Method and system for detecting electroacupuncture curative effect of premature ovarian failure patient based on medical image

Abstract

The invention relates to the technical field of image processing, in particular to a premature ovarian failure patient electroacupuncture curative effect detection method and system based on a medical image, and the method comprises the steps: obtaining a suspected follicle degree and a suspected ovarian matrix degree of each pixel point according to the local gray level change of each pixel point in a gray level image of an ovarian medical image; according to the suspected follicle degree and the suspected ovarian matrix degree of each pixel point, obtaining a suspected follicle region, a suspected ovarian matrix region and other regions, and according to local features of each region, obtaining credibility of each region; according to the gradient direction distribution in the local range of each pixel point and the credibility of the area where each pixel point is located, the self-adaptive h parameter of each pixel point in the NLM algorithm is obtained, then the NLM algorithm is used for denoising the gray level image, the electroacupuncture treatment effect of the premature ovarian failure patient is detected according to the denoised gray level image, and the premature ovarian failure treatment effect of the premature ovarian failure patient is detected according to the electroacupuncture treatment effect of the premature ovarian failure patient. The accuracy of detecting the electroacupuncture curative effect of the premature ovarian failure patient is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and particularly to a method and system for detecting the electroacupuncture efficacy of premature ovarian failure patients based on medical images. Background Art

[0002] Premature ovarian failure is a disease that seriously affects women's reproductive health and physical and mental health. Premature ovarian failure not only leads to a decline in female fertility, but also causes a series of endocrine disorders, seriously affecting the quality of life of patients. In recent years, electroacupuncture treatment for premature ovarian failure has gradually attracted attention. Electroacupuncture applies a weak current at specific acupoints, stimulates nerve afferent fibers, regulates the neuroendocrine system, improves ovarian function, and displays the number of follicles and ovarian morphology through B-ultrasound medical images of the ovaries to evaluate the efficacy of electroacupuncture in treating premature ovarian failure. However, in un-denoised B-ultrasound medical images, noise may mask ovarian morphological changes and interfere with follicle number detection, resulting in a deviation between the efficacy evaluation result and the true result. Therefore, before analyzing the B-ultrasound medical images of the ovaries, it is necessary to first perform denoising processing on the B-ultrasound medical images of the ovaries.

[0003] Under the traditional method, the non-local means denoising algorithm (NLM algorithm) is selected to denoise B-ultrasound medical images. However, when using the NLM algorithm to denoise B-ultrasound medical images, the Gaussian function smoothing parameter (h parameter) of each pixel point is usually set according to empirical values. A larger h parameter has a lower requirement for similarity, and more pixel blocks will be included in the weighted average calculation, resulting in a stronger denoising effect, but it will cause loss of image details; on the contrary, a smaller h parameter has a higher requirement for similarity, and can retain more image details, but it will cause a weakening of the denoising effect.

[0004] Therefore, how to adaptively obtain the h parameter of each pixel point to improve the denoising effect of B-ultrasound medical images of the ovaries, and thus improve the accuracy of detecting the electroacupuncture efficacy of premature ovarian failure patients has become an urgent problem to be solved. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method and system for detecting the electroacupuncture efficacy of premature ovarian failure patients based on medical images to solve the problem of how to adaptively obtain the h parameter of each pixel point to improve the denoising effect of B-ultrasound medical images of the ovaries, and thus improve the accuracy of detecting the electroacupuncture efficacy of premature ovarian failure patients.

[0006] In a first aspect, embodiments of the present invention provide a method for detecting the electroacupuncture efficacy of premature ovarian failure patients based on medical images, and the method includes the following steps: Obtain the grayscale image of the ovarian medical image of the premature ovarian failure patient, and obtain the suspected follicle degree and suspected ovarian stroma degree of each pixel according to the gray value distribution of the pixels within the local range of each pixel in the grayscale image; According to the suspected follicle degree and suspected ovarian stroma degree of each pixel, divide the grayscale image into a suspected follicle area, a suspected ovarian stroma area, and other areas. According to the gradient of the boundary pixels of each suspected follicle area and the shape regularity of each suspected follicle area, obtain the credibility of each suspected follicle area. According to the credibility of the adjacent areas of each suspected ovarian stroma area, obtain the credibility of each suspected ovarian stroma area; According to the distribution of the gradient directions of the pixels within the local range of each pixel in the grayscale image, the credibility of the suspected follicle area, the credibility of the suspected ovarian stroma area, and the preset credibility of other areas, obtain the adaptive h parameter of each pixel in the NLM algorithm. According to the adaptive h parameter of each pixel, use the NLM algorithm to denoise the grayscale image to obtain the denoised grayscale image, and detect the electroacupuncture efficacy of the premature ovarian failure patient according to the denoised grayscale image.

[0007] Preferably, the obtaining the suspected follicle degree and suspected ovarian stroma degree of each pixel according to the gray value distribution of the pixels within the local range of each pixel in the grayscale image includes: For any pixel in the grayscale image, construct a preset number of windows with different scales centered on the any pixel, calculate the mean gray value and the standard deviation of the gray values of all pixels in each window, calculate the product between the reciprocal of the mean gray value and the reciprocal of the standard deviation of the gray value corresponding to each window to obtain the follicle eigenvalue of each window, and linearly normalize the mean of the follicle eigenvalues of all windows to obtain the suspected follicle degree of the any pixel.

[0008] Preferably, the obtaining the suspected follicle degree and suspected ovarian stroma degree of each pixel according to the gray value distribution of the pixels within the local range of each pixel in the grayscale image further includes: Calculate the product between the reciprocal of the mean gray value and the standard deviation of the gray value corresponding to each window of the any pixel to obtain the ovarian stroma eigenvalue of each window, and linearly normalize the mean of the ovarian stroma eigenvalues of all windows to obtain the suspected ovarian stroma degree of the any pixel.

[0009] Preferably, the dividing the grayscale image into a suspected follicle area, a suspected ovarian stroma area, and other areas according to the suspected follicle degree and suspected ovarian stroma degree of each pixel includes: Cluster according to the suspected follicle degree of each pixel point in the grayscale image to obtain at least one suspected follicle cluster. Calculate the mean value of the suspected follicle degrees of all corresponding pixel points in each of the suspected follicle clusters, which is denoted as the mean suspected follicle degree. If the mean suspected follicle degree of any one of the suspected follicle clusters is greater than or equal to the preset suspected follicle degree threshold, then the region formed by all pixel points in the any one of the suspected follicle clusters in the grayscale image is denoted as the suspected follicle region; Cluster according to the suspected ovarian stroma degree of each pixel point in the grayscale image to obtain at least one suspected ovarian stroma cluster. Calculate the mean value of the suspected ovarian stroma degrees of all corresponding pixel points in each of the suspected ovarian stroma clusters, which is denoted as the mean suspected ovarian stroma degree. If the mean suspected ovarian stroma degree of any one of the suspected ovarian stroma clusters is greater than or equal to the preset suspected ovarian stroma degree threshold, then the region formed by all pixel points in the any one of the suspected ovarian stroma clusters in the grayscale image is denoted as the suspected ovarian stroma region; Denote the region in the grayscale image other than the suspected follicle region and the suspected ovarian stroma region as other regions.

[0010] Preferably, the obtaining the credibility of each suspected follicle region according to the gradient of the boundary pixel points of each suspected follicle region and the shape regularity of each suspected follicle region includes: For any one of the suspected follicle regions, calculate the difference between the constant 1 and the circularity of the any one of the suspected follicle regions, calculate the reciprocal of the sum of the difference and the preset constant to obtain the circularity-like degree of the any one of the suspected follicle regions; Obtain the boundary pixel points of the any one of the suspected follicle regions. In the grayscale image, calculate the mean value of the grayscale values of a preset number of pixel points on the right side of the normal direction of any one of the boundary pixel points to obtain the right grayscale mean value, calculate the mean value of the grayscale values of a preset number of pixel points on the left side of the normal direction of the any one of the boundary pixel points to obtain the left grayscale mean value, calculate the absolute value of the difference between the right grayscale mean value and the left grayscale mean value, and calculate the product of the gradient value of the any one of the boundary pixel points and the absolute value of the difference to obtain the clarity of the any one of the boundary pixel points; Calculate the mean value of the clarities of all boundary pixel points of the any one of the suspected follicle regions, and take the product of the mean value of the clarities and the circularity-like degree as the credibility of the any one of the suspected follicle regions.

[0011] Preferably, the obtaining the credibility of each suspected ovarian stroma region according to the credibility of the adjacent regions of each suspected ovarian stroma region includes: For any suspected ovarian stroma region, in the grayscale image, obtain at least one pixel point adjacent to the boundary pixel points of the any suspected ovarian stroma region, denoted as the target pixel point, and remove the any suspected ovarian stroma region in the region where all target pixel points are located, to obtain at least one adjacent region of the any suspected ovarian stroma region; For any adjacent region, according to the gradient of the boundary pixel points of the any adjacent region and the shape regularity of the any adjacent region, obtain the credibility of the any adjacent region, denoted as the follicle credibility, and calculate the product between the mean value of the suspected follicle degree of all pixel points in the any adjacent region and the follicle credibility of the any adjacent region, to obtain the follicle possibility degree of the any adjacent region; Obtain the follicle possibility degree of each adjacent region of the any suspected ovarian stroma region, and select the maximum value among the follicle possibility degrees of all adjacent regions as the credibility of the any suspected ovarian stroma region.

[0012] Preferably, obtaining the adaptive h parameter of each pixel point in the NLM algorithm according to the distribution of the gradient directions of the pixel points in the local range of each pixel point in the grayscale image, the credibility of the suspected follicle region, the credibility of the suspected ovarian stroma region, and the preset credibility of other regions, includes: For any pixel point in the grayscale image, if the any pixel point belongs to the suspected follicle region, then calculate the product between the suspected follicle degree of the any pixel point and the credibility of the suspected follicle region where the any pixel point is located, to obtain the key degree of the any pixel point; If the any pixel point belongs to the suspected ovarian stroma region, then calculate the product between the suspected ovarian stroma degree of the any pixel point and the credibility of the suspected ovarian stroma region where the any pixel point is located, to obtain the key degree of the any pixel point; If the any pixel point belongs to both the suspected ovarian stroma region and the suspected follicle region, then obtain the maximum value between the suspected follicle degree and the suspected ovarian stroma degree of the any pixel point. If the maximum value is the suspected follicle degree, then calculate the product between the suspected follicle degree and the credibility of the suspected follicle region where the any pixel point is located, to obtain the key degree of the any pixel point. If the maximum value is the suspected ovarian stroma degree, then calculate the product between the suspected ovarian stroma degree and the credibility of the suspected ovarian stroma region where the any pixel point is located, to obtain the key degree of the any pixel point; If the any pixel point belongs to other regions, then use the preset credibility of other regions as the key degree of the any pixel point; Obtain the adaptive h parameter of any pixel point in the NLM algorithm according to the key degree of any pixel point and the distribution of the gradient directions of the pixel points within the local range of any pixel point.

[0013] Preferably, the obtaining of the adaptive h parameter of any pixel point in the NLM algorithm according to the key degree of any pixel point and the distribution of the gradient directions of the pixel points within the local range of any pixel point includes: In a grayscale image, with any pixel point as the center, construct a target window with a preset size, obtain the gradient direction entropy according to the gradient directions of all pixel points within the target window, and linearly normalize the product between the gradient value of any pixel point and the reciprocal of the gradient direction entropy to obtain the edge degree of any pixel point. Perform weighted summation on the edge degree and the key degree of any pixel point to obtain a weighted summation result, calculate the difference between the preset maximum h parameter and the preset minimum h parameter to obtain the preset h parameter range difference, and calculate the product between the preset h parameter range difference and the weighted summation result to obtain the h parameter adjustment coefficient of any pixel point. Calculate the difference between the preset maximum h parameter and the h parameter adjustment coefficient of any pixel point to obtain the adaptive h parameter of any pixel point in the NLM algorithm.

[0014] Preferably, the preset credibility of other regions is set to 0.

[0015] In a second aspect, an electroacupuncture efficacy detection system for premature ovarian failure patients based on medical images according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements an electroacupuncture efficacy detection method for premature ovarian failure patients based on medical images as described in the first aspect.

[0016] The beneficial effects of the embodiments of the present invention compared with the prior art are: The present invention obtains the grayscale image of the ovarian medical image of a patient with premature ovarian failure, and obtains the suspected follicle degree and the suspected ovarian stroma degree of each pixel point according to the distribution of the grayscale values of the pixel points within the local range of each pixel point in the grayscale image; according to the suspected follicle degree and the suspected ovarian stroma degree of each pixel point, the grayscale image is divided into a suspected follicle area, a suspected ovarian stroma area, and other areas, and according to the gradient of the boundary pixel points of each suspected follicle area and the shape regularity of each suspected follicle area, the credibility of each suspected follicle area is obtained, and according to the credibility of the adjacent areas of each suspected ovarian stroma area, the credibility of each suspected ovarian stroma area is obtained; according to the distribution of the gradient directions of the pixel points within the local range of each pixel point in the grayscale image, the credibility of the suspected follicle area, the credibility of the suspected ovarian stroma area, and the preset credibility of the other areas, the adaptive h parameter of each pixel point in the NLM algorithm is obtained, and according to the adaptive h parameter of each pixel point, the grayscale image is denoised by using the NLM algorithm to obtain the denoised grayscale image, and the electroacupuncture efficacy of the patient with premature ovarian failure is detected according to the denoised grayscale image. Among them, according to the local features (suspected follicle degree and suspected ovarian stroma degree) of each pixel point in the grayscale image, the suspected follicle area and the suspected ovarian stroma area are obtained, and according to the features of each suspected follicle area and each suspected ovarian stroma area, the credibility of each suspected follicle area and each suspected ovarian stroma area is obtained to exclude the interference of the noise area; according to the credibility of the area where each pixel point is located and the local features of each pixel point, the adaptive h parameter of each pixel point in the NLM algorithm is obtained, and then according to the adaptive h parameter of each pixel point, the grayscale image is denoised by using the NLM algorithm, and while denoising, the features of the key areas (follicle area and ovarian stroma area) are retained, the denoising effect of the B-ultrasound medical image of the ovary is improved, and further the accuracy of detecting the electroacupuncture efficacy of the patient with premature ovarian failure is improved. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings.

[0018] Figure 1 It is a flowchart of a method for detecting the electroacupuncture efficacy of a patient with premature ovarian failure based on a medical image provided in Embodiment 1 of the present invention. Detailed Embodiments

[0019] Embodiments of the present disclosure will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are intended to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0020] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above accompanying drawings are used to distinguish similar objects and do not necessarily have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0021] In order to illustrate the technical solution of the present invention, it will be described below through specific embodiments.

[0022] See Figure 1 , which is a method flow chart of a method for detecting the electroacupuncture efficacy of patients with premature ovarian failure based on medical images provided in the first embodiment of the present invention. As Figure 1 shown, the method may include: Step S101, obtain a grayscale image of the ovarian medical image of a patient with premature ovarian failure, and obtain the suspected follicle degree and suspected ovarian stroma degree of each pixel point according to the gray value distribution of the pixel points within the local range of each pixel point in the grayscale image.

[0023] Obtain the ovarian medical image of a patient with premature ovarian failure after electroacupuncture treatment through a B-ultrasound machine. Since the efficacy of a patient with premature ovarian failure after electroacupuncture treatment needs to be detected in combination with the number of follicles and ovarian morphology in the ovarian medical image, and there may be noise in the ovarian medical image, the noise may mask the ovarian morphological changes and interfere with the detection of the number of follicles, resulting in the deviation of the efficacy detection result from the true result. Therefore, before analyzing the ovarian medical image, it is necessary to perform denoising processing on the ovarian medical image.

[0024] In the traditional method, the NLM algorithm is selected to denoise the ovarian medical image. However, the Gaussian function smoothing parameter (h parameter) of each pixel point in the ovarian medical image is usually set according to empirical values. A larger h parameter has a lower requirement for similarity, and more pixel blocks will be included in the weighted average calculation, resulting in a stronger denoising effect but loss of details; on the contrary, a smaller h parameter has a higher requirement for similarity, and more details can be retained, but the denoising effect will be weakened.

[0025] Therefore, in the embodiments of the present invention, first, OpenCV is used to grayscale the ovarian medical image to obtain the grayscale image of the ovarian medical image of the premature ovarian failure patient. Then, the local change characteristics of each pixel point in the grayscale image are analyzed to obtain the adaptive h parameter of each pixel point in the NLM algorithm, so as to improve the denoising effect of the B-ultrasound medical image of the ovary, and further improve the accuracy of detecting the curative effect after electroacupuncture treatment of the patient for premature ovarian failure. Among them, using OpenCV to grayscale the ovarian medical image is a prior art and will not be elaborated here.

[0026] Since the curative effect of premature ovarian failure patients after electroacupuncture treatment needs to be detected in combination with the number of follicles and ovarian morphology in the ovarian medical image, first, the follicle region and ovarian stroma region in the grayscale image need to be determined. For the pixel points at the edges of the follicle region and ovarian stroma region, a smaller h parameter should be assigned to retain the edge details; for the pixel points inside the follicle region and ovarian stroma region, a larger h parameter should be assigned to appropriately denoise while retaining the morphological characteristics within the region and eliminating noise interference.

[0027] Considering that follicles usually appear as dark areas, in the grayscale image, the grayscale value of the follicle region is low, and the grayscale values of the pixel points in the region are evenly distributed. The ovarian stroma consists of the cortex and medulla. There are follicles and dense connective tissues in the cortex, and the medulla contains loose connective tissues, abundant blood vessels, nerves, lymphatic vessels, and smooth muscle fibers, etc. Different tissue components have different reflections and absorptions of ultrasound, resulting in the overall low echo of the ovarian stroma region. In the grayscale image, the grayscale value of the ovarian stroma region is low, and the grayscale values of the pixel points in the region are unevenly distributed. Also considering that there are not only follicle regions and ovarian stroma regions in the ovarian medical image, but also other tissue regions or noise interference regions that present high echoes, which are shown as high grayscale values in the grayscale image. Therefore, in the embodiments of the present invention, according to the local grayscale characteristics of each pixel point in the grayscale image, the degree of suspected follicle and the degree of suspected ovarian stroma of each pixel point are obtained, which are used to initially judge the possibility of each pixel point belonging to the follicle region and ovarian region while reducing the interference of other tissue regions or noise interference regions, and further determine the follicle region and ovarian stroma region in the grayscale image.

[0028] Taking the i-th pixel point in the grayscale image as an example, the specific method for obtaining the degree of suspected follicle of the i-th pixel point is as follows: Construct three windows with different scales centered on the \(i\)-th pixel. In the embodiments of the present invention, the scales of the three windows are set to \(5\times5\), \(7\times7\), and \(9\times9\), which is not limited herein. The implementer can set the number of windows and the window scales according to the specific scenario. Calculate the mean and standard deviation of the gray values of all pixel points in each of the windows, calculate the product between the reciprocal of the mean gray value and the reciprocal of the standard deviation of the gray value corresponding to each of the windows, obtain the follicle feature value of each of the windows, and perform linear normalization on the mean of the follicle feature values of all windows to obtain the suspected follicle degree of any of the pixel points. Among them, linear normalization is a prior art and will not be elaborated herein.

[0029] In one embodiment, the formula for calculating the suspected follicle degree of the \(i\)-th pixel point is: Wherein, represents the suspected follicle degree of the \(i\)-th pixel point, represents the number of all windows corresponding to the \(i\)-th pixel point (in the embodiments of the present invention, \(N = 3\)), represents the number of all pixel points in the window with scale \(n\), represents the gray value of the \(j\)-th pixel point in the window with scale \(n\) of the \(i\)-th pixel point, represents the standard deviation of the gray values of all pixel points in the window with scale \(n\) of the \(i\)-th pixel point, and norm() represents the linear normalization function.

[0030] It should be noted that, the smaller, the more the \(i\)-th pixel point conforms to the dark area feature, that is, the more the \(i\)-th pixel point conforms to the feature of the follicle area. Furthermore, the larger, the greater the possibility that the \(i\)-th pixel point is a pixel point in the follicle area; the smaller, the more uniform the distribution of the gray values of the pixel points in the window corresponding to the \(i\)-th pixel point. Furthermore, the larger, the greater the possibility that the \(i\)-th pixel point is a pixel point in the follicle area.

[0031] Furthermore, obtain the suspected ovarian stroma degree of the \(i\)-th pixel point. Specifically: Calculate the product between the reciprocal of the mean gray value and the standard deviation corresponding to each of the windows of the \(i\)-th pixel point to obtain the ovarian stroma feature value of each of the windows, and perform linear normalization on the mean of the ovarian stroma feature values of all windows to obtain the suspected ovarian stroma degree of the \(i\)-th pixel point.

[0032] In one embodiment, the formula for calculating the suspected ovarian stroma degree of the \(i\)-th pixel point is: Wherein, represents the suspected ovarian stroma degree of the i-th pixel point represents the number of all windows corresponding to the i-th pixel point (in the embodiment of the present invention, N = 3) represents the number of all pixel points within the window with scale n represents the gray value of the j-th pixel point within the window with scale n of the i-th pixel point represents the standard deviation of the gray values of all pixel points within the window with scale n of the i-th pixel point, and norm() represents the linear normalization function

[0033] It should be noted that the smaller it is, the more the i-th pixel point conforms to the hypoechoic feature of the ovarian stroma region, and thus the larger it is, the greater the possibility that the i-th pixel point is a pixel point in the ovarian stroma region the larger it is, the more uneven the distribution of the gray values of the pixel points within the window corresponding to the i-th pixel point, and thus the larger it is, the greater the possibility that the i-th pixel point is a pixel point in the ovarian stroma region

[0034] Thus, the suspected follicle degree and the suspected ovarian stroma degree of the i-th pixel point are obtained. Similarly, the suspected follicle degree and the suspected ovarian stroma degree of each pixel point in the grayscale image are obtained for subsequent determination of the follicle region and the ovarian stroma region in the grayscale image

[0035] Step S102: According to the suspected follicle degree and the suspected ovarian stroma degree of each pixel point, divide the grayscale image into a suspected follicle region, a suspected ovarian stroma region, and other regions. According to the gradient of the boundary pixel points of each suspected follicle region and the shape regularity of each suspected follicle region, obtain the credibility of each suspected follicle region. According to the credibility of the adjacent regions of each suspected ovarian stroma region, obtain the credibility of each suspected ovarian stroma region

[0036] After obtaining the suspected follicle degree and the suspected ovarian stroma degree of each pixel point in the grayscale image through step S101, preliminarily determine the suspected follicle region and the suspected ovarian stroma region in the grayscale image. Specifically Using the mean shift clustering algorithm, cluster according to the suspected follicle degree of each pixel point in the grayscale image to obtain at least one suspected follicle clustering cluster. There is no limitation here. Implementers can select the clustering algorithm according to the specific scenario. Calculate the mean of the suspected follicle degrees of all corresponding pixel points in each of the suspected follicle clustering clusters, denoted as the mean of the suspected follicle degree. According to experimental statistics, in the embodiments of the present invention, the preset suspected follicle degree threshold is set to 0.8. There is no limitation here. Implementers can set it according to the specific scenario. If the mean of the suspected follicle degree of any one of the suspected follicle clustering clusters is greater than or equal to 0.8, then the region formed by all pixel points in the any one of the suspected follicle clustering clusters in the grayscale image is denoted as the suspected follicle region; Using the mean shift clustering algorithm, cluster according to the suspected ovarian stroma degree of each pixel point in the grayscale image to obtain at least one suspected ovarian stroma clustering cluster. Calculate the mean of the suspected ovarian stroma degrees of all corresponding pixel points in each of the suspected ovarian stroma clustering clusters, denoted as the mean of the suspected ovarian stroma degree. According to experimental statistics, in the embodiments of the present invention, the preset suspected ovarian stroma degree threshold is set to 0.8. There is no limitation here. Implementers can set it according to the specific scenario. If the mean of the suspected ovarian stroma degree of any one of the suspected ovarian stroma clustering clusters is greater than or equal to 0.8, then the region formed by all pixel points in the any one of the suspected ovarian stroma clustering clusters in the grayscale image is denoted as the suspected ovarian stroma region; The region in the grayscale image other than the suspected follicle region and the suspected ovarian stroma region is denoted as the other region. Among them, the mean shift clustering algorithm is prior art and will not be elaborated here.

[0037] Since there may be interference regions in the B-ultrasound medical images of the ovary that are similar to the characteristics of the follicle region or the ovarian stroma region, in the embodiments of the present invention, the local characteristics of the suspected follicle region and the suspected ovarian stroma region in the grayscale image are combined to respectively obtain the credibility of the suspected follicle region and the suspected ovarian stroma region, so as to improve the accuracy of the determination of the follicle region or the ovarian stroma region.

[0038] Considering that the follicle region is approximately circular and has a clear boundary, and the ovarian stroma region and the follicle region belong to the ovarian region, so the ovarian stroma region is adjacent to the follicle region, while the interference region generally has an irregular shape and a blurred boundary. According to the above characteristics, in the embodiments of the present invention, first, according to the gradient of the boundary pixel points of each suspected follicle region and the shape regularity of each suspected follicle region, obtain the credibility of each suspected follicle region. Specifically: For any one of the suspected follicle regions, calculate the difference between the constant 1 and the circularity of the any one of the suspected follicle regions, and calculate the reciprocal of the sum of the difference and the preset constant to obtain the circularity degree of the any one of the suspected follicle regions; Binarize any of the suspected follicle regions, and use the findContours function to extract the outermost pixel points of any of the suspected follicle regions, that is, the boundary pixel points of any of the suspected follicle regions. In the grayscale image, use the Sobel operator to obtain the normal direction of any boundary pixel point, calculate the mean value of the grayscale values of a preset number of pixel points on the right side of the normal direction of any boundary pixel point to obtain the right grayscale mean value, calculate the mean value of the grayscale values of a preset number of pixel points on the left side of the normal direction of any boundary pixel point to obtain the left grayscale mean value, calculate the absolute value of the difference between the right grayscale mean value and the left grayscale mean value, and calculate the product of the gradient value of any boundary pixel point and the absolute value of the difference to obtain the clarity of any boundary pixel point. Among them, set the preset number to 3, which is not limited here, and the implementer can set it according to the specific scenario. Binarization, the findContours function, and using the Sobel operator to obtain the normal direction of pixel points are prior arts and will not be elaborated here; Calculate the clarity mean value of all boundary pixel points of any of the suspected follicle regions, and use the product of the clarity mean value and the circularity as the credibility of any of the suspected follicle regions.

[0039] In one embodiment, taking the f-th suspected follicle region as an example, the calculation formula for the credibility of the f-th suspected follicle region is: Among them, represents the credibility of the f-th suspected follicle region, represents the circularity of the f-th suspected follicle region, represents a preset constant used to prevent the denominator from being 0. In the embodiments of the present invention, is set, which is not limited here, and the implementer can set it according to the specific scenario, represents the gradient value of the b-th boundary pixel point of the f-th suspected follicle region, represents the mean value of the grayscale values of a preset number (in the embodiments of the present invention, the preset number is 3) of pixel points on the left side of the normal direction of b boundary pixel points, that is, the left grayscale mean value, represents the mean value of the grayscale values of a preset number (in the embodiments of the present invention, the preset number is 3) of pixel points on the right side of the normal direction of b boundary pixel points, that is, the right grayscale mean value, represents the number of all boundary pixel points of the f-th suspected follicle region, represents the absolute value symbol.

[0040] It should be noted that The larger it is, the closer the f-th suspected follicle region is to a circle. Furthermore, The larger it is, the higher the probability that the f-th suspected follicle region belongs to the follicle region; The larger it is, and The larger it is, indicating that the boundary of the f-th suspected follicle region is clearer, and thus The larger it is, the higher the probability that the f-th suspected follicle region belongs to the follicle region.

[0041] Similarly, obtain the credibility of all suspected follicle regions in the grayscale image.

[0042] Furthermore, obtain the credibility of all suspected ovarian stroma regions in the grayscale image. Taking the h-th suspected ovarian stroma region as an example, since the ovarian stroma region and the follicle region both belong to the ovarian region and are adjacent to each other, in the embodiments of the present invention, according to the method for obtaining the credibility of the suspected follicle region, obtain the follicle possibility degree of each adjacent region of the h-th suspected ovarian stroma region in the grayscale image, and then obtain the credibility of the h-th suspected ovarian stroma region according to the follicle possibility degree of each adjacent region. Specifically: Binarize the h-th suspected ovarian stroma region, and use the findContours function to extract the outermost pixel points of the h-th suspected ovarian stroma region, that is, the boundary pixel points of the h-th suspected ovarian stroma region. In the grayscale image, obtain at least one pixel point adjacent to the boundary pixel points of the h-th suspected ovarian stroma region, denoted as the target pixel point, and remove the h-th suspected ovarian stroma region in all regions where the target pixel points are located to obtain at least one adjacent region of the h-th suspected ovarian stroma region; For any adjacent region, according to the method for obtaining the credibility of the suspected follicle region described above, obtain the credibility of the any adjacent region, denoted as the follicle credibility, according to the gradient of the boundary pixel points of the any adjacent region and the shape regularity of the any adjacent region, and calculate the product between the mean value of the suspected follicle degrees of all pixel points in the any adjacent region and the follicle credibility of the any adjacent region to obtain the follicle possibility degree of the any adjacent region.

[0043] In one embodiment, taking the x-th adjacent region of the h-th suspected ovarian stroma region as an example, the calculation formula for the follicle possibility degree of the x-th adjacent region of the h-th suspected ovarian stroma region is: Wherein, represents the follicle possibility degree of the x-th adjacent region of the h-th suspected ovarian stroma region, represents the mean value of the suspected follicle degrees of all pixel points in the x-th adjacent region of the h-th suspected ovarian stroma region, represents the follicle credibility of the x-th adjacent region of the h-th suspected ovarian stroma region.

[0044] It should be noted that The larger it is, the more the x-th adjacent region conforms to the characteristics of the pixel points in the follicle region. Furthermore The larger it is, the greater the possibility that the x-th adjacent region is the follicle region, and the higher the possibility that the h-th suspected ovarian stroma region belongs to the suspected ovarian stroma region; The larger it is, the greater the circular similarity of the x-th adjacent region, and the clearer the boundary, the more it conforms to the characteristics of the follicle region. Furthermore The larger it is, the greater the possibility that the x-th adjacent region is the follicle region, and the higher the possibility that the h-th suspected ovarian stroma region belongs to the suspected ovarian stroma region.

[0045] Similarly, obtain the follicle possibility degree of each of the adjacent regions of the h-th suspected ovarian stroma region. Among the follicle possibility degrees of all adjacent regions, select the maximum value as the credibility of the h-th suspected ovarian stroma region, denoted as .

[0046] Similarly, obtain the credibility of each suspected ovarian stroma region in the grayscale image.

[0047] Step S103: According to the distribution of the gradient directions of the pixel points within the local range of each pixel point in the grayscale image, the credibility of the suspected follicle region, the credibility of the suspected ovarian stroma region, and the preset credibility of other regions, obtain the adaptive h parameter of each pixel point in the NLM algorithm. According to the adaptive h parameter of each pixel point, use the NLM algorithm to denoise the grayscale image to obtain the denoised grayscale image, and detect the electroacupuncture efficacy of premature ovarian failure patients based on the denoised grayscale image.

[0048] After obtaining the credibility of each suspected follicle region and each suspected ovarian stroma region in the grayscale image according to step S102, according to the suspected follicle degree and suspected ovarian stroma degree of each pixel point in the grayscale image, and the credibility of the region where each pixel point is located, obtain the key degree of each pixel point. Furthermore, according to the key degree of each pixel point, obtain its adaptive h parameter in the NLM algorithm. Specifically: For the i-th pixel point in the grayscale image, if the i-th pixel point belongs to the suspected follicle region, then calculate the product of the suspected follicle degree of the i-th pixel point and the credibility of the suspected follicle region where the i-th pixel point is located to obtain the key degree of the i-th pixel point, denoted as , that is , where represents the suspected follicle degree of the i-th pixel point, Represents the credibility of the suspected follicle area where the \(i\)-th pixel is located. The greater the degree of suspicion of the \(i\)-th pixel as a follicle and the greater the credibility of the suspected follicle area where the \(i\)-th pixel is located, the greater the likelihood that the \(i\)-th pixel belongs to the follicle area; If the \(i\)-th pixel belongs to the suspected ovarian stroma area, then calculate the product between the degree of suspicion of the \(i\)-th pixel as ovarian stroma and the credibility of the suspected ovarian stroma area where the \(i\)-th pixel is located to obtain the key degree of the \(i\)-th pixel, denoted as That is, where, represents the degree of suspicion of the \(i\)-th pixel as ovarian stroma, represents the credibility of the suspected ovarian stroma area where the \(i\)-th pixel is located. The greater the degree of suspicion of the \(i\)-th pixel as ovarian stroma and the greater the credibility of the suspected ovarian stroma area where the \(i\)-th pixel is located, the greater the likelihood that the \(i\)-th pixel belongs to the ovarian stroma area; Since the suspected ovarian stroma area and the suspected follicle area are obtained by clustering based on the degree of suspicion of each pixel as ovarian stroma and follicle, it is possible that a certain pixel belongs to both the suspected ovarian stroma area and the suspected follicle area. If the \(i\)-th pixel belongs to both the suspected ovarian stroma area and the suspected follicle area, then obtain the maximum value between the degree of suspicion of the \(i\)-th pixel as a follicle and the degree of suspicion of the \(i\)-th pixel as ovarian stroma. If the maximum value is the degree of suspicion of the \(i\)-th pixel as a follicle, then calculate the product between the degree of suspicion of the \(i\)-th pixel as a follicle and the credibility of the suspected follicle area where the \(i\)-th pixel is located to obtain the key degree of the \(i\)-th pixel, that is If the maximum value is the degree of suspicion of the \(i\)-th pixel as ovarian stroma, then calculate the product between the degree of suspicion of the \(i\)-th pixel as ovarian stroma and the credibility of the suspected ovarian stroma area where the \(i\)-th pixel is located to obtain the key degree of the \(i\)-th pixel, that is ; Since the electroacupuncture efficacy of premature ovarian failure patients is detected based on the number of follicles and ovarian morphology, and other areas are irrelevant to the detection of the electroacupuncture efficacy of premature ovarian failure patients, therefore, if the \(i\)-th pixel belongs to other areas, then use the preset credibility of other areas as the key degree of the \(i\)-th pixel. In the present invention, the preset credibility of other areas is set to 0, that is, if the \(i\)-th pixel belongs to other areas, then the key degree of the \(i\)-th pixel , which is not limited here, and the implementer can set it according to the specific scenario.

[0049] At this point, the criticality of the ith pixel point is obtained. The higher the criticality of the ith pixel point, the higher the possibility that it belongs to the follicle area or the ovarian stroma area. Since the efficacy of electroacupuncture in patients with premature ovarian failure is detected based on the number of follicles and the morphology of the ovaries, it is also necessary to retain the detailed features of the suspected follicle area and the suspected ovarian stroma area in the grayscale image. If the ith pixel point is an edge pixel point of the suspected follicle area or the suspected ovarian stroma area, it should be assigned a smaller h parameter to retain the edge details. Since the edges of the follicle area and the ovarian stroma area are relatively smooth, in an embodiment of the present invention, the edge degree of the ith pixel point is obtained based on the distribution of the gradient direction of the ith pixel point in its local range. Specifically: In the grayscale image, a 5×5 target window is constructed with the ith pixel as the center. There is no restriction here. The implementer can set the size of the target window according to the specific scenario. The gradient direction entropy is obtained according to the gradient direction of all pixels in the target window. The product of the gradient value of the ith pixel and the inverse of the gradient direction entropy is linearly normalized to obtain the edge degree of the ith pixel.

[0050] In one embodiment, the calculation formula for the edge degree of the i-th pixel is: in, Indicates the edge degree of the i-th pixel, Represents the gradient value of the i-th pixel, It represents the gradient directional entropy of all pixels in the target window of the i-th pixel, and norm() is a linear normalization function.

[0051] It should be noted that The smaller it is, the more consistent the distribution of the gradient direction of the pixels in the target window is. The larger it is, the greater the possibility that the i-th pixel is an edge pixel of the suspected follicle area or the suspected ovarian stroma area; The larger the value, the more dramatic the change of the gray value of the i-th pixel in the local range of the i-th pixel, and the greater the edge feature of the i-th pixel. The larger the value is, the greater the possibility that the i-th pixel is an edge pixel of the suspected follicle region or the suspected ovarian stroma region.

[0052] Furthermore, according to the edge degree and criticality of the i-th pixel, the adaptive h parameter of the i-th pixel in the NLM algorithm is obtained, specifically: Perform a weighted sum of the edge degree and the key degree of the i-th pixel to obtain a weighted sum result. Calculate the difference between the preset maximum h parameter and the preset minimum h parameter to obtain the preset h parameter range difference. Calculate the product between the preset h parameter range difference and the weighted sum result to obtain the h parameter adjustment coefficient of the i-th pixel; Calculate the difference between the preset maximum h parameter and the h parameter adjustment coefficient of the i-th pixel to obtain the adaptive h parameter of the i-th pixel in the NLM algorithm.

[0053] In one embodiment, the calculation formula for the adaptive h parameter of the i-th pixel is: Wherein, represents the adaptive h parameter of the i-th pixel, represents the preset maximum h parameter, represents the preset minimum h parameter, represents the weight, represents the edge degree of the i-th pixel, represents the key degree of the i-th pixel.

[0054] It should be noted that according to experimental statistics, in the embodiments of the present invention, set , , which is not limited here. Implementers can set according to specific scenarios. is the weight for representing the edge degree, represents the weight of the key degree. Since the adaptive h parameter of the i-th pixel in the NLM algorithm mainly depends on its key degree, in the embodiments of the present invention, set , that is , which is not limited here. Implementers can set according to specific scenarios. The greater the key degree of the i-th pixel and the greater the edge degree, it indicates that the i-th pixel is more likely to be an edge pixel of the suspected follicle area or the suspected ovarian stroma area, and a smaller h parameter should be assigned to retain the edge details. Furthermore the smaller, the smaller the adaptive h parameter of the i-th pixel.

[0055] Similarly, obtain the adaptive h parameters of all pixels in the grayscale image. According to the adaptive h parameter of each pixel, use the NLM algorithm to denoise the grayscale image to obtain the denoised grayscale image. In the denoised grayscale image, detect the electroacupuncture efficacy of patients with premature ovarian failure according to the number of follicles and the morphology of the ovarian stroma. Among them, detecting the electroacupuncture efficacy of patients with premature ovarian failure according to the number of follicles and the morphology of the ovarian stroma is the prior art and will not be elaborated here.

[0056] In summary, the present invention obtains the grayscale image of the ovarian medical image of a patient with premature ovarian failure, and obtains the suspected follicle degree and suspected ovarian stroma degree of each pixel point according to the distribution of the grayscale values of the pixel points within the local range of each pixel point in the grayscale image; according to the suspected follicle degree and suspected ovarian stroma degree of each pixel point, the grayscale image is divided into a suspected follicle area, a suspected ovarian stroma area, and other areas, and the credibility of each suspected follicle area is obtained according to the gradient of the boundary pixel points of each suspected follicle area and the shape regularity of each suspected follicle area, and the credibility of each suspected ovarian stroma area is obtained according to the credibility of the adjacent areas of each suspected ovarian stroma area; according to the distribution of the gradient directions of the pixel points within the local range of each pixel point in the grayscale image, the credibility of the suspected follicle area, the credibility of the suspected ovarian stroma area, and the preset credibility of the other areas, the adaptive h parameter of each pixel point in the NLM algorithm is obtained, and according to the adaptive h parameter of each pixel point, the grayscale image is denoised by using the NLM algorithm to obtain the denoised grayscale image, and the electroacupuncture efficacy of the patient with premature ovarian failure is detected according to the denoised grayscale image. Among them, according to the local features (suspected follicle degree and suspected ovarian stroma degree) of each pixel point in the grayscale image, the suspected follicle area and the suspected ovarian stroma area are obtained, and the credibility of each suspected follicle area and the suspected ovarian stroma area is obtained according to the characteristics of each suspected follicle area and the suspected ovarian stroma area to exclude the interference of the noise area; according to the credibility of the area where each pixel point is located and the local features of each pixel point, the adaptive h parameter of each pixel point in the NLM algorithm is obtained, and then according to the adaptive h parameter of each pixel point, the grayscale image is denoised by using the NLM algorithm, and the features of the key areas (follicle area and ovarian stroma area) are retained while denoising, improving the denoising effect of the B-ultrasound medical image of the ovary, and further improving the accuracy of detecting the electroacupuncture efficacy of the patient with premature ovarian failure.

[0057] Based on the same inventive concept as the above method, an embodiment of the present invention further provides a detection system for the electroacupuncture efficacy of a patient with premature ovarian failure based on a medical image, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of any one of the above methods for detecting the electroacupuncture efficacy of a patient with premature ovarian failure based on a medical image are implemented.

[0058] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for detecting the electroacupuncture efficacy of premature ovarian failure patients based on medical images, characterized in that, The method for detecting the electroacupuncture efficacy of premature ovarian failure patients based on medical images includes: Obtain the grayscale image of the ovarian medical image of a premature ovarian failure patient, and according to the gray value distribution of the pixel points within the local range of each pixel point in the grayscale image, obtain the suspected follicle degree and the suspected ovarian stroma degree of each pixel point; According to the suspected follicle degree and the suspected ovarian stroma degree of each pixel point, divide the grayscale image into a suspected follicle area, a suspected ovarian stroma area, and other areas. According to the gradient of the boundary pixel points of each suspected follicle area and the shape regularity of each suspected follicle area, obtain the credibility of each suspected follicle area. According to the credibility of the adjacent areas of each suspected ovarian stroma area, obtain the credibility of each suspected ovarian stroma area; According to the distribution of the gradient directions of the pixel points within the local range of each pixel point in the grayscale image, the credibility of the suspected follicle area, the credibility of the suspected ovarian stroma area, and the preset credibility of other areas, obtain the adaptive h parameter of each pixel point in the NLM algorithm. According to the adaptive h parameter of each pixel point, use the NLM algorithm to denoise the grayscale image to obtain the denoised grayscale image, and detect the electroacupuncture efficacy of premature ovarian failure patients according to the denoised grayscale image.

2. The electroacupuncture efficacy detection method for premature ovarian failure patients based on medical images according to claim 1, wherein, The obtaining of the suspected follicle degree and the suspected ovarian stroma degree of each pixel point according to the gray value distribution of the pixel points within the local range of each pixel point in the grayscale image includes: For any pixel point in the grayscale image, construct a preset number of windows with different scales centered on the any pixel point, calculate the mean gray value and the standard deviation of the gray values of all pixel points within each window, calculate the product of the reciprocal of the mean gray value and the reciprocal of the standard deviation of the gray values corresponding to each window to obtain the follicle eigenvalue of each window, and linearly normalize the mean of the follicle eigenvalues of all windows to obtain the suspected follicle degree of the any pixel point.

3. The electroacupuncture efficacy detection method for premature ovarian failure patients based on medical images according to claim 2, wherein The obtaining of the suspected follicle degree and the suspected ovarian stroma degree of each pixel point according to the gray value distribution of the pixel points within the local range of each pixel point in the grayscale image further includes: Calculate the product of the reciprocal of the mean gray value and the standard deviation of the gray values corresponding to each window of the any pixel point to obtain the ovarian stroma eigenvalue of each window, and linearly normalize the mean of the ovarian stroma eigenvalues of all windows to obtain the suspected ovarian stroma degree of the any pixel point.

4. The electroacupuncture efficacy detection method for premature ovarian failure patients based on medical images according to claim 1, wherein The dividing of the grayscale image into a suspected follicle area, a suspected ovarian stroma area, and other areas according to the suspected follicle degree and the suspected ovarian stroma degree of each pixel point includes: Cluster according to the suspected follicle degree of each pixel point in the grayscale image to obtain at least one suspected follicle clustering cluster, calculate the mean of the suspected follicle degrees of all pixel points corresponding to each suspected follicle clustering cluster, denoted as the mean suspected follicle degree. If the mean suspected follicle degree of any suspected follicle clustering cluster is greater than or equal to the preset suspected follicle degree threshold, then the area formed by all pixel points in the any suspected follicle clustering cluster in the grayscale image is denoted as the suspected follicle area; Cluster according to the suspected ovarian stroma degree of each pixel point in the grayscale image to obtain at least one suspected ovarian stroma cluster. Calculate the mean value of the suspected ovarian stroma degrees of all corresponding pixel points in each of the suspected ovarian stroma clusters, which is denoted as the mean value of the suspected ovarian stroma degree. If the mean value of the suspected ovarian stroma degree of any suspected ovarian stroma cluster is greater than or equal to the preset suspected ovarian stroma degree threshold, then the region formed by all pixel points in the grayscale image in the any suspected ovarian stroma cluster is denoted as the suspected ovarian stroma region; Denote the region in the grayscale image other than the suspected follicle region and the suspected ovarian stroma region as other regions.

5. The electroacupuncture efficacy detection method for premature ovarian failure patients based on medical images according to claim 1, wherein The obtaining of the credibility of each suspected follicle region according to the gradient of the boundary pixel points of each suspected follicle region and the shape regularity of each suspected follicle region includes: For any suspected follicle region, calculate the difference between the constant 1 and the circularity of the any suspected follicle region, and calculate the reciprocal of the sum of the difference and the preset constant to obtain the roundness of the any suspected follicle region; Obtain the boundary pixel points of the any suspected follicle region. In the grayscale image, calculate the mean value of the grayscale values of a preset number of pixel points on the right side of the normal direction of any boundary pixel point to obtain the right grayscale mean value, calculate the mean value of the grayscale values of a preset number of pixel points on the left side of the normal direction of the any boundary pixel point to obtain the left grayscale mean value, calculate the absolute value of the difference between the right grayscale mean value and the left grayscale mean value, and calculate the product of the gradient value of the any boundary pixel point and the absolute value of the difference to obtain the clarity of the any boundary pixel point; Calculate the mean value of the clarity of all boundary pixel points of the any suspected follicle region, and take the product of the mean value of the clarity and the roundness as the credibility of the any suspected follicle region.

6. The electroacupuncture efficacy detection method for premature ovarian failure patients based on medical images according to claim 5, wherein, The obtaining of the credibility of each suspected ovarian stroma region according to the credibility of the adjacent regions of each suspected ovarian stroma region includes: For any suspected ovarian stroma region, in the grayscale image, obtain at least one pixel point adjacent to the boundary pixel points of the any suspected ovarian stroma region, which is denoted as the target pixel point. Remove the any suspected ovarian stroma region in the region where all target pixel points are located to obtain at least one adjacent region of the any suspected ovarian stroma region; For any adjacent region, obtain the credibility of the any adjacent region according to the gradient of the boundary pixel points of the any adjacent region and the shape regularity of the any adjacent region, which is denoted as the follicle credibility. Calculate the product of the mean value of the suspected follicle degrees of all pixel points in the any adjacent region and the follicle credibility of the any adjacent region to obtain the follicle possibility degree of the any adjacent region; Obtain the follicle possibility degree of each adjacent region of the any suspected ovarian stroma region, and select the maximum value among the follicle possibility degrees of all adjacent regions as the credibility of the any suspected ovarian stroma region.

7. The electroacupuncture efficacy detection method for premature ovarian failure patients based on medical images according to claim 1, wherein Obtaining an adaptive h parameter of each pixel point in the NLM algorithm according to the distribution of the gradient directions of the pixel points in the local range of each pixel point in the grayscale image, the credibility of the suspected follicle region, the credibility of the suspected ovarian stroma region, and the preset credibility of other regions, includes: For any pixel point in the grayscale image, if the any pixel point belongs to the suspected follicle region, calculate the product of the suspected follicle degree of the any pixel point and the credibility of the suspected follicle region where the any pixel point is located, to obtain the key degree of the any pixel point; If the any pixel point belongs to the suspected ovarian stroma region, calculate the product of the suspected ovarian stroma degree of the any pixel point and the credibility of the suspected ovarian stroma region where the any pixel point is located, to obtain the key degree of the any pixel point; If the any pixel point belongs to both the suspected ovarian stroma region and the suspected follicle region, obtain the maximum value of the suspected follicle degree and the suspected ovarian stroma degree of the any pixel point. If the maximum value is the suspected follicle degree, calculate the product of the suspected follicle degree and the credibility of the suspected follicle region where the any pixel point is located, to obtain the key degree of the any pixel point. If the maximum value is the suspected ovarian stroma degree, calculate the product of the suspected ovarian stroma degree and the credibility of the suspected ovarian stroma region where the any pixel point is located, to obtain the key degree of the any pixel point; If the any pixel point belongs to other regions, use the preset credibility of the other regions as the key degree of the any pixel point; According to the key degree of the any pixel point and the distribution of the gradient directions of the pixel points in the local range of the any pixel point, obtain the adaptive h parameter of the any pixel point in the NLM algorithm.

8. The method for detecting the electroacupuncture efficacy of premature ovarian failure patients based on medical images according to claim 7, wherein The obtaining the adaptive h parameter of the any pixel point in the NLM algorithm according to the key degree of the any pixel point and the distribution of the gradient directions of the pixel points in the local range of the any pixel point, includes: In the grayscale image, with the any pixel point as the center, construct a target window with a preset size. According to the gradient directions of all pixel points in the target window, obtain the gradient direction entropy, and perform linear normalization on the product of the gradient value of the any pixel point and the reciprocal of the gradient direction entropy, to obtain the edge degree of the any pixel point; Perform weighted summation on the edge degree and the key degree of the any pixel point to obtain a weighted summation result. Calculate the difference between the preset maximum h parameter and the preset minimum h parameter to obtain the preset h parameter range difference. Calculate the product of the preset h parameter range difference and the weighted summation result to obtain the h parameter adjustment coefficient of the any pixel point; Calculate the difference between the preset maximum h parameter and the h parameter adjustment coefficient of the any pixel point to obtain the adaptive h parameter of the any pixel point in the NLM algorithm.

9. The method for detecting the electroacupuncture curative effect of premature ovarian failure patients based on medical images according to claim 7, characterized in that, The preset credibility of the other regions is set to 0.

10. An electroacupuncture efficacy detection system for premature ovarian failure patients based on medical images, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for detecting the electroacupuncture efficacy of premature ovarian failure patients based on medical images according to any one of claims 1-9.

Images