Method and system for auxiliary diagnosis of polyovarian syndrome based on image recognition

Abstract

The invention relates to the technical field of ovarian ultrasound image enhancement, in particular to a polyovarian syndrome auxiliary diagnosis method and system based on image recognition. The method comprises the following steps: segmenting an ovary ultrasound image of a patient, analyzing the pixel gray level and distance distribution of an ovary region, evaluating the image recognition degree, clustering, and recognizing a potential follicle region. Any region is selected as a reference region, and the sinus degree is calculated according to the boundary gray level change and the internal gradient of the region; evaluating the follicle arrangement degree in combination with the position distribution and the neighbor distance; and calculating the change degree of the multi-capsule sample through the area proportion. And integrating the three indexes of follicle areas of all patients to mark to-be-enhanced levels, thereby enhancing the image and assisting in diagnosis of the polycystic ovarian syndrome. According to the method, targeted image enhancement can be carried out on all potential follicle regions, the omission ratio is reduced, and thus a relatively accurate diagnosis result is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ovarian ultrasound image enhancement, and in particular to a multi-ovary syndrome auxiliary diagnosis method and system based on image recognition. BACKGROUND

[0002] Polycystic ovary syndrome (PCOS) is a common endocrine and metabolic disease in women of childbearing age. The main symptoms are abnormal increase of androgen, persistent anovulation and multiple cysts in the ovary, which have a lasting impact on the patient's physiology and psychology. The evaluation criteria for PCOS usually include three aspects: clinical symptoms, laboratory tests and imaging examinations. Imaging examination is to observe the appearance of the ovary using ultrasound images to determine whether it has PCOS manifestations. Therefore, identifying the ovarian ultrasound image is one of the core steps in the auxiliary diagnosis of PCOS.

[0003] In reality, due to the abdominal wall thickness and fat interference of the patient's abdomen, it is difficult for relevant personnel to choose the best section or scan angle when performing 2D ultrasound examination on the ovary, which leads to poor improvement of the obtained ovarian ultrasound image and ultimately cannot obtain accurate diagnostic results. SUMMARY

[0004] In order to solve the technical problem that due to the abdominal wall thickness and fat interference of the patient's abdomen, it is difficult for relevant personnel to choose the best section or scan angle when performing 2D ultrasound examination on the ovary, which leads to poor improvement of the obtained ovarian ultrasound image and ultimately cannot obtain accurate diagnostic results, the purpose of the present application is to provide a multi-ovary syndrome auxiliary diagnosis method and system based on image recognition, and the technical solution adopted is as follows:

[0005] A multi-ovary syndrome auxiliary diagnosis method based on image recognition, the method comprising:

[0006] obtaining ovarian ultrasound images of different patients;

[0007] The ovarian ultrasound image is image segmented to obtain an ovary region and other regions; an image recognition degree of the ovarian ultrasound image is obtained according to a pixel gray scale distribution and a distance distribution in the ovary region and the other regions; the ovarian ultrasound image is clustered according to the image recognition degree to obtain potential follicle regions; an optional potential follicle region is taken as a reference region; a follicle degree of the reference region is obtained according to a gray scale change feature in a direction from a centroid of the reference region to a boundary of the reference region and a gradient distribution feature in the reference region; a follicle arrangement degree of the reference region is obtained according to a position distribution of the reference region in the ovary region and a distance between the reference region and two potential follicle regions closest to the reference region; a polycystic change degree of the reference region is obtained according to an area proportion of the reference region in the ovary region; the potential follicle regions of different patients are marked according to the follicle degree, the follicle arrangement degree and the polycystic change degree of each potential follicle region of the patients, and a to-be-enhanced level of each potential follicle region is obtained.

[0008] The ovarian ultrasound image is enhanced according to the to-be-enhanced level of each potential follicle region, and polycystic ovary syndrome is assisted in diagnosis according to the enhanced ovarian ultrasound image.

[0009] Further, the image recognition degree obtaining method comprises:

[0010] The image recognition degree is obtained according to an image recognition degree calculation formula, and the image recognition degree calculation formula is as follows:

[0011]

[0012] In the formula, the image recognition degree of the ovarian ultrasound image is represented by D; the number of pixel points in the other regions is represented by N; the gray scale value of the i-th pixel point in the other regions is represented by G; the shortest distance between the i-th pixel point in the other regions and the ovary region boundary is represented by S; the number of pixel points in the ovary region is represented by M; the gray scale value of the i-th pixel point in the ovary region is represented by G; the shortest distance between the i-th pixel point in the ovary region and the ovary region boundary is represented by S; and the gray scale standard deviation of all pixel points in the ovary region is represented by D.

[0013] Further, the potential follicle region obtaining method comprises:

[0014] A standard resolution of the ovarian ultrasound image is obtained, and a standard bandwidth of the ovarian ultrasound image is obtained according to the standard resolution.​​​​​​​​​​​​

[0015] The bandwidth correction coefficient is obtained by negatively normalizing the image recognition level.

[0016] The corrected bandwidth is obtained based on the standard bandwidth and bandwidth correction factor of the ovarian ultrasound image;

[0017] Clustering of pixels in ovarian ultrasound images based on the modified bandwidth yields all connected components;

[0018] All connected regions with a circularity greater than a preset first threshold are considered as potential follicle regions.

[0019] Furthermore, methods for obtaining the degree of sinusoids include:

[0020] Starting from the centroid pixel of the reference area, extend in multiple preset directions until stopping at the edge of the reference area to obtain multiple preset direction line segments;

[0021] The degree of sinus symptom is obtained according to the formula for calculating sinus symptom severity, which is shown below:

[0022]

[0023] In the formula, Indicates the degree of sinusoids in the reference area; Indicates the number of line segments in the preset direction; Indicates the first The maximum gradient value of a preset direction line segment; Indicates the first The first-order difference mean of gray levels of all pixels in a preset direction line segment.

[0024] Furthermore, methods for obtaining the degree of follicle arrangement include:

[0025] The degree of follicle arrangement is obtained using a formula shown below:

[0026]

[0027] In the formula, Indicates the degree of follicle arrangement in the reference region; This represents the shortest distance between the centroid pixel of the reference region and the edge of the ovarian region. Indicates the maximum inscribed circle radius of the reference region; This represents the distance between the centroid pixel of the reference region and the centroid pixel of the nearest first potential follicle region; This represents the radius of the largest inscribed circle of the region closest to the first potential follicle. a distance between the center pixel of the reference region and the center pixel of the second nearest potential follicle region; a maximum inscribed circle radius of the second nearest potential follicle region.

[0028] Further, the method for obtaining the polycystic change degree comprises:

[0029] calculating a ratio between the area of the reference region and the area of the ovary region as a first ratio; and performing negative correlation normalization on the first ratio to obtain the polycystic change degree of the reference region.

[0030] Further, the method for obtaining the to-be-enhanced level of each potential follicle region comprises:

[0031] the sinusoidal degree, the follicle arrangement degree, and the polycystic change degree of the potential follicle region are collectively referred to as follicle region features.

[0032] In the ovary ultrasound images of all the patients, abnormal thresholds of all the potential follicle regions in all the follicle region features are calculated according to the box plot method;

[0033] a follicle region feature greater than the abnormal threshold is recorded as a value 1, and a follicle region feature less than the abnormal threshold is recorded as a value 0; the corresponding values of all the potential follicle regions of each potential follicle region are added to calculate the to-be-enhanced level of each potential follicle region.

[0034] An image recognition-based polycystic ovary syndrome auxiliary diagnosis system, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above-mentioned image recognition-based polycystic ovary syndrome auxiliary diagnosis method when executing the computer program.

[0035] A computer readable storage medium, the computer readable storage medium storing a computer program, and the computer program being executable on a processor to implement the steps of the above-mentioned image recognition-based polycystic ovary syndrome auxiliary diagnosis method.

[0036] A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above-mentioned image recognition-based polycystic ovary syndrome auxiliary diagnosis method when executing the computer program.

[0037] The present application has the following advantages:

[0038] This invention first acquires ovarian ultrasound images from different patients. Since the diagnosis of polyovarian syndrome primarily relies on the ovarian region within the ultrasound images, image segmentation is necessary before deep processing to distinguish the ovarian region from other regions. Due to significant individual differences among patients, to improve the accuracy of follicle identification, the recognizability of the ultrasound images must first be analyzed. Therefore, the image recognition level of the ovarian ultrasound image is obtained based on the pixel grayscale distribution and distance distribution between the ovarian region and other regions. After determining the image recognition level, preliminary extraction of the follicular region within the ovary is required, i.e., identification of potential follicular regions. Based on the recognition level of the ovarian ultrasound images from different patients, an adaptive scale can be adopted to extract potential follicular regions; if the recognition level is high, further extraction is performed. To avoid extracting too many non-follicular areas, the scale is reduced during image extraction. If the recognition rate is low, the scale is increased to avoid missing true follicular areas. Therefore, an adaptive scale can be used to extract potential follicular areas based on the recognition level of the ovarian ultrasound images of different patients. If the recognition rate is high, the scale is reduced during extraction to avoid extracting too many non-follicular areas; if the recognition rate is low, the scale is increased to avoid missing true follicular areas. Based on the degree of sinusoidal follicle formation, follicular arrangement, and polycystic changes of the potential follicular areas in different patients, all potential follicular areas of each patient are marked, and the enhancement level of each potential follicular area is obtained. The ovarian ultrasound image is then enhanced according to the enhancement level of each potential follicular area. The enhanced ovarian ultrasound image is used for auxiliary diagnosis of polyovary syndrome. This invention can perform targeted image enhancement on all potential follicular areas, reducing the false negative rate and thus obtaining more accurate diagnostic results. Attached Figure Description

[0039] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A flowchart of an image recognition-based auxiliary diagnostic method for multiple ovarian syndrome provided in one embodiment of the present invention;

[0041] Figure 2 This is a block diagram of an image recognition-based auxiliary diagnostic system for multiple ovarian syndrome, provided as an embodiment of the present invention. Detailed Implementation

[0042] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an image recognition-based auxiliary diagnostic method and system for multiple ovarian syndrome proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0044] The following description, in conjunction with the accompanying drawings, details the specific scheme of the image recognition-based auxiliary diagnostic method for multiple ovarian syndrome provided by this invention.

[0045] Please see Figure 1 This illustrates an image recognition-based auxiliary diagnostic method for multiple ovarian syndrome provided by an embodiment of the present invention, the method comprising:

[0046] Step S1: Obtain ovarian ultrasound images from different patients.

[0047] The embodiments of the present invention are mainly applied to the image enhancement of ovarian ultrasound images of patients, so the first step is to acquire ovarian ultrasound images of different patients for comprehensive analysis.

[0048] In one embodiment of the present invention, an ultrasound device is used to perform an ultrasound examination on a patient. The ultrasound device is ensured to be functioning properly. An appropriate ultrasound probe is selected based on the patient's condition, and the angle is adjusted to obtain the optimal view of the ovary, resulting in an initial ultrasound image of the patient's ovary. This initial ultrasound image is then transmitted to an image processing system to obtain an ovarian ultrasound image for subsequent analysis.

[0049] Step S2: Segment the ovarian ultrasound image to obtain the ovarian region and other regions; determine the image recognition level of the ovarian ultrasound image based on the pixel grayscale distribution and distance distribution between the ovarian region and other regions; cluster the ovarian ultrasound image based on the image recognition level to obtain potential follicle regions; select one potential follicle region as a reference region; determine the sinusoidal degree of the reference region based on the grayscale change characteristics from the centroid to the boundary direction and the gradient distribution characteristics within the reference region; determine the follicle arrangement degree of the reference region based on its position within the ovarian region and the distance between the reference region and the two nearest potential follicle regions; determine the degree of polycystic changes in the reference region based on its area proportion within the ovarian region; and mark all potential follicle regions for each patient based on the sinusoidal degree, follicle arrangement degree, and polycystic change degree of the potential follicle regions, obtaining the enhancement level for each potential follicle region.

[0050] In reality, the diagnosis of polyovarian syndrome mainly relies on the ovarian region in ovarian ultrasound images. Therefore, image segmentation is necessary before in-depth processing to distinguish the ovarian region from other regions (other structures in the pelvic cavity). Currently, the most commonly used method in medicine is the U-Net-based semantic segmentation model, which has high accuracy. Therefore, in one embodiment of this invention, the U-Net-based semantic segmentation model is used to identify and label the ovarian region in the image for subsequent in-depth computational analysis of the ovary's interior.

[0051] Due to significant individual differences among patients, the ultrasound examination method and image quality can be directly affected. Furthermore, ovarian ultrasound images inherently have low resolution, so follicle identification is greatly influenced by patient characteristics. For example, there are differences between various parts of transvaginal and transabdominal ultrasound images, and variations in fat distribution can also lead to different image quality. To improve the accuracy of follicle identification, the identifiability of the ultrasound image needs to be analyzed first. Therefore, in this embodiment of the invention, the image recognition level of the ovarian ultrasound image is obtained based on the pixel grayscale distribution and distance distribution between the ovarian region and other regions.

[0052] Preferably, in one embodiment of the present invention, the method for obtaining the image recognition level includes:

[0053] The image recognition level is obtained according to the image recognition level calculation formula, which is shown below:

[0054]

[0055] In the formula, This indicates the image recognition level of the ovarian ultrasound image; This indicates the number of pixels in other areas; Indicates the first in other regions The grayscale value of each pixel; Indicates the first in other areas The shortest distance between each pixel and the boundary of the ovarian region; This indicates the number of pixels within the ovarian region; Indicates the first ovarian region The grayscale value of each pixel; Indicates the first ovarian region The shortest distance between each pixel and the boundary of the ovarian region; This represents the standard deviation of grayscale values ​​for all pixels within the ovarian region.

[0056] In the image recognition accuracy calculation formula, the larger the gray value of a pixel, the greater its sharpness. Since pixels closer to the ovarian region boundary have higher sharpness, the reciprocal of the distance to the ovarian region boundary is used as the weight of the pixel's gray value. The weighted average gray value of pixels in other regions is then calculated. Weighted gray mean of pixels in the ovarian region The ratio between these two regions is important because the ovary is soft tissue, which has relatively lower reflectivity and grayscale values ​​in ultrasound images. Therefore, a larger ratio indicates a more significant pixel difference between the ovarian region and other regions, resulting in greater image recognition of the ovarian ultrasound image. The standard deviation of grayscale values ​​for all pixels within the ovarian region is also considered. The larger the value, the less local ambiguity in the ovarian region, resulting in a clearer ovarian ultrasound image, meaning a greater degree of image recognition in the ovarian ultrasound image.

[0057] After determining the image recognition level, it is necessary to perform preliminary extraction of the follicular regions in the ovary, that is, to identify potential follicular regions, which may include actual follicular regions or other non-follicular regions.

[0058] Preferably, in one embodiment of the present invention, the method for obtaining the potential follicle region includes:

[0059] The standard resolution of the ovarian ultrasound image is obtained, and the standard bandwidth of the ovarian ultrasound image is obtained based on the standard resolution. In one embodiment of the present invention, the standard bandwidth is [missing information]. It should be noted that standard bandwidth is a technical means well known to those skilled in the art, and will not be elaborated here; the image recognition degree is negatively correlated and normalized to obtain the bandwidth correction coefficient.

[0060] The corrected bandwidth is obtained based on the standard bandwidth and bandwidth correction factor of the ovarian ultrasound image. In one embodiment of the present invention, the calculation formula for the corrected bandwidth is as follows:

[0061]

[0062] In the formula, Indicates the corrected bandwidth; This indicates the image recognition level of the ovarian ultrasound image; This represents the nearest integer function; This represents the normalization function.

[0063] The pixels of the ovarian ultrasound image are clustered based on the modified bandwidth to obtain all connected components. In one embodiment of the present invention, the Mean-Shift clustering algorithm is used for this operation. It should be noted that the Mean-Shift clustering algorithm is a well-known technique to those skilled in the art and will not be described in detail here.

[0064] All connected regions with a circularity greater than a preset first threshold are considered as potential follicle regions. It should be noted that circularity is a technique well-known to those skilled in the art and will not be elaborated upon here. The preset first threshold is set to 0.6, and can be set by the user without limitation.

[0065] Based on the recognition level of ovarian ultrasound images from different patients, an adaptive scale can be used to extract potential follicular regions. If the recognition level is high, the scale is reduced during extraction to avoid extracting too many non-follicular areas; if the recognition level is low, the scale is increased to avoid missing true follicular regions. Therefore, in this embodiment of the invention, ovarian ultrasound images are first clustered according to the image recognition level to obtain potential follicular regions. Three features are set in the potential follicular regions: the degree of sinusoidal morphology, the degree of follicular arrangement, and the degree of polycystic changes to select true follicular regions.

[0066] First, based on the gray-level variation characteristics of the reference region from the centroid to the boundary, and the gradient distribution characteristics within the reference region, the sinus degree of the reference region is obtained. Preferably, in one embodiment of the present invention, the method for obtaining the sinus degree includes:

[0067] Starting from the centroid pixel of the reference region, the line segments extend in multiple preset directions until they reach the edge of the reference region, thus obtaining multiple preset directional line segments. In one embodiment of the present invention, the multiple preset directions are set as the four directions directly above, below, to the left, and to the right of the centroid pixel of the reference region. It should be noted that the preset directions and their number can be set arbitrarily and are not limited here.

[0068] The degree of sinus symptom is obtained according to the formula for calculating sinus symptom severity, which is shown below:

[0069]

[0070] In the formula, Indicates the degree of sinusoids in the reference area; Indicates the number of line segments in the preset direction; Indicates the first The maximum gradient value of a preset direction line segment; Indicates the first The first-order difference mean of gray levels of all pixels in a preset direction line segment.

[0071] In the formula for calculating the degree of sinusoidal follicle formation, since sinusoidal follicles are formed by the accumulation of fluid inside the follicle, and the fluid cavity usually exhibits a low grayscale (hypoechoic region), there will be obvious grayscale change points within the follicle, i.e., the pixel point corresponding to the maximum gradient value. The larger the maximum gradient value of a preset direction line segment, the more obvious the sinus-like features of the preset direction line segment, and the greater the degree of sinus-likeness of the preset direction line segment. The larger the value, the more likely it is to be the first. The more obvious the grayscale change of a preset direction line segment, and the more obvious the grayscale increase from the center to the edge, the more obvious the sinus-like feature of the preset direction line segment, that is, the greater the sinus-like degree of the preset direction line segment. Perform the same analysis on all preset direction line segments and calculate the average value to obtain the sinus-like degree of the reference area.

[0072] Secondly, based on the location distribution of the reference region within the ovarian region and the distance between the reference region and the two nearest potential follicle regions, the degree of follicle arrangement in the reference region is obtained. Preferably, in one embodiment of the present invention, the method for obtaining the degree of follicle arrangement includes:

[0073] The degree of follicle arrangement is obtained using a formula shown below:

[0074]

[0075] In the formula, Indicates the degree of follicle arrangement in the reference region; This represents the shortest distance between the centroid pixel of the reference region and the edge of the ovarian region. Indicates the maximum inscribed circle radius of the reference region; This represents the distance between the centroid pixel of the reference region and the centroid pixel of the nearest first potential follicle region; This represents the radius of the largest inscribed circle of the region closest to the first potential follicle. This represents the distance between the centroid pixel of the reference region and the centroid pixel of the nearest second potential follicle region; This represents the radius of the largest inscribed circle of the region closest to the second potential follicle.

[0076] In the formula for calculating the degree of follicle arrangement, since a large number of immature follicles under the ovarian capsule are arranged in a ring, the closer the reference area is to the edge of the ovary, the better. The smaller the value, the more likely the reference area is located near the ovary; the smaller the distance between the reference area and the two nearest potential follicle areas, the better. , The smaller the value, the more likely the reference area is to be arranged in a ring. That is, the closer the reference area is to the edge of the ovary and the more closely it is arranged with other surrounding areas, the more likely it is to be a characteristic of a large number of immature follicles arranged in a ring. In this case, the degree of follicle arrangement in the reference area is considered to be greater.

[0077] Finally, the degree of polycystic changes in the reference region is obtained based on the area ratio of the reference region within the ovarian region. Preferably, in one embodiment of the present invention, the method for obtaining the degree of polycystic changes includes:

[0078] The ratio between the area of ​​the reference region and the area of ​​the ovarian region is calculated as the first ratio; the first ratio is then negatively correlated and normalized to obtain the degree of polycystic changes in the reference region. In one embodiment of the present invention, the formula for calculating the degree of polycystic changes is as follows:

[0079]

[0080] In the formula, Indicates the degree of multicystic changes in the reference region; Indicates the area of ​​the reference region; This indicates the area of ​​the ovarian region.

[0081] In the formula for calculating the degree of polycystic changes, because the follicles of patients with polycystic ovary syndrome arrest during the developmental stage and fail to develop into normal dominant follicles, their follicles are characterized by being small and numerous, with a diameter much smaller than that of mature follicles. However, their ovaries may enlarge due to stromal hyperplasia. The smaller the reference area is compared to the ovarian region, the more likely it is to be a small follicle, and the greater the degree of polycystic changes in the reference area.

[0082] Based on the degree of sinusoidal morphology, follicular arrangement, and polycystic changes in the potential follicular regions of different patients, all potential follicular regions of each patient were marked to obtain the level of enhancement required for each potential follicular region.

[0083] Preferably, in one embodiment of the present invention, the method for obtaining the enhancement level of each potential follicle region includes:

[0084] The degree of sinusoidal morphology, follicular arrangement, and polycystic changes in potential follicular regions are collectively referred to as follicular region characteristics. In all patients' ovarian ultrasound images, the abnormal thresholds of all potential follicular regions in all ovarian ultrasound images based on all follicular region characteristics were calculated using the box plot method. It should be noted that the box plot method is a technique well-known to those skilled in the art and will not be elaborated upon here.

[0085] Follicular region features greater than the abnormal threshold are recorded as value 1, and follicular region features less than the abnormal threshold are recorded as value 0. The corresponding values ​​of all potential follicular regions for each potential follicular region are added together to calculate the enhancement level of each potential follicular region.

[0086] Step S3: Enhance the ovarian ultrasound images according to the enhancement level required for each potential follicular region; use the enhanced ovarian ultrasound images to assist in the diagnosis of multiple ovary syndrome.

[0087] The above steps obtained the enhancement level of each potential follicle region. The higher the enhancement level, the stronger the pathological manifestation of the potential follicle region, and the larger the enhancement scale needs to be during enhancement to further highlight the pathological manifestation. A smaller enhancement level may be a noise area (such as artifact) in the image, and the enhancement scale should be smaller.

[0088] Therefore, in one embodiment of the present invention, the enhancement scale is adjusted by controlling the shearing threshold in adaptive histogram equalization; a larger shearing threshold results in a larger enhancement scale. In one embodiment of the present invention, the commonly used shearing threshold for medical images is... The calculated level of potential follicle regions to be strengthened is... Therefore, the shearing threshold is mapped one-to-one with the region order. If the order is 0, the shearing threshold is 2; if the order is 1, the shearing threshold is 3, and so on. Finally, the shearing threshold of all potential follicle regions is obtained. Based on this threshold, the ovarian ultrasound image is subjected to adaptive histogram equalization, so that each potential follicle region receives a different enhancement effect, which can highlight and enhance key areas while avoiding image distortion.

[0089] After obtaining enhanced ultrasound images of the patient's ovaries, relevant personnel can be assisted in diagnosing polycystic ovary syndrome (PCOS) by analyzing the images to determine whether they conform to the polycystic ovary morphology criteria of the Rotterdam criteria. Combining the patient's clinical manifestations and hormone test results, it can be determined whether the patient has PCOS. If two of the criteria are met, a diagnosis can be made, thus completing the auxiliary diagnosis of PCOS.

[0090] In summary, the following steps were taken: Ovarian ultrasound images were acquired from different patients; these images were segmented to identify the ovarian region and other regions; the image recognition level of the ovarian ultrasound images was determined based on the pixel grayscale distribution and distance distribution between the ovarian region and other regions; the ovarian ultrasound images were clustered based on the image recognition level to identify potential follicular regions; one potential follicular region was selected as a reference region; the sinusoidal degree of the reference region was determined based on the grayscale change characteristics from the centroid to the boundary direction and the gradient distribution characteristics within the reference region; the follicular arrangement degree of the reference region was determined based on its position within the ovarian region and the distance between the reference region and the two nearest potential follicular regions; the degree of polycystic changes in the reference region was determined based on its area proportion within the ovarian region; all potential follicular regions of each patient were marked according to the sinusoidal degree, follicular arrangement degree, and polycystic change degree of their potential follicular regions, and the enhancement level of each potential follicular region was determined; the ovarian ultrasound images were enhanced according to the enhancement level of each potential follicular region; and the enhanced ovarian ultrasound images were used for auxiliary diagnosis of polyovary syndrome.

[0091] A second objective of one embodiment of the present invention is to provide an image recognition-based auxiliary diagnostic system for polyovary syndrome. This system includes a memory, a processor, and a computer program. The memory stores the corresponding computer program, and the processor runs the corresponding computer program. When the computer program runs in the processor, it can implement the methods described in steps S1-S3, specifically including:

[0092] Image acquisition module 101 is used to acquire ovarian ultrasound images of different patients;

[0093] Image analysis module 102 is used to segment ovarian ultrasound images to obtain the ovarian region and other regions; to obtain the image recognition degree of the ovarian ultrasound image based on the pixel grayscale distribution and distance distribution of the ovarian region and other regions; to cluster the ovarian ultrasound image based on the image recognition degree to obtain potential follicle regions; to select one potential follicle region as a reference region; to obtain the sinusoidal degree of the reference region based on the grayscale change characteristics from the centroid to the boundary direction and the gradient distribution characteristics within the reference region; to obtain the follicle arrangement degree of the reference region based on the positional distribution of the reference region within the ovarian region and the distance between the reference region and the two nearest potential follicle regions; to obtain the polycystic change degree of the reference region based on the area proportion of the reference region within the ovarian region; and to mark all potential follicle regions of each patient based on the sinusoidal degree, follicle arrangement degree, and polycystic change degree of the potential follicle regions, thereby obtaining the enhancement level of each potential follicle region.

[0094] The auxiliary diagnostic module 103 is used to enhance ovarian ultrasound images according to the enhancement level of each potential follicle region; and to perform auxiliary diagnosis of multiple ovarian syndrome based on the enhanced ovarian ultrasound images.

[0095] A third objective of this invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aforementioned scoliosis identification method.

[0096] The fourth objective of this invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned scoliosis identification method.

[0097] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0098] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for auxiliary diagnosis of multiple ovarian syndrome based on image recognition, characterized in that, The method includes: Obtain ovarian ultrasound images from different patients; The ovarian ultrasound image is segmented to obtain the ovarian region and other regions. The image recognition level of the ovarian ultrasound image is obtained based on the pixel grayscale distribution and distance distribution between the ovarian region and other regions. The ovarian ultrasound image is clustered based on the image recognition level to obtain potential follicle regions. One potential follicle region is selected as a reference region. The sinusoidal degree of the reference region is obtained based on the grayscale change characteristics from the centroid to the boundary direction and the gradient distribution characteristics within the reference region. The follicle arrangement degree of the reference region is obtained based on its position within the ovarian region and the distance between the reference region and the two nearest potential follicle regions. The degree of polycystic changes in the reference region is obtained based on its area proportion within the ovarian region. Based on the sinusoidal degree, follicle arrangement degree, and polycystic change degree of the potential follicle regions in different patients, all potential follicle regions of each patient are marked to obtain the enhancement level for each potential follicle region. The ovarian ultrasound images are enhanced according to the enhancement level of each potential follicular region; the enhanced ovarian ultrasound images are used to assist in the diagnosis of multiple ovarian syndrome.

2. The method for auxiliary diagnosis of multiple ovarian syndrome based on image recognition according to claim 1, characterized in that, The method for obtaining the image recognition level includes: The image recognition degree is obtained according to the image recognition degree calculation formula, which is as follows: In the formula, This indicates the image recognition level of the ovarian ultrasound image; This indicates the number of pixels in other areas; Indicates the first in other regions The grayscale value of each pixel; Indicates the first in other regions The shortest distance between each pixel and the boundary of the ovarian region; This indicates the number of pixels within the ovarian region; Indicates the first ovarian region The grayscale value of each pixel; Indicates the first ovarian region The shortest distance between each pixel and the boundary of the ovarian region; This represents the standard deviation of grayscale values ​​for all pixels within the ovarian region.

3. The method for auxiliary diagnosis of multiple ovarian syndrome based on image recognition according to claim 1, characterized in that, The method for obtaining the potential follicle region includes: Obtain the standard resolution of the ovarian ultrasound image, and obtain the standard bandwidth of the ovarian ultrasound image based on the standard resolution; The image recognition level is negatively correlated and normalized to obtain the bandwidth correction coefficient. The corrected bandwidth is obtained based on the standard bandwidth and the bandwidth correction coefficient of the ovarian ultrasound image; Cluster the pixels of the ovarian ultrasound image according to the modified bandwidth to obtain all connected components; All connected regions with a circularity greater than a preset first threshold are considered as potential follicle regions.

4. The method for auxiliary diagnosis of multiple ovarian syndrome based on image recognition according to claim 1, characterized in that, The method for obtaining the degree of sinusoids includes: Starting from the centroid pixel of the reference area, extend in multiple preset directions until stopping at the edge of the reference area to obtain multiple preset direction line segments; The degree of sinus symptom is obtained according to the formula for calculating the degree of sinus symptom, which is as follows: In the formula, Indicates the degree of sinusoids in the reference area; Indicates the number of line segments in the preset direction; Indicates the first The maximum gradient value of a preset direction line segment; Indicates the first The first-order difference mean of gray levels of all pixels in a preset direction line segment.

5. The method for auxiliary diagnosis of multiple ovarian syndrome based on image recognition according to claim 1, characterized in that, The methods for obtaining the degree of follicle arrangement include: The degree of follicle arrangement is obtained according to the formula for calculating the degree of follicle arrangement, which is shown below: In the formula, Indicates the degree of follicle arrangement in the reference region; This represents the shortest distance between the centroid pixel of the reference region and the edge of the ovarian region. Indicates the maximum inscribed circle radius of the reference region; This represents the distance between the centroid pixel of the reference region and the centroid pixel of the nearest first potential follicle region; This represents the radius of the largest inscribed circle of the region closest to the first potential follicle. This represents the distance between the centroid pixel of the reference region and the centroid pixel of the nearest second potential follicle region; This represents the radius of the largest inscribed circle of the region of the second nearest potential follicle.

6. The method for auxiliary diagnosis of multiple ovarian syndrome based on image recognition according to claim 1, characterized in that, The method for obtaining the degree of polycystic changes includes: The ratio between the area of ​​the reference region and the area of ​​the ovarian region is calculated as the first ratio; the first ratio is negatively normalized to obtain the degree of polycystic changes in the reference region.

7. The method for auxiliary diagnosis of multiple ovarian syndrome based on image recognition according to claim 1, characterized in that, The methods for determining the enhancement level for each potential follicular region include: The sinus-like degree, follicular arrangement degree, and polycystic change degree of potential follicular regions are collectively referred to as follicular region characteristics. In all patients’ ovarian ultrasound images, the abnormal threshold of all potential follicular regions in all follicular region features was calculated using the box plot method. Follicular region features greater than the abnormal threshold are recorded as value 1, and follicular region features less than the abnormal threshold are recorded as value 0. The corresponding values ​​of all potential follicular regions for each potential follicular region are added together to calculate the enhancement level of each potential follicular region.

8. A diagnostic system for polyovary syndrome based on image recognition, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the image recognition-based auxiliary diagnostic method for multiple ovarian syndrome as described in any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the image recognition-based auxiliary diagnostic method for multiple ovarian syndrome as described in any one of claims 1 to 7.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the image recognition-based auxiliary diagnostic method for multiple ovarian syndrome as described in any one of claims 1 to 7.

Images