An intelligent segmentation system for kidney wound surfaces

Abstract

The present invention provides an intelligent kidney wound segmentation system, which relates to the technical field of image data processing. The intelligent kidney wound segmentation system includes: an image enhancement module; an edge blur index acquisition module; a wound rule index acquisition module; and a wound segmentation index acquisition module. By combining the obtained edge data of the kidney image and the preset reference image edge data, the present invention obtains the edge blur index. Then, according to the obtained kidney wound rule data and the preset reference wound rule data, the wound rule index is obtained. Finally, by combining the obtained wound segmentation data and the preset reference segmentation data, the wound segmentation index is obtained, achieving the effect of improving the accuracy of kidney wound segmentation and solving the problem of low correlation between kidney wound segmentation and wound edges in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and particularly to an intelligent kidney wound segmentation system. Background Art

[0002] Intelligent segmentation of kidney wounds is an important research direction in the field of medical image processing, and is particularly significant in clinical applications such as kidney wound suture reconstruction and trauma treatment. With the continuous progress of medical technology, kidney surgery is increasingly tending towards minimally invasive and precise. Traditional surgical methods often rely on the experience and intuition of doctors, while the intelligent kidney wound segmentation technology based on surgical video images can provide doctors with more accurate and objective surgical guidance. Through accurate segmentation of kidney wounds, doctors can more accurately understand the degree, scope, and nature of kidney injuries, thereby formulating a more scientific and reasonable treatment plan. The application prospect of kidney wound segmentation is broad. Clinically, it can provide doctors with more objective and accurate diagnostic bases and treatment plans; in scientific research, it can provide important data support for the research of kidney diseases; in medical education, it can be used as a teaching tool to help students better understand kidney anatomy and pathological changes. Identification of kidney wounds can help doctors more accurately evaluate the condition and formulate wound reconstruction strategies for surgical planning.

[0003] Existing systems mainly achieve segmentation based on the pixel values or feature differences between different regions in the image. The boundary usually shows a mutation of local image features such as pixel grayscale, texture, and color. By detecting this mutation information, effective segmentation of the image can be achieved.

[0004] However, in the process of implementing the technical solutions of the embodiments of the present invention, it is found that the above technologies have at least the following technical problems:

[0005] In the prior art, due to the possible blurriness or irregularity of the boundary of the kidney wound area, the problem that the correlation between kidney wound segmentation and the wound edge is not high occurs. Summary of the Invention

[0006] The embodiments of the present invention provide an intelligent kidney wound segmentation system, which solves the problem that the correlation between kidney wound segmentation and the wound edge is not high in the prior art, and realizes the improvement of the accuracy of kidney wound segmentation.

[0007] An embodiment of the present invention provides an intelligent segmentation system for kidney wound surfaces, including: an image enhancement module, an edge blur index acquisition module, a wound surface regularity index acquisition module, and a wound surface segmentation index acquisition module. Among them, the image enhancement module is used to obtain a kidney wound surface image through a preset medical imaging device, and perform image enhancement processing on the kidney wound surface image to obtain a characteristic wound surface image and kidney image edge data; the edge blur index acquisition module is used to obtain an edge blur index based on the obtained kidney image edge data and preset reference image edge data obtained from a preset database, and combine the preset edge threshold obtained from the preset database to judge whether to perform image adjustment. If image adjustment is not performed, the function of the wound surface regularity index acquisition module is executed. The edge blur index is used to evaluate the edge blur degree of the characteristic wound surface image; the wound surface regularity index acquisition module is used to obtain a wound surface regularity index based on the obtained kidney wound surface regularity data and preset reference wound surface regularity data obtained from a preset database, and combine the preset wound surface regularity threshold obtained from the database to judge whether to execute the function of the wound surface segmentation index acquisition module. The wound surface regularity index is used to evaluate the edge regularity of the characteristic wound surface image; the wound surface segmentation index acquisition module is used to perform segmentation processing on the characteristic wound surface image to obtain a segmented wound surface image and wound surface segmentation data, and combine the preset reference segmentation data obtained from a preset database to obtain a wound surface segmentation index, and perform segmentation adjustment based on the wound surface segmentation index and the preset segmentation threshold obtained from the preset database. The wound surface segmentation index is used to evaluate the segmentation of the characteristic wound surface image.

[0008] Optionally, the specific process of obtaining the characteristic wound surface image is as follows: scanning the kidney area through a preset medical imaging device to obtain a kidney wound surface image, and the preset medical imaging device includes computed tomography, magnetic resonance imaging, and ultrasonic imaging devices; performing image enhancement processing on the kidney wound surface image to obtain a characteristic wound surface image, and the image enhancement processing includes contrast enhancement, edge enhancement, and region enhancement.

[0009] Optionally, the specific process of obtaining the edge blur index based on the obtained kidney image edge data and preset reference image edge data obtained from a preset database is as follows: extracting edge data from the characteristic wound surface image to obtain kidney image edge data, and the kidney image edge data includes a first gray value, a second gray value, and a pixel point distance, and the kidney image edge data is used to describe the edge contour of the kidney wound surface in the characteristic wound surface image; obtaining a preset gradient threshold and a preset gray value change rate threshold from a preset database, and combining the kidney image edge data to obtain an edge gradient amplitude deviation and a gray value change rate deviation; obtaining a preset gradient deviation threshold and a preset gray value change rate deviation threshold from a preset database, and combining the edge gradient amplitude deviation and the gray value change rate deviation to obtain an edge blur index.

[0010] Optionally, the edge blur index is calculated using the following formula:

[0011]

[0012] In the formula, α i represents the edge blur index of the i-th pixel point on the characteristic wound surface image, where i = 1, 2,..., n - 1, i represents the serial number of the pixel point on the characteristic wound surface image, n represents the total number of pixel points on the characteristic wound surface image, R i represents the deviation of the gray value change rate of the i-th pixel point on the characteristic wound surface image, G i represents the deviation of the edge gradient amplitude of the i-th pixel point on the characteristic wound surface image, I i represents the first gray value of the i-th pixel point on the characteristic wound surface image, Ii‘ represents the second gray value of the i-th pixel point on the characteristic wound surface image, ΔX i represents the pixel point distance corresponding to the i-th pixel point on the characteristic wound surface image, G0 represents a preset gradient threshold, R0 represents a preset gray value change rate threshold, ΔR represents a preset gray value change rate deviation threshold, ΔG represents a preset gradient deviation threshold, represents the first-order derivative of the gray value corresponding to the i-th pixel point in the x-axis direction, represents the first-order derivative of the gray value corresponding to the i-th pixel point in the y-axis direction, and e represents the natural constant.

[0013] Optionally, the specific process of determining whether to perform image adjustment by combining the preset edge threshold obtained from the preset database is as follows: Q1, determine whether the edge blur index is not higher than the preset edge threshold. When the edge blur index is not higher than the preset edge threshold, no image adjustment is performed; otherwise, execute Q2; Q2, perform sharpening processing on the characteristic wound surface image. When the monitored edge blur index is not higher than the preset edge threshold, end the image adjustment; otherwise, execute Q3; Q3, perform noise removal processing on the characteristic wound surface image. When the monitored edge blur index is not higher than the preset edge threshold, end the image adjustment; otherwise, the system prompts the preset personnel to re-acquire the kidney wound surface image.

[0014] Optionally, the specific process of obtaining the kidney wound surface rule data is as follows: Obtain the kidney wound surface rule data through a preset image measurement device, and the kidney wound surface rule data includes the tangent angle, the curve arc length, and the edge point distance.

[0015] Optionally, the specific process for obtaining the wound surface regularity index is as follows: Obtain a preset curvature deviation threshold from a preset database, and obtain a curvature influence degree by combining a preset curvature threshold, kidney wound surface regularity data, and the preset curvature deviation threshold; Obtain a preset distance standard deviation deviation threshold from the preset database, and obtain a distance standard deviation influence degree by combining a preset distance standard deviation threshold, kidney wound surface regularity data, and the preset distance standard deviation deviation threshold; Obtain a preset weight from the preset database, and combine the obtained curvature influence degree and distance standard deviation influence degree to obtain the wound surface regularity index. The preset weight includes a preset regularity first weight and a preset regularity second weight.

[0016] Optionally, the specific process for segmenting the characteristic wound surface image to obtain a segmented wound surface image and wound surface segmentation data is as follows: Obtain the wound surface segmentation data through a three-dimensional slicer. The wound surface segmentation data includes an intersection area, a union area, the number of positive class pixels, and the number of marked positive class pixels. The intersection area represents the area of the overlapping part of the wound surface area and the manually marked wound surface area in the segmented wound surface image. The union area represents the area covered after combining the wound surface area and the manually marked wound surface area in the segmented wound surface image.

[0017] Optionally, the specific process for obtaining the wound surface segmentation index is as follows: Obtain a preset intersection over union threshold from a preset database, and obtain the difference between the preset intersection over union threshold and the ratio of the intersection area to the union area by combining the intersection area and the union area; Obtain a preset segmentation precision threshold from the preset database, and obtain the difference between the preset segmentation precision threshold and the ratio of the number of positive class pixels to the number of marked positive class pixels by combining the number of positive class pixels and the number of marked positive class pixels; Obtain a first segmentation weight and a preset second segmentation weight from the preset database, and combine the preset intersection over union threshold and the difference between the ratio of the intersection area to the union area, the preset segmentation precision threshold and the difference between the ratio of the number of positive class pixels to the number of marked positive class pixels to obtain the wound surface segmentation index; The preset reference segmentation data includes a preset segmentation precision threshold, a preset intersection over union threshold, a preset first segmentation weight, and a preset second segmentation weight, and combines the wound surface segmentation data to obtain the wound surface segmentation index.

[0018] Optionally, the specific process for performing segmentation adjustment based on the wound surface segmentation index and a preset segmentation threshold obtained from a preset database is as follows: M1, Determine whether the wound surface segmentation index is not lower than the preset segmentation threshold. When the wound surface segmentation index is not lower than the preset segmentation threshold, no segmentation adjustment is performed. Otherwise, continue to execute M2; M2, Obtain the preset segmentation threshold from the preset database, send a prompt to a preset person to gradually reduce the segmentation area by a preset multiple of the preset segmentation threshold. When the monitored wound surface segmentation index is not lower than the preset segmentation threshold, end the segmentation adjustment. Otherwise, continue to execute M3; M3, Perform region merging. When the monitored wound surface segmentation index is not lower than the preset segmentation threshold, end the segmentation adjustment. Otherwise, feedback to the preset person to re-obtain the kidney wound surface image.

[0019] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:

[0020] 1. An edge blur index is obtained from the edge data of the kidney image and the preset reference image edge data. Then, a wound surface rule index is obtained according to the acquired kidney wound surface rule data and the preset reference wound surface rule data. Finally, a wound surface segmentation index is obtained by combining the acquired wound surface segmentation data and the preset reference segmentation data, thereby improving the comprehensiveness of obtaining the kidney wound surface segmentation data, and further improving the accuracy of kidney wound surface segmentation, effectively solving the problem of low correlation between kidney wound surface segmentation and wound surface edges in the prior art.

[0021] 2. It is determined whether to perform image adjustment based on the preset edge threshold and the edge blur index. Then, it is determined whether to execute the function of the wound surface segmentation index acquisition module by combining the preset wound surface rule threshold and the wound surface rule index. Finally, segmentation adjustment is performed based on the wound surface segmentation index and the preset segmentation threshold, thereby achieving comprehensive adjustment and further improving the reliability of kidney wound surface area segmentation.

[0022] 3. Edge data of the kidney image is obtained by extracting edge data from the characteristic wound surface image. Then, kidney wound surface rule data is obtained by a preset image measurement device. Finally, wound surface segmentation data is obtained by a three-dimensional slicer, thereby improving the real-time performance of obtaining kidney wound surface segmentation data and further improving the accuracy of obtaining kidney wound surface segmentation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0024] Figure 1 It is a schematic structural diagram of a kidney wound surface intelligent segmentation system provided in the embodiments of the present invention;

[0025] Figure 2 It is a statistical chart of the change of the wound surface rule index provided in the embodiments of the present invention;

[0026] Figure 3 It is a postoperative kidney wound surface diagram provided in the embodiments of the present invention;

[0027] Figure 4 It is a kidney diagram before wound surface segmentation provided in the embodiments of the present invention;

[0028] Figure 5The kidney diagram after wound segmentation provided by the embodiment of the present invention;

[0029] Figure 6 The surgical process splitting framework diagram provided by the embodiment of the present invention;

[0030] Figure 7 The surgical stage diagram provided by the embodiment of the present invention;

[0031] Figure 8 The surgical step diagram provided by the embodiment of the present invention. Detailed implementation manners

[0032] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0033] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings understood by those of ordinary skill in the art to which the present invention pertains. The "first", "second" and similar terms used in the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. Similarly, the terms such as "a", "an" or "the" do not denote a quantity limitation, but mean that there is at least one. The terms such as "comprising" or "including" mean that the elements or items appearing before this term cover the elements or items listed after this term and their equivalents, without excluding other elements or items. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

[0034] It should be noted that the "upper", "lower", "left", "right", "front", "rear", etc. used in the present invention are only used to represent relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0035] In an embodiment of the present invention, by providing an intelligent segmentation system for kidney wound surfaces, the problem that the association degree between the segmentation of kidney wound surfaces and the wound surface edges in the prior art is not high is solved. The kidney wound surface image is obtained through a preset medical imaging device, and then the edge blur index is obtained based on the kidney image edge data and the preset reference image edge data, and it is judged whether image adjustment is required. Then, the wound surface rule index is obtained based on the obtained kidney wound surface rule data and the preset reference wound surface rule data, and it is judged whether to execute the function of the wound surface segmentation index acquisition module. Finally, the wound surface segmentation index is obtained by combining the obtained wound surface segmentation data and the preset reference segmentation data, and segmentation adjustment is performed, thereby improving the accuracy of kidney wound surface segmentation.

[0036] To better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings of the specification and specific implementation manners.

[0037] Such as Figure 1As shown in the figure, it is a schematic structural diagram of an intelligent kidney wound segmentation system provided by an embodiment of the present invention, including: an image enhancement module, an edge blur index acquisition module, a wound rule index acquisition module, and a wound segmentation index acquisition module. Among them, the image enhancement module is used to obtain a kidney wound image through a preset medical imaging device, perform image enhancement processing on the kidney wound image to obtain a characteristic wound image and kidney image edge data. The image enhancement processing is used to enhance the wound characteristics of the kidney wound image. The characteristic wound image represents the kidney wound image after image enhancement processing, and the kidney image edge data is used to describe the edge characteristics of the characteristic wound image. The edge blur index acquisition module is used to obtain an edge blur index based on the acquired kidney image edge data and the preset reference image edge data obtained from a preset database, and combine the preset edge threshold obtained from the preset database to determine whether to perform image adjustment. If image adjustment is not performed, the function of the wound rule index acquisition module is executed. The edge blur index is used to evaluate the edge blur degree of the characteristic wound image, and image adjustment is used to adjust the edge blur index to not be higher than the preset edge threshold. The wound rule index acquisition module is used to obtain a wound rule index based on the acquired kidney wound rule data and the preset reference wound rule data obtained from a preset database, and combine the preset wound rule threshold obtained from the database to determine whether to execute the function of the wound segmentation index acquisition module. The wound rule index is used to evaluate the edge rule situation of the characteristic wound image. The wound segmentation index acquisition module is used to perform segmentation processing on the characteristic wound image to obtain a segmented wound image and wound segmentation data, and combine the preset reference segmentation data obtained from a preset database to obtain a wound segmentation index, and perform segmentation adjustment based on the wound segmentation index and the preset segmentation threshold obtained from the preset database. The segmented wound image represents the characteristic wound image after segmentation processing. The wound segmentation index is used to evaluate the segmentation situation of the characteristic wound image. The segmentation processing is used to segment the characteristic wound image through an image segmentation algorithm, and the segmentation adjustment is used to adjust the wound segmentation index to not be lower than the preset segmentation threshold.

[0038] In this embodiment, the edge blur index is used to evaluate the edge blur degree of the characteristic wound surface image. If the edge blur index is higher than the preset edge threshold, image adjustment may be required to improve the image quality and make its edges clearer. The wound surface regularity index is used to evaluate the edge regularity of the characteristic wound surface image. If the wound surface regularity index is lower than the preset wound surface regularity threshold, further processing is required to improve the segmentation result. The wound surface segmentation index is used to evaluate the segmentation of the characteristic wound surface image. Segmentation adjustment ensures that the wound surface segmentation index is not lower than the preset segmentation threshold to ensure the accuracy of segmentation. These three indices are interrelated and act together in the process of kidney segmentation. The accuracy of segmentation can be evaluated through the edge blur index and the wound surface regularity index, while the wound surface segmentation index is a direct indicator for evaluating the segmentation effect. The preset reference image edge data includes a preset gradient threshold, a preset gray value change rate threshold, a preset gradient deviation threshold, and a preset gray value change rate deviation threshold. The preset reference wound surface regularity data includes a preset curvature threshold, a preset distance standard deviation threshold, a preset curvature deviation threshold, and a preset distance standard deviation deviation threshold. The preset reference segmentation data includes a preset segmentation precision threshold and a preset intersection over union threshold. The preset reference image edge data, the preset reference wound surface regularity data, and the preset reference segmentation data are obtained from the preset database, achieving an improvement in the accuracy of kidney wound surface segmentation.

[0039] It should be understood that after obtaining an edge blur index not higher than the preset edge threshold after image adjustment, the function of the wound surface regularity index module is continued to be executed. It is judged whether to execute the function of the wound surface segmentation index acquisition module in combination with the preset wound surface regularity threshold obtained from the database. When the wound surface regularity index is not lower than the preset wound surface regularity threshold, the function of the wound surface segmentation index acquisition module is continued to be executed. Segmentation adjustment is performed to obtain a wound surface segmentation index not lower than the preset segmentation threshold.

[0040] Optionally, the specific process for obtaining the characteristic wound surface image is as follows: The kidney wound surface image is obtained by scanning the kidney area with a preset medical imaging device, and the preset medical imaging device includes computed tomography, magnetic resonance imaging, and ultrasonic imaging devices. The kidney wound surface image is subjected to image enhancement processing to obtain the characteristic wound surface image. The image enhancement processing includes contrast enhancement, edge enhancement, and region enhancement. Contrast enhancement means enhancing the contrast of the kidney wound surface image through the histogram equalization method. Edge enhancement means highlighting the edge contour of the kidney wound surface in the kidney wound surface image through the Sobel edge detection algorithm. Region enhancement means enhancing the characteristics of the preset kidney wound surface area through the local histogram equalization method. A rectangular coordinate system has been added to the characteristic wound surface image.

[0041] In this embodiment, the contrast of the kidney wound image is enhanced by the histogram equalization method, making the gray-scale difference between different tissues in the image more obvious, which is helpful for subsequent edge detection and region segmentation. The kidney wound image is converted into a grayscale image, and the histogram equalization function in the image processing library is used to calculate the histogram of the input grayscale image, making the histogram distribution of the image uniform, thereby improving the contrast of the image. The Sobel edge detection algorithm is used to highlight the edge contour of the kidney wound in the kidney wound image. The Sobel operator is used to convolve the image in the horizontal and vertical directions to calculate the gradient magnitude. Edge enhancement helps to clarify the boundary of the kidney, which is very important for the segmentation algorithm because accurate edge information can guide the segmentation algorithm to more accurately identify the target area. The local histogram equalization method enhances the characteristics of the preset kidney wound area. Histogram equalization is applied to each local area (such as the kidney wound area), and histogram equalization is performed on each block, and then methods such as bilinear interpolation are used to smooth the transition between blocks. The processed local areas are merged back into the original image to form a complete enhanced image. Region enhancement helps to improve the contrast and details within a specific region, enabling the segmentation algorithm to better distinguish different tissues and structures within the kidney. In the field of kidney wound segmentation, contrast enhancement provides a better basis for edge detection, and the result of edge detection provides guidance for region enhancement, enabling region enhancement to be more accurately applied to the required regions; the accuracy of kidney wound segmentation is improved.

[0042] Optionally, the specific process of obtaining the edge blur index based on the obtained kidney image edge data and the preset reference image edge data obtained from the preset database is as follows: Edge data extraction is performed on the feature wound image to obtain kidney image edge data. The kidney image edge data includes a first gray value, a second gray value, and a pixel point distance. Edge data extraction means extracting the kidney image edge data in the feature wound image through the Sobel edge detection algorithm. The kidney image edge data is used to describe the edge contour of the kidney wound in the feature wound image. The first gray value represents the gray value of the pixel point on the feature wound image, the second gray value represents the gray value of the adjacent pixel point of the pixel point corresponding to the first gray value, and the pixel point distance represents the distance between the pixel points corresponding to the first gray value and the second gray value; The preset gradient threshold and the preset gray value change rate threshold are obtained from the preset database, and the edge gradient magnitude deviation and the gray value change rate deviation are obtained in combination with the kidney image edge data. The gray value change rate deviation represents the difference between the preset gray value change rate threshold and the gray value change rate on the feature wound image, and the edge gradient magnitude deviation represents the difference between the preset gradient threshold and the edge gradient magnitude on the feature wound image; The preset gradient deviation threshold and the preset gray value change rate deviation threshold are obtained from the preset database, and the edge blur index is obtained in combination with the edge gradient magnitude deviation and the gray value change rate deviation.

[0043] Among them, the edge blurring index is calculated using the following formula:

[0044]

[0045]

[0046] In the formula, α i represents the edge blurring index of the i-th pixel on the characteristic wound surface image, i = 1, 2,..., n - 1, i represents the pixel number on the characteristic wound surface image, n represents the total number of pixels on the characteristic wound surface image, R i represents the deviation of the gray value change rate of the i-th pixel on the characteristic wound surface image, G i represents the deviation of the edge gradient amplitude of the i-th pixel on the characteristic wound surface image, I i represents the first gray value of the i-th pixel on the characteristic wound surface image, I i‘ represents the second gray value of the i-th pixel on the characteristic wound surface image, ΔX i represents the pixel distance corresponding to the i-th pixel on the characteristic wound surface image, G0 represents the preset gradient threshold, R0 represents the preset gray value change rate threshold, ΔR represents the preset gray value change rate deviation threshold, ΔG represents the preset gradient deviation threshold, represents the first-order derivative of the gray value corresponding to the i-th pixel in the x-axis direction, represents the first-order derivative of the gray value corresponding to the i-th pixel in the y-axis direction, e represents the natural constant.

[0047] In this embodiment, the preset gradient threshold is represented by directly obtaining the maximum value of the edge gradient data in the historical period in the preset database, the preset gray value change rate threshold is represented by directly obtaining the maximum value of the gray value change rate data in the historical period in the preset database, the preset gradient deviation threshold is represented by directly obtaining the average value of the gradient deviation data in the historical period in the preset database, and the preset gray value change rate deviation threshold is represented by directly obtaining the average value of the gray value change rate deviation data in the historical period in the preset database. In the field of kidney image segmentation, the interaction of these three parameters, namely the first gray value, the second gray value, and the pixel distance, is crucial for accurately describing the edge contour of the kidney wound surface. The first gray value represents the gray value of the pixel on the characteristic wound surface image, which reflects the brightness information of the pixel in the image. The second gray value represents the gray value of the adjacent pixel, and the adjacent pixel refers to the pixel with the smallest pixel distance corresponding to the first gray value. The comparison between it and the first gray value can reveal the brightness change between pixels, and this change is often an indication of the existence of an edge. The pixel distance provides the spatial distribution information of these changes, which helps to more accurately locate the edge; thus improving the accuracy of kidney wound surface segmentation.

[0048] Specifically, the algorithm of this embodiment combines the edge data of the kidney image for comprehensive analysis to obtain the edge blur index. There is a common adjustment association among the edge data of the kidney image in this formula. For example, the difference between the first gray value and the second gray value reflects the degree of gray value change at the edge. The first gray value, the second gray value, and the pixel distance jointly determine the accuracy and continuity of the edge. When the pixel distance increases, this spatial correlation weakens, and the gray value difference between adjacent pixels increases. As the second gray value and the first gray value increase, when the increase amplitude of the first gray value is less than that of the second gray value, the edge blur index decreases, which means that the second gray value is inversely proportional to the edge blur index. The gray value change rate reflects the severity of the gray value change at the edge. The smaller the change rate, the blurrier the edge. The gray value change rate reflects the difference degree of the gray values of the pixels on both sides of the edge, and the edge gradient amplitude is the quantitative representation of this difference in the edge direction.

[0049] Specifically, assume that the deviation R of the gray value change rate i ranges from 0 to 10, and the deviation G of the edge gradient amplitude i ranges from 0 to 50. The preset deviation threshold ΔR of the gray value change rate is 5, and the preset gradient deviation threshold ΔG is 25. The change statistical table of the edge blur index is shown in Table 1:

[0050] Table 1 Change Statistical Table of Edge Blur Index

[0051] <![CDATA[Deviation R of gray value change rate i > <![CDATA[Edge gradient magnitude deviation G i > <![CDATA[Edge blur index α i > 8.66 46.02 2.6627 5.83 38.79 2.2847 2.33 26.26 1.6735 ... ... ...

[0052] As can be seen from Table 1, as the deviation R of the gray value change rate i and the deviation G of the edge gradient amplitude i gradually decrease, the edge blur index gradually decreases, which means that the edge blur degree of the characteristic wound surface image gradually decreases. In the kidney wound surface segmentation, accurate boundary recognition is the key to improving the segmentation accuracy. Clear edge information is crucial for determining the position and shape of the wound surface; the accuracy of kidney wound surface segmentation is improved.

[0053] Optionally, the specific process of determining whether to perform image adjustment in combination with a preset edge threshold obtained from a preset database is as follows: Q1, determine whether the edge blur index is not higher than the preset edge threshold. When the edge blur index is not higher than the preset edge threshold, no image adjustment is performed; otherwise, execute Q2. Q2, perform sharpening processing on the feature wound surface image. When the monitored edge blur index is not higher than the preset edge threshold, end the image adjustment; otherwise, execute Q3. The sharpening processing is used to enhance the edge information in the feature wound surface image to improve clarity. Q3, perform noise removal processing on the feature wound surface image. When the monitored edge blur index is not higher than the preset edge threshold, end the image adjustment; otherwise, the system prompts the preset personnel to re-obtain the kidney wound surface image. The noise removal processing is used to reduce and eliminate the random noise in the feature wound surface image.

[0054] In this embodiment, the preset edge threshold is represented by directly obtaining the average value of the gray variance data in the historical period in the preset database. The purpose of the sharpening processing is to enhance the edge information in the feature wound surface image to improve the clarity of the image. The noise removal processing aims to reduce and eliminate the random noise in the feature wound surface image. Noise may interfere with the edge recognition of the segmentation algorithm, resulting in inaccurate segmentation results. Gaussian filtering is a linear smoothing filter suitable for removing Gaussian noise. Median filtering is a non-linear filtering technique especially suitable for removing salt-and-pepper noise. In the field of kidney wound surface segmentation, image enhancement processing is a key step to improve segmentation accuracy. The kidney wound surface image obtained by the preset medical imaging device can significantly improve the quality of the feature wound surface image after image enhancement processing, thereby improving the accuracy and efficiency of segmentation. Read the feature wound surface image using an image processing library, create a sharpening kernel and perform a convolution operation with the image to highlight the edge information. The Laplacian operator is a second-order derivative operator commonly used for image sharpening. It can enhance the edge part of the image and make the image look clearer. Remove salt-and-pepper noise through median filtering and remove Gaussian noise through Gaussian filtering or bilateral filtering, achieving an improvement in the accuracy of kidney wound surface segmentation.

[0055] Optionally, the specific process of obtaining the wound rule index based on the obtained kidney wound rule data and the preset reference wound rule data obtained from the preset database is as follows: Obtain the kidney wound rule data through a preset imaging measurement device. The kidney wound rule data includes the tangent angle, curve arc length, and edge point distance. The preset imaging measurement device includes a CT scanner and a three-dimensional slicer. The edge point distance represents the distance between the marked edge point on the edge curve of the characteristic wound image and the center point of the preset edge curve; Obtain the preset curvature threshold and the preset distance standard deviation threshold from the preset database, and combine the kidney wound rule data to obtain the curvature deviation and the distance standard deviation deviation. The curvature deviation represents the difference between the preset curvature threshold and the curvature of the edge curve in the characteristic wound image, and the distance standard deviation deviation represents the difference between the preset distance standard deviation threshold and the distance standard deviation of the edge curve in the characteristic wound image; Obtain the preset curvature deviation threshold and the preset distance standard deviation deviation threshold from the preset database, and combine the curvature deviation and the distance standard deviation deviation to obtain the curvature influence degree and the distance standard deviation influence degree. The curvature influence degree is used to evaluate the influence degree of the curvature deviation on the wound rule index, and the distance standard deviation influence degree is used to evaluate the influence degree of the distance standard deviation deviation on the wound rule index; Obtain the preset rule first weight and the preset rule second weight from the preset database, and combine the curvature influence degree and the distance standard deviation influence degree to obtain the wound rule index. The preset rule first weight is used to evaluate the influence degree of the curvature influence degree on the wound rule index, and the preset rule second weight is used to evaluate the influence degree of the distance standard deviation influence degree on the wound rule index.

[0056] Among them, the wound rule index is calculated using the following formula:

[0057] β f = tanh(k f *ρ1 + D f *ρ2);

[0058]

[0059] In the formula, β f represents the wound rule index of the edge curve corresponding to the f-th wound on the characteristic wound image, f = 1, 2,..., b. f represents the number of the wound on the characteristic wound image, and b represents the total number of wounds on the characteristic wound image. k f represents the curvature influence degree of the edge curve corresponding to the f-th wound on the characteristic wound image, D f represents the distance standard deviation influence degree of the edge curve corresponding to the f-th wound on the characteristic wound image, θ vf represents the tangent angle corresponding to the v-th marked edge point on the edge curve corresponding to the f-th wound on the characteristic wound image, v = 1, 2,..., a. v represents the number of the marked edge points on the edge curve, and a represents the total number of the marked edge points on the edge curve, Sf denotes the arc length of the edge curve corresponding to the f-th wound surface on the characteristic wound surface image, d vf denotes the edge point distance corresponding to the v-th marked edge point on the edge curve corresponding to the f-th wound surface on the characteristic wound surface image, k0 denotes a preset curvature threshold, d0 denotes a preset distance standard deviation threshold, Δk denotes a preset curvature deviation threshold, Δd denotes a preset distance standard deviation deviation threshold, ρ1 denotes a preset regularity first weight, ρ2 denotes a preset regularity second weight denotes with respect to S f the first derivative of, where e represents the natural constant

[0060] In this embodiment, the preset curvature deviation threshold is obtained from a preset database, and the curvature influence degree is obtained by combining the preset curvature threshold, the kidney wound surface rule data, and the preset curvature deviation threshold; the preset distance standard deviation deviation threshold is obtained from a preset database, and the distance standard deviation influence degree is obtained by combining the preset distance standard deviation threshold, the kidney wound surface rule data, and the preset distance standard deviation deviation threshold; the preset weights are obtained from a preset database, and the wound surface rule index is obtained by combining the obtained curvature influence degree and the distance standard deviation influence degree. The preset weights include a preset regularity first weight and a preset regularity second weight. The preset curvature threshold is represented by directly obtaining the maximum value of the curvature data in the historical period in the preset database, the preset distance standard deviation threshold is represented by directly obtaining the maximum value of the distance standard deviation data in the historical period in the preset database, the preset curvature deviation threshold is represented by directly obtaining the average value of the curvature deviation data in the historical period in the preset database, the preset distance standard deviation deviation threshold is represented by directly obtaining the average value of the distance standard deviation deviation data in the historical period in the preset database. A CT scanner is used to obtain the kidney wound surface image, and a three-dimensional slicer is used for image reconstruction. An edge detection algorithm is used to extract the edge. The direction of the tangent is calculated at each point on the edge curve, and the tangent angle can be calculated by the angle between the tangent direction and the horizontal line. The arc length of the curve is obtained by accumulating the distances between the edge points; the improvement of the accuracy of kidney wound surface segmentation is realized

[0061] Specifically, the sum of the preset regular first weight and the preset regular second weight is 1. For example, the preset regular first weight is 0.5 and the preset regular second weight is 0.5. The preset regular first weight is the weight corresponding to the preset curvature value in the preset database, which is a value representing the influence degree of the curvature value on the wound surface regularity index. When in use, the weight corresponding to the preset curvature value can be directly obtained from the preset database, and its corresponding relationship can be a pre-set mapping relationship. For example, in the kidney wound surface segmentation training set, a mapping set is formed between the curvature and the weight corresponding to the preset curvature value in the preset database. The real-time intersection area is input into the mapping set to obtain the corresponding weight, and the mapping relationship therein can be a one-to-one or many-to-one relationship. In this example, its value range is [0, 1].

[0062] The algorithm of this embodiment combines the kidney wound surface regular data for comprehensive analysis to obtain the wound surface regularity index. There is a common adjustment association among the kidney wound surface regular data in this formula. For example, the tangent angle reflects the rate of change of the direction of the edge curve at a specific point, and the curve arc length describes the length of the edge curve, which is directly related to the actual size and shape of the kidney wound surface. The change of the tangent angle will affect the measurement of the curve arc length because the arc length is the result of integrating along the edge curve, and the edge point distance is related to both the tangent angle and the curve arc length because it describes the position of the edge point relative to the center point. As the tangent angle and the edge point distance increase, the curve arc length decreases, and the wound surface regularity index increases, which means that the tangent angle and the edge point distance are directly proportional to the wound surface regularity index. The greater the curvature deviation, the more irregular the shape of the edge curve, and the greater the impact on the wound surface regularity index. The greater the distance standard deviation deviation, the more uneven the distribution of the edge points, which will also increase the irregularity of the wound surface. When the shape of the edge curve is distorted and the distribution of the edge points is extremely uneven (the distance standard deviation deviation is large), the curvature deviation and the distance standard deviation deviation may increase simultaneously, jointly aggravating the irregularity of the wound surface. The curve arc length is inversely proportional to the wound surface regularity index;

[0063] Specifically, assume the curvature influence degree k f ranges from 0 to 2, the distance standard deviation influence degree D f ranges from 0 to 2, the preset regular first weight ρ1 is 0.5, and the preset regular second weight ρ2 is 0.5. As Figure 2 shown, it is the change statistical chart of the wound surface regularity index provided by the embodiment of the present invention. As Figure 2As shown, when the curvature influence degree is 2, as the distance standard deviation influence degree increases, the wound surface regularity index gradually increases. When the distance standard deviation influence degree is 1, as the curvature influence degree increases, the wound surface regularity index gradually increases, indicating that the edge regularity of the characteristic wound surface image is gradually improved. In the field of kidney wound surface segmentation, by analyzing the change trends of the curvature influence degree and the distance standard deviation influence degree, the boundary of the kidney can be more accurately identified and located, achieving an improvement in the accuracy of kidney wound surface segmentation.

[0064] Optionally, perform segmentation processing on the characteristic wound surface image to obtain a segmented wound surface image and wound surface segmentation data, and combine the preset reference segmentation data obtained from the preset database to obtain a wound surface segmentation index. The specific process is as follows: Obtain the wound surface segmentation data through a three-dimensional slicer. The wound surface segmentation data includes the intersection area, union area, number of positive class pixels, and marked number of positive class pixels. The intersection area represents the area of the overlapping part of the wound surface area in the segmented wound surface image and the manually marked wound surface area. The union area represents the area covered after combining the wound surface area in the segmented wound surface image and the manually marked wound surface area. Obtain the preset segmentation precision threshold, preset intersection-over-union threshold, preset first segmentation weight, and preset second segmentation weight from the preset database, and combine the wound surface segmentation data to obtain the wound surface segmentation index. The preset first segmentation weight is used to evaluate the influence degree of the segmentation precision on the wound surface segmentation index, and the preset second segmentation weight is used to evaluate the influence degree of the intersection-over-union on the wound surface segmentation index.

[0065] Among them, the wound surface segmentation index is calculated using the following formula:

[0066]

[0067] In the formula, γ p represents the wound surface segmentation index of the p-th segmented wound surface image, p = 1, 2,..., m, p represents the number of the segmented wound surface image, m represents the total number of segmented wound surface images, S 1p represents the corresponding intersection area of the p-th segmented wound surface image, S 2p represents the corresponding union area of the p-th segmented wound surface image, N 1p represents the number of positive class pixels in the p-th segmented wound surface image, N 2p represents the marked number of positive class pixels in the p-th segmented wound surface image, S0 represents the preset intersection-over-union threshold, N0 represents the preset segmentation precision threshold, ω1 represents the preset first segmentation weight, and ω2 represents the preset second segmentation weight.

[0068] In this embodiment, a preset intersection over union (IoU) threshold is obtained from a preset database, and the difference between the preset IoU threshold and the ratio of the intersection area to the union area is obtained by combining the intersection area and the union area; a preset segmentation precision threshold is obtained from the preset database, and the difference between the preset segmentation precision threshold and the ratio of the number of positive-class pixels to the number of labeled positive-class pixels is obtained by combining the number of positive-class pixels and the number of labeled positive-class pixels; a first segmentation weight and a preset second segmentation weight are obtained from the preset database, and a wound surface segmentation index is obtained by combining the preset IoU threshold and the difference between the ratio of the intersection area to the union area, the preset segmentation precision threshold and the difference between the ratio of the number of positive-class pixels to the number of labeled positive-class pixels; the preset reference segmentation data includes the preset segmentation precision threshold, the preset IoU threshold, the preset first segmentation weight, and the preset second segmentation weight, and a wound surface segmentation index is obtained by combining the wound surface segmentation data. The preset precision threshold is represented by directly obtaining the maximum value of the kidney wound surface segmentation precision data in the historical period in the preset database, and the preset IoU threshold is represented by directly obtaining the maximum value of the kidney wound surface IoU data in the historical period in the preset database. An artificially labeled wound surface image is obtained and segmented to obtain a segmented wound surface image. The two images are converted into binary images in a three-dimensional slicer, where the wound surface area is 1 and other areas are 0. A bitwise "AND" operation is performed on the segmented wound surface image and the artificially labeled wound surface image, and the number of pixels with a value of 1 in the result image is calculated to obtain the intersection area. A bitwise "OR" operation is performed on the segmented wound surface image and the artificially labeled wound surface image, and the number of pixels with a value of 1 in the result image is calculated to obtain the union area. The total number of pixels with a value of 1 in the segmented wound surface image, that is, the number of positive-class pixels, is calculated, and the total number of pixels with a value of 1 in the artificially labeled wound surface image, that is, the number of labeled positive-class pixels, is calculated; the accuracy of kidney wound surface segmentation is improved.

[0069] Specifically, the sum of the preset first segmentation weight and the preset second segmentation weight is 1. For example, the preset first segmentation weight is 0.5 and the preset second segmentation weight is 0.5; the preset first segmentation weight is the weight corresponding to the preset intersection area value in the preset database, which represents the value of the influence degree of the intersection area value on the wound surface segmentation index. When used, the weight corresponding to the preset intersection area value can be directly obtained from the preset database, and its corresponding relationship can be a preset mapping relationship. For example, a mapping set is formed between the intersection area of the kidney wound surface segmentation training set and the weight corresponding to the preset intersection area value in the preset database, and the real-time intersection area is input into the mapping set to obtain the corresponding weight, where the mapping relationship can be a one-to-one or many-to-one relationship. In this example, its value range is [0, 1].

[0070] The algorithm of this embodiment combines the wound surface segmentation data for comprehensive analysis to obtain the wound surface segmentation index. There is a common adjustment association among the wound surface segmentation data in this formula. For example, an increase in the intersection area usually means that more positive-class pixels are correctly segmented. In the segmentation result, if pixels in the unlabeled area are wrongly included, then the union area will increase and the precision will decrease. An increase in the number of positive-class pixels usually means that the segmentation algorithm has a stronger ability to recognize the target area, which helps to improve the accuracy and integrity of the segmentation. In the wound surface segmentation index, the number of marked positive-class pixels is used as a reference value to evaluate the integrity of the segmentation result relative to the true annotation. The intersection over union and the precision are both important indicators for measuring the segmentation performance, and they evaluate the quality of the segmentation result from different perspectives. To increase the intersection over union, the precision needs to be reduced to cover more true target areas. As the intersection area and the number of positive-class pixels increase, and the union area and the number of marked positive-class pixels decrease, the wound surface segmentation index gradually increases, indicating that the intersection area and the number of positive-class pixels are directly proportional to the wound surface segmentation index, and the union area and the number of marked positive-class pixels are inversely proportional to the wound surface segmentation index; the accuracy of kidney wound surface segmentation is improved.

[0071] Optionally, the specific process of segmentation adjustment based on the wound surface segmentation index and the preset segmentation threshold obtained from the preset database is as follows: M1, determine whether the wound surface segmentation index is not lower than the preset segmentation threshold. When the wound surface segmentation index is not lower than the preset segmentation threshold, no segmentation adjustment is performed; otherwise, continue to execute M2; M2, obtain the preset segmentation threshold from the preset database, send a prompt to the preset personnel to gradually reduce the segmentation area by a preset multiple of the preset segmentation threshold, and end the segmentation adjustment when the monitored wound surface segmentation index is not lower than the preset segmentation threshold; otherwise, continue to execute M3; M3, perform region merging, and end the segmentation adjustment when the monitored wound surface segmentation index is not lower than the preset segmentation threshold; otherwise, feedback to the preset personnel to re-obtain the kidney wound surface image. Region merging means reducing the degree of discontinuity of the segmentation area through the region merging algorithm.

[0072] In this embodiment, the preset segmentation threshold is represented by the average value of the segmentation data in the preset database, and the preset segmentation threshold is represented by the average value of the wound surface segmentation index data in the preset database. When the wound surface segmentation index is not lower than the preset segmentation threshold, it indicates that the current segmentation effect has reached an acceptable level and the operation stops. If the wound surface segmentation index is still lower than the threshold, it indicates that the segmentation effect is not good. Region merging is a process of reducing the degree of discontinuity of the segmentation area through the region merging algorithm, which helps to improve the coherence and integrity of the segmentation area. The interaction between these two steps lies in that they jointly constitute an iterative process, and by gradually adjusting the segmentation area and region merging, the accuracy of kidney wound surface segmentation is improved.

[0073] Such as Figure 3As shown, it is the postoperative kidney wound surface diagram provided by this embodiment, which is composed of Figure 3 It can be seen that it is the kidney wound surface left after nephrectomy. There is a layer of capsule on the surface of the kidney, and the divided part is the wound surface. The kidney wound surface and the rupture boundary of the renal capsule are marked in the figure.

[0074] Such as Figure 4 As shown, it is the kidney diagram before wound surface segmentation provided by the embodiment of the present invention, which is composed of Figure 4 It can be seen that it shows the state of the kidney before wound surface segmentation. During the kidney wound surface segmentation, as the operation progresses, the wound surface may be affected by factors such as the traction of surrounding tissues and the movement of surgical instruments, resulting in changes in the shape, size, and position of the wound surface. At this time, the edge points and edge curves also change correspondingly.

[0075] Such as Figure 5 As shown, it is the kidney diagram after wound surface segmentation provided by the embodiment of the present invention, which is composed of Figure 5 It can be seen that it shows the state of the kidney after wound surface segmentation.

[0076] Such as Figure 6 As shown, it is the split framework diagram of the surgical process provided by the embodiment of the present invention, which is composed of Figure 6 It can be seen that it shows the overall process of the robot-assisted partial nephrectomy operation.

[0077] Such as Figure 7 As shown, it is the surgical stage diagram provided by the embodiment of the present invention, which is composed of Figure 7 It can be seen that it shows the surgical stage and the corresponding marked ID, access point, and marked position.

[0078] Such as Figure 8 As shown, it is the surgical step diagram provided by the embodiment of the present invention, which is composed of Figure 8 It can be seen that it shows the specific steps of the operation and the corresponding ID, surgical stage, start state, and end state.

[0079] In summary, in the embodiment of the present invention, the edge blur index is obtained through the kidney image edge data and the preset reference image edge data. Then, the wound surface rule index is obtained according to the obtained kidney wound surface rule data and the preset reference wound surface rule data. Finally, the wound surface segmentation index is obtained by combining the obtained wound surface segmentation data and the preset reference segmentation data, thereby realizing the improvement of the comprehensiveness of obtaining the kidney wound surface segmentation data, and further realizing the improvement of the accuracy of the kidney wound surface segmentation, effectively solving the problem of low correlation between the kidney wound surface segmentation and the wound surface edge in the prior art.

[0080] The following points need to be explained:

[0081] (1) The attached drawings of the embodiment of the present invention only relate to the structures involved in the embodiment of the present invention, and other structures can refer to the general design.

[0082] (2) For clarity, in the drawings used to describe the embodiments of the present invention, the thickness of layers or regions is enlarged or reduced, that is, these drawings are not drawn to actual scale. It will be understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element can be "directly" on or under the other element or intervening elements may be present.

[0083] (3) Without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0084] The above are only the specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A kidney wound intelligent segmentation system, characterized in that: It includes image enhancement module, edge fuzzy index acquisition module, wound surface rule index acquisition module and wound surface segmentation index acquisition module: The image enhancement module is used to obtain a renal wound surface image through a preset medical imaging device, and perform image enhancement processing on the renal wound surface image to obtain a characteristic wound surface image and renal image edge data; the image enhancement processing is used to enhance the wound surface characteristics of the renal wound surface image, the characteristic wound surface image represents the renal wound surface image after image enhancement processing, and the renal image edge data is used to describe the edge characteristics of the characteristic wound surface image; The edge blur index acquisition module is used to obtain the edge blur index according to the acquired kidney image edge data and the preset reference image edge data obtained from the preset database, and determine whether to perform image adjustment in combination with the preset edge threshold obtained from the preset database. If the image adjustment is not performed, the wound surface rule index acquisition module function is executed. The edge blur index is used to evaluate the edge blur degree of the characteristic wound surface image, and the image adjustment is used to adjust the edge blur index to be no higher than the preset edge threshold; The wound surface rule index acquisition module is used to acquire the wound surface rule index according to the acquired kidney wound surface rule data and the preset reference wound surface rule data acquired from the preset database, and determine whether to execute the wound surface segmentation index acquisition module function in combination with the preset wound surface rule threshold value acquired from the database, and the wound surface rule index is used to evaluate the edge rule situation of the characteristic wound surface image; The wound surface segmentation index acquisition module is used to perform segmentation processing on the characteristic wound surface image to obtain a segmented wound surface image and wound surface segmentation data, and to obtain a wound surface segmentation index in combination with preset reference segmentation data obtained from a preset database, and to perform segmentation adjustment based on the wound surface segmentation index and a preset segmentation threshold value obtained from the preset database, wherein the segmented wound surface image represents the characteristic wound surface image after segmentation processing, the wound surface segmentation index is used to evaluate the segmentation status of the characteristic wound surface image, the segmentation processing is used to segment the characteristic wound surface image through an image segmentation algorithm, and the segmentation adjustment is used to adjust the wound surface segmentation index to be not less than the preset segmentation threshold value; When the edge blur index is obtained after image adjustment and is not higher than the preset edge threshold, the wound rule index module acquisition function continues to be executed; combined with the preset wound rule threshold obtained from the database, it is determined whether to execute the wound segmentation index acquisition module function. When the wound rule index is not lower than the preset wound rule threshold, the wound segmentation index acquisition module function continues to be executed.

2. The intelligent renal wound segmentation system according to claim 1, characterized in that: The specific process of obtaining the characteristic wound surface image is as follows: Scanning the kidney area by using a preset medical imaging device to obtain a kidney wound surface image, wherein the preset medical imaging device includes a computer tomography, a magnetic resonance imaging, and an ultrasound imaging device; Performing image enhancement processing on the kidney wound surface image to obtain a characteristic wound surface image, wherein the image enhancement processing includes contrast enhancement, edge enhancement and region enhancement.

3. The intelligent renal wound segmentation system according to claim 1, characterized in that: The specific process of obtaining the edge blur index based on the acquired kidney image edge data and the preset reference image edge data obtained from the preset database is as follows: Extracting edge data of the characteristic wound surface image to obtain kidney image edge data, wherein the kidney image edge data includes a first gray value, a second gray value, and a pixel point distance, and the kidney image edge data is used to describe the edge contour of the kidney wound surface in the characteristic wound surface image; Obtaining a preset gradient threshold and a preset gray value change rate threshold from a preset database, and obtaining an edge gradient amplitude deviation and a gray value change rate deviation in combination with kidney image edge data; A preset gradient deviation threshold and a preset gray value change rate deviation threshold are obtained from a preset database, and an edge blur index is obtained by combining the edge gradient amplitude deviation and the gray value change rate deviation.

4. The intelligent renal wound segmentation system according to claim 3, characterized in that: The edge blur index is calculated using the following formula: In the formula, α i represents the edge blur index of the i-th pixel on the characteristic wound surface image, i=1,2,...,n-1, i represents the number of the pixel on the characteristic wound surface image, n represents the total number of pixels on the characteristic wound surface image, R i G represents the gray value change rate deviation of the i-th pixel on the characteristic wound image. i Indicates the edge gradient amplitude deviation of the i-th pixel on the characteristic wound surface image, I i Represents the first gray value of the i-th pixel on the characteristic wound surface image, I i‘ Indicates the second gray value of the i-th pixel on the characteristic wound image, ΔX i represents the pixel distance corresponding to the i-th pixel on the characteristic wound image, G0 represents the preset gradient threshold, R0 represents the preset gray value change rate threshold, ΔR represents the preset gray value change rate deviation threshold, ΔG represents the preset gradient deviation threshold, It represents the first-order derivative of the gray value corresponding to the i-th pixel in the x-axis direction. It represents the first-order derivative of the gray value corresponding to the i-th pixel in the y-axis direction, and e represents a natural constant.

5. The intelligent renal wound segmentation system according to claim 1, characterized in that: The specific process of determining whether to perform image adjustment in combination with the preset edge threshold obtained from the preset database is as follows: Q1, judging whether the edge blur index is not higher than the preset edge threshold, if the edge blur index is not higher than the preset edge threshold, no image adjustment is performed, otherwise Q2 is executed; Q2, sharpen the characteristic wound surface image. When the monitored edge blur index is not higher than the preset edge threshold, the image adjustment ends. Otherwise, Q3 is executed. Q3, perform noise removal on the characteristic wound surface image, and end image adjustment when the monitored edge blur index is not higher than the preset edge threshold. Otherwise, the system prompts the preset personnel to reacquire the kidney wound surface image.

6. The intelligent renal wound segmentation system according to claim 1, characterized in that: The specific process of obtaining the kidney wound surface rule data is as follows: The kidney wound surface regular data is obtained by a preset image measurement device, wherein the kidney wound surface regular data includes tangent angle, curve arc length, and edge point distance.

7. The intelligent renal wound segmentation system according to claim 6, characterized in that: The specific process of obtaining the wound surface rule index is as follows: Obtaining a preset curvature deviation threshold from a preset database, and obtaining a curvature influence by combining the preset curvature threshold, kidney wound surface rule data, and the preset curvature deviation threshold; Obtaining a preset distance standard deviation threshold from a preset database, and obtaining a distance standard deviation influence by combining the preset distance standard deviation threshold, kidney wound rule data, and the preset distance standard deviation threshold; The preset weights are obtained from the preset database, and the wound surface regularity index is obtained by combining the obtained curvature influence and distance standard deviation influence. The preset weights include a preset regularity first weight and a preset regularity second weight.

8. The intelligent renal wound segmentation system according to claim 1, characterized in that: The specific process of segmenting the characteristic wound surface image to obtain the segmented wound surface image and wound surface segmentation data is as follows: The wound segmentation data is obtained through a three-dimensional slicer, and the wound segmentation data includes an intersection area, a union area, a number of positive pixels, and a number of labeled positive pixels. The intersection area represents the area of ​​the overlapping part of the wound area in the segmented wound image and the manually annotated wound area, and the union area represents the area of ​​the covered part after the wound area in the segmented wound image and the manually annotated wound area are combined.

9. The intelligent renal wound segmentation system according to claim 8, characterized in that: The specific process of obtaining the wound segmentation index is as follows: Obtain a preset intersection-and-union ratio threshold from a preset database, and obtain the preset intersection-and-union ratio threshold and the difference between the ratio of the intersection area and the union area by combining the intersection area and the union area; Obtain a preset segmentation accuracy threshold from a preset database, and obtain the difference between the preset segmentation accuracy threshold and the ratio of the number of positive pixels to the number of labeled positive pixels by combining the number of positive pixels and the number of labeled positive pixels; Obtaining a first segmentation weight and a preset second segmentation weight from a preset database, and combining a preset intersection-to-union ratio threshold and a difference between the ratio of the intersection area and the union area, a preset segmentation accuracy threshold and a difference between the ratio of the number of positive pixels and the number of labeled positive pixels to obtain a wound segmentation index; The preset reference segmentation data includes a preset segmentation accuracy threshold, a preset intersection-over-union threshold, a preset first segmentation weight and a preset second segmentation weight, and is combined with the wound surface segmentation data to obtain a wound surface segmentation index.

10. The intelligent renal wound segmentation system according to claim 1, characterized in that: The specific process of performing segmentation adjustment based on the wound surface segmentation index and the preset segmentation threshold obtained from the preset database is as follows: M1, judging whether the wound segmentation index is not lower than the preset segmentation threshold. If the wound segmentation index is not lower than the preset segmentation threshold, no segmentation adjustment is performed, otherwise, M2 is continued; M2, obtaining a preset segmentation threshold from a preset database, sending a prompt to the preset personnel to gradually reduce the segmentation area by a preset multiple of the preset segmentation threshold, and ending the segmentation adjustment when the monitored wound segmentation index is not lower than the preset segmentation threshold, otherwise continue to execute M3; M3, performs regional merging. When the monitored wound segmentation index is not lower than the preset segmentation threshold, the segmentation adjustment is terminated. Otherwise, feedback is given to the preset personnel to reacquire the kidney wound image.

Images