Enhanced analysis method for medical images of obstetrics and gynecology department
By performing regional division and connection domain analysis on uterine B-ultrasound images, combining abnormalities and distribution complexity, and using different colors for enhancement, the problem of insufficient processing accuracy under the global enhancement method is solved and the image analysis effect is improved.
Patent Information
- Application Number
- CN202411578801.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-07
- Publication Date
- 2024-12-03
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, the global enhancement method of uterine B-ultrasound images leads to insufficient processing accuracy, poor image enhancement effect, and poor analysis effect, especially when multiple abnormal features in the uterus exist.
By obtaining the grayscale histogram of the uterine B-ultrasound image, it is divided into uterus, ovary and fallopian tube areas, the connection domain is analyzed, the initial abnormality and distribution complexity of each connection domain are calculated, and the influence coefficient and indicators to be enhanced are combined. Use different colors for image enhancement.
The processing accuracy and image analysis effect of uterine B-ultrasound images are improved, errors caused by unified processing are avoided, and the visualization effect of the image is enhanced, making abnormal characteristics in different regions more obvious.
Smart Images

Figure CN119067859A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image enhancement, and in particular to an enhanced analysis method for gynecological and obstetric medical images. Background Art
[0002] In recent years, B-ultrasound has become an important part of medical imaging, and more and more medical examinations are based on B-ultrasound. The means of enhancing B-ultrasound images are constantly updated, and the main ones used are histogram equalization and pseudo-color image processing.
[0003] Uterine B-ultrasound examination is particularly important in obstetrics and gynecology examinations. In related technologies, uterine B-ultrasound images are enhanced based on global enhancement. Since the B-ultrasound images in the uterus involve multiple types of abnormal features, high-precision B-ultrasound images are required. However, performing global enhancement on them will most likely lose image details, resulting in insufficient processing accuracy of uterine B-ultrasound images and poor image enhancement effects. In addition, global enhancement will make the enhanced image more blurred, resulting in poor image analysis effects. Summary of the invention
[0004] In order to solve the technical problems of insufficient processing accuracy, poor image enhancement effect, and poor image analysis effect when using global enhancement for processing in the related art, the present invention provides an enhancement analysis method for obstetrics and gynecology medical images, and the technical scheme adopted is as follows: The present invention proposes an enhanced analysis method for obstetrics and gynecology medical images, the method comprising: Acquire a uterine B-ultrasound image to be processed, obtain a grayscale histogram of the uterine B-ultrasound image, and divide the uterine B-ultrasound image to be processed into three areas to be tested according to the distribution of grayscale values in the grayscale histogram, wherein the areas to be tested include a uterine area, an ovarian area, and a fallopian tube area; Performing a connected domain analysis on the area to be tested to obtain a connected domain, determining the initial abnormality of the connected domain according to the grayscale features of each connected domain, the grayscale features of all connected domains connected to the connected domain, and the grayscale features of the area to be tested where the connected domain is located; determining the distribution complexity of each pixel in the connected domain according to the grayscale distribution of all pixels within a preset neighborhood range centered on each pixel in the connected domain, wherein the grayscale feature is the grayscale value of all pixels in the corresponding interval; Determine the influence coefficient of the connected domain according to the distribution complexity of all pixels in the connected domain and the initial abnormality of the connected domain; determine the index to be enhanced of the tested area by combining the influence coefficients of all connected domains in any tested area; determine the abnormality index of the connected domain according to the influence coefficient of the connected domain and the index to be enhanced of the tested area where the connected domain is located; Based on the abnormal index, each connected domain is enhanced using different colors to obtain an enhanced image.
[0005] Furthermore, the uterine B-ultrasound image to be processed is divided into three areas to be tested according to the distribution of gray values in the gray histogram, including: The two most prominent peak points in the grayscale histogram are respectively recorded as the first peak point and the second peak point, wherein the grayscale value corresponding to the first peak point is smaller than the grayscale value corresponding to the second peak point; Taking the first peak point as the starting point, based on the grayscale histogram, the grayscale levels are gradually reduced from the first peak point to the left and the pixels are marked as the pixels to be fitted at the corresponding grayscale levels. Ellipse fitting is performed on each level of pixels to be fitted, and the ovarian area is determined according to the fitting results; Performing regional fitting on the pixel points between the first peak point and the second peak point, and taking the maximum area obtained by the regional fitting as the uterus area; The area excluding the uterine area and the ovarian area was regarded as the fallopian tube area.
[0006] Further, determining the ovarian region according to the fitting result includes: The fitting coefficient of the ellipse fitting was calculated, and the fitting was stopped when the fitting coefficient reached the maximum value, and the corresponding ellipse area was taken as the ovarian area.
[0007] Furthermore, determining the initial abnormality degree of the connected domain according to the grayscale features of each connected domain, the grayscale features of all connected domains connected to the connected domain, and the grayscale features of the area to be tested in which the connected domain is located, includes: Take any connected domain as the block to be analyzed, and calculate the gray value mean of the block to be analyzed as the mean value to be analyzed; All other connected domains adjacent to the block to be analyzed are regarded as adjacent blocks, and the gray value mean of each adjacent block is calculated as the adjacent mean; Calculate the variance of the mean to be analyzed and the adjacent mean as the first abnormal coefficient; Calculate the mean gray value of all pixels in the tested area where the connected domain is located as the regional mean; Calculate the absolute value of the difference between the mean to be analyzed and the regional mean, and perform normalization processing on the absolute value of the difference to obtain a second abnormal coefficient; An initial abnormality degree of the connected domain is obtained according to the first abnormality coefficient and the second abnormality coefficient.
[0008] Further, obtaining the initial abnormality degree of the connected domain according to the first abnormality coefficient and the second abnormality coefficient includes: A normalized value of the product of the first abnormality coefficient and the second abnormality coefficient is calculated to obtain an initial abnormality degree of the connected domain.
[0009] Furthermore, determining the distribution complexity of each pixel point in the connected domain according to the grayscale distribution of all pixels within a preset neighborhood range centered on each pixel point in the connected domain includes: Calculate the grayscale value variance of all pixels within the preset neighborhood as the neighborhood variance; Calculate the number of all pixels within the preset neighborhood whose grayscale values are less than the center point as the number of anomalies; and use the ratio of the number of anomalies to the number of all pixels within the preset neighborhood as the distribution complexity weight; The normalized value of the product of the neighborhood variance and the distribution complexity weight is used as the distribution complexity.
[0010] Furthermore, determining the influence coefficient of the connected domain according to the distribution complexity of all pixels in the connected domain and the initial abnormality degree of the connected domain includes: The sum of squares of grayscale deviations from the mean of the complexity of the distribution of all pixels in the connected domain is calculated based on the formula for the sum of squares of deviations from the mean as the influence weight; The normalized value of the product of the influence weight and the initial abnormality degree is calculated as the influence coefficient of the connected domain.
[0011] Furthermore, combining the influence coefficients of all connected domains in any region to be tested to determine the index to be enhanced of the region to be tested includes: The sum of squares of grayscale deviations from the mean of the influence coefficients of all connected domains in the tested area is calculated based on the formula for the sum of squares of grayscale deviations from the mean, and the sum of squares of grayscale deviations from the mean is normalized to obtain the enhancement index of the tested area.
[0012] Furthermore, the influence coefficient of the connected domain and the to-be-enhanced index of the tested area where the connected domain is located are both positively correlated with the abnormal index of the connected domain, and the value of the abnormal index of the connected domain is a normalized value.
[0013] Furthermore, based on the abnormal index, each connected domain is enhanced using different colors to obtain an enhanced image, including: Dividing the abnormal indicators into different indicator ranges based on a preset division threshold; Different colors are used to enhance the connected domains within different index ranges to obtain enhanced images.
[0014] The present invention has the following beneficial effects: The present invention is applied to the field of image enhancement technology. By obtaining a grayscale histogram of a uterine B-ultrasound image, and according to the distribution of grayscale values in the grayscale histogram, the uterine B-ultrasound image to be processed is divided into a uterine region, an ovarian region and a fallopian tube region. By analyzing different regions to be tested respectively, it is possible to effectively perform adaptive image processing on different regions to be tested according to the echo characteristics and reflection ability differences of different tissues, avoid errors caused by unified processing, and improve processing accuracy. Based on the grayscale features of a connected domain in the region to be tested, the grayscale features of all connected domains connected to the connected domain, and the grayscale features of the region to be tested where the connected domain is located, an abnormality analysis is performed on the connected domain, and a complexity analysis is performed on each pixel point based on the grayscale distribution of the neighborhood of the pixel point in the connected domain. Combined with the abnormality analysis, the image processing method can be used to analyze the abnormality of the connected domain. The initial abnormal degree obtained by analysis and the distribution complexity obtained by complexity analysis are used to determine the influence coefficient of the connected domain, which can effectively analyze the distribution of the connected domain as a whole and the individual pixel points in the connected domain from multiple angles, ensure that the influence coefficient can accurately characterize the importance of the connected domain itself, and further obtain the indicators to be enhanced in the area to be tested, which can ensure the accuracy and reliability of the indicators to be enhanced; combine the influence coefficient and the indicators to be enhanced to determine the abnormal indicators of the connected domain, and then based on the abnormal indicators, use different colors to perform color enhancement on each connected domain to obtain an enhanced image, and perform color enhancement processing of different color tones on different connected domains through abnormal indicators, so that the image analysis results can be intuitively displayed according to the different colors in the enhanced image, and the image analysis effect is enhanced. In summary, the present invention can adaptively perform image analysis based on different regions, effectively improve the processing accuracy of the uterine B-ultrasound image to be processed, and then improve the image enhancement effect, so that the obtained enhanced image has a better image analysis effect. 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 enhanced analysis method for obstetrics and gynecology medical 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 enhanced analysis method of obstetrics and gynecology medical images proposed by the present invention, its specific implementation method, structure, characteristics 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 enhanced analysis method of obstetrics and gynecology medical images provided by the present invention is described in detail below with reference to the accompanying drawings.
[0020] See also Figure 1 , which shows a flow chart of an enhanced analysis method for obstetrics and gynecology medical images provided by an embodiment of the present invention, the method comprising: S101: Acquire a uterine B-ultrasound image to be processed, obtain a grayscale histogram of the uterine B-ultrasound image, and divide the uterine B-ultrasound image to be processed into three areas to be tested according to the distribution of grayscale values in the grayscale histogram, wherein the areas to be tested include a uterine area, an ovarian area, and a fallopian tube area.
[0021] A specific use scenario of the present invention is to perform image analysis on the uterine B-ultrasound image to be processed according to the distribution of pixel points in the uterine B-ultrasound image to be processed, thereby assisting in improving the visual effect of the uterine B-ultrasound image to be processed.
[0022] In an embodiment of the present invention, a corresponding B-ultrasound detection instrument can be used to perform B-ultrasound detection on the patient's uterus, and then, an original uterine B-ultrasound image is obtained, and the original uterine B-ultrasound image is semantically segmented and background-removed to obtain a uterine B-ultrasound image to be processed. It can be understood that since the B-ultrasound image itself is a grayscale image, the grayscale distribution analysis of the uterine B-ultrasound image to be processed can be directly performed to obtain a grayscale histogram of the uterine B-ultrasound image to be processed.
[0023] Among them, the grayscale histogram shows the grayscale distribution of pixel points in the uterine B-ultrasound image to be processed. In the grayscale histogram, the peak point is the point with the most grayscale distribution. The embodiment of the present invention can divide the uterine B-ultrasound image to be processed according to the grayscale characteristics of the uterine area, ovarian area and fallopian tube area in the B-ultrasound image.
[0024] Further, in some embodiments of the present invention, according to the distribution of grayscale values in the grayscale histogram, the uterine B-ultrasound image to be processed is divided into three areas to be tested, including: the two most prominent peak points in the grayscale histogram are respectively recorded as the first peak point and the second peak point, wherein the grayscale value corresponding to the first peak point is less than the grayscale value corresponding to the second peak point; taking the first peak point as the starting point, based on the grayscale histogram, starting from the first peak point, the grayscale level decreases to the left and marks the pixel points as the pixel points to be fitted at the corresponding grayscale level, performing ellipse fitting on each level of the pixel points to be fitted, and calculating the corresponding fitting coefficient, stopping fitting when the fitting coefficient reaches the maximum value, and taking the corresponding ellipse area as the ovarian area; performing regional fitting on the pixel points between the first peak point and the second peak point, and taking the maximum area obtained by regional fitting as the uterine area; taking the area other than the uterine area and the ovarian area as the fallopian tube area.
[0025] It should be noted that, according to the imaging characteristics of B-ultrasound, the echo characteristics and reflection ability differences of different tissues will lead to different grayscale values. Therefore, the grayscale values presented in the uterine B-ultrasound image are different depending on the hardness of the uterus, ovaries and fallopian tubes: the follicles or cysts in the ovaries are often fluid in nature, appearing as a larger echo-free area with a relatively low grayscale value; the fallopian tubes are usually difficult to display directly in B-ultrasound images. When there is fluid accumulation in the fallopian tubes, the fallopian tubes can appear as a thin, echo-free or low-grayscale structure; the endometrium is highly echoic and appears as an echo-free area in the uterus, but the grayscale value of the uterus is higher than that of the ovaries and fallopian tubes.
[0026] Among them, since the uterus occupies a large area in the uterine B-ultrasound image, it is manifested as two most prominent peak points in the grayscale histogram. Therefore, in an embodiment of the present invention, the two most prominent peak points in the grayscale histogram are respectively recorded as the first peak point and the second peak point, wherein the grayscale value corresponding to the first peak point is less than the grayscale value corresponding to the second peak point, and the pixel points between the first peak point and the second peak point are regionally fitted, and the area obtained by regional fitting is used as the uterine area.
[0027] Among them, since the grayscale value of ovarian tissue is usually lower than that of the uterus, the grayscale value distribution of the pixels in the ovarian area in the grayscale histogram is mostly concentrated on the left side of the uterine area. Therefore, the embodiment of the present invention uses the first peak point as the maximum value of the grayscale interval of the ovarian area. Since the shape of the ovarian area is usually elliptical or oval, this embodiment uses the grayscale corresponding to the first peak point as the starting grayscale, and starts to decrease grayscale to the left based on the grayscale corresponding to the first peak point and mark the pixels. For each additional grayscale pixel, the least squares method is used to perform ellipse fitting on the marked pixels, and the fitting coefficient of each fitting is calculated. When the fitting coefficient just reaches the maximum value, the marking is stopped, and the area formed by the marked pixels at this time is the ovarian area. Among them, ellipse fitting is a fitting method well known to those skilled in the art. When the fitting coefficient is the largest, it can be said that the fitting of the corresponding pixel is closest to the elliptical shape, and thus, the corresponding area is taken as the ovarian area.
[0028] In this embodiment, the background area has been eliminated through threshold segmentation, and the uterine area and ovarian area have been obtained. Since the fallopian tube is usually difficult to display directly in the B-ultrasound image and difficult to judge through the grayscale histogram, this embodiment will exclude the uterine area and ovarian area. The area is recorded as the fallopian tube area.
[0029] Thus, the uterine B-ultrasound image to be processed is divided into three test areas, namely, the uterine area, the ovarian area and the fallopian tube area. Then, the three test areas can be analyzed respectively.
[0030] S102: Performing a connected domain analysis on the area to be tested to obtain a connected domain, determining the initial abnormality degree of the connected domain according to the grayscale characteristics of each connected domain, the grayscale characteristics of all connected domains connected to the connected domain, and the grayscale characteristics of the area to be tested in which the connected domain is located; determining the distribution complexity of each pixel in the connected domain according to the grayscale distribution of all pixels within a preset neighborhood with each pixel in the connected domain as the center, wherein the grayscale characteristic is the grayscale value of all pixels in the corresponding interval.
[0031] In the embodiment of the present invention, connected domain analysis is a technology well known in the art, which divides pixels with the same or similar grayscale values into a connected domain to facilitate image analysis.
[0032] Further, in some embodiments of the present invention, the initial abnormality degree of the connected domain is determined according to the grayscale features of each connected domain, the grayscale features of all connected domains connected to the connected domain, and the grayscale features of the area to be tested in which the connected domain is located, including: taking any connected domain as a block to be analyzed, and calculating the grayscale value mean of the block to be analyzed as the mean to be analyzed; taking all other connected domains adjacent to the block to be analyzed as adjacent blocks, and respectively calculating the grayscale value mean of each adjacent block as the adjacent mean; calculating the variance of the mean to be analyzed and the adjacent mean as the first abnormality coefficient; calculating the grayscale value mean of all pixels in the area to be tested in which the connected domain is located as the regional mean; calculating the absolute value of the difference between the mean to be analyzed and the regional mean, and normalizing the absolute value of the difference to obtain the second abnormality coefficient; and obtaining the initial abnormality degree of the connected domain according to the first abnormality coefficient and the second abnormality coefficient.
[0033] In one embodiment of the present invention, the normalization processing can be specifically, for example, maximum and minimum value normalization processing, and the normalization in subsequent steps can all adopt maximum and minimum value normalization processing. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of numerical values, which will not be described in detail.
[0034] Among them, any connected domain is taken as a block to be analyzed, and the gray value mean of all pixels in the block to be analyzed is calculated as the mean value to be analyzed, and the mean value to be analyzed can be used as the gray feature of the block to be analyzed.
[0035] In the embodiment of the present invention, all other connected domains adjacent to the block to be analyzed are regarded as adjacent blocks, wherein adjacent means having a partial boundary connected to the block to be analyzed. In the embodiment of the present invention, those satisfying this condition are regarded as adjacent blocks. Since the connected domains are distributed relatively evenly under normal circumstances, and in abnormal areas, their abnormal conditions will be different from those of the surrounding connected domains, the local area can be accurately analyzed overall through the adjacent blocks. That is to say, in the embodiment of the present invention, the variance of the mean to be analyzed and the adjacent mean is calculated as the first abnormality coefficient. The larger the first abnormality coefficient is, the greater the possibility of abnormality of the corresponding block to be analyzed is.
[0036] In an embodiment of the present invention, the absolute value of the difference between the mean to be analyzed and the regional mean is calculated, and the absolute value of the difference is normalized to obtain a second anomaly coefficient, wherein, when the absolute value of the difference is larger, it can be indicated that the connected domain has a larger difference from the region to be measured, and the difference may be an abnormal difference. Therefore, the absolute value of the difference is normalized to obtain the second anomaly coefficient.
[0037] In the embodiment of the present invention, after obtaining the first abnormality coefficient and the second abnormality coefficient, the initial abnormality degree of the connected domain can be obtained by combining the first abnormality coefficient and the second abnormality coefficient.
[0038] Further, in some embodiments of the present invention, obtaining the initial abnormality degree of the connected domain according to the first abnormality coefficient and the second abnormality coefficient includes: calculating a normalized value of the product of the first abnormality coefficient and the second abnormality coefficient to obtain the initial abnormality degree of the connected domain.
[0039] Since the larger the first abnormality coefficient and the second abnormality coefficient are, the more abnormal the corresponding connected domain is, therefore, the first abnormality coefficient and the second abnormality coefficient are both positively correlated with the abnormality degree of the connected domain, the present invention calculates the product of the first abnormality coefficient and the second abnormality coefficient, and performs normalization processing to obtain the initial abnormality degree.
[0040] It is understandable that it is not possible to accurately and effectively analyze abnormal situations based only on the grayscale values of pixels in each connected domain and the local grayscale distribution. Therefore, the embodiment of the present invention introduces the distribution complexity for comprehensive analysis.
[0041] Furthermore, in some embodiments of the present invention, the distribution complexity of each pixel in the connected domain is determined based on the grayscale distribution of all pixels in a preset neighborhood with each pixel in the connected domain as the center, including: calculating the grayscale value variance of all pixels in the preset neighborhood as the neighborhood variance; calculating the number of all pixels in the preset neighborhood whose grayscale values are less than the center point as the number of anomalies; taking the ratio of the number of anomalies to the number of all pixels in the preset neighborhood as the distribution complexity weight; and taking the normalized value of the product of the neighborhood variance and the distribution complexity weight as the distribution complexity.
[0042] The preset neighborhood range may be, for example, an eight-neighborhood range, or other neighborhood ranges, such as a 24-neighborhood range, may be selected according to actual needs, and there is no limitation to this.
[0043] In the embodiment of the present invention, it can be understood that since the abnormal area usually presents a fuzzy grayscale characteristic in the B-ultrasound image, and the grayscale value is lower than that in the normal situation, therefore, in the embodiment of the present invention, the distribution complexity is analyzed by calculating the neighborhood variance and the number of anomalies, that is, the ratio of the number of anomalies to the number of all pixels within the preset neighborhood range is used as the distribution complexity weight; the normalized value of the product of the neighborhood variance and the distribution complexity weight is used as the distribution complexity.
[0044] Among them, the distribution complexity represents the complexity of the corresponding pixel distribution. The embodiment of the present invention calculates the distribution complexity through the grayscale variance of the surrounding neighborhood pixels and the number of surrounding neighborhood pixels with low grayscale values. The greater the distribution complexity, the more abnormal the local range of the pixel is. Therefore, the embodiment of the present invention can analyze the connected domain in combination with the distribution complexity of the pixel and the initial abnormality of the connected domain. For the specific analysis process, please refer to the subsequent embodiments.
[0045] S103: determining the influence coefficient of the connected domain according to the distribution complexity of all pixels in the connected domain and the initial abnormality of the connected domain; determining the index to be enhanced of the tested area by combining the influence coefficients of all connected domains in any tested area; determining the abnormality index of the connected domain according to the influence coefficient of the connected domain and the index to be enhanced of the tested area where the connected domain is located.
[0046] Furthermore, in some embodiments of the present invention, the influence coefficient of the connected domain is determined based on the distribution complexity of all pixels in the connected domain and the initial abnormality degree of the connected domain, including: calculating the sum of squared grayscale deviations from the mean of the distribution complexity of all pixels in the connected domain based on the sum of squared deviations from the mean formula as the influence weight; and calculating the normalized value of the product of the influence weight and the initial abnormality degree as the influence coefficient of the connected domain.
[0047] That is to say, in an embodiment of the present invention, the mean square sum of the grayscale mean square sum of the distribution complexity of all pixels in the connected domain can be used to calculate the influence weight, wherein the mean square sum of the grayscale mean square sum of the distribution complexity of all pixels in the connected domain is a calculation method well known to those skilled in the art. By calculating the mean square sum of the grayscale mean square sum of the distribution complexity of all pixels in the connected domain, the deviation of the distribution complexity from the average value can be quantified to obtain a more accurate and effective influence weight. Then, the influence coefficient of the connected domain is calculated based on the influence weight and the initial abnormality. Since the larger the influence weight, the more serious the deviation of the distribution complexity of all pixels in the corresponding connected domain, and since the larger the initial abnormality, the more serious the abnormal features in the corresponding connected domain, therefore, the present invention calculates the normalized value of the product of the influence weight and the initial abnormality, and uses the normalized value as the influence coefficient of the connected domain. The larger the influence coefficient, the greater the influence of the area corresponding to the connected domain on the overall analysis result, and the more attention it needs to pay.
[0048] Furthermore, in some embodiments of the present invention, the index to be enhanced of the area to be tested is determined in combination with the influence coefficients of all connected domains in any area to be tested, including: calculating the sum of squared grayscale deviations from the mean of the influence coefficients of all connected domains in the area to be tested based on the sum of squared grayscale deviations from the mean formula, and normalizing the sum of squared grayscale deviations from the mean to obtain the index to be enhanced of the area to be tested.
[0049] After obtaining the influence coefficient of each connected domain, the influence coefficients of all connected domains in the tested area can be combined to perform an overall analysis of the tested area. In a normal tested area, due to the differences in echo characteristics and reflection capabilities of different tissues, it is impossible to perform a unified image analysis. The embodiment of the present invention analyzes the connected domains of different tested areas separately to avoid the error caused by the overall analysis. That is to say, in the process of image enhancement, by calculating the enhanced index of each tested area separately for different tested areas, it is possible to facilitate the subsequent adaptive image enhancement of each tested area and improve the image enhancement effect.
[0050] Further, the embodiment of the present invention can combine the influence coefficient of the connected domain and the index to be enhanced of the area to be tested where the connected domain is located, and analyze to obtain the abnormal index of the connected domain. It can be understood that the influence coefficient of the connected domain and the index to be enhanced of the area to be tested where the connected domain is located are both positively correlated with the abnormal index of the connected domain, and the value of the abnormal index of the connected domain is a normalized value.
[0051] Among them, a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. The specific relationship can be a multiplication relationship, an addition relationship, or the power of an exponential function, which is determined by actual application; a negative correlation means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It can be a subtraction relationship, a division relationship, etc., which is determined by actual application.
[0052] S104: Based on the abnormality index, each connected domain is enhanced using different colors to obtain an enhanced image.
[0053] Furthermore, in some embodiments of the present invention, based on the abnormal indicators, different colors are used to perform color enhancement on each connected domain to obtain an enhanced image, including: dividing the abnormal indicators into different indicator ranges based on a preset division threshold; using different colors to perform color enhancement on the connected domains within different indicator ranges to obtain an enhanced image.
[0054] Among them, the preset division threshold is the endpoint value of the division range of the abnormal index. The preset division threshold can be specifically 0.25, 0.5, and 0.75, for example. That is to say, the abnormal index is divided into four index ranges: [0, 0.25), [0.25, 0.5), [0.5, 0.75), and [0.75, 1]. Different abnormal indicators represent the abnormal state of the corresponding connected domain. The larger the abnormal index, the more abnormal the corresponding connected domain. Therefore, the present invention can perform color enhancement on different indicator ranges based on different colors.
[0055] For example, in an embodiment of the present invention, the uterine B-ultrasound image to be processed can be converted into an image in red-green-blue (RGB) format, and cyan is used for color enhancement in the connected domain within the index range of [0, 0.25), green is used for color enhancement in the connected domain within the index range of [0.25, 0.5), yellow is used for color enhancement in the connected domain within the index range of [0.5, 0.75), and red is used for color enhancement in the connected domain within the index range of [0.75, 1]. The enhancement process is mainly to improve the tone curve of the corresponding color in the corresponding connected domain, so that the image as a whole presents the corresponding tone information. Through image enhancement processing, the abnormal area is made more obvious, thereby improving the convenience of image analysis.
[0056] The present invention is applied to the field of image enhancement technology. By obtaining a grayscale histogram of a uterine B-ultrasound image, and according to the distribution of grayscale values in the grayscale histogram, the uterine B-ultrasound image to be processed is divided into a uterine region, an ovarian region and a fallopian tube region. By analyzing different regions to be tested respectively, it is possible to effectively perform adaptive image processing on different regions to be tested according to the echo characteristics and reflection ability differences of different tissues, avoid errors caused by unified processing, and improve processing accuracy. Based on the grayscale features of a connected domain in the region to be tested, the grayscale features of all connected domains connected to the connected domain, and the grayscale features of the region to be tested where the connected domain is located, an abnormality analysis is performed on the connected domain, and a complexity analysis is performed on each pixel point based on the grayscale distribution of the neighborhood of the pixel point in the connected domain. Combined with the abnormality analysis, the image processing method can be used to analyze the abnormality of the connected domain. The initial abnormal degree obtained by analysis and the distribution complexity obtained by complexity analysis are used to determine the influence coefficient of the connected domain, which can effectively analyze the distribution of the connected domain as a whole and the individual pixel points in the connected domain from multiple angles, ensure that the influence coefficient can accurately characterize the importance of the connected domain itself, and further obtain the indicators to be enhanced in the area to be tested, which can ensure the accuracy and reliability of the indicators to be enhanced; combine the influence coefficient and the indicators to be enhanced to determine the abnormal indicators of the connected domain, and then based on the abnormal indicators, use different colors to perform color enhancement on each connected domain to obtain an enhanced image, and perform color enhancement processing of different color tones on different connected domains through abnormal indicators, so that the image analysis results can be intuitively displayed according to the different colors in the enhanced image, and the image analysis effect is enhanced. In summary, the present invention can adaptively perform image analysis based on different regions, effectively improve the processing accuracy of the uterine B-ultrasound image to be processed, and then improve the image enhancement effect, so that the obtained enhanced image has a better image analysis effect.
[0057] 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.
[0058] 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 enhanced analysis method for obstetrics and gynecology medical images, characterized in that: The method comprises: Acquire a uterine B-ultrasound image to be processed, obtain a grayscale histogram of the uterine B-ultrasound image, and divide the uterine B-ultrasound image to be processed into three areas to be tested according to the distribution of grayscale values in the grayscale histogram, wherein the areas to be tested include a uterine area, an ovarian area, and a fallopian tube area; Performing a connected domain analysis on the area to be tested to obtain a connected domain, determining the initial abnormality of the connected domain according to the grayscale features of each connected domain, the grayscale features of all connected domains connected to the connected domain, and the grayscale features of the area to be tested where the connected domain is located; determining the distribution complexity of each pixel in the connected domain according to the grayscale distribution of all pixels within a preset neighborhood range centered on each pixel in the connected domain, wherein the grayscale feature is the grayscale value of all pixels in the corresponding interval; Determine the influence coefficient of the connected domain according to the distribution complexity of all pixels in the connected domain and the initial abnormality of the connected domain; determine the index to be enhanced of the tested area by combining the influence coefficients of all connected domains in any tested area; determine the abnormality index of the connected domain according to the influence coefficient of the connected domain and the index to be enhanced of the tested area where the connected domain is located; Based on the abnormal indicators, each connected domain is enhanced using different colors to obtain an enhanced image; Determining the initial abnormality degree of the connected domain according to the grayscale features of each connected domain, the grayscale features of all connected domains connected to the connected domain, and the grayscale features of the area to be tested where the connected domain is located, comprises: Take any connected domain as the block to be analyzed, and calculate the gray value mean of the block to be analyzed as the mean value to be analyzed; All other connected domains adjacent to the block to be analyzed are regarded as adjacent blocks, and the gray value mean of each adjacent block is calculated as the adjacent mean; Calculate the variance of the mean to be analyzed and the adjacent mean as the first abnormal coefficient; Calculate the mean gray value of all pixels in the tested area where the connected domain is located as the regional mean; Calculate the absolute value of the difference between the mean to be analyzed and the regional mean, and perform normalization processing on the absolute value of the difference to obtain a second abnormal coefficient; Obtaining an initial abnormality degree of the connected domain according to the first abnormality coefficient and the second abnormality coefficient; The step of obtaining the initial abnormality degree of the connected domain according to the first abnormality coefficient and the second abnormality coefficient includes: Calculating a normalized value of the product of the first abnormality coefficient and the second abnormality coefficient to obtain an initial abnormality degree of the connected domain; Determining the distribution complexity of each pixel in the connected domain according to the grayscale distribution of all pixels in a preset neighborhood with each pixel in the connected domain as the center includes: Calculate the grayscale value variance of all pixels within the preset neighborhood as the neighborhood variance; Calculate the number of all pixels within the preset neighborhood whose grayscale values are less than the center point as the number of anomalies; and use the ratio of the number of anomalies to the number of all pixels within the preset neighborhood as the distribution complexity weight; The normalized value of the product of the neighborhood variance and the distribution complexity weight is used as the distribution complexity.
2. The enhanced analysis method for obstetrics and gynecology medical images according to claim 1, characterized in that: According to the distribution of gray values in the gray histogram, the uterine B-ultrasound image to be processed is divided into three areas to be tested, including: The two most prominent peak points in the grayscale histogram are respectively recorded as the first peak point and the second peak point, wherein the grayscale value corresponding to the first peak point is smaller than the grayscale value corresponding to the second peak point; Taking the first peak point as the starting point, based on the grayscale histogram, the grayscale levels are gradually reduced from the first peak point to the left and the pixels are marked as the pixels to be fitted at the corresponding grayscale levels. Ellipse fitting is performed on each level of pixels to be fitted, and the ovarian area is determined according to the fitting results; Performing regional fitting on the pixel points between the first peak point and the second peak point, and taking the maximum area obtained by the regional fitting as the uterus area; The area excluding the uterine area and the ovarian area was regarded as the fallopian tube area.
3. The enhanced analysis method for obstetrics and gynecology medical images according to claim 2, characterized in that: Determining the ovarian region according to the fitting result includes: The fitting coefficient of the ellipse fitting was calculated, and the fitting was stopped when the fitting coefficient reached the maximum value, and the corresponding ellipse area was taken as the ovarian area.
4. The enhanced analysis method for obstetrics and gynecology medical images according to claim 1, characterized in that: The method of determining the influence coefficient of the connected domain according to the distribution complexity of all pixels in the connected domain and the initial abnormality degree of the connected domain includes: The sum of squares of grayscale deviations from the mean of the complexity of the distribution of all pixels in the connected domain is calculated based on the formula for the sum of squares of deviations from the mean as the influence weight; The normalized value of the product of the influence weight and the initial abnormality degree is calculated as the influence coefficient of the connected domain.
5. The enhanced analysis method for obstetrics and gynecology medical images according to claim 1, characterized in that: The step of combining the influence coefficients of all connected domains in any region to be tested to determine the index to be enhanced in the region to be tested includes: The sum of squares of grayscale deviations from the mean of the influence coefficients of all connected domains in the tested area is calculated based on the formula for the sum of squares of grayscale deviations from the mean, and the sum of squares of grayscale deviations from the mean is normalized to obtain the enhancement index of the tested area.
6. The enhanced analysis method for obstetrics and gynecology medical images according to claim 1, characterized in that: The influence coefficient of the connected domain and the to-be-enhanced index of the tested area where the connected domain is located are both positively correlated with the abnormal index of the connected domain, and the value of the abnormal index of the connected domain is a normalized value.
7. The enhanced analysis method for obstetrics and gynecology medical images according to claim 1, characterized in that: Based on the abnormal index, each connected domain is enhanced by using different colors to obtain an enhanced image, including: Dividing the abnormal indicators into different indicator ranges based on a preset division threshold; Different colors are used to enhance the connected domains within different index ranges to obtain enhanced images.
Citation Information
Cited By
Obstetrics and gynecology image analysis system based on deep learning
CN121837841A
