Network image processing method and system based on fusion growth point optimization algorithm and Unet

Abstract

The present invention relates to the technical field of computer vision and deep learning algorithm optimization, specifically to an image processing method and system based on a fusion growth point optimization algorithm and a Unet network, comprising the following steps: collecting lumbar CT images, extracting the grayscale value and HU value of each pixel in the image, marking the lumbar vertebrae region to obtain density information of the lumbar vertebrae region, analyzing the lumbar vertebrae gradient and segmenting the initial region, obtaining basic region information points therefrom and classifying them, and generating preliminary structural features of the lumbar vertebrae. The present invention, by collecting lumbar CT images and extracting the grayscale value and HU value of pixels, obtains density information of the lumbar vertebrae region and performs preliminary segmentation, can accurately analyze the gradient changes in the region, and provides a reliable basis for the preliminary image region. In this process, by dynamically adjusting the image rotation angle and analyzing the difference in the boundary normal angle, the deviation of the lumbar vertebrae morphology is successfully corrected, thereby ensuring the consistency of the morphology in the subsequent processing process.

Description

Technical Field

[0001] The present invention relates to the technical field of computer vision and deep learning algorithm optimization, and in particular to a method and system for image processing based on a fusion growing point optimization algorithm and Unet network. Background Art

[0002] The field of computer vision and deep learning algorithm optimization encompasses the analysis and processing of image and video data using computer algorithms and artificial intelligence methods. Core aspects of this technical field include image feature extraction, pattern recognition, object detection, and segmentation, aiming to extract meaningful information from large-scale image data through automated methods. Systematic research in computer vision encompasses multi-level algorithm development, including classic algorithms such as edge detection and texture analysis, as well as the rapidly developing deep learning-based model optimization methods, such as convolutional neural networks and transfer learning, which are widely used in medical image analysis, autonomous driving, security monitoring, and other fields.

[0003] Among them, image processing based on the fusion growth point optimization algorithm and Unet network refers to a method for processing orthopedic medical images using a fusion optimization algorithm and a deep learning network. This patent subject is aimed at the problem of high-precision regional segmentation that needs to be achieved in orthopedic images. It uses the fusion growth point optimization algorithm to improve the adaptability of the initial parameter settings, and combines the multi-scale feature extraction capabilities of the Unet network model to complete the specific regional segmentation processing of orthopedic images. By introducing an algorithm-based weight initialization and learning rate adjustment mechanism, combined with the cascade feature extraction method in the network, the effective separation and enhancement processing of regional features of orthopedic images can be achieved. This method completes the structured processing of image data by combining optimization strategies with deep learning models.

[0004] Although existing technologies have achieved good results in some basic algorithms when processing orthopedic images, due to the complexity of large-scale image data and the inaccuracy of feature extraction, they are often unable to make effective adjustments and corrections when faced with morphological deviations or irregular structures, resulting in certain errors in the segmentation results. In terms of initial parameter settings, existing technologies have poor adaptability to different orthopedic images and often rely on fixed parameters. This may lead to overfitting or underfitting in actual applications, affecting the final processing effect. Especially for images with strong curved surface features, existing technologies are insufficient in the extraction and optimization of curved surface details, and are prone to blurred or discontinuous boundaries, which makes the final image segmentation unable to provide accurate positioning or recognition in certain key areas. In addition, traditional methods rarely use dynamic adjustment mechanisms during the processing process and cannot flexibly respond to different image characteristics. Therefore, it is difficult to operate stably in various complex image scenes, affecting the versatility and practicality of the overall system. Summary of the Invention

[0005] In order to solve the technical problems existing in the prior art, the embodiment of the present invention provides a network image processing method and system based on the fusion growth point optimization algorithm and Unet. The technical solution is as follows:

[0006] Based on the fusion growth point optimization algorithm and Unet network image processing method, the following steps are included:

[0007] S1: Collect lumbar CT images, extract the grayscale value and HU value of each pixel in the image, mark the lumbar vertebrae to obtain lumbar bone density information, analyze the lumbar bone gradient and segment the initial region, obtain basic regional information points and classify them, and generate preliminary structural features of the lumbar vertebrae;

[0008] S2: extracting the central axis and boundary normal in the image from the preliminary structural features of the lumbar vertebrae, obtaining the angle difference between the central axis and the boundary normal, and dynamically adjusting the rotation angle of the image to correct the lumbar vertebrae morphological deviation, thereby generating an angle-corrected lumbar vertebrae region;

[0009] S3: Analyzing the surface characteristics of the angle-corrected lumbar vertebrae region using a differential geometry optimization algorithm, calculating local curvature changes based on the surface characteristics, identifying surface features of the lumbar vertebrae surface, mapping the surface features to a Unet network and extracting surface boundary information, optimizing the smoothness and connectivity of the surface boundary, and generating surface feature optimization results;

[0010] S4: identifying the gradient change of each pixel point according to the surface feature optimization result, marking the key points of the lumbar vertebrae region, clustering the key points to distinguish the lumbar vertebrae region from other tissue regions outside the lumbar vertebrae region, determining the boundary of the lumbar vertebrae region and segmenting it, and generating a lumbar vertebrae region segmentation boundary map;

[0011] S5: Enhance the lumbar vertebrae edge according to the lumbar vertebrae region segmentation boundary map, adjust the segmentation region with reference to the basic region information points, fill the boundary gaps and similar lumbar vertebrae features based on the adjusted segmentation region, iteratively correct and smooth the boundary, and generate an optimized lumbar vertebrae segmentation map.

[0012] The improvements of the present invention are that the preliminary structural features of the lumbar vertebrae include corner points, connection points, bending points, the central axis of the lumbar vertebrae area, and the boundary normal of the lumbar vertebrae area; the lumbar vertebrae area after angle correction includes the lumbar vertebrae area after rotation angle adjustment; the surface feature optimization results include the surface characteristics and surface boundary information of the lumbar vertebrae area; the lumbar vertebrae area segmentation boundary map includes the refined lumbar vertebrae area and the segmentation boundary; the optimized lumbar vertebrae segmentation map includes clear edges, adjusted segmentation areas, expanded segmentation areas, segmentation areas after noise removal, and corrected and smoothed boundaries.

[0013] The present invention has the following improvements:

[0014] S101: Collecting a lumbar spine CT image of the patient, extracting the grayscale value and HU value of each pixel in the CT image, comparing the pixel grayscale value with a preset grayscale threshold, marking lumbar vertebrae regions with grayscale thresholds higher than the preset grayscale threshold, screening regions with HU values exceeding the preset HU threshold and extracting density data, thereby generating density information of the marked lumbar vertebrae regions;

[0015] S102: analyzing the gradient change rate of the marked lumbar vertebrae region according to the density information of the marked lumbar vertebrae region, segmenting the initial region marked with the grayscale threshold with reference to the gradient information of the lumbar vertebrae region, obtaining basic region information points of the lumbar vertebrae according to the segmented grayscale regions, and generating basic region information points of the lumbar vertebrae region;

[0016] S103: Identify the positions of the corresponding corner points, connection points, and bending points in the basic information points of the lumbar vertebrae region, determine the preliminary structure of the lumbar vertebrae according to the position information of each point, and generate preliminary structural features of the lumbar vertebrae.

[0017] The present invention has the following improvements:

[0018] S201: extracting the central axis and boundary normal of the lumbar vertebra region from the preliminary structural features of the lumbar vertebra, locating the contour of the lumbar vertebra, analyzing the direction and position differences between the central axis and the boundary normal, and generating central axis and boundary normal information;

[0019] S202: Obtaining angle difference values between the central axis and the boundary normal from the central axis and boundary normal information, and generating angle difference results by referring to the difference analysis of each angle and the influence of the morphological changes of the lumbar vertebrae region;

[0020] S203: Dynamically adjust the rotation angle of the lumbar vertebra image according to the angle difference result, use image processing technology to perform rotation correction on the image, adjust the morphological deviation of the original shooting angle, mark the asymmetric area of the lumbar vertebra CT image, and generate the lumbar vertebra area after angle correction.

[0021] The present invention has the following improvements:

[0022] S301: performing geometric analysis on the angle-corrected lumbar vertebrae region to obtain surface information of the lumbar vertebrae region, analyzing surface features of the lumbar vertebrae region using a differential geometry optimization algorithm, the surface features including curvature and surface shape, and generating surface feature information of the lumbar vertebrae region;

[0023] S302: Calculating local curvature values of the lumbar vertebrae surface area based on the surface feature information of the lumbar vertebrae area, marking lumbar vertebrae areas exceeding a preset curvature threshold, and generating local curvature change information of the marked lumbar vertebrae areas;

[0024] S303: Identify the surface features of the lumbar vertebrae based on the local curvature change information of the lumbar vertebrae area, map the surface features onto the Unet network and extract the surface boundary information, optimize the smoothness and connectivity of the surface boundary, and generate a surface feature optimization result.

[0025] The present invention has the following improvements: for the local curvature value of the lumbar vertebrae surface area, the formula is used:

[0026] ;

[0027] Calculate the local curvature value of the lumbar vertebral surface area ;

[0028] in, 、 are the first and second curvature radii of the key points in the lumbar vertebrae area in the main curvature direction, 、 To adjust the first curvature radius and the second radius of curvature The correction factor, is the morphological flexibility factor, is the curvature change, is the change in curvature angle.

[0029] The present invention has the following improvements:

[0030] S401: using the Shi-Tomasi corner detection algorithm based on the surface feature optimization result to identify the gradient change of each pixel, analyzing the change amplitude of adjacent pixels, comparing the grayscale difference value of each adjacent pixel, and marking the key point area of the lumbar vertebrae region based on the comparison result;

[0031] S402: performing spatial clustering calculation on the key point region using a K-means clustering algorithm, identifying key points and grouping them according to position and gradient differences, distinguishing the lumbar vertebrae region from other tissue regions outside the lumbar vertebrae region based on the grouped key points, and generating a lumbar vertebrae region classification result;

[0032] S403: Compare the grayscale difference and gradient information according to the classification results of the lumbar vertebrae region, divide the boundary of each lumbar vertebrae region by morphological gradient, use the watershed algorithm to segment the boundary of the lumbar vertebrae region, refine the segmented lumbar vertebrae region, and generate a lumbar vertebrae region segmentation boundary map.

[0033] The present invention has the following improvements: for the classification results of the lumbar vertebrae area, the formula is used:

[0034] ;

[0035] in, Represents key points and cluster centers The weighted Euclidean distance between and The key points and cluster centers are The coordinate value of the dimension, and are the gradient information of key points and cluster centers respectively, is the gradient weighting factor, is the number of dimensions.

[0036] The present invention is improved in that the specific steps of S5 are as follows:

[0037] S501: optimizing the clarity of the lumbar vertebrae edge in the segmented region using a region growing method based on the lumbar vertebrae region segmentation boundary map, analyzing contour information of the segmented region and identifying changes in adjacent pixels, determining differences in lumbar vertebrae edge clarity, and smoothing the edge based on gradient changes in adjacent pixels to generate an optimized edge clarity result;

[0038] S502: Analyzing the position and distribution of the basic region information points based on the edge definition optimization result and referring to the basic region information points, and adjusting the range of the lumbar vertebra segmentation region to fill the original boundary gaps and generate an adjusted segmentation region;

[0039] S503: Analyze the grayscale differences of adjacent pixels according to the adjusted segmented area and merge similar lumbar vertebrae features to remove similar lumbar vertebrae feature noise, use a morphological algorithm to expand and merge, iteratively correct and smooth the boundaries, and generate an optimized lumbar vertebrae segmentation map.

[0040] Based on the fusion growth point optimization algorithm and the Unet network image processing system, the system includes:

[0041] The data preprocessing module is used to collect lumbar CT images, extract the grayscale value and HU value of each pixel, mark the lumbar vertebrae and obtain the density information of the lumbar vertebrae, analyze the grayscale value to calculate the regional gradient, segment and mark the initial region, extract and classify the basic regional information points including corner points, connection points, and bending points, and generate the preliminary structural features of the lumbar vertebrae;

[0042] a feature extraction module for extracting the central axis and boundary normal in the image based on the preliminary structural features of the lumbar vertebrae, obtaining the angular difference between the central axis and the boundary normal and dynamically adjusting the image rotation angle, correcting the morphological deviation of the original shooting angle, and generating an angle-corrected lumbar vertebrae region;

[0043] An angle correction module is used to analyze the surface characteristics of the lumbar vertebrae region after angle correction using a differential geometry optimization algorithm, calculate the change in local curvature, identify surface features and extract surface boundary information of the lumbar vertebrae region, optimize the smoothness and connectivity of the surface boundary, and generate a surface feature optimization result;

[0044] a segmentation optimization module for identifying gradient changes and marking key points of the lumbar vertebrae region based on the surface feature optimization results, classifying the key points using a spatial clustering method, distinguishing the lumbar vertebrae from other tissue regions, identifying region boundaries and performing segmentation, and generating a segmentation boundary map of the lumbar vertebrae region;

[0045] The boundary correction module is used to optimize the edges of the segmented area based on the lumbar vertebrae area segmentation boundary map, analyze the basic area information points and adjust their position and distribution, expand and fill the boundary gaps in the segmented area, merge similar features by judging the grayscale differences of adjacent pixels and remove noise, correct and smooth the boundaries, and generate an optimized lumbar vertebrae segmentation map.

[0046] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0047] In the present invention, by collecting lumbar CT images and extracting the grayscale value and HU value of the pixels, the density information of the lumbar vertebrae area is obtained and preliminary segmentation is performed, which can accurately analyze the gradient changes in the area and provide a reliable basis for the preliminary area of the image. In this process, the deviation of the lumbar vertebrae morphology is successfully corrected by dynamically adjusting the image rotation angle and analyzing the difference in the boundary normal angle, thereby ensuring the consistency of the morphology in the subsequent processing process. The differential geometry optimization algorithm is applied to further analyze the surface characteristics, identify and optimize the surface features of the lumbar vertebrae, so that the surface information is more accurately mapped to the network, and then the surface boundary is optimized to ensure boundary smoothness and connectivity. Through the analysis of gradient changes and key point clustering, the segmentation of the lumbar vertebrae area and other tissues can be refined, the bone area can be clearly defined, and the accuracy of the segmentation is enhanced. Finally, through iterative correction and smoothing of the segmented area, the optimized segmentation map can better reflect the characteristics of the lumbar vertebrae and reduce the errors caused by inaccurate initial parameters. Compared with existing technologies, this segmentation processing method makes the segmentation results more accurate and stable through the integration of optimization adjustment and deep learning. Especially in complex orthopedic images, it can effectively improve the adaptability and robustness of processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 is a flow chart of the method of the present invention;

[0050] Figure 2 This is a detailed flow chart of step S1 of the present invention;

[0051] Figure 3 This is a schematic diagram of a detailed process of step S2 of the present invention;

[0052] Figure 4 This is a detailed flow chart of step S3 of the present invention;

[0053] Figure 5 This is a detailed flow chart of step S4 of the present invention;

[0054] Figure 6 This is a detailed flow chart of step S5 of the present invention;

[0055] Figure 7 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0056] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0057] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0058] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0059] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0060] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0061] See also Figure 1 The embodiment of the present invention provides a network image processing method based on the fusion growth point optimization algorithm and Unet, including the following steps:

[0062] S1: Collect lumbar CT images, extract the grayscale value and HU value of each pixel in the image, mark the lumbar vertebrae to obtain lumbar bone density information, analyze the lumbar bone gradient and segment the initial region, obtain basic regional information points and classify them, and generate preliminary structural features of the lumbar vertebrae;

[0063] S2: Extract the central axis and boundary normal in the image from the preliminary structural features of the lumbar vertebrae, obtain the angular difference between the central axis and the boundary normal, and dynamically adjust the rotation angle of the image to correct the lumbar vertebrae morphological deviation and generate the angle-corrected lumbar vertebrae region;

[0064] S3: The surface characteristics of the angle-corrected lumbar vertebrae region are analyzed using a differential geometry optimization algorithm. The local curvature change is calculated based on the surface characteristics, and the surface features of the lumbar vertebrae surface are identified. The surface features are mapped to the Unet network and the surface boundary information is extracted. The smoothness and connectivity of the surface boundary are optimized to generate the surface feature optimization results.

[0065] S4: Identify the gradient change of each pixel based on the surface feature optimization results, mark the key points of the lumbar vertebrae region, cluster the key points to distinguish the lumbar vertebrae region from other tissue regions outside the lumbar vertebrae region, determine the boundary of the lumbar vertebrae region and segment it, and generate a lumbar vertebrae region segmentation boundary map;

[0066] S5: Enhance the lumbar vertebrae edge according to the lumbar vertebrae segmentation boundary map, adjust the segmentation region with reference to the basic region information points, fill the boundary gaps and similar lumbar vertebrae features based on the adjusted segmentation region, iteratively correct and smooth the boundary, and generate an optimized lumbar vertebrae segmentation map;

[0067] The preliminary structural features of the lumbar vertebrae include corner points, connection points, bending points, the central axis of the lumbar vertebrae area, and the boundary normal of the lumbar vertebrae area. The lumbar vertebrae area after angle correction includes the lumbar vertebrae area after rotation angle adjustment. The surface feature optimization results include the surface characteristics and surface boundary information of the lumbar vertebrae area. The lumbar vertebrae area segmentation boundary map includes the refined lumbar vertebrae area and the segmentation boundary. The optimized lumbar vertebrae segmentation map includes clear edges, adjusted segmentation areas, expanded segmentation areas, segmentation areas after noise removal, and corrected and smoothed boundaries.

[0068] See also Figure 2 , the specific steps of S1 are as follows:

[0069] S101: Collecting a lumbar spine CT image of the patient, extracting the grayscale value and HU value of each pixel in the CT image, comparing the pixel grayscale value with a preset grayscale threshold, marking lumbar vertebrae regions with grayscale thresholds higher than the preset grayscale threshold, screening regions with HU values exceeding the preset HU threshold and extracting density data, thereby generating density information of the marked lumbar vertebrae regions;

[0070] After obtaining a CT image of the patient's lumbar spine, the grayscale and HU values of each pixel in the image need to be extracted using image processing software. A grayscale threshold is set, typically based on the CT device and patient size. For example, if the grayscale threshold is set to 200, only pixels with grayscale values greater than 200 will be marked as lumbar bone areas. If the proportion of pixels in the image with grayscale values greater than 200 is high, these areas may be lumbar bone areas. Next, areas with higher grayscale values are filtered and further filtered based on HU values. HU values generally range from 100 to 1000, with lumbar bone values typically greater than 250. This process ensures that only denser areas (i.e., lumbar bone areas) are marked and their density data extracted.

[0071] S102: analyzing the gradient change rate of the marked lumbar vertebrae region based on the density information of the marked lumbar vertebrae region, segmenting the initial region marked with the grayscale threshold with reference to the gradient information of the lumbar vertebrae region, obtaining basic region information points of the lumbar vertebrae based on the segmented grayscale regions, and generating basic region information points of the lumbar vertebrae region;

[0072] The density information of the marked lumbar vertebrae area is analyzed by gradient change rate. The gradient change rate can be obtained by calculating the rate of change of the grayscale value of each area. The location with a larger gradient is usually the boundary or where the structure changes significantly. For example, when the grayscale value of a part of the area jumps from a lower value to a higher value, it indicates that there is a significant structural change in this area. Assuming that the set grayscale threshold is 200, if the grayscale value of the adjacent area jumps from less than 200 to more than 200, it means that there is a significant density change in these areas, and therefore they can be judged as the boundaries of the lumbar vertebrae. Based on these gradient change rates, the marked area is further segmented to accurately define the range of the lumbar vertebrae. Then, basic information points are extracted from the segmented area, for example, the position of the basic points is determined by geometric features such as angles, edges, and center positions.

[0073] S103: Identify the positions of corresponding corner points, connection points, and bending points in the basic information points of the lumbar vertebrae region, determine the preliminary structure of the lumbar vertebrae based on the position information of each point, and generate preliminary structural features of the lumbar vertebrae;

[0074] Based on the basic information points, the exact positions of corner points, connection points and bending points are identified. Corner points usually appear at the bends, turns or joints of bones. The positions of these points are calibrated by detecting areas with large grayscale changes or edge intensity. For example, if the grayscale value of a certain part of the bone changes rapidly, it indicates that the area is a corner point. The connection point is the intersection of two different areas, usually with small grayscale changes and relatively smooth, indicating that the point may be a connection point. The bending point refers to the turning point of the bone in the curved part, usually with large grayscale changes and significant changes in direction. In actual operation, these key points are extracted through image processing techniques such as edge detection, corner detection and region segmentation, and the geometric features of each point (such as coordinate position, relative distance) are calculated to determine its position in the lumbar vertebrae, and finally generate the preliminary structural features of the lumbar vertebrae.

[0075] See also Figure 3 , the specific steps of S2 are as follows:

[0076] S201: extracting the central axis and boundary normal of the lumbar vertebra region from the preliminary structural features of the lumbar vertebra, locating the contour of the lumbar vertebra, analyzing the direction and position differences between the central axis and the boundary normal, and generating central axis and boundary normal information;

[0077] The imaging data of the lumbar vertebrae are analyzed, and the outline of the lumbar vertebrae is identified from the image using skeleton extraction or image segmentation technology. The central axis of the lumbar vertebrae is extracted from the image. The central axis is generally a straight line or curve connecting the key areas of the lumbar vertebrae to ensure that the longitudinal structure of the lumbar vertebrae is captured. For the boundary normal, it is necessary to calculate the direction line perpendicular to the boundary based on the edge of the lumbar vertebrae. These normals can be obtained by calculating the gradient direction of each pixel point. A large gradient value usually corresponds to the edge of the structure, and the boundary normal is determined by this information. When analyzing the direction difference between the central axis and the boundary normal, the geometric changes of the lumbar vertebrae can be judged by comparing the angle difference between the two.

[0078] S202: Obtaining angle difference values between the central axis and the boundary normal from the central axis and boundary normal information, and generating angle difference results by referring to the difference analysis of each angle and the influence of the morphological changes of the lumbar vertebrae region;

[0079] By calculating the angle between the central axis and the boundary normal, the angle difference is determined. To do this, it is necessary to extract the normal direction between each pair of key points and use the vector dot product formula to calculate the angle difference. Each pair of angle difference values reflects the morphological changes of the lumbar vertebrae around that point. If the angle difference is large, it means that there may be large morphological variation or asymmetry in the area. Then, an angle threshold is set to determine whether these differences are significant. For example, an angle difference threshold of 10 degrees is set. When the angle difference between the two normals is greater than this value, it is considered that the lumbar vertebrae morphology in this area has changed significantly. By analyzing the angle difference results, possible diseased areas, curvature or abnormal postures in the lumbar vertebrae can be further identified.

[0080] S203: Dynamically adjusting the rotation angle of the lumbar vertebrae image according to the angle difference result, performing rotation correction on the image using image processing technology, adjusting the morphological deviation of the original shooting angle, marking the asymmetric area of the lumbar vertebrae CT image, and generating an angle-corrected lumbar vertebrae area;

[0081] Based on the angle difference results, the required rotation angle is calculated. For example, if the angle difference is greater than a set threshold (such as 10 degrees), image rotation adjustment is required. The calculation of the rotation angle is usually based on the coordinate axis of the image data, and the image is rotated and corrected using a rotation matrix. In this process, the rotation center of the image is first determined, and the rotation angle is calculated. After the rotation adjustment, the asymmetric areas in the image can be detected by morphological analysis methods. Through image processing techniques such as threshold segmentation and edge detection, the areas that are inconsistent with the standard morphology after rotation correction are marked. These asymmetric areas may be caused by angle problems during image acquisition or irregularities in the natural morphology of the lumbar spine.

[0082] See also Figure 4, the specific steps of S3 are as follows:

[0083] S301: performing geometric analysis on the angle-corrected lumbar vertebrae region to obtain surface information of the lumbar vertebrae region, analyzing surface features of the lumbar vertebrae region using a differential geometry optimization algorithm, the surface features including curvature and surface shape, and generating surface feature information of the lumbar vertebrae region;

[0084] The lumbar vertebrae are subjected to geometric analysis after angle correction, usually using 3D modeling technology to extract surface information of the lumbar spine from the imaging data. This surface information helps determine the three-dimensional morphology of the lumbar vertebrae for subsequent differential geometry analysis. Through differential geometry optimization algorithms, various surface features can be extracted, including curvature and surface shape. The curvature calculation method is usually based on local features of the surface, such as the normal direction, the tangent plane, and the changes in adjacent points on the surface, and is calculated using the gradient descent method. Surface features can provide important information about changes in lumbar vertebrae morphology, especially in lesions, degenerative diseases, or abnormal postures. Through calculation and analysis, surface feature information of the lumbar vertebrae can be generated.

[0085] S302: Calculating local curvature values of the lumbar vertebrae surface area based on the surface feature information of the lumbar vertebrae area, marking lumbar vertebrae areas exceeding a preset curvature threshold, and generating local curvature change information of the marked lumbar vertebrae areas;

[0086] Based on the surface feature information of the lumbar vertebrae area, according to the formula:

[0087] ;

[0088] Calculate the local curvature value of the lumbar vertebral surface area ;

[0089] in, 、 are the first and second curvature radii of the key points in the lumbar vertebrae area in the main curvature direction, 、 To adjust the first curvature radius and the second radius of curvature The correction factor is usually 、 The setting range is [0, 1], = =1 is the default setting. When there is no special requirement, using this value means that the curvature radius will not be corrected and the original curvature value will be used directly. In some complex structural areas, such as the joints of the lumbar spine, the curvature may be significantly different in the two directions. In this case, different correction coefficients can be set for the two main curvature radii. 、 Adjust separately, allowing for more detailed adjustments based on actual conditions. is the morphological flexibility factor, The setting range of varies from 0 to 1. Larger values (such as 0.5 to 1) are usually applied to the boundary analysis of the lumbar region to ensure that the curvature can obtain sufficient weight in the classification. By experimenting with the curvature values of different regions, The value of can be selected according to the characteristics of the lumbar vertebrae. For example, if the curvature of the L2 region varies greatly, set = 0.7 can help to accurately delineate boundaries and optimize segmentation, is the curvature change, is the change in curvature angle, The local curvature value is the degree of curvature of the surface at a certain point. It is obtained by calculating the curvature of the point in two directions. It is usually obtained using geometric analysis techniques or surface analysis methods based on differential geometry. The curvature value can be obtained through image analysis tools, such as in 3D image processing software, through surface fitting or direct extraction of surface features. (Principal curvature radius 1) is the curvature radius of the point in the first principal curvature direction. The curvature radius is a measure of the curvature of the surface. The smaller the curvature radius, the greater the curvature of the surface at that point. It can be obtained by fitting a local surface or using the Gaussian curvature calculation method, by observing the curvature changes around the point, such as through image processing or scanning equipment surface fitting function, (Main curvature radius 2) is the radius of curvature of the point in the direction of the second principal curvature, similar to ,and The directions are different, usually obtained by fitting the quadratic curve of the local surface. Representatives and The curvature in the vertical direction, and (Correction coefficient) is used to correct the curvature radius to improve the accuracy of the model. It is determined by experiments on actual data of local areas. In specific cases, such as when the lumbar vertebrae have complex morphology, these coefficients can be appropriately adjusted to reflect more accurate surface features. (Morphological flexibility factor) is a factor used to describe the morphological changes of a surface. It is mainly adjusted by analyzing the curvature and shape changes of the surface. It can be obtained through statistical analysis. For example, when conducting research on the lumbar spine area, this value can be set based on the statistical results of multiple patient data to reflect the overall trend. (Curvature change) is the change in curvature in a certain direction, usually obtained by calculating the difference in curvature between adjacent points or adjacent areas, for example, by differential calculation of a specific area or by changes in image data. (Curvature angle change) refers to the angle change in the curvature direction, which can be obtained by measuring the angle change on a curve or surface. It reflects the turning changes of the surface of the lumbar vertebrae at different positions. This value is usually obtained through angle calculation in image analysis, especially in the process of surface reconstruction.

[0090] For example, in the analysis of lumbar spine CT images, the extracted parameter values are as follows:- mm, mm, , , , mm, Based on these parameters, the local curvature value is calculated by the formula as follows:

[0091] ;

[0092] The results show that the surface changes greatly at this point, and by introducing the correction coefficient and morphological flexibility factor, it can better adapt to the complex surface characteristics of the lumbar vertebrae, providing a more accurate reference for further morphological analysis and treatment.

[0093] S303: Identifying surface features of the lumbar vertebrae based on local curvature variation information of the lumbar vertebrae region, mapping the surface features onto a Unet network, extracting surface boundary information, optimizing the smoothness and connectivity of the surface boundary, and generating surface feature optimization results;

[0094] The surface features of the lumbar vertebrae are extracted based on the local curvature change information. The surface features of different areas of the lumbar vertebrae are identified by the previously calculated local curvature values. These areas usually represent the key structural parts of the lumbar vertebrae. For example, in the L2 area of the lumbar vertebrae, the local curvature value is , showing significant curvature changes in the area, indicating the risk of fracture or deformation. Then, these surface features are mapped to the Unet network, and the network's powerful image segmentation and feature extraction capabilities are used to extract accurate surface boundary information. Through the Unet network, the boundaries of the lumbar vertebrae area can be accurately identified in the image, and regions with different curvature changes can be further segmented. After the surface boundary is extracted, the smoothness and connectivity of the boundary are further optimized to eliminate unnecessary noise and fractures, ensuring the smoothness and integrity of the boundary. For example, for the boundary of the L3 area, after optimization, the smoothness of the surface is improved from the initial curvature value Upgrade to , connectivity is improved, avoiding data loss or discontinuity.

[0095] See also Figure 5 , the specific steps of S4 are as follows:

[0096] S401: Based on the surface feature optimization results, the Shi-Tomasi corner detection algorithm is used to identify the gradient change of each pixel, analyze the change amplitude of adjacent pixels, compare the grayscale difference values of each adjacent pixel, and mark the key point area of the lumbar vertebrae according to the comparison results;

[0097] Based on the surface feature optimization results, the image is processed using the Shi-Tomasi corner detection algorithm, which uses gradient changes between pixels to identify corners in the image. During corner detection, the gradient of each pixel is calculated, typically measuring the magnitude of the gradient change by measuring the grayscale difference in the area surrounding the pixel. After detecting areas with large gradient changes, the grayscale differences between adjacent pixels are further analyzed and compared with a set threshold. By comparing grayscale differences, lumbar vertebrae with significant morphological features can be identified, allowing key point regions to be extracted.

[0098] S402: Using a K-means clustering algorithm to perform spatial clustering calculation on the key point area, identifying key points and grouping them according to position and gradient differences, distinguishing the lumbar vertebrae area from other tissue areas outside the lumbar vertebrae area based on the grouped key points, and generating a lumbar vertebrae area classification result;

[0099] Refer to the lumbar vertebrae area classification results, according to the formula:

[0100] ;

[0101] in, Represents key points and cluster centers The weighted Euclidean distance between 、 The key points and cluster centers are The coordinate value of the dimension, 、 are the gradient information of key points and cluster centers respectively, is the gradient weighting factor, The range is usually from 0 to 100. Large values (such as 10 to 50) are usually used to process areas with complex morphology and large gradient changes, while smaller values (such as 0.1 to 1) are suitable for areas with smoother morphology and less gradient changes. The value of can be determined experimentally. For example, in the L3 region of the lumbar vertebrae, it is observed that within the normal curvature range of the lumbar region, =10, the surface boundary can be better separated from other soft tissues, so this value is selected. is the number of dimensions, Represents key points and cluster centers The distance between them is used to assign key points to the nearest cluster center. The smaller the distance, the more likely the key point belongs to the cluster center. This calculation is based on the extension of Euclidean distance, including position difference and gradient difference. 、 Represent key points and cluster centers In the Coordinate values in 3D space, for lumbar vertebrae CT images in 3D space, Usually represents one of the three coordinate axes (X, Y, Z), for example, is the coordinate of the key point on the X axis, is the coordinate on the Y axis, are the coordinates on the Z axis, which are usually directly obtained through image processing or scanning equipment. 、 Represent key points and cluster centers The gradient information is reflected by the grayscale change of the image, indicating the intensity change at a certain point. The larger the gradient value, the more obvious the change around the point, which usually means the turning point of the edge or structure. These gradient information can be calculated by image processing software (such as OpenCV) and the gradient calculation algorithm (such as Sobel or Prewitt operator) can be used to extract the grayscale change of the local area. Is the gradient weighting factor, which is used to control the influence of gradient information when calculating distance. The value of will enhance the role of gradient information in the clustering process, making clustering pay more attention to the areas with large grayscale changes. This value can be determined through experiments, such as by testing different The clustering results are evaluated and the best value is selected. For example, in the lumbar CT image, the selected may better emphasize the border areas, Represents the number of dimensions of the space. For 3D image data, , that is, the coordinates of each point are composed of X, Y, and Z three-dimensional coordinate values. In a two-dimensional image, ; In a three-dimensional image, .

[0102] Through the K-means algorithm, it is necessary to initialize several cluster centers , by calculating the weighted Euclidean distance between each key point and the cluster center, the cluster affiliation of each key point is determined. For example, there is a key point The coordinates on the CT image are (5,4,3), and its gradient value is , and the cluster center The coordinates are (4,3,2), and the gradient value is If you set , then calculate the weighted distance between the point and the cluster center:

[0103] ;

[0104] Through this calculation method, the K-means algorithm will convert this key point Assign to the cluster center with the smallest distance. This process will be repeated many times until the update of the cluster center converges.

[0105] The results show that the clustering results calculated by the K-means clustering algorithm can accurately distinguish the lumbar vertebrae area from other areas based on the position and gradient information of key points.

[0106] S403: comparing grayscale differences and gradient information based on the lumbar vertebrae region classification results, dividing the boundaries of each lumbar vertebrae region using morphological gradients, segmenting the boundaries of the lumbar vertebrae region using a watershed algorithm, refining the segmented lumbar vertebrae region, and generating a lumbar vertebrae region segmentation boundary map;

[0107] Based on the classification results, the grayscale differences and gradient information between different regions are compared. By comparing these differences, the boundaries of the lumbar vertebrae can be clearly identified. Next, the morphological gradient operator is used to extract the lumbar vertebrae boundaries. Morphological gradients help enhance edge information, making the boundaries more distinct. Morphological operations further clarify the location of the lumbar vertebrae boundaries. The extracted boundaries are then segmented using the watershed algorithm. The watershed algorithm analyzes changes in grayscale and gradient information to delineate distinct regions within the image, distinguishing the lumbar vertebrae from surrounding tissue. The key to the watershed algorithm lies in its ability to handle complex images, effectively segmenting regions with complex morphologies and variable boundaries. After segmentation, the segmentation results are refined to eliminate small noise points and breaks, ensuring boundary continuity and accuracy. The resulting lumbar vertebrae segmentation boundary map provides high-quality data support for subsequent lumbar vertebrae analysis, disease detection, and treatment.

[0108] See also Figure 6 , the specific steps of S5 are as follows:

[0109] S501: optimizing the edge clarity of the lumbar vertebrae in the segmented region using a region growing method based on the lumbar vertebrae segmentation boundary map, analyzing the contour information of the segmented region and identifying changes in adjacent pixels, determining differences in the edge clarity of the lumbar vertebrae, and smoothing the edge based on gradient changes in adjacent pixels to generate an optimized edge clarity result;

[0110] Combining a fusion growing point optimization algorithm with Unet network image processing, the algorithm uses region growing to select appropriate seed points and gradually expand the region to optimize the segmentation boundary of the lumbar vertebrae. The growing point algorithm accurately detects high-grayscale lumbar vertebrae edges and, by continuously expanding the coverage area, eliminates boundary blurring caused by segmentation. Edge clarity is related to the grayscale differences between adjacent pixels. Therefore, by analyzing the grayscale variations of adjacent pixels, it is possible to determine which areas have sharp edges and which have blurred edges. After identifying these differences, the algorithm then smooths the blurred boundaries, removing noise and enhancing the contour features of the area.

[0111] S502: Analyzing the position and distribution of the basic region information points based on the edge definition optimization result and referring to the basic region information points, and adjusting the range of the lumbar vertebra segmentation region to fill the original boundary gaps and generate an adjusted segmentation region;

[0112] Based on the optimized edge clarity and combined with the fusion growth point optimization algorithm, the distribution of information points in the basic area is analyzed to identify the key areas of the lumbar vertebrae. By identifying the location of these basic information points, the main structural boundaries of the lumbar vertebrae can be determined, and the scope of the segmentation area can be further adjusted based on this. For example, if certain parts of the original segmentation area are found to be incomplete or have gaps, the algorithm can be adjusted to fill these gaps to ensure that all areas of the lumbar vertebrae are accurately identified and segmented. This adjustment process continuously corrects and improves the segmentation results, making the lumbar vertebrae area more complete and coherent.

[0113] S503: Analyze the grayscale differences of adjacent pixels based on the adjusted segmented area and merge similar lumbar vertebrae features to remove similar lumbar vertebrae feature noise. Use a morphological algorithm to iteratively correct and smooth the boundaries after expansion and merging to generate an optimized lumbar vertebrae segmentation map.

[0114] The grayscale differences between adjacent pixels in the adjusted lumbar vertebrae segmentation region are analyzed. This analysis identifies the clear differences between the lumbar vertebrae bone region and the non-bone region, while removing some noise features that do not belong to the lumbar vertebrae. Similar lumbar vertebrae features are merged to ensure the consistency of the classification results. Next, a morphological algorithm is used to further expand and merge these regions, smoothing the boundaries between regions and removing unnecessary cracks or noise. On this basis, the segmentation of the lumbar vertebrae region is refined through iterative correction and smoothing to ensure high-quality segmentation results. Ultimately, the generated optimized segmentation map can more accurately reflect the actual morphology of the lumbar vertebrae, providing clearer imaging data for subsequent analysis, diagnosis, and treatment.

[0115] See also Figure 7 , based on the fusion growth point optimization algorithm and Unet network image processing system, the system includes:

[0116] The data preprocessing module is used to collect lumbar CT images, extract the grayscale value and HU value of each pixel, mark the lumbar vertebrae and obtain the density information of the lumbar vertebrae, analyze the grayscale value to calculate the regional gradient, segment and mark the initial region, extract and classify the basic regional information points including corner points, connection points, and bending points, and generate the preliminary structural features of the lumbar vertebrae;

[0117] The feature extraction module is used to extract the central axis and boundary normal in the image based on the preliminary structural features of the lumbar vertebrae, obtain the angular difference between the central axis and the boundary normal, and dynamically adjust the image rotation angle to correct the morphological deviation of the original shooting angle and generate the angle-corrected lumbar vertebrae area;

[0118] An angle correction module is used to analyze the surface characteristics of the lumbar vertebrae region after angle correction using a differential geometry optimization algorithm, calculate the change in local curvature, identify surface features, and extract surface boundary information of the lumbar vertebrae region. It optimizes the smoothness and connectivity of the surface boundary and generates surface feature optimization results.

[0119] The segmentation optimization module is used to identify gradient changes and mark key points in the lumbar vertebrae region based on the surface feature optimization results. The module uses spatial clustering methods to classify key points, distinguish the lumbar vertebrae from other tissue regions, identify regional boundaries and perform segmentation, and generate a segmentation boundary map of the lumbar vertebrae region.

[0120] The boundary correction module is used to segment the boundary map of the lumbar vertebrae region, optimize the edges of the segmented region, analyze the information points in the basic region and adjust their position and distribution, expand and fill the boundary gaps in the segmented region, merge similar features by judging the grayscale differences of adjacent pixels and remove noise, perform boundary correction and smoothing, and generate an optimized lumbar vertebrae segmentation map.

[0121] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. Based on the fusion growth point optimization algorithm and Unet network image processing method, it is characterized by: The following steps are involved: S1: Collect lumbar CT images, extract the grayscale value and HU value of each pixel in the image, mark the lumbar vertebrae to obtain lumbar bone density information, analyze the lumbar bone gradient and segment the initial region, obtain basic regional information points and classify them, and generate preliminary structural features of the lumbar vertebrae; S2: extracting the central axis and boundary normal in the image from the preliminary structural features of the lumbar vertebrae, obtaining the angle difference between the central axis and the boundary normal, and dynamically adjusting the rotation angle of the image to correct the lumbar vertebrae morphological deviation, thereby generating an angle-corrected lumbar vertebrae region; S3: Analyzing the surface characteristics of the angle-corrected lumbar vertebrae region using a differential geometry optimization algorithm, calculating local curvature changes based on the surface characteristics, identifying surface features of the lumbar vertebrae surface, mapping the surface features to a Unet network and extracting surface boundary information, optimizing the smoothness and connectivity of the surface boundary, and generating surface feature optimization results; S4: identifying the gradient change of each pixel point according to the surface feature optimization result, marking the key points of the lumbar vertebrae region, clustering the key points to distinguish the lumbar vertebrae region from other tissue regions outside the lumbar vertebrae region, determining the boundary of the lumbar vertebrae region and segmenting it, and generating a lumbar vertebrae region segmentation boundary map; S5: enhancing the lumbar vertebrae edge according to the lumbar vertebrae region segmentation boundary map, analyzing the distribution of information points in the basic region by combining the fusion growth point optimization algorithm, adjusting the segmentation region with reference to the information points in the basic region, filling boundary gaps and similar lumbar vertebrae features based on the adjusted segmentation region, iteratively correcting and smoothing the boundary, and generating an optimized lumbar vertebrae segmentation map; The specific steps of extracting the central axis and boundary normal in the image from the preliminary structural features of the lumbar vertebra, obtaining the angle difference between the central axis and the boundary normal, and dynamically adjusting the rotation angle of the image to correct the lumbar vertebra morphological deviation and generate the angle-corrected lumbar vertebra area are as follows: S201: extracting the central axis and boundary normal of the lumbar vertebra region from the preliminary structural features of the lumbar vertebra, locating the contour of the lumbar vertebra, analyzing the direction and position differences between the central axis and the boundary normal, and generating central axis and boundary normal information; The central axis is a straight line or curve connecting each key area of the lumbar vertebrae, and the boundary normal is a direction line perpendicular to the boundary calculated based on the edge of the lumbar vertebrae; S202: Obtaining angle difference values between the central axis and the boundary normal from the central axis and boundary normal information, and generating angle difference results by referring to the difference analysis of each angle and the influence of the morphological changes of the lumbar vertebrae region; S203: Dynamically adjust the rotation angle of the lumbar vertebra image according to the angle difference result, use image processing technology to perform rotation correction on the image, adjust the morphological deviation of the original shooting angle, mark the asymmetric area of the lumbar vertebra CT image, and generate the lumbar vertebra area after angle correction.

2. The method for processing network images based on the fusion of the growing point optimization algorithm and the Unet according to claim 1 is characterized by: The preliminary structural features of the lumbar vertebrae region include corner points, connection points, bending points, the central axis of the lumbar vertebrae region, and the boundary normal of the lumbar vertebrae region. The lumbar vertebrae region after angle correction includes the lumbar vertebrae region after rotation angle adjustment. The surface feature optimization results include the surface characteristics and surface boundary information of the lumbar vertebrae region. The lumbar vertebrae region segmentation boundary map includes the refined lumbar vertebrae region and the segmentation boundary. The optimized lumbar vertebrae segmentation map includes clear edges, adjusted segmentation areas, expanded segmentation areas, segmentation areas after noise removal, and corrected and smoothed boundaries.

3. The method for processing network images based on the fusion of the growing point optimization algorithm and the Unet according to claim 1 is characterized by: The specific steps for collecting lumbar CT images, extracting the grayscale value and HU value of each pixel in the image, marking the lumbar vertebrae to obtain lumbar bone density information, analyzing the lumbar bone gradient and segmenting the initial region, obtaining basic regional information points and classifying them, and generating preliminary structural features of the lumbar vertebrae are as follows: S101: Collecting a lumbar spine CT image of the patient, extracting the grayscale value and HU value of each pixel in the CT image, comparing the pixel grayscale value with a preset grayscale threshold, marking lumbar vertebrae regions with grayscale thresholds higher than the preset grayscale threshold, screening regions with HU values exceeding the preset HU threshold and extracting density data, thereby generating density information of the marked lumbar vertebrae regions; S102: analyzing the gradient change rate of the marked lumbar vertebrae region according to the density information of the marked lumbar vertebrae region, segmenting the initial region marked with the grayscale threshold with reference to the gradient information of the lumbar vertebrae region, obtaining basic region information points of the lumbar vertebrae according to the segmented grayscale regions, and generating basic region information points of the lumbar vertebrae region; S103: Identify the positions of the corresponding corner points, connection points, and bending points in the basic information points of the lumbar vertebrae region, determine the preliminary structure of the lumbar vertebrae according to the position information of each point, and generate preliminary structural features of the lumbar vertebrae.

4. The method for processing network images based on the fusion of the growing point optimization algorithm and the Unet network according to claim 1 is characterized by: The surface characteristics of the angle-corrected lumbar vertebrae region are analyzed using a differential geometry optimization algorithm. Local curvature changes are calculated based on the surface characteristics, surface features of the lumbar vertebrae surface are identified, the surface features are mapped to a Unet network, and surface boundary information is extracted. The smoothness and connectivity of the surface boundary are optimized to generate the surface feature optimization results. The specific steps are as follows: S301: performing geometric analysis on the angle-corrected lumbar vertebrae region to obtain surface information of the lumbar vertebrae region, analyzing surface features of the lumbar vertebrae region using a differential geometry optimization algorithm, the surface features including curvature and surface shape, and generating surface feature information of the lumbar vertebrae region; S302: Calculating local curvature values of the lumbar vertebrae surface area based on the surface feature information of the lumbar vertebrae area, marking lumbar vertebrae areas exceeding a preset curvature threshold, and generating local curvature change information of the marked lumbar vertebrae areas; S303: Identify the surface features of the lumbar vertebrae based on the local curvature change information of the lumbar vertebrae area, map the surface features onto the Unet network and extract the surface boundary information, optimize the smoothness and connectivity of the surface boundary, and generate a surface feature optimization result.

5. The method for processing network images based on the fusion of the growing point optimization algorithm and the Unet according to claim 4 is characterized by: For the local curvature value of the lumbar vertebral surface area, the formula is used: ； Calculate the local curvature value of the lumbar vertebral surface area ; in, represents the local curvature value, 、 are the curvature radius of the point in the direction of the principal curvature, 、 To adjust the correction factor for the curvature radius, is the morphological flexibility factor, is the curvature change, is the change in curvature angle.

6. The method for processing network images based on the fusion of the growing point optimization algorithm and the Unet according to claim 1 is characterized by: According to the surface feature optimization results, the gradient change of each pixel point is identified, and the key points of the lumbar vertebrae region are marked. The lumbar vertebrae region is distinguished from other tissue regions outside the lumbar vertebrae region based on the clustering key points. The boundary of the lumbar vertebrae region is determined and segmented. The specific steps of generating the lumbar vertebrae region segmentation boundary map are as follows: S401: using the Shi-Tomasi corner detection algorithm based on the surface feature optimization result to identify the gradient change of each pixel, analyzing the change amplitude of adjacent pixels, comparing the grayscale difference value of each adjacent pixel, and marking the key point area of the lumbar vertebrae region based on the comparison result; S402: performing spatial clustering calculation on the key point region using a K-means clustering algorithm, identifying key points and grouping them according to position and gradient differences, distinguishing the lumbar vertebrae region from other tissue regions outside the lumbar vertebrae region based on the grouped key points, and generating a lumbar vertebrae region classification result; S403: Compare the grayscale difference and gradient information according to the classification results of the lumbar vertebrae region, divide the boundary of each lumbar vertebrae region by morphological gradient, use the watershed algorithm to segment the boundary of the lumbar vertebrae region, refine the segmented lumbar vertebrae region, and generate a lumbar vertebrae region segmentation boundary map.

7. The method for processing network images based on the fusion of the growing point optimization algorithm and the Unet according to claim 6, characterized in that: For the lumbar vertebrae area classification results, the formula is used: ； Calculate the distance between each key point area and the cluster center ; in, Represents key points and cluster centers The weighted Euclidean distance between and The key points and cluster centers are The coordinate value of the dimension, and are the gradient information of key points and cluster centers respectively, is the gradient weighting factor, is the number of dimensions.

8. The method for processing network images based on the fusion of growing point optimization algorithm and Unet according to claim 1 is characterized by: The specific steps of enhancing the lumbar vertebrae edge according to the lumbar vertebrae region segmentation boundary map, adjusting the segmentation region with reference to the basic region information points, filling the boundary gaps and similar lumbar vertebrae features based on the adjusted segmentation region, and iteratively correcting and smoothing the boundary to generate an optimized lumbar vertebrae segmentation map are as follows: S501: optimizing the clarity of the lumbar vertebrae edge in the segmented region using a region growing method based on the lumbar vertebrae region segmentation boundary map, analyzing contour information of the segmented region and identifying changes in adjacent pixels, determining differences in lumbar vertebrae edge clarity, and smoothing the edge based on gradient changes in adjacent pixels to generate an optimized edge clarity result; S502: Analyzing the position and distribution of the basic region information points based on the edge definition optimization result and referring to the basic region information points, and adjusting the range of the lumbar vertebra segmentation region to fill the original boundary gaps and generate an adjusted segmentation region; S503: Analyze the grayscale differences of adjacent pixels according to the adjusted segmented area and merge similar lumbar vertebrae features to remove similar lumbar vertebrae feature noise, use a morphological algorithm to expand and merge, iteratively correct and smooth the boundaries, and generate an optimized lumbar vertebrae segmentation map.

9. Based on the fusion growth point optimization algorithm and Unet network image processing system, it is characterized by: The system according to any one of claims 1 to 8, wherein the system comprises: The data preprocessing module collects lumbar CT images, extracts the grayscale value and HU value of each pixel, marks the lumbar vertebrae and obtains the density information of the lumbar vertebrae, analyzes the grayscale value to calculate the regional gradient, segments and marks the initial region, extracts and classifies the basic regional information points including corner points, connection points, and inflection points, and generates the preliminary structural features of the lumbar vertebrae. The feature extraction module extracts the central axis and boundary normal in the image based on the preliminary structural features of the lumbar vertebrae, obtains the angular difference between the central axis and the boundary normal, and dynamically adjusts the image rotation angle to correct the morphological deviation of the original shooting angle and generate an angle-corrected lumbar vertebrae region; The angle correction module analyzes the surface characteristics of the lumbar vertebrae region after angle correction using a differential geometry optimization algorithm, calculates changes in local curvature, identifies surface features, extracts surface boundary information of the lumbar vertebrae region, optimizes the smoothness and connectivity of the surface boundary, and generates surface feature optimization results; The segmentation optimization module identifies gradient changes and marks key points of the lumbar vertebrae region based on the surface feature optimization results, classifies the key points using a spatial clustering method, distinguishes the lumbar vertebrae from other tissue regions, identifies regional boundaries and performs segmentation, and generates a lumbar vertebrae region segmentation boundary map; The boundary correction module is based on the lumbar vertebrae segmentation boundary map, optimizes the edges of the segmented area, analyzes the basic area information points and adjusts their position and distribution, expands and fills the boundary gaps in the segmented area, merges similar features by judging the grayscale differences of adjacent pixels and removes noise, corrects and smoothes the boundaries, and generates an optimized lumbar vertebrae segmentation map.

Images