Gynecological endocrine index analysis method and system based on image recognition

Abstract

The application relates to the field of gynecological endocrinology and discloses a gynecological endocrine index analysis method and system based on image recognition, which comprises the following steps: processing an ovary ultrasound image, automatically recognizing an ovary region and cutting, dividing the cut image into a plurality of grid regions; analyzing and extracting the texture direction of each grid region, evaluating the texture consistency, judging whether the follicle shape is complete or not; if the texture consistency is lower than a threshold value, the region is removed; in the screened follicle candidate region, a dynamic region growth method is used for adaptive segmentation to complete follicle labeling; a multi-scale roundness matching function is used to evaluate the roundness characteristics of the follicle to determine the regularity of the follicle shape; the regular follicle is subjected to edge sharpness analysis to calculate the volume and area of the follicle; the regular follicle is classified and counted; based on the spatial distribution of the follicle, local density analysis is carried out to generate an endocrine atlas feature vector of the ovary. The application has the advantages of improving the diagnosis and treatment efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of gynecological endocrinology, in particular to a gynecological endocrine index analysis method and system based on image recognition. BACKGROUND

[0002] In the clinical diagnosis of gynecological endocrine diseases, especially polycystic ovary syndrome, polycystic-like changes in the ovary are one of the important imaging indicators. Currently, the number of follicles is usually counted manually by two-dimensional ultrasound images to determine whether it meets the diagnostic criteria of polycystic ovary morphology. However, the process is highly dependent on the subjective judgment of the doctor, especially when the number of follicles is close to the diagnostic threshold, the accuracy of counting has a decisive influence on the final diagnosis. The current manual counting method is prone to misjudgment, omission or recounting in the case of low image resolution, unclear follicle boundary or dense follicle distribution, which seriously affects the accuracy and consistency of the diagnosis. In addition, the image acquisition conditions vary greatly in the clinic, resulting in significant differences in image features between different patients, further increasing the difficulty of manual recognition. Therefore, it is necessary to design a gynecological endocrine index analysis method and system based on image recognition to improve the efficiency of diagnosis and treatment. SUMMARY

[0003] In view of the deficiencies of the prior art, the present application provides a gynecological endocrine index analysis method and system based on image recognition, which has the advantage of improving the efficiency of diagnosis and treatment and solving the problems in the background art.

[0004] To achieve the above-mentioned purpose of improving the efficiency of diagnosis and treatment, the present application provides the following technical scheme: a gynecological endocrine index analysis method based on image recognition, comprising the following steps:

[0005] By automatic recognition of the ovary region in the ovary ultrasound image, the boundary of the ovary ultrasound image is cropped to obtain a plurality of grid regions;

[0006] By analyzing the consistency of the texture direction in the grid region, it is determined whether the follicle shape is complete, the follicle with incomplete shape is excluded, and the follicle candidate region is determined;

[0007] After determining the follicle candidate region, an adaptive segmentation method based on dynamic region growth is used to label the follicle, and the speed and threshold of region growth are adaptively adjusted by analyzing the local image features;

[0008] A multi-scale roundness matching function is used to evaluate the roundness characteristics of the follicle in the follicle candidate region to determine the regularity of the follicle shape;

[0009] The regular follicle is subjected to edge sharpness analysis and combined with the follicle shape and spatial distribution to classify and count the follicles;

[0010] According to the spatial distribution of follicles and combined with local density analysis, the ovarian endocrine map feature vector is generated.

[0011] Preferably, the process of determining whether the follicle morphology is complete is:

[0012] By calculating the main direction angles of all pixels in the grid area, and calculating the mean and mean square error;

[0013] If the mean square error is less than the threshold, it means that the texture direction of the grid area is consistent, indicating that the follicle morphology is complete;

[0014] If the mean square error is greater than or equal to the threshold, it means that the texture direction of the grid area is inconsistent, indicating that the follicle morphology is incomplete.

[0015] Preferably, the process of marking follicles is:

[0016] In the selected follicle candidate area, pixels with stable brightness, edge intensity and texture direction are selected as initial seed points;

[0017] Set the initial grayscale similarity threshold and define the region growth criteria, including the grayscale difference between adjacent pixels and the current region, texture similarity, and edge continuity;

[0018] In the process of region growing, the neighborhood pixels around the seed point are traversed. If the neighborhood pixels meet the similarity criteria, they are incorporated into the current region and used as new seed points.

[0019] Dynamically adjust the similarity threshold according to local image features to adapt to grayscale changes and texture fluctuations in the image;

[0020] When all neighboring pixels do not meet the growth conditions, the region growth terminates;

[0021] Finally, a closed outline of the candidate follicle region is formed and the region is marked as the follicle region.

[0022] Preferably, the process of adaptively adjusting the speed and threshold of region growth according to local image features is as follows:

[0023] Apply edge detection algorithm to the candidate follicle area to extract edge gradient information and obtain edge intensity map;

[0024] In the edge intensity map, pixels with significant edge response and high texture direction consistency are selected as initial seed points;

[0025] For each seed point, calculate the grayscale difference, texture direction difference and edge gradient change between its surrounding pixels and the marked pixels in the current area;

[0026] According to the degree of difference, determine whether the adjacent pixels meet the region expansion conditions. If so, they are merged into the current segmentation region and added to the seed point queue;

[0027] During the growth process, the similarity threshold and growth step size are dynamically adjusted according to the average grayscale, texture direction consistency and edge density of the pixels in the current area;

[0028] The growing process continues to iterate until all the pixels in the neighborhood of the seed point do not meet the expansion condition, thus forming the initial segmentation area of ​​the follicle.

[0029] Preferably, the process of determining whether adjacent pixels meet the region expansion condition is as follows:

[0030] During the region growing process, if the grayscale difference between the neighboring pixel and the current region is less than the preset grayscale similarity threshold, the pixel is considered to be sufficiently similar to the current region and meets the region expansion conditions. It is added to the current segmented region and continues to participate in region growing as a new seed point.

[0031] If the grayscale difference is greater than or equal to the similarity threshold, the pixel is considered to be too different from the current area and is excluded from the expansion area and not included in the segmentation area.

[0032] Preferably, the process of determining the regularity of follicle morphology is:

[0033] Extract boundary information from the follicle segmentation region obtained by the region growing process, and use edge detection algorithm to obtain the contour edge of the follicle region;

[0034] Fit and sample boundary contours at multiple scales to construct a multi-scale circularity matching function;

[0035] The roundness values ​​were calculated at different scales, and the scale with the smallest fitting error was selected as the best evaluation scale;

[0036] The roundness value is compared with the preset threshold. If the roundness is greater than the threshold, the shape is considered regular, otherwise it is irregular.

[0037] A gynecological endocrine index analysis system based on image recognition, comprising:

[0038] Ovarian region recognition module: responsible for processing the original ovarian ultrasound image, automatically identifying the ovarian region and cropping it;

[0039] Follicle morphology assessment module: Analyzes and extracts the texture direction of each mesh area and evaluates the texture consistency of the area;

[0040] Follicle segmentation module: responsible for adaptive segmentation within the selected follicle candidate area;

[0041] Ovarian follicle shape determination module: the roundness characteristics of the ovarian follicle are evaluated by applying a multi-scale roundness matching function;

[0042] Secretion profile generation module: based on the spatial distribution of the ovarian follicle, local density analysis is performed to evaluate the distribution pattern of the ovarian follicle.

[0043] Compared with the prior art, the gynecological endocrine index analysis method and system based on image recognition provided by the present application have the following beneficial effects:

[0044] 1. The gynecological endocrine index analysis method based on image recognition can efficiently and accurately identify and analyze the shape of the ovarian follicle in the ovarian ultrasound image, which helps to improve the accuracy of ovarian disease diagnosis, especially in the early diagnosis of endocrine abnormality related diseases such as polycystic ovary syndrome, and has important clinical value.

[0045] 2. The grid region texture direction consistency analysis combined with the adaptive segmentation method of dynamic region growth effectively improves the accuracy of the determination of the shape of the ovarian follicle, and can automatically screen out the ovarian follicle with complete shape, reducing the possibility of manual intervention and misjudgment, and improving the automation level.

[0046] 3. The application of the multi-scale roundness matching function can more accurately evaluate the regularity of the shape of the ovarian follicle, avoiding the problem of poor sensitivity of traditional methods to irregular ovarian follicle shape, and further improving the accuracy of ovarian follicle counting and classification.

[0047] 4. The classification and counting by combining edge sharpness analysis with ovarian follicle spatial distribution can provide detailed information about the state and spatial distribution characteristics of the ovarian follicle, thereby providing comprehensive ovarian function evaluation information for clinicians, and further helping to develop more accurate treatment plans.

[0048] 5. The generated ovarian endocrine profile feature vector can provide quantitative analysis of the ovarian endocrine level, which helps to track the endocrine changes of patients and provides data support for clinical treatment effect evaluation and decision-making.

[0049] 6. Efficient and automated ovarian follicle analysis is realized, human factors are reduced, and the method can be widely applied in different patient groups, has good universality and stability, and provides strong support for large-scale clinical screening and telemedicine. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The figure is a schematic diagram of the method of the present application;

[0051] Figure 2 The figure is a schematic diagram of the structure of the present application. DETAILED DESCRIPTION

[0052] With reference to the drawings and embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0053] Embodiment 1

[0054] Please refer to Figure 1 As shown in the drawings, the gynecological endocrine index analysis method based on image recognition described in the embodiments of the present application comprises the following steps:

[0055] S1: The boundary of the ovary ultrasound image is cropped through automatic recognition of the ovary region in the ovary ultrasound image, and a plurality of grid regions are obtained.

[0056] The ovary ultrasound image is obtained by a conventional two-dimensional gray-scale ultrasound imaging device, the image is saved in a DICOM format or a standard image format, and is imported into the system for analysis through an image acquisition module. In order to realize the localization and fine processing of image analysis, the ovary region in the image needs to be automatically recognized first.

[0057] The automatic recognition of the ovary region adopts a medical image segmentation algorithm based on deep learning, such as U-Net, DeepLab or an improved network structure thereof. The network takes the ultrasound image as input, extracts multi-scale image features through an encoder, gradually restores the spatial resolution through a decoder, and finally outputs a pixel-level segmentation mask of the ovary region. A traditional method based on image gray threshold, edge detection and morphological operation is selected to realize the extraction of the ovary candidate region. The recognition result is used to guide the subsequent cropping and feature extraction operations.

[0058] After the ovary region is recognized, the boundary is cropped, the image region within the minimum circumscribed rectangle is extracted, and the image region is taken as the target region for subsequent analysis. In order to enhance the description ability of the local features of the image, the cropped image region is divided into a plurality of grid regions. The grid division adopts an equal interval division strategy, that is, according to the set number of rows and columns or the grid size, the ovary region is divided into sub-regions with consistent size in proportion. Each grid region is taken as an independent analysis unit, which is used for the calculation and evaluation of texture direction, consistency, edge feature and other indicators, thereby providing a basis for the screening of the follicle candidate region.

[0059] S2: The texture direction of each grid region is extracted, the consistency of the texture direction of the grid region is evaluated, and whether the follicle shape is complete is judged; if the consistency of the texture direction of the grid region is lower than a preset threshold, the follicle shape is considered to be incomplete, and the grid region is excluded from the candidate region.

[0060] The process of judging whether the follicle morphology is complete in S2 is:

[0061] The process of judging whether the follicle morphology is complete is:

[0062] The principal direction angle of all pixel points in a grid area is calculated, and the mean value and mean square error are counted;

[0063] The principal direction angle of all pixel points in a grid area is counted, and the mean value and mean square error are calculated and A texture consistency threshold T is set to distinguish whether the texture direction is consistent, and the judgment is made according to the size of the mean square error:

[0064] If <T, it indicates that the texture direction of the grid area is consistent, indicating that the follicle morphology is complete;

[0065] If ≥T, it indicates that the texture direction of the grid area is not consistent, indicating that the follicle morphology is not complete;

[0066] It should be noted that the role of judging whether the follicle morphology is complete is:

[0067] Role one: used to remove abnormal areas caused by poor image quality, tissue overlap, artifact interference or edge occlusion, etc., so as to ensure that the subsequent follicle segmentation and counting analysis process is based on effective follicle areas with complete morphological features and clear boundaries, thereby improving the segmentation accuracy and the reliability of endocrine index extraction;

[0068] Role two: follicles with complete morphology are easier to be geometrically fitted, which is convenient for quantifying their development state and maturity, and has key significance for evaluating ovarian function and assisting in the diagnosis of gynecological diseases such as polycystic ovary syndrome.

[0069] ​The technical solution of this embodiment is to determine whether the follicle morphology is complete by extracting the texture direction of the grid area and evaluating its consistency. The specific process is as follows: first, the main direction angles of all pixels in the grid area are calculated, and their mean and mean square deviation are counted; then, a texture consistency threshold is set, and the integrity of the follicle morphology is determined based on the size of the mean square deviation. When the mean square deviation is lower than the threshold, it indicates that the texture direction is consistent and the follicle morphology is complete; when the mean square deviation is higher than the threshold, it indicates that the texture direction is inconsistent, the follicle morphology may be incomplete, and the relevant area will be eliminated. This method can effectively eliminate abnormal areas caused by factors such as poor image quality, tissue overlap, or artifacts, thereby ensuring the accuracy of subsequent follicle segmentation and counting analysis, and improving the reliability of endocrine index extraction. At the same time, complete follicle morphology is easier to perform geometric fitting, which helps to quantify the developmental status and maturity of the follicle, and is of great significance for ovarian function assessment and auxiliary diagnosis of gynecological diseases such as polycystic ovary syndrome.

[0070] Example 2

[0071] like Figure 1 As shown, the gynecological endocrine index analysis method based on image recognition also includes the following steps:

[0072] S3: After determining the candidate follicle regions, an adaptive segmentation method based on dynamic region growing is used to mark the follicles. The initial segmentation is completed by analyzing the edge and texture features of the follicles, and the region growing speed and threshold are adaptively adjusted according to the local image features.

[0073] In S3, an adaptive segmentation method based on dynamic region growing is used to mark follicles. The process is as follows:

[0074] Within the selected follicle candidate region, pixels with stable brightness, edge strength, and texture direction are selected as initial seed points. Pixels with relatively high brightness and clarity are selected as candidate seed points, as these points usually represent the edge or core area of ​​the follicle and have better segmentation effects. The edge of the follicle region is detected using an edge detection algorithm, and pixels that are clearly visible in the edge strength map are selected. Pixels with stronger edge strength tend to have higher resolution and accuracy.

[0075] Set the initial grayscale similarity threshold and define the region growth criteria, including the grayscale difference between adjacent pixels and the current region, texture similarity, and edge continuity; set a threshold to control the grayscale difference between the seed point and the surrounding neighboring pixels. This threshold indicates whether there is similarity between the seed point and the adjacent pixels, ensuring that only those pixels similar to the current region are included in the expanded region; define the region growth criteria, including:

[0076] Gray level difference: the difference of gray level between the adjacent pixel and the marked region of the current region. If the difference is less than the set threshold, the pixel belongs to the current region.

[0077] Texture similarity: by calculating the texture direction consistency between the adjacent pixel and the pixel in the current region, if the texture direction is similar, the pixel meets the expansion condition.

[0078] Edge continuity: check whether the edge between the adjacent pixel and the marked region is continuous. The pixel with strong edge continuity is usually more suitable to join the current region to form a more complete follicle contour.

[0079] In the region growing process, the neighborhood pixels around the seed point are traversed, and if the field pixel meets the similarity criterion, it is incorporated into the current region and serves as a new seed point. In the region growing process, the neighborhood pixels around each seed point are traversed to check whether they meet the set similarity criterion. The similarity calculation of the neighborhood pixel includes:

[0080] Gray level difference: calculate the gray level difference between the seed point and the neighborhood pixel, and if the gray level difference is less than the preset threshold, the pixel is considered to belong to the current region.

[0081] Texture direction difference: compare the texture direction of the seed point and the neighborhood pixel, and if the texture direction difference is small, the pixel meets the expansion condition.

[0082] Edge gradient change: calculate the edge gradient change of the neighborhood pixel, and if the gradient change of the neighborhood pixel is small, it means that the edge continuity with the current region is good, which meets the expansion condition.

[0083] Add the pixel meeting the condition to the current region: if the neighborhood pixel meets the similarity criterion, it is incorporated into the current segmentation region, and the pixel is used as a new seed point to continue participating in the region growing.

[0084] Adjust the similarity threshold dynamically according to the local image features to adapt to the gray level changes and texture fluctuations in the image; in the region growing process, the gray level and texture in the image may change, so the similarity threshold needs to be adjusted dynamically. The method is:

[0085] Local gray level change: when the gray level changes greatly in the image region, the gray level similarity threshold is appropriately increased to avoid misjudgment.

[0086] Texture fluctuation: adjust the texture similarity threshold according to the texture fluctuation of the local region to ensure that the region growing process does not mistakenly expand to unrelated regions.

[0087] Edge density change: in the edge region, there may be a large change in edge density, so the threshold is appropriately adjusted to adapt to this change and ensure the continuity of the edge.

[0088] When all the neighborhood pixels do not meet the growth condition, the region growth terminates; during the expansion process, the region growth stops when all the neighborhood pixels do not meet the similarity criterion. This means that the current region has reached its maximum expansion range and cannot continue to merge more qualified pixels.

[0089] Finally, a closed candidate follicle region profile is formed, and the region is marked as a follicle region; after the expansion process of region growth, a closed follicle region is finally formed. The boundary of this region is completely labeled in the image and conforms to the actual morphology of the follicle, and the region is marked as a follicle and provides a clear follicle profile for subsequent analysis.

[0090] The initial segmentation is completed by analyzing the edge features and texture features of the follicle in S3, and the speed and threshold process of region growth are adaptively adjusted according to the local image features as follows:

[0091] An edge detection algorithm is applied to the follicle candidate region to extract edge gradient information, and an edge intensity map is obtained;

[0092] In the edge intensity map, pixels with significant edge response and high texture direction consistency are selected as initial seed points;

[0093] For each seed point, the gray level difference, texture direction difference and edge gradient change between the surrounding pixels and the marked pixels in the current region are calculated;

[0094] According to the difference degree, it is judged whether the adjacent pixels meet the region expansion condition, if yes, they are merged into the current segmentation region, and the seed point queue is added;

[0095] During the growth process, the similarity threshold and the growth step are dynamically adjusted according to the average gray level, texture direction consistency and edge density of the pixels in the current region;

[0096] The growth process continues to iterate until all the seed point neighborhood pixels do not meet the expansion condition, thereby forming the initial segmentation region of the follicle.

[0097] An edge detection algorithm is applied to the follicle candidate region to extract edge gradient information, and an edge intensity map is obtained; at the same time, Gabor filter is used to extract the main texture direction, texture energy and consistency index of the region, forming a complete texture feature map;

[0098] Pixels with significant edge intensity and stable texture direction are selected as seed points. The growth criterion is defined as: the average edge intensity difference, gray value difference and texture direction difference between the pixel to be expanded and the current growth region should be lower than the corresponding threshold, so as to be included in the current region;

[0099] For each seed point, the gray difference between its surrounding pixels and the marked region pixels in the current region is calculated, and the gray difference calculation formula is:

[0100] ;

[0101] In the formula, is the gray value of the pixel point (x, y) to be judged, is the gray value of the seed point, is the gray difference between the pixel and the seed point;

[0102] In some preferred embodiments, the pixel gray values in the seed point neighborhood and the current region are compared:

[0103] In the region growing process, if the gray difference between the neighborhood pixels and the current region is less than the preset gray similarity threshold, it is considered that the pixel has sufficient similarity with the current region, meets the region expansion condition, is added to the current segmentation region, and continues to participate in the region growing as a new seed point.

[0104] If the gray difference is greater than or equal to the similarity threshold, it is considered that the pixel has too large difference with the current region, and is excluded from the expanded region and not included in the segmentation region.

[0105] S4: Apply a multi-scale roundness matching function to the labeled follicle candidate region to evaluate the matching degree of the roundness characteristics of the candidate region with the standard circle, and determine the regularity of the follicle shape.

[0106] The process of evaluating the matching degree of the roundness characteristics of the candidate region with the standard circle in S4 to determine the regularity of the follicle shape is:

[0107] Extract the boundary information in the follicle segmentation region obtained through the region growing process, use an edge detection algorithm to extract the contour edge of the follicle region, and perform smoothing processing thereon;

[0108] Analyze the boundary of the follicle through multiple scales, define a multi-scale roundness matching function, and compare the function based on the geometric characteristics of the standard circle, and the formula is:

[0109] ;

[0110] In the formula, represents the roundness under the scale , is the boundary length of the candidate region, is the radius for calculating the roundness; the function measures the similarity between the follicle region and the standard circle, and the closer the roundness is to 1, the closer the shape is to the ideal circle;

[0111] At different scales, calculate the roundness matching degree, and select the most suitable scale for roundness evaluation. By adjusting the scale, it is ensured that the small changes in follicle morphology and the circular characteristics at different sizes can be captured, and the judgment of morphological changes due to a single scale is avoided. The roundness matching degree at different scales can be obtained by the following steps: adjusting the radius of the candidate region at each scale, and calculating the corresponding boundary length; calculating the roundness at the scale to obtain the roundness matching degree results at multiple scales;

[0112] According to the calculated roundness value, a preset threshold is set to judge the regularity of follicle morphology;

[0113] In some preferred embodiments, the roundness matching value is compared with the preset threshold:

[0114] If the roundness matching value is greater than or equal to the pre-auditing threshold, it indicates that the follicle morphology is regular;

[0115] If the roundness matching value is less than the pre-auditing threshold, it indicates that the follicle morphology is irregular.

[0116] S5: For the follicle evaluated as regular by the roundness matching function, edge sharpness analysis is performed, and the volume and area are calculated according to the morphological characteristics, the regular follicle region is classified, the follicles are classified into different categories according to their size, morphology and distribution, and the counting of follicles is completed.

[0117] The classification of regular follicles and the counting of follicles in S5 are as follows:

[0118] The edge sharpness of the follicle region evaluated as regular by the roundness matching function is analyzed, the follicle boundary is extracted by applying the Canny edge detection algorithm to the follicle region, and the gradient intensity, continuity and curvature of the edge are further calculated to evaluate whether the follicle edge is smooth and clear. The edge sharpness formula is:

[0119] ;

[0120] In the formula, E represents the edge sharpness, N is the number of edge pixels, is the gradient value of the ith edge pixel point;

[0121] After confirming that the follicle morphology is regular and the edge is clear, the volume and area of the follicle are calculated, and the area of the follicle is estimated by the number of pixel points in the binary image;

[0122] According to the size, morphology and distribution characteristics of the follicle, the follicle is classified into different categories. The specific classification criteria include:

[0123] Size classification: according to the area and volume of the follicle, the follicle is classified into small follicle, medium follicle and large follicle, and classified according to the size range.

[0124] Morphological classification: According to the roundness, edge smoothness and other characteristics of the follicle, the follicle is further subdivided into regular and irregular morphological follicles.

[0125] Distribution classification: According to the spatial distribution of follicles in the ovary, it is determined whether the follicles are evenly distributed, concentrated in a certain area of the ovary, and the distribution density. For example, follicles can be classified as evenly distributed follicles, clustered follicles, or single follicles.

[0126] By labeling and classifying each type of follicle region, the counting of follicles is completed. The counting method is: traversing all the follicle regions that meet the conditions, recording the classification results of each follicle; counting each type of follicle separately to obtain the number of each type of follicle; counting the total number of all follicles and the number of follicles under each classification, providing a basis for subsequent endocrine analysis.

[0127] Through the above technical solutions, the volume and area of the follicle can be accurately calculated, and the follicle can be classified based on its morphology, size, distribution and other characteristics, providing detailed basis for subsequent follicular development analysis, endocrine index evaluation and clinical decision-making. In addition, the accurate counting and classification of follicles can also help doctors better understand the diagnosis process of ovarian function status and polycystic ovary syndrome.

[0128] S6: Based on the spatial distribution of regular follicles, local density analysis is performed to evaluate the distribution pattern of follicles and generate an endocrine atlas feature vector of the ovary.

[0129] The process of local density analysis based on the spatial distribution of regular follicles in S6 is: the spatial position of the follicle is extracted from the classified regular follicle region, and each regular follicle is identified by its center coordinate in the ovary image. The spatial coordinate of the follicle can be converted from the pixel coordinate system of the image to the actual spatial coordinate system to obtain the accurate position of the follicle in the ovary.

[0130] The purpose of local density analysis of the spatial distribution of regular follicles is to evaluate the distribution pattern of follicles in the ovary region. By calculating the number of other follicles within a certain range around the center point of a follicle in a local window, the local density value of each follicle is obtained.

[0131] The process of evaluating the distribution pattern of follicles and generating an endocrine atlas feature vector of the ovary in S6 is:

[0132] By statistically analyzing the local density of all follicles, the spatial distribution pattern of follicles is analyzed. According to the local density value, the follicles are divided into the following distribution patterns:

[0133] Uniform distribution: The follicles are evenly distributed in the ovary region, and the local density changes little.

[0134] Clustered distribution: follicles are densely distributed within certain areas of the ovary, with higher local density.

[0135] Scattered distribution: follicles are distributed in a scattered manner within the ovary, with significant fluctuations in local density.

[0136] Based on the results of local density analysis, generate an ovarian endocrine map feature vector, which is used to describe the distribution characteristics and density patterns of follicles in the ovary. The steps are:

[0137] Feature extraction: extract a set of feature values through local density values, spatial distribution density of follicles, distribution type, etc. These feature values include:

[0138] Mean follicle density: reflects the overall density level of follicle distribution in the ovary.

[0139] Follicle density standard deviation: describes the dispersion degree of follicle distribution.

[0140] Distribution pattern category: according to the analysis results of local density, mark the distribution pattern of follicles, such as uniform distribution, clustered distribution or scattered distribution.

[0141] Combine the extracted feature values into a feature vector, which can include multiple dimensions such as local density values of follicles, distribution pattern labels, spatial positions, etc., to form a comprehensive ovarian endocrine map feature vector.

[0142] Through the technical solution of this embodiment, the spatial distribution characteristics of follicles can be accurately analyzed, and the local density and distribution pattern of follicles can be evaluated, thereby providing more accurate evaluation of the endocrine state of the ovary. The generated ovarian endocrine map feature vector provides strong data support for subsequent ovarian health analysis, disease prediction and diagnosis, and can serve as an important basis for clinical doctors to make decisions.

[0143] Example 3

[0144] Please refer to Figure 2 As shown in the figure, the gynecological endocrine index analysis system based on image recognition according to the embodiments of the present application comprises:

[0145] Ovary region recognition module: responsible for processing the original ovary ultrasound image, automatically recognizing the ovary region and performing cropping;

[0146] Follicle morphology evaluation module: analyze and extract the texture direction of each grid region, and evaluate the texture consistency of the region;

[0147] Follicle segmentation module: responsible for adaptive segmentation in the screened follicle candidate region;

[0148] Oocyte follicle shape determination module: the roundness feature of the oocyte follicle is evaluated by applying a multi-scale roundness matching function;

[0149] Secretion profile generation module: the distribution pattern of the oocyte follicle is evaluated based on local density analysis of the spatial distribution of the oocyte follicle.

[0150] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one entity or action from another, without necessarily requiring or implying any actual relationship or order between or among such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0151] While the embodiments of the application have been illustrated and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and alterations can be made therein without departing from the spirit and scope of the application, which is defined by the appended claims and their equivalents.

Claims

1. A gynecological endocrine index analysis method based on image recognition, characterized in that: The following steps are involved: By automatically identifying the ovarian region in the ovarian ultrasound image, the boundary of the ovarian ultrasound image is clipped to obtain several grid regions; By analyzing the consistency of texture direction within the grid area, the integrity of the follicle morphology is determined, and incomplete follicles are eliminated to determine the candidate follicle area; The process of judging whether the follicle morphology is complete is as follows: By calculating the main direction angles of all pixels in the grid area and counting the mean and mean square error, a texture consistency threshold is set to distinguish whether the texture direction is consistent, and the judgment is made based on the mean square error; If the mean square error is less than the threshold, it means that the texture direction of the grid area is consistent, indicating that the follicle morphology is complete; If the mean square error is greater than or equal to the threshold, it means that the texture direction of the grid area is inconsistent, indicating that the follicle morphology is incomplete; After determining the candidate follicle regions, an adaptive segmentation method based on dynamic region growing is used to mark the follicles. The speed and threshold of region growing are adaptively adjusted by analyzing local image features. A multi-scale circularity matching function was applied to evaluate the circularity of follicles within the candidate follicle region and determine the regularity of follicle morphology. The process of determining the regularity of follicle morphology is: Extract boundary information from the follicle segmentation region obtained by the region growing process, and use edge detection algorithm to obtain the contour edge of the follicle region; Fit and sample boundary contours at multiple scales to construct a multi-scale circularity matching function; The roundness values ​​were calculated at different scales, and the scale with the smallest fitting error was selected as the best evaluation scale; Compare the roundness value with the preset threshold. If the roundness is greater than the threshold, the shape is considered regular, otherwise it is irregular. The edge clarity of regular follicles was analyzed and combined with the follicle morphology and spatial distribution to classify and count the follicles; According to the spatial distribution of follicles and combined with local density analysis, the ovarian endocrine map feature vector is generated.

2. The gynecological endocrine index analysis method based on image recognition according to claim 1, characterized in that: The process of labeling follicles is as follows: In the selected follicle candidate area, pixels with stable brightness, edge intensity and texture direction are selected as initial seed points; Set the initial grayscale similarity threshold and define the region growth criteria, including the grayscale difference between adjacent pixels and the current region, texture similarity, and edge continuity; In the process of region growing, the neighborhood pixels around the seed point are traversed. If the neighborhood pixels meet the similarity criteria, they are incorporated into the current region and used as new seed points. Dynamically adjust the similarity threshold according to local image features to adapt to grayscale changes and texture fluctuations in the image; When all neighboring pixels do not meet the growth conditions, the region growth terminates; Finally, a closed outline of the candidate follicle region is formed and the region is marked as the follicle region.

3. The gynecological endocrine index analysis method based on image recognition according to claim 2, characterized in that: The process of adaptively adjusting the speed and threshold of region growth according to local image features is as follows: Apply edge detection algorithm to the candidate follicle area to extract edge gradient information and obtain edge intensity map; In the edge intensity map, pixels with significant edge response and high texture direction consistency are selected as initial seed points; For each seed point, calculate the grayscale difference, texture direction difference and edge gradient change between its surrounding pixels and the marked pixels in the current area; According to the degree of difference, determine whether the adjacent pixels meet the region expansion conditions. If so, they are merged into the current segmentation region and added to the seed point queue; During the growth process, the similarity threshold and growth step size are dynamically adjusted according to the average grayscale, texture direction consistency and edge density of the pixels in the current area; The growing process continues to iterate until all the pixels in the neighborhood of the seed point do not meet the expansion condition, thus forming the initial segmentation area of ​​the follicle.

4. The gynecological endocrine index analysis method based on image recognition according to claim 3, characterized in that: The process of judging whether adjacent pixels meet the region expansion conditions is as follows: During the region growing process, if the grayscale difference between the neighboring pixel and the current region is less than the preset grayscale similarity threshold, the pixel is considered to be sufficiently similar to the current region and meets the region expansion conditions. It is added to the current segmented region and continues to participate in region growing as a new seed point. If the grayscale difference is greater than or equal to the similarity threshold, the pixel is considered to be too different from the current area and is excluded from the expansion area and not included in the segmentation area.

5. A gynecological endocrine index analysis system based on image recognition, applied to the method according to any one of claims 1 to 4, characterized in that: include: Ovarian region recognition module: responsible for processing the original ovarian ultrasound image, automatically identifying the ovarian region and cropping it; Follicle morphology assessment module: Analyzes and extracts the texture direction of each mesh area and evaluates the texture consistency of the area; Follicle segmentation module: responsible for adaptive segmentation within the selected follicle candidate area; Follicle morphology determination module: evaluates the roundness characteristics of follicles by applying a multi-scale roundness matching function; Secretion map generation module: performs local density analysis based on the spatial distribution of follicles to evaluate the distribution pattern of follicles.

Images