not only is the feasibility of the target polar body region determination process and the clarity of the judgment criteria ensured, but the stability and repeatability of polar body identification results under complex microscopic backgrounds are also improved, thus providing a reliable regional localization basis for subsequent intracytoplasmic sperm injection (ICSI) operations.

[0088] Example 2:

[0089] In this embodiment, microscopic digital images of oocytes are first acquired. These images are obtained by a microscopic imaging device from the oocytes in the culture dish and transmitted to the image processing unit. The resolution of the acquired images is 1280×960 pixels, and the spatial calibration coefficient is 0.31 μm per pixel. To reduce the impact of uneven illumination, background noise, and local defocusing on the subsequent polar body localization results, the microscopic digital images are preprocessed. Specifically, median filtering, grayscale normalization, and local contrast enhancement are sequentially applied to the original images to make the grayscale transitions between the zona pellucida, cytoplasm, and peripheral gaps clearer and improve the stability of boundary extraction.

[0090] After preprocessing, the outline structure of the oocyte is identified. Specifically, the peripheral edge is first extracted based on gray-level gradient, and then the inner edge of the zona pellucida and the outer edge of the cytoplasm are obtained through morphological closure and smoothing fitting. The inner edge of the zona pellucida is used to characterize the boundary of the zona pellucida near the periphery gap, and the outer edge of the cytoplasm is used to characterize the boundary of the cytoplasm near the periphery gap. Further, the center of the cytoplasmic region is used as the center of the oocyte, and the center coordinates of the sample image are determined to be 512.4, 498.7. The equivalent radius of the inner edge of the zona pellucida is 61.5 μm, and the equivalent radius of the outer edge of the cytoplasm is 49.1 μm, so the band-like area between them can be determined as the periphery gap region. The target polar body region located within the periphery gap region can be obtained from the previous screening steps. In this embodiment, the screening process among multiple effective candidate regions is not carried out. Instead, the polar body orientation information and the polar body center position are further determined based on the known target polar body region. Figure 5 As shown in the figure, the oocyte center, the inner edge of the zona pellucida, the outer edge of the cytoplasm, the peripheral space region, and the target polar body region located within this region are illustrated. The target polar body region is located within a reasonable spatial range between the inner edge of the zona pellucida and the outer edge of the cytoplasm, which can serve as the geometric basis for determining the subsequent overlapping interval, midpoint, and polar body orientation line.

[0091] With the target polar body region already determined, the overlapping region corresponding to this region is first determined. Specifically, using the oocyte center as the polar coordinate reference point, the circumferential coverage of the target polar body region is projected onto the inner edge of the zona pellucida and the outer edge of the cytoplasm, respectively, and the circumferential range jointly covered by both is taken as the overlapping region. For the sample image in this embodiment, the starting angle of the overlapping region is 64.2° and the ending angle is 81.8°. Since the overlapping region represents the circumferential range of the target polar body region that simultaneously has a related coverage relationship between the inner edge of the zona pellucida and the outer edge of the cytoplasm, it can be used as the geometric basis for subsequently determining the polar body azimuth line. Figure 5 The diagram also uses a fan-shaped band to represent the circumferential overlap range of 64.2° to 81.8°, which is used to characterize the dominant circumferential coverage of the target polar body region within the peripheral gap.

[0092] To accurately determine the midpoint of the overlapping region, this embodiment calculates it based on the arc length parameter. Since the overlapping region lies between the inner edge of the zona pellucida and the outer edge of the cytoplasm, the equivalent median radius of the peripheral gap region is defined as... . The average of the equivalent radius of the inner edge of the zona pellucida and the equivalent radius of the outer edge of the cytoplasm is used to characterize the equivalent circumferential radius of the overlapping region within the peripheral gap. Let the arc length function at any circumferential position within the overlapping region be:

[0093]

[0094] In the formula, Indicates the starting angle from the overlapping interval Accumulate along the circumferential path to the angle The arc length distance at the location, in μm; Indicates the starting angle of the overlapping interval; For integration variables; This represents the equivalent radius function at the corresponding circumferential position. When the circumferential range of the target polar body region is short and the thickness variation of the peripheral gap is small, it can be... Approximate as a constant Then there is

[0095]

[0096] Since the midpoint of the overlapping interval satisfies that the arc lengths to the starting point and the ending point are equal, therefore let

[0097]

[0098] In the formula, This represents the angle parameter corresponding to the midpoint of the overlapping interval. Let represent the terminating angle of the overlapping interval. Substituting the previous equation and simplifying, we get...

[0099]

[0100]

[0101] The above derivation shows that the midpoint of the overlapping interval is not a simple empirical position, but is determined by the geometric constraint that the arc lengths on both sides of the circumferential path are equal. Therefore, it can more stably characterize the dominant orientation of the target polar body region.

[0102] In this embodiment, take

[0103]

[0104] Substituting the real data, we get

[0105]

[0106]

[0107]

[0108] Further, the total angular span of the overlapping interval was obtained.

[0109]

[0110]

[0111] The corresponding total arc length is

[0112]

[0113] Therefore, the half-arc length corresponding to the midpoint is

[0114]

[0115] Therefore, the midpoint of the overlapping interval, 73.0°, corresponds to the point where the circumferential path advances 8.495 μm from the initial angle of 64.2°. After obtaining the midpoint of the overlapping interval, the radial line passing through this midpoint and connecting the inner edge of the zona pellucida and the outer edge of the cytoplasm is determined as the polar body azimuth line. Specifically, at the position with a circumferential parameter of 73.0°, the corresponding boundary points on the inner edge of the zona pellucida and the corresponding boundary points on the outer edge of the cytoplasm are extracted, and these two are connected to form the polar body azimuth line. Since this radial line passes through the center of the target polar body region in the circumferential direction and simultaneously connects the two boundaries of the peripheral gap, it can characterize the orientation of the polar body relative to the overall outline of the oocyte. Figure 6 As shown in the figure, the polar body azimuth line is shown by a thicker black solid line, and the starting angle boundary line and the ending angle boundary line of the overlapping area are shown by gray dashed lines. The starting angle of 64.2°, the midpoint angle of 73.0° and the ending angle of 81.8° are given by the outer leader line. Figure 6 The text also indicates the total arc length of the overlapping section as approximately 16.99 μm and the half-length arc length as approximately 8.495 μm, used to illustrate the process of determining the midpoint angle.

[0116] Subsequently, the intersection between the polar body azimuth line and the target polar body region is calculated. If the polar body azimuth line forms only one continuous intersection with the target polar body region, the midpoint of this intersection can be directly determined as the center position of the polar body. In this embodiment, due to local undulations at the boundary of the target polar body region, the polar body azimuth line forms three separate intersections with it, with lengths of 4.8μm, 7.1μm, and 5.6μm, respectively. To filter out the intersection that truly corresponds to the polar body from multiple intersections, the projection range of each intersection in the circumferential direction is first determined. To avoid degenerating a single radial line into a single angle point, in this embodiment, the circumferential projection range of the intersection is calculated based on the neighboring pixel set of the intersection. Specifically, a local neighboring region of a preset width is extracted centered on each intersection, and then the coverage of this local neighboring region in the circumferential direction is projected to the inner edge of the transparent zone and the outer edge of the cytoplasm, thereby determining the circumferential projection range corresponding to each intersection. Then, it is determined whether the circumferential projection range corresponding to each intersection includes the midpoint 73.0° of the overlapping interval. After screening, the circumferential projection range of all three intersection segments covers the area around 73.0°, and therefore they were all retained as candidate intersection segments. Figure 7 As shown in the figure, the relative positions of the first, second, and third intersection segments within the peripheral gap area are illustrated along the polar body azimuth line. The length, total distance, and auxiliary comprehensive evaluation value of each intersection segment are marked to demonstrate the subsequent selection process for target intersection segments.

[0117] If multiple candidate intersection segments remain, the sum of the distances from the endpoints of each candidate intersection segment to the outer edge of the cytoplasm and the inner edge of the zona pellucida is further calculated. Let the shortest distance from the endpoint closest to the outer edge of the cytoplasm in the j-th candidate intersection segment to the outer edge of the cytoplasm be... The shortest distance from the endpoint near the inner edge of the zona pellucida to the inner edge of the zona pellucida is Then the total distance of the candidate intersection segment Defined as

[0118]

[0119] In the formula, Indicates the first The sum of the boundary matching distances of the candidate intersection segments, in μm; Used to characterize the degree of adhesion between the inner endpoint of the junction and the outer edge of the cytoplasm; It is used to characterize the degree of fit between the outer end point of the intersection and the inner edge of the transparent band. The smaller the value, the more completely the intersection segment is positioned within the reasonable area between the outer edge of the cytoplasm and the inner edge of the zona pellucida, and therefore the more likely it is to correspond to the true polar body. Furthermore, to assist in verifying the priority of multiple candidate intersection segments, a normalized evaluation index can be introduced.

[0120]

[0121] In the formula, Indicates the first The auxiliary comprehensive evaluation value of each candidate intersection segment; This represents the minimum sum of distances among all candidate intersection segments; Indicates the first The length of each candidate intersection segment; Indicates the maximum length among all candidate intersection segments; and Let represent the weight coefficients of the distance matching factor and the length factor, respectively, and satisfy . In this embodiment, it is preferred to take... , . It serves only as an auxiliary verification indicator and does not replace the main sorting rule of ascending order of total distance and descending order of intersection length.

[0122] Since all candidate intersection segments are located on the same polar body azimuth line, and the two ends of the corresponding intersection segments extend towards the outer edge of the cytoplasm and the inner edge of the zona pellucida, respectively, the sum of the distances from the two ends of the candidate intersection segment to the outer edge of the cytoplasm and the inner edge of the zona pellucida can be expressed as the difference between the peripheral gap thickness and the length of the candidate intersection segment. In this embodiment, the peripheral gap thickness is:

[0123]

[0124] For the three candidate intersection segments, we have:

[0125]

[0126]

[0127]

[0128] Therefore, it is evident that the second intersection segment has the smallest total distance; thus, according to the main sorting rule, the second intersection segment should be prioritized as the target intersection segment. Further...

[0129]

[0130]

[0131] Substituting the auxiliary comprehensive evaluation indicators, we get:

[0132]

[0133]

[0134]

[0135] The results above show that the second intersection segment has the highest auxiliary comprehensive evaluation value, consistent with the main ranking result obtained by ascending the total distance, thus further verifying the rationality of the second intersection segment as the target intersection segment. If two or more candidate intersection segments have the same total distance, they are then ranked in descending order of intersection segment length, with the longer intersection segment being selected as the target intersection segment, thereby avoiding the priority selection of shorter, fragmented intersection segments. Correspondingly, Figure 7 The second intersection segment not only has the smallest total distance of 5.3 μm among the three, but also has a length of 7.1 μm and an auxiliary comprehensive evaluation value of 1.000. Therefore, it was finally determined as the target intersection segment in the illustrated results.

[0136] After the target intersection segment is determined, its midpoint is designated as the center of the polar body. To ensure consistency between the coordinate results and the aforementioned center coordinates, azimuth, radius range, and spatial calibration parameters, in this embodiment, the midpoint of the target intersection segment is located 55.0 μm from the center of the oocyte along the polar body azimuth line. The corresponding pixel radius is...

[0137]

[0138] Therefore, the coordinates of the polar body center are:

[0139]

[0140]

[0141] To verify that the midpoint is located inside the peripheral gap region, the half-length of the distance from the midpoint between the two endpoints along the polar body azimuth line is calculated based on the target intersection length of 7.1 μm.

[0142]

[0143] Therefore, the radius of the inner endpoint of the target intersection is obtained as follows:

[0144]

[0145] The radius of the outer endpoint of the target intersection is

[0146]

[0147] because

[0148]

[0149] Therefore, the target segment is located entirely between the outer edge of the cytoplasm and the inner edge of the zona pellucida, consistent with the spatial range of the aforementioned peripheral gap region. Finally, the polar body orientation information corresponding to 73.0° and the polar body center positions corresponding to 564.3 and 668.4° are output as polar body localization results for guidance in subsequent intracytoplasmic sperm injection (ICSI) procedures. Figure 8 As shown in the figure, the location of the oocyte center, the inner edge of the zona pellucida, the outer edge of the cytoplasm, the polar body azimuth line, the target intersection, and the polar body center are illustrated using local image planar coordinates. The coordinates of the polar body center are labeled as 564.3, 668.4. The target intersection is located on the polar body azimuth line and falls completely within the peripheral gap region, with a polar body azimuth angle of 73.0°, thus corresponding to the final localization result in this embodiment.

[0150] The above processing method first determines the midpoint based on the principle of equal arc length in overlapping intervals, then constructs the polar body azimuth line using this midpoint. Subsequently, candidate intersection segments are screened by combining the circumferential projection of the intersection segment's neighborhood, and the target intersection segment is optimized by using the sum of the distances from the intersection segment's endpoint to the outer edge of the cytoplasm and the inner edge of the zona pellucida. This method can stably determine the polar body's azimuth information and center position even when the polar body azimuth line passes through multiple local intersection segments in the target polar body region. Therefore, the process of determining the polar body's center position not only has a clear geometric basis and feasibility, but also improves the accuracy and repeatability of polar body localization results under complex microscopic image conditions.

Claims

1. A method for locating the polar body of an oocyte in intracytoplasmic sperm injection (ICSI) based on image recognition, characterized in that... include: Obtaining digital microscopic images of oocytes; The inner edge of the zona pellucida, the outer edge of the cytoplasm, and the peripheral space region located between the two are determined based on the aforementioned digital microscopic images; Candidate regions are extracted within the peripheral gap region; it is then determined whether each candidate region is located within the peripheral gap region, whether it is not connected to the cytoplasmic region, and whether it is not connected to the outer region of the zona pellucida. Candidate regions located within the peripheral gap region that are not connected to the cytoplasm region and are not connected to the outer region of the zona pellucida are identified as valid candidate regions; the target polar body region is determined based on the valid candidate regions; the polar body center position and polar body orientation information are determined based on the target polar body region, and the polar body positioning result is output.

2. The method for locating oocyte polar bodies in intracytoplasmic sperm injection based on image recognition according to claim 1, characterized in that... Each connected region within the periphery gap region is projected circumferentially and associated with the first corresponding segment on the inner edge of the zona pellucida and the second corresponding segment on the outer edge of the cytoplasm, respectively. Determine the overlapping intervals of the first corresponding segment and the second corresponding segment in the circumferential direction; determine the connected regions with the overlapping intervals as candidate regions.

3. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 1, characterized in that... When there are multiple valid candidate regions, corresponding points are selected at preset circumferential intervals within the overlapping intervals corresponding to each valid candidate region, and corresponding points on the inner edge of the zona pellucida and the outer edge of the cytoplasm are connected to form a radial line segment group; the number of radial line segments intersecting with each valid candidate region is counted; the radial line segments are arranged in descending order according to the number of radial line segments, and the first and only valid candidate region is determined as the target polar body region; when there are ties for the first position, the target polar body region is determined by arranging them in descending order according to the length of the corresponding overlapping interval.

4. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 3, characterized in that... Along the circumferential direction of the overlapping intervals corresponding to each of the effective candidate regions, corresponding points at the same circumferential position are selected at a preset circumferential interval on the inner edge of the zona pellucida and the outer edge of the cytoplasm, and the corresponding points at the inner edge of the zona pellucida and the outer edge of the cytoplasm at the same circumferential position are connected respectively to form the radial line segment group.

5. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 3, characterized in that... The number of radial segments refers to the number of radial segments that intersect with the same valid candidate region; radial segments that intersect with multiple valid candidate regions are not included in the number of radial segments of each valid candidate region.

6. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 3, characterized in that... When determining the target polar body region, a preset circumferential direction is used as the positive direction, and the opposite direction is used as the negative direction; corresponding points are selected along the positive and negative directions to form a first radial line segment group and a second radial line segment group; a first candidate result and a second candidate result are determined based on the first radial line segment group and the second radial line segment group, respectively; when the first candidate result and the second candidate result are the same, they are determined as the target polar body region; when they are different, the target polar body region is determined according to the length of the overlapping interval.

7. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 1, characterized in that... Determine the midpoint of the overlapping interval corresponding to the target polar body region; determine the radial line passing through the midpoint and connecting the inner edge of the zona pellucida and the outer edge of the cytoplasm as the polar body azimuth line; determine the midpoint of the intersection of the polar body azimuth line and the target polar body region as the polar body center position.

8. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 7, characterized in that... Determine the starting and ending points of the overlapping interval along the circumferential direction; determine the position where the arc length distance from the starting and ending points along the circumferential path of the overlapping interval is equal to that of the starting and ending points as the midpoint of the overlapping interval.

9. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 7, characterized in that... When the polar body azimuth line forms multiple intersections with the target polar body region, the projection range of each intersection in the circumferential direction is determined; the intersection whose circumferential projection range includes the midpoint of the overlapping interval is determined as a candidate intersection, and the polar body center position is determined based on the candidate intersection.

10. The method for locating the polar body of an oocyte in intracytoplasmic sperm injection based on image recognition according to claim 9, characterized in that... When there are multiple candidate intersection segments, calculate the sum of the distances from the endpoints of each candidate intersection segment to the inner edge of the zona pellucida and the outer edge of the cytoplasm; sort them in ascending order according to the sum of the distances, and determine the first and only candidate intersection segment as the target intersection segment; when there are ties for the first position, sort them in descending order according to the length of the candidate intersection segments to determine the target intersection segment; determine the midpoint of the target intersection segment as the polar body center position.

Images