Obstetrical image intelligent processing method

Abstract

The invention relates to the technical field of fetal ultrasonic image enhancement, in particular to an intelligent processing method for an obstetrical image. The method comprises the steps of collecting a fetus ultrasonic image; obtaining a brain probability factor, a heart probability factor and an abdomen probability factor of the pixel points by combining gray distribution and gradient distribution of the pixel points in a preset direction according to good regional gray features and gradient features of the brain tissue, the heart tissue and the abdomen tissue in the image, and further obtaining a brain region, a heart region and an abdomen region; and obtaining an image matching degree according to the probability factor difference and the position distribution in different images so as to enhance the images. According to the method, the real form of the key feature area of the fetus can be well restored, the influence of interference factors on the image precision is reduced, and related personnel can diagnose the fetus condition more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of fetal ultrasonic image enhancement, and in particular to an intelligent processing method for obstetric images. Background Art

[0002] In the field of medical image processing, the processing of obstetric images has always been an important research direction. As the most common imaging examination method in obstetrics, ultrasound examination can obtain many key information about the fetus and pregnant women in the early and middle stages of pregnancy, including embryonic development, organ health, amniotic fluid volume, placental function, etc. Although ultrasound examination has the advantages of being non-invasive, convenient, and real-time, its imaging quality is often affected by many factors. The obtained ultrasound images have low resolution and are subject to noise and artifacts, and they also need to rely on certain image processing operations to enhance the image quality.

[0003] Using ultrasound images to diagnose whether the fetus has developmental abnormalities is one of its important functions, but poor image quality may result in unclear imaging of some tissues or structures of the fetus, making it difficult for doctors to make accurate judgments. Non-local mean filtering is a processing method commonly used for ultrasound image enhancement. This method uses the pixel values ​​of similar areas in the image to perform weighted averaging on each pixel, thereby effectively removing noise and retaining image details. This method has certain advantages in ultrasound image enhancement, but when applied to fetal ultrasound images, the algorithm will blur the edge structure to a certain extent, and the filtering effect on complex areas is not strong, making some subtle structures too blurred or distorted, which ultimately leads to misdiagnosis and missed diagnosis when relevant personnel analyze key fetal features (such as ventricle size, heart morphology, etc.), and the processing effect is not ideal. Summary of the invention

[0004] In order to solve the technical problem that when enhancing fetal ultrasound images using a non-local mean filtering algorithm, the edge structure will be blurred, making some subtle structures too blurred or distorted, and ultimately leading to misdiagnosis or missed diagnosis when relevant personnel analyze key fetal features, and the processing effect is not ideal, the purpose of the present invention is to provide an intelligent processing method for obstetric images, and the technical scheme adopted is as follows: an intelligent processing method for obstetric images, the method comprises: collecting a preset number of fetal ultrasound images; all the fetal ultrasound images constitute a fetal ultrasound image set; any pixel point in any fetal ultrasound image is selected as a reference pixel point of a reference fetal image; according to the regional grayscale features and regional gradient features of brain tissue, heart tissue and abdominal tissue in the reference fetal image respectively, and the gradient distribution and grayscale distribution of the reference pixel point in all preset directions, obtain the reference pixel point belonging to the brain group The invention relates to a method for enhancing the image matching degree of a fetal image by using a brain probability factor of the reference pixel belonging to the heart tissue, a heart probability factor of the reference pixel belonging to the heart tissue, and an abdominal probability factor of the reference pixel belonging to the abdominal tissue; obtaining the brain region, heart region and abdominal region in each fetal ultrasound image according to the brain probability factor, the heart probability factor and the abdominal probability factor of each pixel in each fetal ultrasound image; the brain region, the heart region and the abdominal region are used as the key feature regions in the fetal ultrasound image; obtaining the image matching degree between the reference fetal image and each other fetal ultrasound image according to the difference in position distribution of the key feature regions in the reference fetal image and each other fetal ultrasound image, and the difference in brain probability factor, heart probability factor and abdominal probability factor of the pixels in the key feature regions; enhancing the reference fetal image according to the image matching degree between the reference fetal image and each other fetal ultrasound image.

[0005] Furthermore, the method for obtaining the brain probability factor includes: obtaining the possibility that each preset direction of the reference pixel point can cross the edge of the brain tissue according to the gradient distribution of the reference pixel point in each preset direction; taking the preset direction whose possibility of crossing the edge of the brain tissue is greater than a preset first threshold as the brain tissue judgment direction; counting the number of brain tissue judgment directions in the preset direction; and obtaining the brain probability factor according to the brain probability factor calculation formula, wherein the brain probability factor calculation formula is as follows: In the formula, The brain probability factor indicating that the reference pixel belongs to brain tissue; Indicates the number of directions judged by brain tissue at the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the reference pixel a preset first number of pixel points in a preset direction; Indicates the reference pixel The first The grayscale difference between a pixel and the next pixel.

[0006] Furthermore, the method for obtaining the possibility that each preset direction of the reference pixel point can cross the edge of the brain tissue includes: obtaining the possibility according to a possibility calculation formula, and the possibility calculation formula is as follows: In the formula, Indicates the reference pixel The degree to which a given direction can cross the edge of brain tissue; Indicates the reference pixel The maximum value of the gradient in a preset direction; Indicates the reference pixel The second largest value of the gradient in the preset direction; represents the maximum value of the gradient on the reference fetal image; Indicates the reference pixel The distance between the pixel corresponding to the maximum gradient value and the pixel corresponding to the second maximum gradient value in a preset direction; Represents the normalization function.

[0007] Furthermore, the method for obtaining the heart probability factor includes: calculating the possibility that each preset direction of the reference pixel point can cross the edge of the heart tissue; obtaining the heart tissue judgment direction of the reference pixel point according to the possibility that each preset direction of the reference pixel point can cross the edge of the heart tissue; taking the midpoint of the pixel points corresponding to the two maximum gradient values ​​of the reference pixel point in each preset direction as the symmetry center point of the reference pixel point in each preset direction; obtaining the reference symmetric point of the reference pixel point in each preset direction according to the position of the reference pixel point and the symmetry center point in each preset direction; obtaining the heart tissue judgment direction of the reference symmetric point according to the possibility that each preset direction of the reference symmetric point can cross the edge of the heart tissue; obtaining the heart probability factor according to the heart probability factor calculation formula, and the heart probability factor calculation formula is as follows: In the formula, A cardiac probability factor indicating that the reference pixel belongs to cardiac tissue; Indicates the number of cardiac tissue determination directions of the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the reference symmetry point The number of cardiac tissue determination directions in the preset directions.

[0008] Furthermore, the method for obtaining the abdominal probability factor includes: obtaining the abdominal tissue judgment direction of the reference pixel point according to the possible degree of each preset direction of the reference pixel point being able to cross the edge of the abdominal tissue; using the Otsu method to divide the preset area of ​​the reference pixel point into two parts to obtain all high gray value areas; obtaining the abdominal probability factor according to the abdominal probability factor calculation formula, and the abdominal probability factor calculation formula is as follows: In the formula, Abdominal probability factor indicating that the reference pixel belongs to abdominal tissue; Indicates the number of abdominal tissue determination directions of the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the number of high gray value areas within the preset area of ​​the reference pixel point; Indicates the number of pixels in each high gray value area within the preset area of ​​the reference pixel; Indicates the reference pixel point in the preset area. In the high gray value area The gray value of each pixel; Indicates the maximum area of ​​the high grayscale value region within the preset area of ​​the reference pixel point.

[0009] Furthermore, the method for acquiring the key feature area includes: taking the pixel points whose brain probability factor is greater than the preset second threshold as brain tissue pixel points; taking the pixel points whose heart probability factor is greater than the preset second threshold as heart tissue pixel points; taking the pixel points whose abdomen probability factor is greater than the preset second threshold as abdominal tissue pixel points; taking the area composed of all brain tissue pixel points as the brain area; taking the area composed of all heart tissue pixel points as the heart area; taking the area composed of all abdominal tissue pixel points as the abdominal area; taking the brain area, the heart area and the abdominal area as the key feature areas.

[0010] Furthermore, the method for obtaining the image matching degree includes: taking any fetal ultrasound image other than the reference fetal image as a comparison fetal image; obtaining the key feature similarity between the reference fetal image and the comparison fetal image based on the brain probability factor difference, heart probability factor difference and abdomen probability factor difference of the pixel points in the key feature area between the reference fetal image and the comparison fetal image; obtaining the body structure similarity between the reference fetal image and the comparison fetal image based on the key feature area position distribution difference between the reference fetal image and the comparison fetal image; the brain probability factor, heart probability factor and abdomen probability factor in the key feature area are collectively referred to as probability factors; calculating the probability factor mean of all pixel points in all key feature areas in the reference fetal image; obtaining the image matching degree according to the image matching degree calculation formula, and the image matching degree calculation formula is as follows: In the formula, Indicates the degree of image matching between the reference fetal image and the comparison fetal image; Indicates the similarity of key features between the reference fetal image and the comparison fetal image; Represents the mean probability factor of all pixels in all key feature areas in the reference fetal image; Indicates a preset second threshold; Indicates the similarity of body structure between the reference fetal image and the comparison fetal image.

[0011] Furthermore, the method for obtaining the key feature similarity includes: calculating the mean difference in brain probability factors of pixels in the brain area between the reference fetal image and the comparison fetal image, calculating the mean difference in heart probability factors of pixels in the heart area between the reference fetal image and the comparison fetal image, and calculating the mean difference in abdominal probability factors of pixels in the abdominal area between the reference fetal image and the comparison fetal image; averaging the mean difference in brain probability factors, the mean difference in heart probability factors, and the mean difference in abdominal probability factors between the reference fetal image and the comparison fetal image, and performing a negative correlation normalization operation to obtain the key feature similarity between the reference fetal image and the comparison fetal image.

[0012] Furthermore, the method for obtaining the body structure similarity includes: obtaining the center of gravity of the brain region, the center of gravity of the heart region, and the center of gravity of the abdomen region of the reference fetal image and the comparison fetal image; in each fetal ultrasound image, obtaining the line between the center of gravity of the brain region and the center of gravity of the heart region as the first line, and obtaining the line between the center of gravity of the heart region and the center of gravity of the abdomen region as the second line; taking the angle between the first line and the second line as the first angle; obtaining the body structure similarity according to the body structure similarity calculation formula between the reference fetal image and the comparison fetal image, and the body structure similarity calculation formula is as follows: In the formula, Indicates the similarity of body structure between the reference fetal image and the comparison fetal image; represents the length of the first line in the reference fetal image; represents the length of the second line in the reference fetal image; represents the length of the first line in the contrast fetal image; represents the length of the second line in the contrast fetal image; represents the first angle in the reference fetal image; Indicates the first angle in the contrast fetal image.

[0013] Furthermore, the reference fetal image is enhanced according to the degree of image matching between the reference fetal image and each other fetal ultrasound image, including: calculating the degree of image matching between the reference fetal image and each other fetal ultrasound image in the fetal ultrasound image set; selecting a preset first number of other fetal ultrasound images with the greatest image matching degree as reference filtered images; and performing non-local mean filtering on the reference fetal image according to the preset first number of reference filtered images to obtain an enhanced fetal image.

[0014] The present invention has the following beneficial effects: the present invention collects a preset number of fetal ultrasound images; since in the fetal ultrasound images, the fetal brain, heart, spine and gastrointestinal area are important image areas for discovering potential abnormalities of the fetus, it is very important for relevant personnel to accurately judge the fetal condition, so it is necessary to accurately extract the brain area, heart area and abdomen area in the fetal ultrasound image, and use them as key feature areas; since the brain area, heart area and abdomen area each have relatively obvious regional grayscale features and regional gradient features, the gradient distribution and grayscale distribution of the pixel point in all preset directions are combined to obtain the brain probability factor and the reference pixel point belonging to the brain tissue. Consider the cardiac probability factor that the pixel belongs to cardiac tissue and the abdominal probability factor that the reference pixel belongs to abdominal tissue; because the conventional non-local mean filtering algorithm will make the important anatomical structures of the fetus with a large number of significant edges blurred and difficult to identify, the key feature areas of the reference fetal image are matched in the fetal ultrasound image set to achieve non-local mean filtering of multiple images; and the similarity of the fetal shape is calculated according to the proportion and angle of the line segments formed by the centers of these three areas, and the image performance similarity of the three key feature areas is obtained by combining the comprehensive probability factor of each area, and finally the image matching degree between every two fetal images is obtained. The present invention can better restore the true morphology of the key feature areas of the fetus, reduce the impact of interference factors on image accuracy, and enable relevant personnel to diagnose the fetal condition more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 A flow chart of an intelligent processing method for obstetric images provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the intelligent processing method of obstetric images proposed by the present invention, its specific implementation, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0018] Unless defined otherwise, 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 belongs.

[0019] The specific scheme of the intelligent processing method for obstetric images provided by the present invention is described in detail below with reference to the accompanying drawings.

[0020] See also Figure 1 , which shows an intelligent processing method for obstetric images provided by an embodiment of the present invention, the method comprising: step S1: collecting a preset number of fetal ultrasound images; all fetal ultrasound images constitute a fetal ultrasound image set.

[0021] The embodiment of the present invention is mainly applied to the fetal ultrasound image enhancement scenario, so the fetal ultrasound image of the parturient is firstly acquired; in order to facilitate the subsequent enhancement of the fetal image through similar structures in different fetal ultrasound images, the embodiment of the present invention collects a preset number of fetal ultrasound images.

[0022] In one embodiment of the present invention, an ultrasound probe is used to obtain ultrasound images of fetuses of different parturients, and the images are uploaded to an image processing module in an ultrasound device for processing, and finally stored in a corresponding image storage module. It should be noted that the preset number is set to 100, and the preset number can be set by oneself, and is not limited here.

[0023] Step S2: Select any pixel point in any fetal ultrasound image as a reference pixel point of the reference fetal image; obtain a brain probability factor that the reference pixel point belongs to the brain tissue, a heart probability factor that the reference pixel point belongs to the heart tissue, and an abdominal probability factor that the reference pixel point belongs to the abdominal tissue according to the regional grayscale features and regional gradient features of the brain tissue, heart tissue and abdominal tissue in the reference fetal image, respectively, and the gradient distribution and grayscale distribution of the reference pixel point in all preset directions; obtain the brain area, heart area and abdominal area in each fetal ultrasound image according to the brain probability factor, heart probability factor and abdominal probability factor of each pixel point in each fetal ultrasound image; the brain area, heart area and abdominal area are used as key feature areas in the fetal ultrasound image.

[0024] In fetal ultrasound images, the fetal brain, heart, spine and gastrointestinal tract are important image areas for discovering potential fetal abnormalities. It is very important for relevant personnel to accurately judge the condition of the fetus, so it is necessary to accurately extract the brain area, heart area and abdominal area in the fetal ultrasound image.

[0025] In the fetal brain region, the intraventricular fluid (such as cerebrospinal fluid) reflects ultrasound weakly, so it appears as a low grayscale area in the fetal ultrasound image; while the brain tissue is mainly composed of water and soft tissue, ultrasound is reflected and scattered more in these areas, so it appears as a high grayscale area in the fetal ultrasound image. In addition, the acoustic characteristics of the intraventricular fluid are relatively uniform, and the pixel distribution is relatively consistent; while the surface of the brain tissue has more details (such as gyri and sulci), and the pixel distribution is richer. Therefore, in an embodiment of the present invention, the brain probability factor that the reference pixel belongs to the brain tissue can be obtained based on the regional grayscale characteristics and regional gradient characteristics of the brain tissue in the reference fetal image, as well as the gradient distribution and grayscale distribution of the reference pixel in all preset directions.

[0026] Preferably, in one embodiment of the present invention, the method for obtaining the brain probability factor includes: obtaining the possibility that each preset direction of the reference pixel can cross the edge of the brain tissue according to the gradient distribution of the reference pixel in each preset direction; in one embodiment of the present invention, the preset direction is set to the four directions of the reference pixel, and the preset direction can be set by itself, which is not limited here. The possibility degree calculation formula is as follows: In the formula, Indicates the reference pixel The degree to which a given direction can cross the edge of brain tissue; Indicates the reference pixel The maximum value of the gradient in a preset direction; Indicates the reference pixel The second largest value of the gradient in the preset direction; represents the maximum value of the gradient on the reference fetal image; Indicates the reference pixel The distance between the pixel corresponding to the maximum gradient value and the pixel corresponding to the second maximum gradient value in a preset direction; Represents the normalization function.

[0027] In the probability calculation formula, the reference pixel The maximum value of the gradient in the preset direction With the second largest value of the gradient They correspond to the outer edge of the brain tissue and the inner edge of the ventricle, respectively, and the gradient values ​​of the pixels on the two edges are close to the maximum gradient value in the fetal ultrasound image, so The larger the The smaller the size, the closer each pixel point of the chain is to the distance between the outer edge of the brain tissue and the edge of the ventricular wall in reality, indicating that the The greater the probability of crossing the edge of brain tissue in a preset direction.

[0028] The preset direction whose possibility of crossing the edge of brain tissue is greater than the preset first threshold is used as the brain tissue judgment direction; in one embodiment of the present invention, the preset first threshold is set to 0.7. It should be noted that the preset first threshold can be set by oneself and is not limited here.

[0029] The number of brain tissue judgment directions in the preset direction is counted; the brain probability factor is obtained according to the brain probability factor calculation formula, and the brain probability factor calculation formula is as follows: In the formula, The brain probability factor indicating that the reference pixel belongs to brain tissue; Indicates the number of directions judged by brain tissue at the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the reference pixel a preset first number of pixel points in a preset direction; Indicates the reference pixel The first The grayscale difference between a pixel and the next pixel.

[0030] In the brain probability factor calculation formula, The larger the average grayscale change value in the preset direction, the greater the The richer the grayscale changes in the preset directions, the more the average grayscale change values ​​of all preset directions are analyzed to obtain , The larger it is, the richer the grayscale changes in all preset directions of the reference pixel, and the more likely the reference pixel is to be located in the brain tissue; the larger the proportion of the number of brain tissue judgment directions of the reference pixel, the more likely the reference pixel is to be located in the brain tissue; the greater the brain probability factor of the reference pixel belonging to the brain tissue.

[0031] The fetal heart is smaller than the brain and is usually located in the middle of the body. The grayscale distribution of the heart region is very similar to that of the brain. The myocardium outside the heart shows clear boundaries and has a higher grayscale. The interior is mainly composed of four chambers (ventricles and atria) with a lower grayscale. However, the heart chambers are symmetrical in pairs, and adjacent chambers are basically symmetrical in position and size. Therefore, in an embodiment of the present invention, the heart probability factor that the reference pixel belongs to the heart tissue is obtained based on the regional grayscale characteristics and regional gradient characteristics of the heart tissue in the reference fetal image, as well as the gradient distribution and grayscale distribution of the reference pixel in all preset directions.

[0032] Preferably, in one embodiment of the present invention, the method for obtaining the heart probability factor includes: calculating the possibility that each preset direction of the reference pixel point can cross the edge of the heart tissue according to the same calculation steps as the above-mentioned brain probability factor; and using the preset direction whose possibility of crossing the edge of the heart tissue is greater than a preset first threshold as the heart tissue judgment direction of the reference pixel point.

[0033] The midpoint of the pixel points corresponding to the two maximum gradient values ​​of the reference pixel point in each preset direction is used as the symmetry center point of the reference pixel point in each preset direction; the reference symmetry point of the reference pixel point in each preset direction is obtained according to the position of the reference pixel point and the symmetry center point in each preset direction; the heart tissue judgment direction of the reference symmetry point is obtained according to the possibility of the reference symmetry point crossing the edge of the heart tissue in each preset direction; the heart tissue judgment direction of the reference symmetry point is obtained according to the same screening method.

[0034] The heart probability factor is obtained according to the heart probability factor calculation formula. The heart probability factor calculation formula is as follows: In the formula, A cardiac probability factor indicating that the reference pixel belongs to cardiac tissue; Indicates the number of cardiac tissue determination directions of the reference pixel; Indicates the number of preset directions of reference pixels; Indicates The number of cardiac tissue determination directions of reference symmetric points in a preset direction.

[0035] In the abdominal probability factor calculation formula, if the number of cardiac tissue judgment directions of the reference pixel point is greater, it means that the reference pixel point is more likely to be in the cardiac chamber; The more the number of cardiac tissue judgment directions of the reference symmetric points in the preset directions is, the The more likely the reference symmetric point in the preset direction is to be in the heart chamber, the reference symmetric points of the reference pixel point in all preset directions are analyzed to obtain , The larger it is, the more likely it is that the reference pixel point is in the heart chamber, and the greater the cardiac probability factor that the reference pixel point belongs to the heart tissue.

[0036] The abdomen is located in the lower part of the fetal body, and the lower part of the abdomen is mainly covered by the intestines, so the fetal intestines are identified to represent the lower part of the entire body. The intestines in the ultrasound image also have similar characteristics to the brain and heart, that is, the periphery is the intestinal wall with high grayscale values, and the interior is amniotic fluid or other fluids with low grayscale values. In addition, the intestines of the fetus may accumulate a certain amount of gas during development, and the echo of the gas is strong, so the intestinal area will show more irregular shapes of high grayscale value areas, mixed in low grayscale value liquid areas. Therefore, in the embodiment of the present invention, the abdominal probability factor that the reference pixel belongs to the abdominal tissue is obtained based on the regional grayscale characteristics and regional gradient characteristics of the abdominal tissue in the reference fetal image, and the gradient distribution and grayscale distribution of the reference pixel in all preset directions.

[0037] Preferably, in one embodiment of the present invention, the method for obtaining the abdominal probability factor includes: calculating the possibility that each preset direction of the reference pixel point can cross the edge of the abdominal tissue according to the same calculation steps as the above-mentioned brain probability factor; and taking the preset direction whose possibility of crossing the edge of the abdominal tissue is greater than a preset first threshold as the abdominal tissue judgment direction of the reference pixel point.

[0038] By using the Otsu method to perform a binary segmentation on the preset area of ​​the reference pixel point, all areas of the reference fetal image can be divided into two types of areas, wherein the type area with a higher gray value is used as a high gray value area, and the type area with a lower gray value is used as a low gray value area, thereby obtaining all high gray value areas; in the embodiment of the present invention, the preset area is set to be centered on the reference pixel point, The preset area can be set voluntarily and is not limited here. Among them, the Otsu method is a technical means well known to those skilled in the art and will not be described in detail here.

[0039] The abdominal probability factor is obtained according to the abdominal probability factor calculation formula. The abdominal probability factor calculation formula is as follows: In the formula, Abdominal probability factor indicating that the reference pixel belongs to abdominal tissue; Indicates the number of abdominal tissue determination directions of the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the number of high gray value areas within the preset area of ​​the reference pixel point; Indicates the number of pixels in each high gray value area within the preset area of ​​the reference pixel; Indicates the reference pixel point in the preset area. In the high gray value area The gray value of each pixel; Indicates the maximum area of ​​the high grayscale value region within the preset area of ​​the reference pixel point.

[0040] In the abdominal probability factor calculation formula, if the number of abdominal tissue judgment directions of the reference pixel point is greater, it means that the reference pixel point is more likely to be in the abdominal area, and the abdominal probability factor of the reference pixel point belonging to the abdominal tissue is greater. Since the gray value of the gas area in the fetal intestine is very high, the more high gray value areas in the preset area, that is, The larger it is, the larger the grayscale value of the high grayscale value area is. At this time, it is more likely that there is intestinal gas in the preset area of ​​the reference pixel point, and the reference pixel point is more likely to be in the abdominal area, that is, the greater the abdominal probability factor of the reference pixel point belonging to the abdominal tissue; the product of the number of high grayscale value areas and the maximum area of ​​the high grayscale value area shows that the reference pixel point is not located in the high grayscale value area with a small number of brain areas, and the maximum area of ​​the high grayscale value area shows that the reference pixel point is not a bright noise pixel point, so The larger it is, the greater the abdominal probability factor that the reference pixel belongs to the abdominal tissue.

[0041] Preferably, in one embodiment of the present invention, the method for acquiring the key feature area includes: taking the pixel points whose brain probability factor is greater than the preset second threshold as brain tissue pixel points; taking the pixel points whose heart probability factor is greater than the preset second threshold as heart tissue pixel points; taking the pixel points whose abdomen probability factor is greater than the preset second threshold as abdominal tissue pixel points; in one embodiment of the present invention, the preset second threshold is set to 0.8, and the preset second threshold can be set by itself and is not limited here.

[0042] The area composed of all brain tissue pixels is taken as the brain area; the area composed of all heart tissue pixels is taken as the heart area; the area composed of all abdominal tissue pixels is taken as the abdominal area; the brain area, heart area and abdominal area are taken as key feature areas.

[0043] Step S3: Obtain the degree of image matching between the reference fetal image and each other fetal ultrasound image based on the difference in position distribution of key feature areas within the reference fetal image and each other fetal ultrasound image, as well as the difference in brain probability factors, heart probability factors and abdomen probability factors of the pixels within the key feature areas; enhance the reference fetal image based on the degree of image matching between the reference fetal image and each other fetal ultrasound image.

[0044] Since the conventional non-local mean filtering algorithm uses the pixel areas with similar grayscale in the same fetal ultrasound image to filter each other to achieve the denoising effect, this method will make the important anatomical structures of the fetus with a large number of significant edges blurred and difficult to identify; therefore, after the key feature areas of the reference fetal image are preliminarily located, the fetal ultrasound images similar to it are matched in the fetal ultrasound image set by the key feature areas of the reference fetal image to achieve non-local mean filtering of multiple images. It is known that the brain, heart, and abdomen refer to the upper, middle, and lower parts of the fetal body respectively. Therefore, the similarity of the fetal body is calculated based on the ratio and angle of the line segments formed by the centers of these three areas, and the image performance similarity of the three key feature areas is obtained by combining the comprehensive probability factor of each area, and finally the image matching degree between every two fetal images is obtained.

[0045] Preferably, in one embodiment of the present invention, the method for acquiring the image matching degree includes: taking any fetal ultrasound image other than the reference fetal image as a comparison fetal image.

[0046] Calculate the mean difference of brain probability factors of pixels in the brain region of the reference fetal image and the comparison fetal image, calculate the mean difference of heart probability factors of pixels in the heart region of the reference fetal image and the comparison fetal image, and calculate the mean difference of abdominal probability factors of pixels in the abdominal region of the reference fetal image and the comparison fetal image; average the mean difference of brain probability factors, the mean difference of heart probability factors, and the mean difference of abdominal probability factors between the reference fetal image and the comparison fetal image, and perform a negative correlation normalization operation to obtain the key feature similarity between the reference fetal image and the comparison fetal image. In one embodiment of the present invention, the key feature similarity calculation formula is as follows: In the formula, represents the similarity of key features between the reference fetal image and the comparison fetal image; represents the mean difference of brain probability factors of pixels in the brain region between the reference fetal image and the comparison fetal image; represents the mean difference of cardiac probability factors of pixels in the cardiac region between the reference fetal image and the comparison fetal image; Represents the mean difference in the abdominal probability factor of the pixels in the abdominal region of the reference fetal image and the comparison fetal image.

[0047] In the key feature similarity calculation formula, the smaller the mean difference in brain probability factors, the mean difference in heart probability factors, and the mean difference in abdomen probability factors between the reference fetal image and the comparison fetal image, the higher the similarity in the key feature areas of the two images, that is, the greater the key feature similarity between the reference fetal image and the comparison fetal image.

[0048] The center of gravity of the brain region, the center of gravity of the heart region, and the center of gravity of the abdomen region of the reference fetal image and the comparison fetal image are obtained; in each fetal ultrasound image, the line between the center of gravity of the brain region and the center of gravity of the heart region is obtained as the first line, and the line between the center of gravity of the heart region and the center of gravity of the abdomen region is obtained as the second line; the angle between the first line and the second line is taken as the first angle.

[0049] The body structure similarity is obtained according to the body structure similarity calculation formula between the reference fetal image and the comparison fetal image. The body structure similarity calculation formula is as follows: In the formula, Indicates the similarity of body structure between the reference fetal image and the comparison fetal image; represents the length of the first line in the reference fetal image; represents the length of the second line in the reference fetal image; represents the length of the first line in the contrast fetal image; represents the length of the second line in the contrast fetal image; represents the first angle in the reference fetal image; Indicates the first angle in the contrast fetal image.

[0050] In the body structure similarity calculation formula, the ratio between the first connecting line and the second connecting line of the reference fetal image is , and the ratio between the first and second lines of the fetal image The smaller the difference between the reference fetal image and the comparison fetal image, the greater the similarity in the overall body proportions of the fetus, and the greater the body structure similarity between the reference fetal image and the comparison fetal image; the smaller the difference in the first angle between the reference fetal image and the comparison fetal image, the more similar the relative positions of the reference fetal image and the comparison fetal image in the key feature area are, and the greater the body structure similarity between the reference fetal image and the comparison fetal image.

[0051] The brain probability factor, heart probability factor and abdomen probability factor in the key feature area are collectively referred to as probability factors; the probability factor mean of all pixel points in all key feature areas in the reference fetal image is calculated.

[0052] The image matching degree is obtained according to the image matching degree calculation formula. The image matching degree calculation formula is as follows: In the formula, Indicates the degree of image matching between the reference fetal image and the comparison fetal image; represents the similarity of key features between the reference fetal image and the comparison fetal image; Represents the mean probability factor of all pixels in all key feature areas in the reference fetal image; Indicates a preset second threshold; Indicates the similarity of body structure between the reference fetal image and the comparison fetal image.

[0053] In the image matching degree calculation formula, It means that the portion of the average probability factor exceeding the preset second threshold accounts for the maximum value that can be exceeded, that is, The larger the ratio between them, the stronger the regional feature performance of the fetus in the reference fetal image in the key feature area. At this time, the greater the weight should be given to the key feature similarity between the reference fetal image and the comparison fetal image, and the smaller the weight should be given to the similarity of the body structure between the reference fetal image and the comparison fetal image, that is, The smaller, the Used to indicate the degree of image matching between the reference fetal image and the comparison fetal image.

[0054] Preferably, in one embodiment of the present invention, the reference fetal image is enhanced according to the degree of image matching between the reference fetal image and each other fetal ultrasound image, including: calculating the degree of image matching between the reference fetal image and each other fetal ultrasound image in the fetal ultrasound image set; selecting a preset first number of other fetal ultrasound images with the greatest degree of image matching as reference filtered images; in this embodiment of the present invention, the preset first number is set to 3. It should be noted that the preset first number can be set voluntarily and is not limited here.

[0055] The reference fetal image is subjected to non-local mean filtering according to a preset first number of reference filtered images, the matching range is expanded to three similar fetal ultrasound images, and a window search is performed on the four fetal ultrasound images when filtering each pixel point of the reference fetal image, so that each pixel point can search for a window with high similarity, thereby improving the denoising effect and better retaining the details, and being able to handle large-scale noise problems such as artifacts, so as to obtain an enhanced fetal image.

[0056] Visualizing enhanced fetal images on display devices helps relevant personnel better analyze fetal development.

[0057] In summary, a preset number of fetal ultrasound images are collected; all fetal ultrasound images constitute a fetal ultrasound image set; any pixel point in any fetal ultrasound image is selected as a reference pixel point of a reference fetal image; according to the regional grayscale characteristics and regional gradient characteristics of brain tissue, heart tissue and abdominal tissue in the reference fetal image, respectively, and the gradient distribution and grayscale distribution of the reference pixel point in all preset directions, the brain probability factor that the reference pixel point belongs to the brain tissue, the heart probability factor that the reference pixel point belongs to the heart tissue and the abdominal probability factor that the reference pixel point belongs to the abdominal tissue are obtained; according to the brain probability factor of each pixel point in each fetal ultrasound image, The heart probability factor and the abdomen probability factor are used to obtain the brain area, heart area and abdomen area in each fetal ultrasound image; the brain area, heart area and abdomen area are used as the key feature areas in the fetal ultrasound image; according to the difference in position distribution of the key feature areas in the reference fetal image and each other fetal ultrasound image, as well as the difference in brain probability factor, heart probability factor and abdomen probability factor of the pixel points in the key feature area, the image matching degree between the reference fetal image and each other fetal ultrasound image is obtained; according to the image matching degree between the reference fetal image and each other fetal ultrasound image, the reference fetal image is enhanced.

[0058] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0059] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent processing method for obstetric images, characterized in that: The method comprises: collecting a preset number of fetal ultrasound images; all the fetal ultrasound images form a fetal ultrasound image set; selecting any pixel point in a fetal ultrasound image as a reference pixel point of a reference fetal image; obtaining a brain probability factor that the reference pixel point belongs to the brain tissue, a heart probability factor that the reference pixel point belongs to the heart tissue, and an abdominal probability factor that the reference pixel point belongs to the abdominal tissue according to the regional grayscale features and regional gradient features of the brain tissue, the heart tissue, and the abdominal tissue in the reference fetal image, respectively, and the gradient distribution and grayscale distribution of the reference pixel point in all preset directions; and obtaining a brain probability factor that the reference pixel point belongs to the heart tissue according to the brain probability factor of each pixel point in each fetal ultrasound image. , the heart probability factor and the abdomen probability factor, to obtain the brain area, heart area and abdomen area in each fetal ultrasound image; the brain area, heart area and abdomen area are used as the key feature areas in the fetal ultrasound image; according to the difference in position distribution of the key feature areas in the reference fetal image and each other fetal ultrasound image, as well as the difference in brain probability factor, heart probability factor and abdomen probability factor of the pixel points in the key feature area, the image matching degree between the reference fetal image and each other fetal ultrasound image is obtained; according to the image matching degree between the reference fetal image and each other fetal ultrasound image, the reference fetal image is enhanced.

2. The intelligent processing method for obstetric images according to claim 1, characterized in that: The method for obtaining the brain probability factor includes: obtaining the possibility that each preset direction of the reference pixel point can cross the edge of the brain tissue according to the gradient distribution of the reference pixel point in each preset direction; taking the preset direction whose possibility of crossing the edge of the brain tissue is greater than a preset first threshold as the brain tissue judgment direction; counting the number of brain tissue judgment directions in the preset direction; and obtaining the brain probability factor according to the brain probability factor calculation formula, wherein the brain probability factor calculation formula is as follows: In the formula, The brain probability factor indicating that the reference pixel belongs to brain tissue; Indicates the number of directions judged by brain tissue at the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the reference pixel a preset first number of pixel points in a preset direction; Indicates the reference pixel The first The grayscale difference between a pixel and the next pixel.

3. The intelligent processing method for obstetric images according to claim 2, characterized in that: The method for obtaining the possibility degree that each preset direction of the reference pixel point can cross the edge of the brain tissue includes: obtaining the possibility degree according to a possibility degree calculation formula, and the possibility degree calculation formula is as follows: In the formula, Indicates the reference pixel The degree to which a given direction can cross the edge of brain tissue; Indicates the reference pixel The maximum value of the gradient in a preset direction; Indicates the reference pixel The second largest value of the gradient in the preset direction; represents the maximum value of the gradient on the reference fetal image; Indicates the reference pixel The distance between the pixel corresponding to the maximum gradient value and the pixel corresponding to the second maximum gradient value in a preset direction; Represents the normalization function.

4. The intelligent processing method for obstetric images according to claim 1, characterized in that: The method for obtaining the heart probability factor includes: calculating the possibility that each preset direction of the reference pixel point can cross the edge of the heart tissue; obtaining the heart tissue judgment direction of the reference pixel point according to the possibility that each preset direction of the reference pixel point can cross the edge of the heart tissue; taking the midpoint of the pixel points corresponding to the two maximum gradient values ​​of the reference pixel point in each preset direction as the symmetry center point of the reference pixel point in each preset direction; obtaining the reference symmetric point of the reference pixel point in each preset direction according to the position of the reference pixel point and the symmetry center point in each preset direction; obtaining the heart tissue judgment direction of the reference symmetric point according to the possibility that each preset direction of the reference symmetric point can cross the edge of the heart tissue; obtaining the heart probability factor according to the heart probability factor calculation formula, and the heart probability factor calculation formula is as follows: In the formula, A cardiac probability factor indicating that the reference pixel belongs to cardiac tissue; Indicates the number of cardiac tissue determination directions of the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the reference symmetry point The number of cardiac tissue determination directions in the preset directions.

5. The intelligent processing method for obstetric images according to claim 1, characterized in that: The method for obtaining the abdominal probability factor includes: obtaining the abdominal tissue judgment direction of the reference pixel point according to the possibility that each preset direction of the reference pixel point can cross the edge of the abdominal tissue; using the Otsu method to divide the preset area of ​​the reference pixel point into two parts to obtain all high gray value areas; obtaining the abdominal probability factor according to the abdominal probability factor calculation formula, and the abdominal probability factor calculation formula is as follows: In the formula, Abdominal probability factor indicating that the reference pixel belongs to abdominal tissue; Indicates the number of abdominal tissue determination directions of the reference pixel; Indicates the number of preset directions of reference pixels; Indicates the number of high gray value areas within the preset area of ​​the reference pixel point; Indicates the number of pixels in each high gray value area within the preset area of ​​the reference pixel; Indicates the reference pixel point in the preset area. In the high gray value area The gray value of each pixel; Indicates the maximum area of ​​the high grayscale value region within the preset area of ​​the reference pixel point.

6. The intelligent processing method for obstetric images according to claim 1, characterized in that: The method for acquiring the key feature area includes: taking the pixel points whose brain probability factor is greater than the preset second threshold as brain tissue pixel points; taking the pixel points whose heart probability factor is greater than the preset second threshold as heart tissue pixel points; taking the pixel points whose abdomen probability factor is greater than the preset second threshold as abdominal tissue pixel points; taking the area composed of all brain tissue pixel points as the brain area; taking the area composed of all heart tissue pixel points as the heart area; taking the area composed of all abdominal tissue pixel points as the abdominal area; taking the brain area, the heart area and the abdominal area as the key feature areas.

7. The intelligent processing method for obstetric images according to claim 1, characterized in that: The method for obtaining the image matching degree includes: taking any fetal ultrasound image other than the reference fetal image as a comparison fetal image; obtaining the key feature similarity between the reference fetal image and the comparison fetal image according to the brain probability factor difference, heart probability factor difference and abdomen probability factor difference of the pixel points in the key feature area between the reference fetal image and the comparison fetal image; obtaining the body structure similarity between the reference fetal image and the comparison fetal image according to the key feature area position distribution difference between the reference fetal image and the comparison fetal image; the brain probability factor, heart probability factor and abdomen probability factor in the key feature area are collectively referred to as probability factors; calculating the probability factor mean of all pixel points in all key feature areas in the reference fetal image; obtaining the image matching degree according to the image matching degree calculation formula, and the image matching degree calculation formula is as follows: In the formula, Indicates the degree of image matching between the reference fetal image and the comparison fetal image; Indicates the similarity of key features between the reference fetal image and the comparison fetal image; Represents the mean probability factor of all pixels in all key feature areas in the reference fetal image; Indicates a preset second threshold; Indicates the similarity of body structure between the reference fetal image and the comparison fetal image.

8. The intelligent processing method for obstetric images according to claim 7, characterized in that: The method for obtaining the key feature similarity includes: calculating the mean difference of brain probability factors of pixels in the brain area between the reference fetal image and the comparison fetal image, calculating the mean difference of heart probability factors of pixels in the heart area between the reference fetal image and the comparison fetal image, and calculating the mean difference of abdominal probability factors of pixels in the abdominal area between the reference fetal image and the comparison fetal image; averaging the mean difference of brain probability factors, the mean difference of heart probability factors and the mean difference of abdominal probability factors between the reference fetal image and the comparison fetal image, and performing a negative correlation normalization operation to obtain the key feature similarity between the reference fetal image and the comparison fetal image.

9. The intelligent processing method for obstetric images according to claim 7, characterized in that: The method for obtaining the body structure similarity includes: obtaining the center of gravity of the brain region, the center of gravity of the heart region, and the center of gravity of the abdomen region of the reference fetal image and the comparison fetal image; in each fetal ultrasound image, obtaining the line between the center of gravity of the brain region and the center of gravity of the heart region as the first line, and obtaining the line between the center of gravity of the heart region and the center of gravity of the abdomen region as the second line; taking the angle between the first line and the second line as the first angle; obtaining the body structure similarity according to the body structure similarity calculation formula between the reference fetal image and the comparison fetal image, and the body structure similarity calculation formula is as follows: In the formula, Indicates the similarity of body structure between the reference fetal image and the comparison fetal image; represents the length of the first line in the reference fetal image; represents the length of the second line in the reference fetal image; represents the length of the first line in the contrast fetal image; represents the length of the second line in the contrast fetal image; represents the first angle in the reference fetal image; Indicates the first angle in the contrast fetal image.

10. The intelligent processing method for obstetric images according to claim 1, characterized in that: The reference fetal image is enhanced according to the degree of image matching between the reference fetal image and each other fetal ultrasound image, including: calculating the degree of image matching between the reference fetal image and each other fetal ultrasound image in the fetal ultrasound image set; selecting a preset first number of other fetal ultrasound images with the greatest degree of image matching as reference filtered images; and performing non-local mean filtering on the reference fetal image according to the preset first number of reference filtered images to obtain an enhanced fetal image.

Images