A method for segmenting the lesion area in kidney ultrasound images

Abstract

The present invention relates to the technical field of regional segmentation, and specifically relates to a method for segmenting lesion regions in kidney ultrasound images. This method obtains the tracking pixel sequence of the feature points in the kidney region of the image to be analyzed, obtains the cyst probability value based on the similarity of the gray-scale change trend between each feature point and its neighboring feature points and the length of the tracking pixel sequence, and then selects the cyst feature points in the kidney region; according to the distance, gray-scale difference and similarity of the gray-scale change trend of the pixel points at the same position in the tracking pixel sequences of any two cyst feature points, the cyst feature points are classified; the regional growth threshold is obtained based on the distance between each cyst feature point and the remaining cyst feature points in its category and the cyst probability value, and the cyst feature points are used as seed points for regional growth to obtain the actual cyst region. This solution effectively improves the segmentation accuracy of cysts in the kidney.

Description

Technical Field

[0001] The present invention relates to the technical field of region segmentation, and particularly relates to a method for segmenting lesion regions in kidney ultrasound images. Background Art

[0002] There are various lesions in the kidney, and the most common lesion is renal cyst. Renal cysts can cause infectious peritonitis, induce hypertension, cause renal failure and canceration, etc. Segmenting the lesion regions in kidney images helps to diagnose common malignant cysts in the urinary system such as renal cysts more quickly and accurately, and is of great significance for the early detection and prevention of kidney diseases.

[0003] Existing methods usually use region segmentation algorithms and neural networks and other methods to extract the renal tumor regions in kidney images. However, due to the fact that the size, shape, quantity and position of renal cysts in the body are often different, and the images of renal cysts and the surrounding renal cortex regions are relatively similar, resulting in blurred cyst edges and difficult to perform precise segmentation, thus there are large errors in the lesion regions segmented from kidney images. Summary of the Invention

[0004] In order to solve the technical problem that tumors in the kidney are not fixed and the tumor edges are blurred, resulting in large errors in the segmentation of cyst regions in the kidney, the purpose of the present invention is to provide a method for segmenting lesion regions in kidney ultrasound images, and the specific technical solution adopted is as follows:

[0005] The present invention proposes a method for segmenting lesion regions in kidney ultrasound images, and the method includes:

[0006] Obtain a continuous number of frames of kidney images of a patient, and record the first frame of kidney image as the image to be analyzed;

[0007] Obtain the kidney region of the image to be analyzed; select feature points from the kidney region, track each feature point in the kidney image, and form a tracking pixel sequence for each feature point from each feature point and its corresponding matching pixel points in the remaining kidney images except the image to be analyzed;

[0008] According to the similarity of the gray change trend between each feature point in the kidney region and its neighboring feature points, and the length of the tracking pixel sequence of each feature point, obtain the cyst probability value of each feature point; select cyst feature points in the kidney region by using the cyst probability value;

[0009] According to the distance, gray difference and similarity of gray change trend of the pixel points at the same position in the tracking pixel sequences of any two cyst feature points, divide the cyst feature points into different categories;

[0010] Obtain the regional growth threshold for each cyst feature point based on the distance between each cyst feature point and the remaining cyst feature points of its category, as well as the possible cyst value of each cyst feature point; based on the regional growth threshold, use each category of cyst feature points as seed points for regional growth to obtain the actual cyst region in the kidney region.

[0011] Further, the method for obtaining the tracking pixel sequence of each feature point includes:

[0012] Arrange all kidney images in chronological order to obtain an image sequence, and the first image in the image sequence is the image to be analyzed;

[0013] For each feature point in the kidney region of the image to be analyzed, set an initial target sequence, and the elements in the initial target sequence are the feature points themselves;

[0014] Respectively record the feature point as the target point, and record the second image in the image sequence as the target image; use the optical flow algorithm to track the target point in the target image. When there is a corresponding matching pixel point for the target point in the target image, record the corresponding matching pixel point of the target point in the target image as the new target point, add the new target point to the target sequence, update the target sequence, and record the adjacent next kidney image of the target image as the new target image until there is no corresponding matching pixel point for the new target point in the new target image or the image sequence is traversed; when there is no corresponding matching pixel point for the new target point in the new target image or the image sequence is traversed, record the target sequence as the tracking pixel sequence of the feature point.

[0015] Further, the obtaining of the possible cyst value of each feature point includes:

[0016] Use the optical flow algorithm to obtain the optical flow vector of each feature point in the kidney region of the image to be analyzed;

[0017] Optionally select a feature point as the analysis point, and record the cosine similarity between the optical flow vector of the analysis point and the feature point closest to it as the gray change similarity value of the analysis point;

[0018] Obtain the possible cyst value of the analysis point according to the gray value of the analysis point, the length of the tracking pixel sequence, and the gray change similarity value; the gray value and the possible cyst value are negatively correlated, and both the length and the gray change similarity value are positively correlated with the possible cyst value.

[0019] Further, the dividing of the cyst feature points into different categories includes:

[0020] Use the optical flow algorithm to obtain the optical flow vectors of the remaining pixel points except the last pixel point in the tracking pixel sequence of each cyst feature point;

[0021] Obtain the characteristic fluctuation value corresponding to two cyst feature points according to the degree of dispersion of the distances between the pixel points at the same positions in the tracking pixel sequences of any two cyst feature points, the degree of dispersion of the gray-scale differences, and the degree of dispersion of the similarity of the optical flow vectors.

[0022] Obtain the same-cyst index corresponding to two cyst feature points according to the distances, gray-scale differences, similarity of the optical flow vectors, and the characteristic fluctuation value of the pixel points at the same positions in the tracking pixel sequences of any two cyst feature points.

[0023] Perform a negative correlation mapping on the same-cyst index to obtain the characteristic distance between any two cyst feature points; cluster all cyst feature points based on the characteristic distance to obtain several clustering clusters; and regard the cyst feature points within each clustering cluster as cyst feature points of the same category.

[0024] Further, the method for obtaining the characteristic fluctuation value includes:

[0025] Optionally select two cyst feature points as the points to be measured, and form pixel pairs of the two points to be measured from the pixel points at the same positions in the tracking pixel sequences of the two points to be measured.

[0026] Record the distance between the two pixel points, the absolute value of the difference in gray-scale values, and the cosine similarity of the optical flow vectors in each pixel pair as the condition value of each pixel pair.

[0027] Obtain the arithmetic mean difference of each of the condition values of all pixel pairs of the two points to be measured, and record it as the condition fluctuation value of each type of the two points to be measured; take the product of all the condition fluctuation values of the two points to be measured as the characteristic fluctuation value of the two points to be measured.

[0028] Further, the method for obtaining the same-cyst index includes:

[0029] Record the mean value of the distances between the two pixel points in all pixel pairs of the two points to be measured as the overall distance between the two points to be measured.

[0030] Record the mean value of the absolute values of the differences in gray-scale values between the two pixel points in all pixel pairs of the two points to be measured as the overall gray-scale difference between the two points to be measured.

[0031] Record the mean value of the cosine similarities of the optical flow vectors between the two pixel points in all pixel pairs of the two points to be measured as the overall gray-scale trend similarity value between the two points to be measured.

[0032] Obtain the initial value of the same cyst for two points to be measured according to the overall distance, the overall gray-level difference, and the overall gray-level trend similarity value; the overall gray-level trend similarity value has a positive correlation with the initial value of the same cyst, and both the overall distance and the overall gray-level difference have a negative correlation with the initial value of the same cyst;

[0033] Perform a negative correlation mapping on the feature fluctuation value, and normalize the weighted result of the mapping result to the initial value of the same cyst to obtain the same cyst index for two points to be measured.

[0034] Further, the obtaining of the regional growth threshold for each cyst feature point includes:

[0035] Obtain the arithmetic mean difference of the distances between each cyst feature point and the remaining cyst feature points of its category, and record it as the distance fluctuation value of each cyst feature point;

[0036] Perform a negative correlation mapping on the distance fluctuation value of each cyst feature point, and normalize the product of the mapping result and the possible value of the cyst to obtain the cyst regional center value of each cyst feature point;

[0037] Based on the cyst regional center value, weight the preset growth threshold to obtain the regional growth threshold for each cyst feature point.

[0038] Further, the obtaining of the kidney region of the image to be analyzed includes:

[0039] Use the maximum inter-class variance method to obtain the segmentation threshold of the image to be analyzed; record the pixel points in the image to be analyzed with gray-level values less than or equal to the segmentation threshold as low-gray-level points;

[0040] For the connected domain composed of low-gray-level points in the image to be analyzed, record the connected domain with non-low-gray-level points inside as the renal cortex suspected region, and the renal cortex suspected region has an outer edge;

[0041] Record the renal cortex suspected region with the largest area as the renal cortex region, and use the closed region formed by the outer edge of the renal cortex region as the kidney region of the image to be analyzed.

[0042] Further, the selection of cyst feature points in the kidney region includes:

[0043] For the feature points in the kidney region, record the feature points with the possible value of the cyst greater than the preset threshold as the cyst feature points in the kidney region.

[0044] Further, the method for selecting feature points from the kidney region is the corner detection algorithm.

[0045] The present invention has the following beneficial effects:

[0046] Renal cysts usually appear in the renal cortex. Small renal cysts are easily confused with the renal cortex area, and the number and size of cysts in the kidney are unequal. It is difficult to extract the cyst area through common region segmentation algorithms and neural networks, etc. To solve the above problems, first, the feature points in the kidney region are tracked in consecutive frame images to obtain a tracking pixel sequence, and the length of the tracking pixel sequence presents the morphological changes of the feature points in consecutive frame ultrasound images; then, the morphological changes of renal cysts in consecutive frame ultrasound images are relatively small compared to those of other kidney regions. According to the similarity of the gray-scale change trends between each feature point and its neighboring feature points, and the length of the tracking pixel sequence of the feature points, the possibility that the feature point is located in the renal cyst position is analyzed to obtain a cyst probability value, and then the cyst feature points in the kidney region are selected. Since there may be multiple cysts in the kidney, the cyst feature points need to be classified; compared with the cyst feature points in different cyst regions, the differences in the gray-scale manifestations of the cyst feature points in the same cyst region are relatively small and the gray-scale change trends are relatively similar. Then, by combining the distance, gray-scale difference, and similarity of the gray-scale change trends of the pixel points at the same position in the tracking pixel sequence of the cyst feature points, the cyst feature points can be accurately divided into different categories. The distance between each cyst feature point and the other cyst feature points in its category, as well as the cyst probability value, can both reflect the position of each cyst feature point in its corresponding cyst region. Using the two to adaptively obtain the region growing threshold to achieve the segmentation of the cyst region, so as to solve the problem that the tumor edge is blurred and cannot be accurately segmented, which helps to improve the adaptability of the region growing algorithm in different regions and ensure the accuracy of tumor region segmentation. Description of the Drawings

[0047] 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 description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0048] Figure 1 It is a step flow chart of a method for segmenting the lesion region of kidney ultrasound images provided by an embodiment of the present invention;

[0049] Figure 2 It is a partial schematic diagram of a kidney image provided by an embodiment of the present invention;

[0050] Figure 3 It is a step flow chart of a method for classifying cyst feature points provided by an embodiment of the present invention;

[0051] Figure 4The system structure diagram of a system for segmenting lesion regions in kidney ultrasound images provided by an embodiment of the present invention;

[0052] Figure 5 The schematic diagram of a computer device of a device for segmenting lesion regions in kidney ultrasound images provided by an embodiment of the present invention. Detailed implementation manners

[0053] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following combines the accompanying drawings and preferred embodiments to detail the specific implementation manners, structures, features, and effects of a method for segmenting lesion regions in kidney ultrasound images proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0055] The following specifically describes the specific solution of a method for segmenting lesion regions in kidney ultrasound images provided by the present invention with reference to the accompanying drawings.

[0056] Embodiment 1:

[0057] The present invention proposes a method for segmenting lesion regions in kidney ultrasound images. Please refer to Figure 1 , which shows the flowchart of the steps of a method for segmenting lesion regions in kidney ultrasound images provided by an embodiment of the present invention. The method includes:

[0058] Step S1: Obtain a continuous number of frames of kidney images of a patient, and record the first frame of kidney image as the image to be analyzed.

[0059] In order to locate and segment the lesion region of the kidney, it is necessary to obtain the ultrasound image of the patient's kidney area. The specific operation is as follows: First, during the detection process, the patient takes a supine position or a lateral position to better display the kidney region; Second, select a curved probe ultrasound device to detect the patient's kidney area, and adjust the parameters of the device to ensure that the kidney is clearly displayed on the display screen of the ultrasound device; Then, place the curved probe at the patient's waist or abdomen position, adjust the angle and pressure of the curved probe to make the kidney completely presented in the image; Finally, the medical staff observes the real-time image on the display screen of the ultrasound device in real time during the detection process. When the kidney structure is clearly visible, continuously press the acquisition button of the device to capture and save a continuous number of frames of images, and record these images as kidney images; Figure 2 The partial schematic diagram of a kidney image provided by an embodiment of the present invention.

[0060] It should be noted that this solution only detects a single kidney, and there is only one kidney region in the kidney image; the images collected by ultrasonic imaging are usually grayscale images, so the kidney image is a grayscale image.

[0061] For the convenience of subsequent analysis, the first kidney image in the collected multiple frames of kidney images is denoted as the image to be analyzed.

[0062] Step S2: Obtain the kidney region of the image to be analyzed; select feature points from the kidney region, track each feature point in the kidney image, and form a tracking pixel sequence for each feature point from each feature point and its corresponding matching pixel points in the remaining kidney images except the image to be analyzed.

[0063] In order to segment the lesion area in the kidney image, it is necessary to extract the kidney region in the kidney image.

[0064] Preferably, in some possible implementation manners of the embodiments of the present invention, the method for obtaining the kidney region includes: using the Otsu method to obtain the segmentation threshold of the image to be analyzed; recording the pixel points in the image to be analyzed with gray values less than or equal to the segmentation threshold as low-gray points; for the connected domains composed of low-gray points in the image to be analyzed, recording the connected domains with non-low-gray points inside as suspected renal cortex regions, and the suspected renal cortex regions have outer edges; recording the largest suspected renal cortex region in terms of area as the renal cortex region, and taking the closed region formed by the outer edges of the renal cortex region as the kidney region of the image to be analyzed. Among them, the Otsu method is a well-known technology to those skilled in the art and will not be elaborated here.

[0065] Since the outer layer of the kidney is composed of cellulose, it presents a high echo feature in the reflection of ultrasonic waves, the renal cortex presents a low echo feature, the renal medulla region presents a high echo feature, and the inside of the kidney is mainly composed of the cortex region and the medulla region. Therefore, in the ultrasonic image, the kidney appears as a group of bright regions wrapped by a circle of low-gray outer layers. Therefore, the connected domains composed of low-gray points with non-low-gray points inside are recorded as suspected renal cortex regions, and the suspected renal cortex regions are hollow inside and have outer edges.

[0066] Since environmental noise and the noise generated during the device acquisition process may also appear as a group of bright regions wrapped by a circle of low-gray outer layers, but the areas of these low-gray outer layer regions are small, the largest suspected renal cortex region in terms of area is the renal cortex region. Because the renal cortex is the outer part of the kidney, the closed region formed by the outer edges of the renal cortex region is taken as the kidney region.

[0067] Renal cysts usually appear in the outer part of the kidney, namely the renal cortex; in ultrasonic images, the renal cortex is continuous with the overall structure of the kidney and presents low echo characteristics in the reflection of ultrasonic waves, while cysts also present low echo characteristics, making it easy for smaller renal cysts to be confused with the renal cortex area and difficult to distinguish and segment. Therefore, the morphological change characteristics of renal cysts in consecutive frame images are used to locate them.

[0068] First, feature points in the kidney region of the image to be analyzed are selected. In this embodiment, a corner detection algorithm is used to extract the feature points in the kidney region, and the feature points are equivalent to optical flow points. Among them, the corner detection algorithm is a well-known technology to those skilled in the art and will not be elaborated here.

[0069] After that, in order to analyze the morphological changes of renal cysts in consecutive frame images, each feature point is tracked in the kidney image; a tracking pixel sequence of each feature point is formed by each feature point and its corresponding matching pixel points in the remaining kidney images except the image to be analyzed.

[0070] Preferably, in some possible implementation manners of the embodiments of the present invention, the method for obtaining the tracking pixel sequence includes: arranging all kidney images in time sequence to obtain an image sequence, and the first image in the image sequence is the image to be analyzed; for each feature point in the kidney region of the image to be analyzed, an initial target sequence of the feature point is set, and the elements in the initial target sequence are the feature point itself; the feature point is respectively denoted as the target point, and the second image in the image sequence is denoted as the target image; the optical flow algorithm is used to track the target point in the target image. When there is a corresponding matching pixel point of the target point in the target image, the corresponding matching pixel point of the target point in the target image is denoted as the new target point, the new target point is added to the target sequence, the target sequence is updated, and the adjacent next kidney image of the target image is denoted as the new target image until there is no corresponding matching pixel point of the new target point in the new target image; when there is no corresponding matching pixel point of the new target point in the new target image or the image sequence traversal is completed, the target sequence is denoted as the tracking pixel sequence of the feature point.

[0071] As an example, assume the image sequence , where are all kidney images; for the feature point a in image A, the initial target sequence of the feature point a is , and the optical flow algorithm is used to track the pixel point a in image B. When there is a corresponding matching pixel point b of the pixel point a in image B, the pixel point b is added to M0 to update the target sequence M0, and the target sequence is obtained.

[0072] When there is a corresponding matching pixel point c of the pixel point b in image C, the pixel point c is added to M1 to update the target sequence M1, and the target sequence Image C is the last image in image sequence P. After the traversal of the image sequence is completed, the target sequence is recorded as the tracking pixel sequence of feature point a.

[0073] When there is no corresponding matching pixel point c for pixel point b in image C, the target sequence is recorded as the tracking pixel sequence of feature point a.

[0074] It should be noted that the matching pixel point corresponding to the target point in the target image is equivalent to the most similar pixel point of the target point in the target image, and can also be regarded as the position of the target point in the target image. In this embodiment, the Lucas-Kanade optical flow algorithm is used to track the target point in the target image. Among them, the Lucas-Kanade optical flow algorithm is a well-known technology for those skilled in the art and will not be elaborated here.

[0075] Step S3: Obtain the cyst possibility value of each feature point according to the similarity of the gray change trend between each feature point and its adjacent feature points in the kidney region, and the length of the tracking pixel sequence of each feature point; select the cyst feature points in the kidney region by using the cyst possibility value.

[0076] When there are multiple cysts in the patient's kidney and the cyst sizes are different, it is difficult to directly segment the cyst region by common region segmentation algorithms. This solution uses the morphological change characteristics of renal cysts in consecutive frame images to locate them.

[0077] When the kidney is under deep breathing or pressure, it will undergo slight displacement due to the change of abdominal pressure. The position and shape of renal cysts are more difficult to be clearly captured in ultrasound images. At the same time, the liquid characteristics of cysts make their shape and position change less. Therefore, the morphological change of renal cysts in consecutive frame ultrasound images is smaller than that of other kidney regions.

[0078] Compared with the feature points in the non-cyst part of the kidney, the gray change trends of the feature points in the cyst region are more similar in consecutive frame ultrasound images; the length of the tracking pixel sequence of the feature points presents the morphological change situation of the feature points in consecutive frame ultrasound images. Combining the similarity of the gray change trend between each feature point and its adjacent feature points, the cyst possibility value is obtained.

[0079] Preferably, in the embodiment of the present invention, the method for obtaining the cyst possibility value includes: using the optical flow algorithm to obtain the optical flow vector of each feature point in the kidney region of the image to be analyzed; arbitrarily selecting a feature point as the analysis point, and recording the cosine similarity between the optical flow vector of the analysis point and the optical flow vector of the feature point closest to it as the gray change similarity value of the analysis point; obtaining the cyst possibility value of the analysis point according to the gray value, the length of the tracking pixel sequence and the gray change similarity value of the analysis point.

[0080] Renal cysts are usually cystic structures filled with fluid. The fluid has a weak ability to emit ultrasonic waves, causing renal cysts to exhibit low echo characteristics in the reflection of ultrasonic waves; the renal cortex is composed of dense renal tubules and blood vessels, and ultrasonic waves will have a strong reflection when passing through these tissues, causing the renal cortex to exhibit medium echo characteristics in the reflection of ultrasonic waves; the renal medulla exhibits high echo characteristics. Therefore, the smaller the grayscale value of a feature point, the greater the possibility that it is in the renal cyst area.

[0081] If the morphological change of a feature point in consecutive frame ultrasound images is greater, it is easier for the optical flow algorithm to have a matching error, resulting in a greater possibility of interruption in the tracking of the feature point in the kidney image, and the shorter the length of the tracking pixel sequence of the feature point. It is known that the morphological change of renal cysts in consecutive frame ultrasound images is smaller than that of other kidney regions. When the length of the tracking pixel sequence of a feature point is longer, the greater the possibility that the feature point is in the renal cyst area.

[0082] The similarity value of the gray-scale change of an analysis point presents the degree of similarity of the gray-scale change trend between the analysis point and the feature point closest to it. Compared with the feature points in the non-cyst part of the kidney, the gray-scale change trends of the feature points in the cyst area are more similar in consecutive frame ultrasound images. If the analysis point and the feature point closest to it are in the same area, the greater the similarity value of the gray-scale change, the greater the possibility that the analysis point is in the renal cyst area.

[0083] The greater the cyst possible value of an analysis point, the greater the possibility that it is in the renal cyst position. Therefore, there is a negative correlation between the grayscale value of the analysis point and the cyst possible value, and there are positive correlations between the length of the tracking pixel sequence of the analysis point and the similarity value of the gray-scale change and the cyst possible value. In the embodiments of the present invention, the grayscale value of the analysis point is subjected to a negative correlation mapping, and the product of the mapping result, the length of the tracking pixel sequence of the analysis point, and the similarity value of the gray-scale change is normalized to obtain the cyst possible value of the analysis point.

[0084] In the embodiments of the present invention, the correlation between the grayscale value of the analysis point, the length of the tracking pixel sequence, the similarity value of the gray-scale change, and the cyst possible value can also be constructed through other basic mathematical operations, which will not be limited and elaborated here.

[0085] It should be noted that in this embodiment, the Lucas-Kanade optical flow algorithm is selected to obtain the optical flow vector of the feature points in the kidney region; the length of the tracking pixel sequence is equal to the total number of elements in the sequence; if there are multiple feature points closest to the analysis point, any one of these feature points is selected to calculate the similarity value of the gray-scale change of the analysis point.

[0086] In a specific implementation manner of the embodiments of the present invention, assuming that the analysis point is the i-th feature point in the kidney region, the cyst possible value of the analysis point is expressed by the formula:

[0087]

[0088] wherein, is the cyst probability value of the i-th feature point in the kidney region; is the length of the tracking pixel sequence of the i-th feature point in the kidney region; is the optical flow vector of the i-th feature point in the kidney region; is the optical flow vector of the feature point closest to the i-th feature point in the kidney region; is the gray-scale change similarity value of the i-th feature point in the kidney region; is the gray-scale value of the i-th feature point in the kidney region; cos is the cosine function; exp is the exponential function with the natural constant as the base; Norm is the normalization function.

[0089] For the feature points in the kidney region, the feature points with cyst probability values greater than the preset threshold are denoted as cyst feature points in the kidney region.

[0090] It should be noted that in the embodiment of the present invention, the preset threshold takes an empirical value of 0.8, and the implementer can set it by himself according to specific situations. The method for obtaining the cyst probability values of all feature points in the kidney region of the image to be analyzed is the same as the method for obtaining the cyst probability value of the analysis point.

[0091] Step S4: Divide the cyst feature points into different categories according to the distances, gray-scale differences, and similarities of gray-scale change trends of the pixel points at the same positions in the tracking pixel sequences of any two cyst feature points.

[0092] When there are multiple cysts in the patient's kidney, the cyst feature points obtained in the above step S3 are in different cyst regions, and it is necessary to classify the cyst feature points, and the cyst feature points belonging to the same cyst region are classified into the same category.

[0093] Since there are differences in the lesion degrees of different cysts, there are significant differences in the gray-scale manifestations and gray-scale change trends of different cyst regions; therefore, compared with the cyst feature points in different cyst regions, the differences in the gray-scale manifestations of the cyst feature points in the same cyst region are smaller and the gray-scale change trends are more similar. At the same time, the distances between the cyst feature points in the same cyst region are smaller than the distances between the cyst feature points in different cyst regions.

[0094] By analyzing the distances, gray-scale differences, and similarities of gray-scale change trends of the pixel points of any two cyst feature points during the entire tracking process, the accuracy of analyzing the morphology and gray-scale changes of the two cyst feature points during the kidney movement is improved, and the cyst feature points belonging to different cyst regions are accurately classified.

[0095] Please refer toFigure 3 , which shows a flowchart of the steps of a method for classifying cyst feature points provided by an embodiment of the present invention. The method includes:

[0096] Step S410: Use the optical flow algorithm to obtain the optical flow vectors of the remaining pixel points except the last pixel point in the tracking pixel sequence of each cyst feature point.

[0097] It should be noted that in this embodiment, the Lucas-Kanade optical flow algorithm is selected to obtain the optical flow vectors of the pixel points in the tracking pixel sequence of each cyst feature point; there is no adjacent next-frame image for the kidney image where the last pixel point in the tracking pixel sequence is located, and tracking cannot continue, resulting in the absence of an optical flow vector for the last pixel point in the tracking pixel sequence.

[0098] Step S420: Obtain the feature fluctuation value corresponding to two cyst feature points according to the degree of dispersion of the distances, the degree of dispersion of the gray-scale differences, and the degree of similarity of the optical flow vectors of the pixel points at the same position in the tracking pixel sequences of any two cyst feature points.

[0099] Compared with the differences between the lesion degrees of different renal cysts, the lesion degrees at different positions of the same renal cyst are relatively close, that is, different cyst feature points in the same cyst have characteristics of relatively close gray scales, relatively close distances, and relatively similar gray-scale change trends. Therefore, the above characteristics of different cyst feature points in the same renal cyst can maintain high stability under continuous optical flow tracking.

[0100] The pixel points at the same position in the tracking pixel sequences of two cyst feature points represent the positions of the two cyst feature points under continuous optical flow tracking; the similarity of the optical flow vectors reflects the similarity degree of the gray-scale change trends of the two pixel points. Combining the distance and gray-scale difference of the two pixel points together represents the distributed gray-scale feature. The degree of dispersion of the above indicators presents the stability degree of the distributed gray-scale feature of the two cyst feature points during the optical flow tracking process, thereby obtaining the feature fluctuation value.

[0101] Preferably, in some possible implementation manners of the embodiments of the present invention, the method for obtaining the feature fluctuation value includes: arbitrarily select two cyst feature points as the points to be measured, and form pixel pairs of the two points to be measured from the pixel points at the same position in the tracking pixel sequences of the two points to be measured; record the distance between the two pixel points, the absolute value of the difference in gray-scale values, and the cosine similarity of the optical flow vectors in each pixel pair as the condition value of each pixel pair; obtain the arithmetic mean difference of each condition value of all pixel pairs of the two points to be measured, and record it as the condition fluctuation value of each type of the two points to be measured; take the product of all types of condition fluctuation values of the two points to be measured as the feature fluctuation value of the two points to be measured.

[0102] The arithmetic mean difference reflects the degree of dispersion of a set of data. The fluctuation value of each condition presents the degree of fluctuation of the distance, gray - scale difference, or similarity of the gray - scale change trend of the pixel points corresponding to the two cyst feature points during continuous optical flow tracking.

[0103] Since the gray - scale distribution characteristics of the cyst feature points in the same cyst are relatively close and the cyst is relatively stable, the gray - scale distribution characteristics of different ultrasonic images under the continuous optical flow tracking of the two cyst feature points are relatively close. Therefore, by comprehensively analyzing all the condition fluctuation values, the overall degree of fluctuation of the gray - scale, distance, and gray - scale change trend of the two points to be measured during the optical flow tracking is obtained.

[0104] It should be noted that the total number of pixel pairs of the two points to be measured is equal to the minimum value of the total number of elements in the tracking pixel sequences of the two points to be measured.

[0105] In a specific implementation manner of the embodiment of the present invention, the feature fluctuation value of the two points to be measured is expressed by the formula:

[0106]

[0107] In the formula, FJ is the feature fluctuation value of the two points to be measured; N is the total number of pixel pairs of the two points to be measured; is the distance between the two pixel points in the nth pixel pair of the two points to be measured. In this embodiment, the Euclidean distance is selected to measure the distance between pixel points; is the mean value of the distances between the two pixel points in all pixel pairs of the two points to be measured; is the absolute value of the difference in gray - scale values between the two pixel points in the nth pixel pair of the two points to be measured; is the mean value of the absolute values of the differences in gray - scale values between the two pixel points in all pixel pairs of the two points to be measured; is the cosine similarity of the optical flow vectors of the two pixel points in the nth pixel pair of the two points to be measured; is the mean value of the cosine similarities of the optical flow vectors of the two pixel points in the remaining pixel pairs except the last pixel pair of the two points to be measured; 、 、 respectively represent a condition value of the nth pixel pair of the two points to be measured, and there are three condition values in total; 、 、 respectively represent a condition fluctuation value of the two points to be measured; is the absolute - value function.

[0108] It should be noted that the smaller the feature fluctuation value FJ is, the more stable the gray - scale difference, distance, and gray - scale change trend at the corresponding positions of the kidney images of two points to be measured during continuous optical flow tracking are. Then the greater the possibility that the two points to be measured are optical flow points of the same cyst, and the greater the credibility of the distributed gray - scale features at the corresponding positions of the kidney images of the two points to be measured during continuous optical flow tracking.

[0109] It should be noted that since there is no optical flow vector for the last pixel point in the tracking pixel sequence, for the pixel pair composed of the last pixel point in the tracking pixel sequence of the point to be measured corresponding to the minimum value of the total number of elements in the tracking pixel sequence and the pixel point at the same position in the tracking pixel sequence of another point to be measured, the conditional value of the cosine similarity of the optical flow vector cannot be obtained for this pixel pair. Then this pixel pair does not participate in the process of obtaining the conditional fluctuation value of the cosine similarity of the optical flow vector.

[0110] Step S430: Obtain the same - cyst index corresponding to two cyst feature points according to the distance, gray - scale difference, similarity of optical flow vectors, and feature fluctuation value of the pixel points at the same position in the tracking pixel sequences of any two cyst feature points.

[0111] Preferably, in some possible implementation manners of the embodiments of the present invention, the method for obtaining the same - cyst index includes: denoting the mean value of the distances between two pixel points in all pixel pairs of two points to be measured as the overall distance between the two points to be measured; denoting the mean value of the absolute values of the differences in gray - scale values between two pixel points in all pixel pairs of two points to be measured as the overall gray - scale difference between the two points to be measured; denoting the mean value of the cosine similarities of the optical flow vectors between two pixel points in all pixel pairs of two points to be measured as the overall gray - scale trend similarity value between the two points to be measured; obtaining the initial same - cyst value of the two points to be measured according to the overall distance, overall gray - scale difference, and overall gray - scale trend similarity value; the overall gray - scale trend similarity value and the initial same - cyst value are in a positive - correlation relationship, and both the overall distance and the overall gray - scale difference are in a negative - correlation relationship with the initial same - cyst value; performing a negative - correlation mapping on the feature fluctuation value, and normalizing the weighted result of the mapping result on the initial same - cyst value to obtain the same - cyst index of the two points to be measured.

[0112] To improve the accuracy of analysis, consider the overall features of the pixel points at the corresponding positions in the kidney images of two points to be measured during continuous optical flow tracking; the overall distance reflects the overall level of the distance at the corresponding positions of the two points to be measured during continuous optical flow tracking, the overall gray - scale difference reflects the overall level of the gray - scale difference at the corresponding positions of the two points to be measured during continuous optical flow tracking, and the overall gray - scale trend similarity value reflects the overall level of the similarity of the gray - scale change trends at the corresponding positions of the two points to be measured during continuous optical flow tracking.

[0113] Due to the differences in the lesion degrees of different cysts, there are significant differences in the gray-scale manifestations and gray-scale change trends of different cyst regions. Therefore, compared with the cyst feature points in different cyst regions, the differences in the gray-scale manifestations of the cyst feature points in the same cyst region are smaller and the gray-scale change trends are more similar. Moreover, the distance between the cyst feature points in the same cyst region is smaller than the distance between the cyst feature points in different cyst regions. Therefore, the overall gray-scale trend similarity value has a positive correlation with the same-cyst initial value, and both the overall distance and the overall gray-scale difference have a negative correlation with the same-cyst initial value.

[0114] Because the smaller the feature fluctuation value, the greater the credibility of the similarity of the distances, gray-scale differences, and gray-scale change trends of the corresponding pixel points of two points to be measured during continuous tracking, and the greater the possibility that the two points to be measured are in the same cyst region. Therefore, it is necessary to perform a negative correlation mapping on the feature fluctuation value, and use the mapping result to weight the same-cyst initial value to obtain the same-cyst index, which is used to measure the possibility that two points to be measured are in the same cyst region.

[0115] In a specific implementation manner of the embodiment of the present invention, the same-cyst index is expressed by the formula:

[0116]

[0117] In the formula, JU is the same-cyst index of two points to be measured; is the mean value of the cosine similarities of the optical flow vectors of two pixel points in the remaining pixel pairs except the last pixel pair of two points to be measured; is the mean value of the distances between two pixel points in all pixel pairs of two points to be measured; is the mean value of the absolute values of the differences in the gray-scale values of two pixel points in all pixel pairs of two points to be measured; FJ is the feature fluctuation value of two points to be measured; is a preset positive number, taking the empirical value 0.1, which is used to prevent the fractional expression from being meaningless due to the denominator being zero; exp is the exponential function with the natural constant as the base; Norm is the normalization function.

[0118] It should be noted that the greater the same-cyst index of two points to be measured, the greater the possibility that the two points to be measured are in the same cyst region. The method for obtaining the same-cyst index of any two cyst feature points is the same as the method for obtaining the same-cyst index of two points to be measured.

[0119] Step S440: Perform a negative correlation mapping on the same-cyst index to obtain the feature distance between any two cyst feature points; cluster all cyst feature points based on the feature distance to obtain several clustering clusters; and use the cyst feature points within each clustering cluster as cyst feature points of the same category.

[0120] Since the larger the same-cyst index of two cyst feature points, the greater the likelihood that they are in the same cyst region, the distance between the two cyst feature points should be smaller during the clustering process to classify the two cyst feature points into the same category. Therefore, it is necessary to perform a negative correlation mapping on the same-cyst index.

[0121] In this embodiment, based on the feature distance between any two cyst feature points, the DBSCAN algorithm is used to cluster all cyst feature points in the kidney region to obtain different clustering clusters. Among them, the DBSCAN algorithm is a well-known technology to those skilled in the art and will not be elaborated here.

[0122] It should be noted that since the value range of the same-cyst index is from 0 to 1, in this embodiment, the difference between the constant 1 and the same-cyst index is used as the feature distance between any two cyst feature points to achieve the negative correlation mapping of the same-cyst index. The number of clustering clusters is equal to the number of cyst regions.

[0123] Step S5: According to the distance between each cyst feature point and the other cyst feature points in its category, and the possible cyst value of each cyst feature point, obtain the region growing threshold of each cyst feature point; based on the region growing threshold, use each type of cyst feature point as a seed point for region growing to obtain the actual cyst regions in the kidney region.

[0124] The distance between each cyst feature point and the other cyst feature points in its category, as well as the possible cyst value, both reflect the position of each cyst feature point in the cyst region where it is located. The preset growing threshold is adjusted by using the position of the cyst feature points in the cyst region to obtain the region growing threshold of the cyst feature point.

[0125] Preferably, in some possible implementation manners of the embodiments of the present invention, the method for obtaining the region growing threshold includes: obtaining the arithmetic mean difference of the distances between each cyst feature point and the other cyst feature points in its category, denoted as the distance fluctuation value of each cyst feature point; performing a negative correlation mapping on the distance fluctuation value of each cyst feature point, normalizing the product of the mapping result and the possible cyst value to obtain the cyst region center value of each cyst feature point; weighting the preset growing threshold based on the cyst region center value to obtain the region growing threshold of each cyst feature point.

[0126] Since the boundary of a renal cyst is relatively blurred, the gray-scale change trend of the pixel points at the edge position of the renal cyst fluctuates more than that of the pixel points at the center position. Then, the possible cyst value of the pixel points at the edge position of the renal cyst is smaller than that of the pixel points at the center position. Therefore, if the possible cyst value is smaller, the cyst feature point is closer to the edge position of the cyst region where it is located; conversely, the cyst feature point is closer to the center position of the cyst region where it is located.

[0127] Compared with the distances between the cyst feature points at the edge positions near the cyst region and the rest of the cyst feature points, the distances between the cyst feature points at the central position near the cyst region and the rest of the cyst feature points are closer, and the distance fluctuations are smaller. The arithmetic mean difference is used to reflect the degree of fluctuation of a set of data. If the distance fluctuation value is smaller, it means that the distance between each cyst feature point and the rest of the cyst feature points in its category is closer, and the smaller the fluctuation, the closer each cyst feature point is to the central position of the cyst region where it is located.

[0128] Therefore, there is a positive correlation between the possible cyst value and the central value of the cyst region, and a negative correlation between the distance fluctuation value and the central value of the cyst region.

[0129] Since the gray-scale change in the edge region is larger and more complex, the seed points in the edge region require more precise control to avoid the misgrowth of other renal tumor regions. While the central region is usually more uniform, the seed points in the central region can tolerate a larger threshold to adapt to the relatively stable regional characteristics. The central value of the cyst region is used to weight the preset growth threshold, so that the threshold of the seed points at the central position of the tumor region is larger, and the threshold of the seed points at the edge position of the tumor region is smaller.

[0130] In a specific implementation manner of the embodiment of the present invention, the regional growth threshold is expressed by the formula:

[0131]

[0132] In the formula, QT is the regional growth threshold of each cyst feature point; is the preset growth threshold, taking the empirical value of 10; UP is the possible cyst value of each cyst feature point; M is the total number of cyst feature points in the category to which each cyst feature point belongs; is the distance between each cyst feature point and the m-th cyst feature point in the rest of its category; is the mean value of the distances between each cyst feature point and the rest of the cyst feature points in its category; is the distance fluctuation value of each cyst feature point; is the central value of the cyst region of each cyst feature point; Norm is the normalization function; exp is the exponential function with the natural constant as the base.

[0133] Based on the regional growth threshold, each type of cyst feature point is used as a seed point, and the region growing algorithm is used for region growing. Each type of cyst feature point grows to obtain an actual cyst region. Among them, the region growing algorithm is a well-known technology to those skilled in the art and will not be elaborated here.

[0134] It should be noted that during the regional growth process of each type of cyst feature point, each cyst feature point, that is, the seed point, is used as the growth point for regional growth. Pixel points whose absolute value of the difference between the gray value of the growth point and the gray value of the pixel points within its eight-neighborhood is less than or equal to the regional growth threshold of the cyst feature point at the initial growth position are used as new growth points for regional growth until the absolute value of the difference between the gray value of all new growth points and the gray value of all pixel points within their eight-neighborhoods is greater than the regional growth threshold of the cyst feature point of the initial growth point, obtaining an actual cyst area.

[0135] This solution adaptively sets the regional growth threshold according to the cyst features and positions shown by the cyst feature points, realizes the segmentation of the cyst area, helps to improve the adaptability of the regional growth algorithm in different regions, and ensures the accuracy of tumor area segmentation.

[0136] So far, the present invention is completed.

[0137] Embodiment 2:

[0138] The present invention proposes a system for segmenting the lesion area of kidney ultrasound images. Please refer to Figure 4 , which shows the system structure diagram of a system for segmenting the lesion area of kidney ultrasound images provided by an embodiment of the present invention. The system includes:

[0139] A data acquisition module 610, configured to acquire a continuous number of frames of kidney images of a patient, and record the first frame of kidney image as the image to be analyzed;

[0140] A feature point tracking module 620, configured to obtain the kidney area of the image to be analyzed; select feature points from the kidney area, track each feature point in the kidney images, and form a tracking pixel sequence of each feature point from each feature point and its corresponding matching pixel points in the remaining kidney images except the image to be analyzed;

[0141] A cyst selection module 630, configured to obtain the cyst possibility value of each feature point according to the similarity of the gray value change trend between each feature point and its adjacent feature points in the kidney area, and the length of the tracking pixel sequence of each feature point; select the cyst feature points in the kidney area by using the cyst possibility value;

[0142] A cyst classification module 640, configured to classify the cyst feature points into different categories according to the distance, gray value difference, and similarity of the gray value change trend of the pixel points at the same position in the tracking pixel sequences of any two cyst feature points;

[0143] The cyst region segmentation module 650 is configured to obtain the region growth threshold for each cyst feature point based on the distance between each cyst feature point and the remaining cyst feature points of its belonging category, and the cyst probability value of each cyst feature point; based on the region growth threshold, use each category of cyst feature points as seed points for region growth to obtain the actual cyst region in the kidney region.

[0144] It should be noted that: for the device provided in the above embodiment, only the division of the above-mentioned functional modules is used for illustration. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the embodiments of a kidney ultrasound image lesion region segmentation system and a kidney ultrasound image lesion region segmentation method provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0145] Embodiment 3:

[0146] Figure 5 It is a schematic diagram of a computer device for a kidney ultrasound image lesion region segmentation device provided by an embodiment of the present invention. Exemplarily, as Figure 5 shown, the computer device includes: a memory 701, a processor 702, and a computer program 703 stored in the memory 701 and running on the processor 702. When the processor 702 executes the computer program 703, the computer device can execute any one of the kidney ultrasound image lesion region segmentation methods introduced above.

[0147] In addition, the embodiments of the present application also protect a device, which may include a memory and a processor. Among them, the memory stores executable program code, and the processor is used to call and execute the executable program code to execute a kidney ultrasound image lesion region segmentation method provided by the embodiments of the present application.

[0148] In this embodiment, the device can be divided into functional modules according to the above method examples. For example, it can correspond to each functional module, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is illustrative, only a logical function division, and there may be other division methods in actual implementation.

[0149] It should be understood that the device provided in this embodiment is used to execute the above-mentioned kidney ultrasound image lesion region segmentation method, so the same effect as the above implementation method can be achieved.

[0150] In the case of adopting an integrated unit, the device may include a processing module and a storage module. Among them, when the device is applied to a device, the processing module can be used to control and manage the actions of the device. The storage module can be used to support the device to execute mutual program codes, etc.

[0151] Among them, the processing module can be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits included in combination with the disclosure of the present application. The processor can also be a combination that realizes computing functions, such as including a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc. The storage module can be a memory.

[0152] It should be noted that: the above-mentioned sequence of embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0153] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments.

[0154] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for segmenting renal ultrasound image lesion areas, characterized in that: The method includes: Acquire a plurality of consecutive frames of kidney images of the patient, and record the first frame of kidney image as the image to be analyzed; Acquire a kidney region of the image to be analyzed; select feature points from the kidney region, track each feature point in the kidney image, and form a tracking pixel sequence for each feature point by each feature point and its corresponding matching pixel points in the remaining kidney images except the image to be analyzed; According to the similarity of the grayscale change trend of each feature point in the kidney region and its neighboring feature points, and the length of the tracking pixel sequence of each feature point, the cyst possibility value of each feature point is obtained; and the cyst feature points in the kidney region are selected using the cyst possibility value; According to the distance, grayscale difference and similarity of grayscale change trend of the pixels at the same position in the tracking pixel sequence of any two cyst feature points, the cyst feature points are divided into different categories; According to the distance between each cyst feature point and the other cyst feature points of the category to which it belongs, and the cyst possible value of each cyst feature point, a regional growth threshold of each cyst feature point is obtained; based on the regional growth threshold, each type of cyst feature point is used as a seed point for regional growth to obtain the actual cyst region in the kidney region; The obtaining of the possible cyst value of each feature point includes: Using an optical flow algorithm to obtain an optical flow vector of each feature point in the kidney region of the image to be analyzed; Select any feature point as the analysis point, and record the cosine similarity between the analysis point and the optical flow vector of the feature point closest to it as the grayscale change similarity value of the analysis point; According to the gray value of the analysis point, the length of the tracking pixel sequence and the gray change similarity value, the cyst possible value of the analysis point is obtained; the gray value and the cyst possible value are negatively correlated, and the length and the gray change similarity value are both positively correlated with the cyst possible value; The step of selecting cyst feature points in the kidney region includes: For the feature points in the kidney region, the feature points whose cyst possibility values ​​are greater than a preset threshold are recorded as cyst feature points in the kidney region.

2. The method for segmenting renal ultrasound lesion regions according to claim 1, characterized in that: The method for acquiring the tracking pixel sequence of each feature point comprises: Arranging all kidney images in time sequence to obtain an image sequence, wherein the first image in the image sequence is the image to be analyzed; For each feature point in the kidney region of the image to be analyzed, an initial target sequence of the feature point is set, and the elements in the initial target sequence are the feature points themselves; The feature points are respectively recorded as target points, and the second image in the image sequence is recorded as the target image; the target point is tracked in the target image using the optical flow algorithm, and when the target point has a corresponding matching pixel point in the target image, the corresponding matching pixel point of the target point in the target image is recorded as a new target point, the new target point is added to the target sequence, the target sequence is updated, and the next kidney image adjacent to the target image is recorded as a new target image, until the new target point has no corresponding matching pixel point in the new target image or the image sequence traversal is completed; when the new target point has no corresponding matching pixel point in the new target image or the image sequence traversal is completed, the target sequence is recorded as a tracking pixel sequence of the feature point.

3. The method for segmenting renal ultrasound image lesion area according to claim 1, characterized in that: The cyst feature points are divided into different categories, including: Using an optical flow algorithm to obtain the optical flow vectors of the remaining pixel points in the tracking pixel sequence of each cyst feature point except the last pixel point; Acquire characteristic fluctuation values ​​corresponding to the two cyst feature points according to the discrete degree of the distance between the pixel points at the same position in the tracking pixel sequence of any two cyst feature points, the discrete degree of the grayscale difference and the discrete degree of the similarity of the optical flow vector; According to the distance between the pixels at the same position in the tracking pixel sequence of any two cyst feature points, the grayscale difference, the similarity of the optical flow vector and the characteristic fluctuation value, the same cyst index corresponding to the two cyst feature points is obtained; The same cyst index is negatively correlated and mapped to obtain a characteristic distance between any two cyst feature points; all cyst feature points are clustered based on the characteristic distance to obtain a plurality of clusters; and the cyst feature points in each cluster are used as cyst feature points of the same category.

4. The method for segmenting renal ultrasound image lesion area according to claim 3, characterized in that: The method for obtaining the characteristic fluctuation value includes: Two cyst feature points are randomly selected as test points, and the pixel pairs of the two test points are formed by the pixel points at the same position in the tracking pixel sequences of the two test points; The distance between the two pixels in each pixel pair, the absolute value of the difference in grayscale values, and the cosine similarity of the optical flow vector are recorded as the conditional value of each pixel pair; The arithmetic mean difference of each condition value of all pixel pairs of the two points to be tested is obtained, and recorded as each condition fluctuation value of the two points to be tested; the product of all condition fluctuation values ​​of the two points to be tested is taken as the characteristic fluctuation value of the two points to be tested.

5. The method for segmenting renal ultrasound image lesion area according to claim 4, characterized in that: The method for obtaining the cyst index comprises: The average of the distances between the two pixels in all pixel pairs of the two test points is recorded as the overall distance between the two test points; The average of the absolute values ​​of the grayscale value differences between the two pixels in all the pixel pairs of the two test points is recorded as the overall grayscale difference between the two test points; The average of the cosine similarities of the optical flow vectors of the two pixel points in all pixel pairs of the two test points is recorded as the overall grayscale trend similarity value of the two test points; According to the overall distance, the overall grayscale difference and the overall grayscale trend similarity value, the same cyst initial values ​​of the two test points are obtained; the overall grayscale trend similarity value is positively correlated with the same cyst initial value, and the overall distance and the overall grayscale difference are negatively correlated with the same cyst initial value; The characteristic fluctuation value is negatively correlated and mapped, and the weighted result of the initial value of the same cyst is normalized by the mapping result to obtain the same cyst index of the two test points.

6. The method for segmenting renal ultrasound image lesion area according to claim 1, characterized in that: The step of obtaining the regional growth threshold of each cyst feature point includes: Obtain the arithmetic mean difference between the distances of each cyst feature point and the other cyst feature points of the category to which it belongs, and record it as the distance fluctuation value of each cyst feature point; Performing negative correlation mapping on the distance fluctuation value of each cyst feature point, normalizing the product of the mapping result and the cyst possible value, and obtaining the cyst area center value of each cyst feature point; The preset growth threshold is weighted based on the cyst region center value to obtain the region growth threshold of each cyst feature point.

7. The method for segmenting renal ultrasound image lesion area according to claim 1, characterized in that: The step of obtaining the kidney region of the image to be analyzed includes: The maximum inter-class variance method is used to obtain the segmentation threshold of the image to be analyzed; the pixel points whose grayscale values ​​in the image to be analyzed are less than or equal to the segmentation threshold are recorded as low grayscale points; For a connected domain formed by low-grayscale points in the image to be analyzed, the connected domain with non-low-grayscale points inside is recorded as a suspected renal cortex region, and the suspected renal cortex region has an outer edge; The largest suspected renal cortex region is recorded as the renal cortex region, and the closed region formed by the outer edge of the renal cortex region is taken as the kidney region of the image to be analyzed.

8. The method for segmenting renal ultrasound lesion regions according to claim 1, characterized in that: The method for selecting feature points from the kidney region is a corner point detection algorithm.

Images