An Adaptive Watershed-Based Method for Locating Pulmonary Segment Boundaries

Abstract

The present invention relates to the technical field of medical image processing. More specifically, it relates to a method for locating the boundaries of lung segments based on an adaptive watershed algorithm, which includes obtaining angiography or high-resolution CT images of the lungs; performing segmentation of the pulmonary segment arteries based on the gray-scale normalized lung images; obtaining an optimized lung segment segmentation result; obtaining the innermost boundary of the lung lobe according to the innermost boundary of the lung segments and the outermost boundary of the lung lobe according to the outermost boundary of the lung segments; determining the gray-scale segmentation threshold corresponding to the outermost boundary and the gray-scale segmentation threshold corresponding to the innermost boundary; determining the average gray scale of the region between the outermost boundary and the innermost boundary of the lung lobe according to the gray-scale segmentation threshold corresponding to the outermost boundary, and determining the gray-scale gradient range of the region between the outermost boundary and the innermost boundary of the lung lobe according to the average gray scale; obtaining the lung segment boundary location result, and using the distribution information of the pulmonary segment arteries to guide the segmentation process, thereby improving the accuracy of segmentation.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and more particularly to a lung segment boundary positioning method based on adaptive watershed division. Background Art

[0002] With the rapid development of medical imaging technology, computer-aided diagnosis (CAD) plays an increasingly important role in the early detection and precise treatment of lung diseases. Precise localization of lung segment boundaries is a key step in achieving accurate diagnosis and personalized treatment. However, due to the complexity and individual variability of the human lung structure, as well as inherent noise and artifacts in medical images, accurately localizing lung segment boundaries remains a significant challenge in medical image processing.

[0003] Traditional lung segmentation methods primarily rely on fixed thresholds or simple region growing algorithms, which often perform poorly when processing images from different patients or under different scanning conditions. Although the introduction of anatomical knowledge-based methods and machine learning techniques in recent years has improved segmentation results to some extent, some urgent issues remain. For example, existing methods are highly dependent on image quality, and segmentation accuracy decreases significantly when processing low-dose CT or lung images with severe lesions. Furthermore, most methods lack sufficient consideration of the lung's anatomical structure, resulting in insufficient anatomical plausibility in the segmentation results.

[0004] Another common problem is the lack of algorithm adaptability. Due to differences in lung structure between patients, algorithms with fixed parameters struggle to adapt to this variability. While some researchers have attempted to introduce adaptive mechanisms, these often only consider the image's grayscale features, neglecting important information such as shape and topology. This results in insufficient algorithm robustness when handling complex cases, making over-segmentation and under-segmentation more likely.

[0005] Furthermore, existing methods face challenges in computational efficiency and clinical practicality. Some high-precision algorithms often require extensive computing resources and time, making them difficult to meet the demands of real-time clinical diagnosis. Meanwhile, some fast algorithms often sacrifice accuracy, failing to provide a sufficiently reliable basis for accurate diagnosis.

[0006] Faced with these challenges, there is an urgent need for a lung segment boundary localization method that can comprehensively consider image features and anatomical structure, is highly adaptable and robust, and takes into account computational efficiency. Summary of the Invention

[0007] The present invention proposes a lung segment boundary positioning method based on adaptive watershed division, which effectively solves the problems of insufficient accuracy, poor adaptability, and unclear anatomical significance in the prior art.

[0008] To solve the above technical problems, the present invention adopts the following technical solutions:

[0009] An adaptive watershed-based lung segment boundary localization method, comprising:

[0010] An acquisition step, comprising:

[0011] Acquiring a contrast image or a high-resolution CT image of the lung;

[0012] Performing gray-scale normalization processing on the lung image;

[0013] A processing step, comprising:

[0014] Based on the lung image after the gray-scale normalization processing, performing lung segment artery segmentation;

[0015] According to the connection relationship between the lung segment arteries, determining a lung lobe division map of 4 lung regions in the left lung and 3 lung regions in the right lung;

[0016] Based on the lung lobe division map, obtaining a lung lobe division table;

[0017] According to the lung lobe division map, determining the left and right ownership of each branch artery to form a lung lobe marking matrix;

[0018] Inside each lung lobe, for each branch artery, using the branch artery as the main trunk, performing region growing to obtain the corresponding connected region;

[0019] Taking the connected region as the initial segmentation result of the lung segment corresponding to the branch artery;

[0020] According to the connection situation between the lung segments and the lung lobe division table, performing merging and optimization processing on adjacent lung segments to obtain an optimized lung segment segmentation result;

[0021] For any lung lobe, determining its innermost lung segment boundary and outermost lung segment boundary;

[0022] According to the innermost lung segment boundary, obtaining the innermost boundary of the lung lobe, and according to the outermost lung segment boundary, obtaining the outermost boundary of the lung lobe;

[0023] Determining the gray-scale segmentation threshold corresponding to the outermost boundary and the gray-scale segmentation threshold corresponding to the innermost boundary;

[0024] According to the gray-scale segmentation threshold corresponding to the outermost boundary, determining the average gray scale of the region between the outermost boundary and the innermost boundary of the lung lobe, and according to the average gray scale, determining the gray-scale gradient range of the region between the outermost boundary and the innermost boundary of the lung lobe;

[0025] Determine the region after threshold segmentation according to the grayscale segmentation threshold corresponding to the innermost boundary, determine the curve parameters of the boundary line of this region according to the length of the boundary line of this region, and determine the smooth coefficient of the lung lobe according to the curve parameters;

[0026] The output step includes:

[0027] Perform final lung lobe boundary segmentation according to the grayscale gradient range and the smooth coefficient to obtain the lung segment boundary localization result.

[0028] Preferably, the lung segment segmentation and optimization processing in the processing step specifically include:

[0029] Initialize the boundary parameters, including the maximum threshold, minimum threshold, step factor, maximum number of iterations, segmentation threshold of the outermost boundary, area of the outermost boundary, segmentation threshold of the innermost boundary, area of the innermost boundary, smooth coefficient;

[0030] Take the sum and average of the current maximum threshold and minimum threshold as the threshold parameter used in this iteration, and perform threshold segmentation on the image;

[0031] Judge whether there are unsegmented lung segments. If so, calculate the boundary lengths of all unsegmented lung segments and obtain the average value of the boundary lengths;

[0032] If the average value is less than the minimum value of all boundary lengths in the region after threshold segmentation and the number of iterations in this iteration is less than the maximum number of iterations, modify the minimum threshold according to the step factor, and return to execute the threshold segmentation step;

[0033] Take the current minimum threshold as the segmentation threshold of the innermost boundary;

[0034] Judge whether there are lung segments that do not reach the minimum area threshold. If so, return to execute the threshold segmentation step;

[0035] Take the current minimum threshold as the segmentation threshold of the outermost boundary;

[0036] Judge whether the area of the outermost boundary is less than the minimum area threshold. If so, modify the segmentation threshold of the outermost boundary to the segmentation threshold of the innermost boundary, and modify the segmentation threshold of the innermost boundary to the segmentation threshold of the outermost boundary in the previous iteration;

[0037] Judge whether the number of final iterations is equal to the maximum number of iterations. If so, output the obtained boundary parameters to obtain the lung segment segmentation result.

[0038] Preferably, the lung segment merging and optimization processing in the processing step specifically includes:

[0039] Initialize the boundary parameters, including the maximum threshold, minimum threshold, step factor, maximum number of iterations, segmentation threshold of the outermost boundary, area of the outermost boundary, segmentation threshold of the innermost boundary, area of the innermost boundary, and smoothing coefficient;

[0040] Take the sum and average of the current maximum threshold and minimum threshold as the threshold parameter for this iteration, and perform threshold segmentation on the image;

[0041] Determine whether there is an unsegmented lung lobe area in the threshold segmentation result. If so, calculate the average gray value of all connected regions and determine the segmentation threshold of the outermost boundary;

[0042] Determine the final image through the weight mixing algorithm using gradient magnitude and curvature magnitude;

[0043] Determine whether the segmentation threshold of the outermost boundary is greater than the current minimum threshold. If so, calculate the average length of all current connected regions;

[0044] If the average length is less than the area of the outermost boundary and the current number of iterations is less than the maximum number of iterations, update the minimum threshold and return to execute the threshold segmentation processing step;

[0045] Calculate the average length of all connected regions. If the average length is greater than the area of the outermost boundary, update the segmentation threshold of the outermost boundary and return to execute the threshold segmentation processing step;

[0046] Save the current minimum threshold;

[0047] Output the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary as the final result of the lung segment merging and optimization processing.

[0048] Preferably, determining the segmentation threshold of the innermost boundary and the segmentation threshold of the outermost boundary in the processing step specifically includes:

[0049] Determine the gray value of the outermost boundary in the lung image after the gray normalization processing, and use the current gray value of the outermost boundary as the segmentation threshold of the outermost boundary;

[0050] Perform reverse gray value sorting. If the average value of the gray values of the three voxel points with the largest gray values in the current lung lobe is less than the segmentation threshold of the innermost boundary, determine the average gray value through the gray value of the outermost boundary and the segmentation threshold of the innermost boundary in the lung image after the gray normalization processing, and then obtain the gray range centered on the average gray value;

[0051] Determine whether the average gray value of the lung lobe area is within the gray range. If so, determine whether the difference between the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary is less than 5;

[0052] If the difference is less than 5, the segmentation threshold of the innermost boundary is the gray level of the current outermost boundary; otherwise, the segmentation threshold of the innermost boundary is the sum of the gray level of the current outermost boundary and 5.

[0053] If the difference between the maximum value and the minimum value of the gray levels of all the voxel points in the lung lobe region exceeds 5, the gray level range is determined by subtracting the gray level of the outermost boundary from the average value of the gray levels of 50% of the voxel points with the maximum gray level values in the lung lobe.

[0054] The gray level range is determined by subtracting the segmentation threshold of the innermost boundary from the maximum value of the gray level values of all the current voxel points in the lung lobe.

[0055] The smoothing coefficient is determined by the segmentation threshold of the current outermost boundary and the segmentation threshold of the current innermost boundary.

[0056] Preferably, the output step specifically includes:

[0057] Segment the lung image after the gray level normalization processing with the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary to obtain two segmented regions.

[0058] Perform threshold segmentation on the lung image after the gray level normalization processing with the segmentation threshold of the innermost boundary and the segmentation threshold of the outermost boundary.

[0059] Perform hole filling on the segmentation result of the previous step to obtain the gray level gradient range.

[0060] Segment all the voxel points between the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary, and update the gray level gradient range with the intensity and position of the voxel points.

[0061] Perform maximum threshold segmentation on the lung image after the gray level normalization processing, and then perform a closing operation on the maximum threshold segmentation result.

[0062] Perform curvature amplitude analysis on the closing result. Gaussian filtering is used for the curvature amplitude analysis. Then, calculate the ratio of the curvature amplitude value of all the voxel points to the average curvature amplitude value, and determine the smoothing coefficient based on this ratio.

[0063] Obtain the final gray level segmentation threshold according to the gray level gradient range and the smoothing coefficient, and then segment the lung image after the gray level normalization processing with this gray level segmentation threshold.

[0064] Adopt local gray level value averaging for region division to obtain the final segmentation result.

[0065] Preferably, in the output step, a model based on machine learning is used to verify the correctness of the boundary marked by the adaptive watershed.

[0066] Preferably, in the processing step, a lung lobe division map and a lung lobe marking matrix are determined according to the blood vessel position information.

[0067] Preferably, in the processing step and the output step, for each segmentation step, morphological filtering and morphological closing operations are performed to eliminate noise of small structures.

[0068] Preferably, in the processing step, a point is selected from the outermost boundary region, the Euclidean distance and the Manhattan distance from this point to the innermost boundary are calculated, and the difference between the two distances is used as the smoothing coefficient.

[0069] Preferably, the method further includes:

[0070] Performing three-dimensional registration on the lung lobe image to obtain a three-dimensionally registered lung lobe image;

[0071] Based on the three-dimensionally registered lung lobe image, calculating characteristic parameters of the surfaces of the lung lobes and lung segments, and completing the lung segment segmentation of the lung lobe image according to the characteristic parameters to obtain a lung segment segmentation result;

[0072] Extracting the lung segment boundaries of the lung segment segmentation result to obtain a boundary curve including a plurality of boundary convex points and boundary concave points;

[0073] Performing a convex hull stitching operation on boundary curves with the same or similar directions through a convex hull algorithm to obtain a closed curve;

[0074] Obtaining a cavity region surrounded by the closed curve, setting boundary convex points according to the volume of the cavity region, and setting lung segment boundary concave points;

[0075] Based on the calculation criterion of the watershed parameter with the set region growth threshold in the watershed algorithm as an adaptive rule, to obtain an optimal lung segment boundary, such that the boundary of the lung segment segmentation region corresponding to the optimal watershed threshold coincides with the lung segment boundary.

[0076] The method of the present invention has the following remarkable technical effects:

[0077] The method for locating lung segment boundaries based on adaptive watershed proposed by the present invention realizes high-precision positioning of lung segment boundaries by innovatively combining image processing technology, anatomical knowledge, and machine learning methods. The core of this method lies in its high adaptability and robustness, and it can effectively cope with image variability under different patients and different scanning conditions. By introducing multi-level adaptive mechanisms, such as adaptive threshold selection, dynamic parameter adjustment, etc., this method greatly improves the accuracy and reliability of the segmentation result.

[0078] In addition, the method of the present invention fully considers the anatomical structure characteristics of the lungs, especially uses the distribution information of the pulmonary segment arteries to guide the segmentation process. This segmentation strategy based on anatomical knowledge not only improves the accuracy of segmentation, but also ensures that the segmentation results have good anatomical significance, providing a reliable basis for subsequent clinical diagnosis and treatment decisions.

[0079] Another significant advantage of the method of the present invention lies in its comprehensiveness and systematicness. Through multiple carefully designed processing steps, such as image preprocessing, preliminary segmentation, boundary optimization, etc., a complete technical solution is formed. These steps are not simply superimposed, but through ingenious design, effective information transfer and complementarity are achieved. For example, the result of the preliminary segmentation provides an important reference for the subsequent boundary optimization, and the result of the boundary optimization in turn helps to improve the overall segmentation effect. This synergistic effect between steps significantly improves the performance of the entire method.

[0080] It is worth mentioning that while ensuring high precision, the method of the present invention also takes into account computational efficiency. By adopting efficient algorithm implementation and reasonable parameter settings, the method can achieve near-real-time processing under ordinary hardware conditions, meeting the requirements of clinical applications. The combination of such high precision and high efficiency greatly enhances the application value of this method in the actual clinical environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 is the overall flowchart of the method of the present invention.

[0082] Figure 2 is the flowchart of the preliminary pulmonary segment segmentation of the present invention.

[0083] Figure 3 is the flowchart of the pulmonary segment merging and optimization of the present invention.

[0084] Figure 4 is the flowchart of the final boundary segmentation of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0085] As Figures 1-4 shown, the present invention provides a method for locating the pulmonary segment boundary based on adaptive watershed. This method realizes the precise location of the pulmonary segment boundary in the lung CT image through multiple steps, providing important support for clinical diagnosis and surgical planning.

[0086] First, this method acquires the angiography or high-resolution CT image of the lungs and performs gray-scale normalization processing on it. This step aims to eliminate the image differences caused by different scanning devices and parameters, laying a foundation for subsequent processing. Preferably, the gray-scale normalization can adopt the linear stretching method to map the image gray-scale values to the range of 0-255.

[0087] Next, the method performs the segmentation of the pulmonary segment arteries of the image. This step is crucial for the subsequent division of the lung lobes and segments. In one embodiment of the present invention, a multi-scale vascular enhancement filtering technique can be used for the segmentation of the pulmonary segment arteries. Specifically, the Frangi filter can be used, and its mathematical expression is:

[0088] ,

[0089] where, are the eigenvalues of the Hessian matrix, are the geometric ratio, cross-sectional ratio, and structure measure respectively. is a control parameter, usually taking values of 0.5, 0.5, 500.

[0090] Based on the connection relationship between the pulmonary segment arteries, the method determines the lobe division map of 4 lung regions in the left lung and 3 lung regions in the right lung. Preferably, a connected component analysis algorithm can be used to implement this step. According to the lobe division map, a lobe division table is obtained. At the same time, the left and right attribution of each branch artery in the lobe division map is determined to form a lobe labeling matrix. These steps provide an anatomical basis for the subsequent pulmonary segment segmentation.

[0091] Inside each lung lobe, for each branch artery, taking this branch artery as the main trunk, the method performs region growing to obtain the corresponding connected region. The key of the region growing algorithm lies in the selection of the growth criterion. In one embodiment of the present invention, the following growth criterion can be used:

[0092] ,

[0093] where, is the gray value of the voxel to be detected, is the average gray value of the seed point neighborhood, is the standard deviation, is a control parameter, usually taking a value of 2.5.

[0094] Taking this connected region as the initial segmentation result of the pulmonary segment corresponding to this branch artery. Using the boundary of the connected region as the boundary of this pulmonary segment. This step realizes the preliminary pulmonary segment division.

[0095] According to the connection situation between the pulmonary segments and the lobe division table, the method performs a merging and optimization process on adjacent pulmonary segments to obtain an optimized pulmonary segment segmentation result. Preferably, a merging threshold can be set. When the boundary length of adjacent pulmonary segments is less than the threshold (such as 10 mm) and the gray value difference is less than a certain threshold (such as 20 HU), a merging operation is performed. This step aims to eliminate the problem of over-segmentation and improve the anatomical rationality of the segmentation result.

[0096] For any lung lobe, this method determines the innermost and outermost pulmonary segment boundaries. Based on the innermost pulmonary segment boundary, the innermost boundary of the lung lobe is obtained, and based on the outermost pulmonary segment boundary, the outermost boundary of the lung lobe is obtained. The area between the innermost and outermost boundaries of the lung lobe is the mediastinal inner boundary of the lung lobe. This step lays the foundation for subsequent fine segmentation.

[0097] This method further determines the gray-scale segmentation threshold corresponding to the outermost boundary and the gray-scale segmentation threshold corresponding to the innermost boundary. According to the gray-scale segmentation threshold corresponding to the outermost boundary, the average gray scale of the area between the outermost and innermost boundaries of the lung lobe is determined, and based on the average gray scale, the gray-scale gradient range of the area between the outermost and innermost boundaries of the lung lobe is determined. Preferably, the Otsu algorithm can be used to automatically determine the optimal threshold.

[0098] According to the gray-scale segmentation threshold corresponding to the innermost boundary, the threshold-segmented area is determined. According to the length of the boundary line of this area, the curve parameter of the boundary line of this area is determined, and based on the curve parameter, the smooth coefficient of the lung lobe is determined. In an embodiment of the present invention, the smooth coefficient can be calculated by the following formula:

[0099] ,

[0100] where, is the actual length of the boundary line, is the radius of the equal-area circle.

[0101] Finally, based on the gray-scale gradient range and the smooth coefficient, the final lung lobe boundary segmentation is performed. Preferably, the level set method can be used to achieve fine segmentation, and its evolution equation is:

[0102] ,

[0103] where, is the level set function, is the edge stopping function, is the dilation coefficient.

[0104] A significant advantage of the method of the present invention lies in its self-adaptability. By introducing multiple adaptive parameters, such as the threshold of region growing, the conditions for pulmonary segment merging, the boundary smooth coefficient, etc., the method can adapt to the individual differences of different patients and the image features under different scanning conditions. This greatly improves the robustness and generality of the method.

[0105] In addition, this method makes full use of the anatomical knowledge of the lungs, especially the distribution characteristics of pulmonary segmental arteries, so that the segmentation results have good anatomical significance. This is of great value for subsequent clinical diagnosis and surgical planning.

[0106] It should be noted that the present method adopts a multi-scale analysis strategy. From lobe division to segment segmentation and then to fine boundary localization, the gradually refined process effectively reduces the risk of error accumulation and improves the accuracy of the final result.

[0107] In practical applications, the present method can be combined with machine learning techniques to further improve the accuracy of boundary localization. For example, a convolutional neural network can be trained to verify and optimize the final boundary position. This strategy of combining traditional image processing techniques and modern artificial intelligence methods represents the future development direction of medical image analysis.

[0108] Generally speaking, the method for segment boundary localization of adaptive watershed provided by the present invention realizes precise localization of segment boundaries through a series of innovative technical means. This method not only has high adaptability and robustness, but also ensures the anatomical rationality of the results, providing strong technical support for the precise diagnosis and personalized treatment of lung diseases.

[0109] The method of the present invention provides an efficient and reliable algorithm for segment segmentation and optimization processing. Specifically, the algorithm includes the following steps:

[0110] First, initialize the boundary parameters. These parameters include the maximum threshold, minimum threshold, step factor, maximum number of iterations, segmentation threshold of the outermost boundary, area of the outermost boundary, segmentation threshold of the innermost boundary, area of the innermost boundary, and smoothing coefficient. Preferably, these initial parameters can be set according to empirical values. For example, the maximum threshold can be set to the maximum gray value of the image, the minimum threshold can be set to the minimum gray value of the image, the step factor can be set to 1, and the maximum number of iterations can be set to 100. The selection of these initial values has an important impact on the convergence speed of the algorithm and the quality of the final result.

[0111] Next, the present method takes the sum and average of the current maximum threshold and minimum threshold as the threshold parameter used in this iteration, and performs threshold segmentation on the image. This step adopts the idea of adaptive thresholding and can better adapt to the characteristics of different images. The mathematical expression of threshold segmentation can be represented as:

[0112] ,

[0113] where is the original image, is the binary image after segmentation, is the threshold, equal to .

[0114] Then, this method determines whether there are unsegmented lung segments. If there are, it calculates the boundary lengths of all unsegmented lung segments and obtains the average value of these boundary lengths. This step aims to evaluate the quality of the current segmentation result.

[0115] Next, this method makes a crucial judgment: if the average value is less than the minimum of all boundary lengths in the region after threshold segmentation and the current iteration count is less than the maximum iteration count, then according to the step factor, the minimum threshold is modified, and the threshold segmentation step is returned for execution. This judgment condition comprehensively considers the segmentation quality and computational efficiency, which is an embodiment of the algorithm's adaptability. Preferably, an empirical threshold, such as 10 mm, can be set as a reference value for the boundary length.

[0116] If the above conditions are not met, this method takes the current minimum threshold as the segmentation threshold for the innermost boundary. This step determines the internal boundary of the lung segment.

[0117] Subsequently, this method determines whether there are lung segments that do not reach the minimum area threshold. If there are, it returns to execute the threshold segmentation step. This step aims to avoid the appearance of overly small lung segments and improve the anatomical rationality of the segmentation result. Preferably, the minimum area threshold can be set to 50 mm².

[0118] If there are no lung segments that do not reach the minimum area threshold, this method takes the current minimum threshold as the segmentation threshold for the outermost boundary. This step determines the external boundary of the lung segment.

[0119] Next, this method determines whether the area of the outermost boundary is less than the minimum area threshold. If so, it modifies the segmentation threshold of the outermost boundary to the segmentation threshold of the innermost boundary, and modifies the segmentation threshold of the innermost boundary to the segmentation threshold of the outermost boundary in the previous iteration. This step is to handle possible abnormal situations and ensure the rationality of the segmentation result.

[0120] Finally, this method determines whether the number of final iterations is equal to the maximum iteration count. If so, it outputs the obtained boundary parameters to get the lung segment segmentation result. This step ensures the convergence of the algorithm and also provides an opportunity for possible manual intervention.

[0121] The method of the present invention also provides an innovative algorithm in the aspect of lung segment merging and optimization processing. The specific steps of this algorithm are as follows:

[0122] First, initialize the boundary parameters, including the maximum threshold, minimum threshold, step factor, maximum iteration count, segmentation threshold of the outermost boundary, area of the outermost boundary, segmentation threshold of the innermost boundary, area of the innermost boundary, and smoothing coefficient. The initialization of these parameters is similar to that described before, but may need to be adjusted according to the specific requirements of merging and optimization.

[0123] Then, the method sums and averages the current maximum threshold and minimum threshold as the threshold parameter used in this iteration, and performs threshold segmentation processing on the image. The mathematical expression of this step is the same as that described before.

[0124] Next, the method determines whether there are unsegmented lung lobe regions in the threshold segmentation result. If so, it calculates the average gray value of all connected regions and determines the segmentation threshold of the outermost boundary. This step aims to evaluate the quality of the current segmentation result and provide a basis for subsequent optimization.

[0125] Subsequently, the method determines the final image through a weighted mixing algorithm using gradient magnitude and curvature magnitude. This step is one of the innovations of this algorithm, which comprehensively considers the gradient information and shape information of the image. Specifically, the following formula can be used:

[0126] ,

[0127] where, is the final image, is the gradient image, is the curvature image, and are weight coefficients, and .

[0128] Next, the method determines whether the segmentation threshold of the outermost boundary is greater than the current minimum threshold. If so, it calculates the average length of all current connected regions. This step is to evaluate the quality of the current segmentation result.

[0129] If the average length is less than the area of the outermost boundary and the current iteration number is less than the maximum iteration number, the minimum threshold is updated, and the threshold segmentation processing step is returned for execution. This conditional judgment comprehensively considers segmentation quality and computational efficiency, and is another manifestation of the algorithm's adaptability.

[0130] If the above condition is not satisfied, the method calculates the average length of all connected regions. If the average length is greater than the area of the outermost boundary, the segmentation threshold of the outermost boundary is updated, and the threshold segmentation processing step is returned for execution. This step aims to further optimize the segmentation result.

[0131] Finally, the method saves the current minimum threshold and outputs the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary as the final result of the lung segment merging and optimization processing.

[0132] The method of the present invention also provides an innovative algorithm for determining the segmentation threshold of the innermost boundary and the outermost boundary. The specific steps of this algorithm are as follows:

[0133] First, determine the gray value of the outermost boundary in the lung image after the gray normalization process, and use the current gray value of the outermost boundary as the segmentation threshold of the outermost boundary. This step provides an initial reference value for subsequent processing.

[0134] Then, the method performs reverse gray value sorting. If the average value of the gray values of the three voxel points with the largest gray values in the current lung lobe is less than the segmentation threshold of the innermost boundary, the average gray value is determined by the gray value of the outermost boundary and the segmentation threshold of the innermost boundary in the lung image after the gray normalization process, and then the gray range centered on the average gray value is obtained. This step aims to adaptively determine a suitable gray range.

[0135] Next, the method determines whether the average gray value of the lung lobe region is within the gray range. If so, it is determined whether the difference between the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary is less than 5. This threshold can be adjusted according to specific application scenarios, and usually the value range is 3 - 7.

[0136] If the difference is less than 5, the segmentation threshold of the innermost boundary is the current gray value of the outermost boundary. Otherwise, the segmentation threshold of the innermost boundary is the sum of the current gray value of the outermost boundary and 5. This step ensures a reasonable difference between the inner and outer boundary thresholds.

[0137] If the difference between the maximum value and the minimum value of the gray values of all voxel points in the lung lobe region exceeds 5, the gray range is determined by subtracting the gray value of the outermost boundary from the average value of the gray values of 50% of the voxel points with the largest gray values in the lung lobe. This step deals with possible extreme cases.

[0138] Then, the method determines the gray range by subtracting the segmentation threshold of the innermost boundary from the maximum value of the gray values of all current voxel points in the lung lobe. This step further refines the determination of the gray range.

[0139] Finally, the method determines the smoothing coefficient based on the current segmentation threshold of the outermost boundary and the current segmentation threshold of the innermost boundary. The calculation of the smoothing coefficient can use the following formula:

[0140] ,

[0141] where, 7] is the smoothing coefficient, is the segmentation threshold of the outermost boundary, is the segmentation threshold of the innermost boundary, and are the maximum and minimum gray values of the image respectively.

[0142] Through this series of steps, the method of the present invention realizes the accurate determination of the segmentation thresholds for the innermost boundary and the outermost boundary, laying a solid foundation for the subsequent localization of the lung segment boundaries. This adaptive threshold determination method can effectively adapt to the image features under different patients and different scanning conditions, greatly improving the robustness and generality of the method.

[0143] Generally speaking, these algorithms provided by the present invention demonstrate unique innovation and high adaptability in lung segment segmentation, optimization processing, and boundary threshold determination. These algorithms not only consider the technical requirements of image processing but also fully combine the characteristics of lung anatomy, thereby being able to produce segmentation results with good anatomical significance. This is of great significance for improving the accuracy of lung disease diagnosis and formulating personalized treatment plans. The method of the present invention adopts an innovative technical route in the final boundary segmentation stage, making full use of the results of the previous processing to achieve high-precision lung segment boundary localization. Specifically, this process includes the following key steps:

[0144] First, the method of the present invention performs segmentation thresholding on the outermost boundary and the innermost boundary of the lung image after gray normalization processing to obtain two segmentation regions. This step lays the foundation for subsequent fine segmentation. Preferably, a double-threshold segmentation technique can be used, and its mathematical expression is as follows:

[0145] ,

[0146] wherein, is the original image, is the binary image after segmentation, and are the segmentation thresholds for the innermost and outermost boundaries respectively. Then, the method of the present invention performs thresholding on the segmentation thresholds of the innermost boundary and the outermost boundary of the lung image after gray normalization processing. This step is a supplement to the previous step, aiming to obtain more comprehensive boundary information.

[0147] Subsequently, in an embodiment of the present invention, hole filling is performed on the segmentation result of the previous step to obtain the gray gradient range. Hole filling is a commonly used morphological operation that can effectively eliminate small holes in the segmentation result and improve the continuity of the boundary. Preferably, a filling algorithm based on connected component analysis can be used, and its complexity is O(n), where n is the number of image pixels.

[0148] Furthermore, the method segments all voxel points between the segmentation thresholds of the outermost boundary and the innermost boundary, and updates the gray gradient range with the intensity and position of the voxel points. This step makes full use of the gray information and spatial information of the image, which helps to improve the accuracy of boundary localization. The update of the gray gradient range can be achieved through the following formula:

[0149] ,

[0150] Among them, is the updated gray gradient, is the gray value of the voxel point, is the distance from the voxel point to the boundary, and are weight coefficients, and .

[0151] Next, the method performs maximum threshold segmentation on the gray-normalized lung image, and then performs a closing operation on the maximum threshold segmentation result. This step aims to obtain a rough lung contour to provide a reference for subsequent fine segmentation. The closing operation can fill small holes near the boundary and make the boundary smoother.

[0152] Then, in a preferred embodiment of the present invention, curvature magnitude analysis is performed on the closing result. The curvature magnitude analysis uses Gaussian filtering, then calculates the ratio of the curvature magnitude value of all voxel points to the average curvature magnitude value, and determines the smooth coefficient based on this ratio. The mathematical expression of this step is as follows:

[0153] ,

[0154] Among them, is the curvature magnitude ratio, is the curvature magnitude of the voxel point, is the average curvature magnitude. The smooth coefficient can be determined by setting a threshold :

[0155] ,

[0156] Finally, the method obtains the final gray segmentation threshold according to the gray gradient range and the smooth coefficient, and then performs segmentation on the gray-normalized lung image with this gray segmentation threshold. Preferably, local gray value averaging is used for region division to obtain the final segmentation result. This step combines the results of all previous processes and realizes high-precision lung segment boundary localization.

[0157] It should be noted that the method of the present invention also uses a machine learning-based model to verify the boundary correctness of the adaptive watershed marking in the output step. This strategy combining traditional image processing techniques and modern artificial intelligence methods greatly improves the accuracy and reliability of boundary localization. Preferably, a convolutional neural network (CNN) can be used as the verification model, and its structure can include multiple convolutional layers, pooling layers, and fully connected layers. The loss function of the model can be designed as:

[0158] ,

[0159] Among them, is the cross-entropy loss, is the Dice coefficient loss, and are weight coefficients.

[0160] In addition, in the processing step of the method of the present invention, the lung lobe division map and the lung lobe labeling matrix are innovatively determined according to the blood vessel position information. This method makes full use of the anatomical characteristics of the lungs and improves the accuracy and anatomical significance of the division results. Specifically, a blood vessel tracking algorithm, such as the Frangi filter combined with the minimum path algorithm, can be used to extract the blood vessel tree structure. Then, based on the branching pattern of the blood vessel tree and combined with prior anatomical knowledge, the boundaries of the lung lobes are determined.

[0161] To further improve the quality of the segmentation results, in the processing step and the output step of the method of the present invention, for each segmentation step, morphological filtering and morphological closing operations are used to eliminate the noise of small structures. These operations can effectively smooth the boundaries, fill small holes, and improve the continuity and smoothness of the segmentation results. Preferably, Gaussian filtering can be used for smoothing processing, and its mathematical expression is:

[0162] ,

[0163] Among them, is the standard deviation of the Gaussian function, usually taking a value of 1-2.

[0164] Finally, an innovation of the present invention is that in the processing step, a point is selected from the outermost boundary region, and the Euclidean distance and Manhattan distance from this point to the innermost boundary are calculated, and the difference between the two distances is used as the smoothness coefficient. This method takes into account the geometric characteristics of the boundary and can better adapt to lung segments of different shapes. Specifically, the calculation formula of the smoothness coefficient can be expressed as:

[0165] ,

[0166] Among them, is the Euclidean distance, is the Manhattan distance. The Euclidean distance and Manhattan distance are respectively defined as:

[0167] ,

[0168] ,

[0169] This calculation method of the smoothness coefficient can effectively reflect the complexity of the boundary, thus providing an important reference for subsequent boundary optimization.

[0170] Generally speaking, the method for locating the pulmonary segment boundary based on adaptive watershed provided by the present invention achieves high-precision location of the pulmonary segment boundary through a series of innovative technical means. This method not only makes full use of image processing technology and machine learning methods, but also combines the anatomical features of the lungs, thereby being able to produce segmentation results with high accuracy and good anatomical significance. This has important practical application value for improving the accuracy of pulmonary disease diagnosis, formulating personalized treatment plans, and assisting surgical planning, etc. The method of the present invention further proposes an innovative technical solution aiming to more precisely locate the pulmonary segment boundary. This solution fully considers the complexity of the pulmonary structure and individual differences, and achieves highly adaptive boundary location through multi-step processing.

[0171] First of all, this method performs three-dimensional registration on the pulmonary lobe image to obtain the three-dimensionally registered pulmonary lobe image. This step is the basis for subsequent processing, which ensures that images collected from different patients or at different time points can be compared and analyzed in the same spatial coordinate system. Preferably, a non-rigid registration algorithm based on mutual information can be adopted, and its objective function can be expressed as:

[0172] ,

[0173] wherein, is the registration transformation, and are the reference image and the floating image respectively, represents the mutual information.

[0174] Next, based on the three-dimensionally registered pulmonary lobe image, this method calculates the characteristic parameters of the pulmonary lobe and pulmonary segment surfaces. These characteristic parameters include but are not limited to curvature, normal vector, surface area, etc. The calculation of the characteristic parameters provides an important basis for subsequent pulmonary segment segmentation. For example, the mean curvature H and the Gaussian curvature K can be calculated by the following formulas:

[0175] ,

[0176] ,

[0177] wherein, and are the principal curvatures.

[0178] According to the calculated characteristic parameters, this method completes the pulmonary segment segmentation of the pulmonary lobe image to obtain the pulmonary segment segmentation result. This step can adopt a feature-based classification algorithm, such as a support vector machine (SVM) or a random forest. Preferably, an ensemble learning method can be used to improve the accuracy of classification.

[0179] Subsequently, the method extracts the lung segment boundaries of the lung segment segmentation results to obtain a boundary curve containing multiple boundary convex points and boundary concave points. This step aims to capture the detailed features of the lung segment boundaries. The extraction of boundary points can be achieved through edge detection algorithms, such as the Canny operator:

[0180] , ,

[0181] where G is the gradient magnitude, is the gradient direction.

[0182] Next, an innovation of the present invention is to perform a convex hull stitching operation on boundary curves with the same or similar directions through a convex hull algorithm to obtain a closed curve. This step can effectively handle complex boundary shapes and improve the continuity and smoothness of the boundaries. The convex hull algorithm can be implemented using the Graham scan method, and its time complexity is O(nlogn).

[0183] Then, the method obtains the cavity region enclosed by the closed curve, sets boundary convex points according to the volume of the cavity region, and sets boundary concave points of the lung segment. This step fully considers the three-dimensional structural characteristics of the lung segment and helps to more accurately describe the lung segment boundaries. The volume of the cavity region can be calculated by the voxel counting method:

[0184] ,

[0185] where I(x, y, z) is a binary image and v is the volume of a single voxel.

[0186] Finally, the core innovation of the method of the present invention lies in calculating the criterion of the watershed parameters by setting the region growing threshold as an adaptive rule in the watershed algorithm to obtain the optimal lung segment boundaries. This adaptive parameter setting method can effectively adapt to the image features under different patients and different scanning conditions, greatly improving the robustness and generality of the method. Specifically, the region growing threshold can be determined by the following formula:

[0187] ,

[0188] where, and are the average gray value and standard deviation of the target region respectively, is an adjustable parameter, and its general value range is 1.5 - 2.5.

[0189] It should be noted that through iterative optimization, the method makes the boundary of the lung segment segmentation region corresponding to the optimal watershed threshold coincide with the lung segment boundary. This process can be achieved by minimizing the following objective function:

[0190] ,

[0191] Among them, is the reference boundary point, is the boundary point of the current segmentation result, is the total number of boundary points.

[0192] In addition, the method of the present invention also includes a supporting lung segment boundary positioning system with an adaptive watershed. The system includes multiple functional modules, and each module is responsible for implementing specific steps in the method. For example, the data loading module is used to load lung CT images and lung contour point sets; the information processing module is used to obtain the lung segment set according to the lung contour point set; the lung segment area map generation module generates a lung segment area map through the region growing algorithm; the gradient map generation module calculates the watershed map of the lung segment area map and the lung segment set through gradient features.

[0193] Specifically, the watershed parameter calculation module optimizes the watershed parameters of the watershed according to the dam characteristics, the ratio of the local water area to the overall water area, and the distance between the dam and the regional centroid, and calculates the watershed spectrum. The core algorithm of this module can be expressed as:

[0194] ,

[0195] Among them, is the optimized watershed parameter, , , respectively represent the dam characteristics, area ratio, and distance characteristics, , , are the corresponding weight coefficients.

[0196] The dam parameter optimization module inputs the pixel point set into the watershed parameter optimization function to obtain the watershed spectrum. The lung segment boundary generation module generates the lung segment boundary and the lung lobe boundary according to the optimal watershed parameters input by the dam parameter optimization module. The boundary optimization module compares the lung lobe boundary with the lung contour point set to obtain the optimized lung contour point set and the final lung lobe boundary.

[0197] Finally, the smoothing processing module performs smoothing processing on the optimized lung contour point set to complete the positioning of the lung segment boundary. The smoothing processing can adopt the locally weighted scatterplot smoothing (LOESS) method, and its mathematical expression is:

[0198] ,

[0199] Among them, is the smoothed coordinate value, is the original coordinate value, , , is the regression coefficient, is the error term.

[0200] Generally speaking, the method and system for positioning the pulmonary segment boundary of the adaptive watershed provided by the present invention achieve high-precision and high-reliability positioning of the pulmonary segment boundary through a series of innovative technical means. This method not only considers the technical requirements of image processing, but also fully combines the characteristics of pulmonary anatomy, so as to be able to produce segmentation results with good anatomical significance. In particular, its adaptive parameter adjustment mechanism enables this method to effectively cope with image variability under different patients and different scanning conditions, and has broad clinical application prospects.

[0201] This high-precision method for positioning the pulmonary segment boundary is of great significance for improving the accuracy of pulmonary disease diagnosis, assisting surgical planning, evaluating treatment effects, etc. For example, in the precise positioning and staging of lung cancer, accurate pulmonary segment boundary information can help doctors better evaluate the location and invasion range of tumors. In the planning of pulmonary segment resection surgery, this method can provide detailed anatomical structure information for surgeons, which helps to formulate the best surgical plan. In addition, in the evaluation of chronic obstructive pulmonary disease (COPD), precise pulmonary segment segmentation can help doctors more accurately quantify the severity and distribution of the disease.

[0202] The above is only the preferred specific implementation manner of the present invention; however, the protection scope of the present invention is not limited thereto; any person familiar with the art within the scope disclosed by the present invention, according to the solution and improvement concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

Claims

1. An adaptive watershed-based lung segment boundary localization method, characterized in that Including: An obtaining step, including: Obtaining a contrast image or a high-resolution CT image of the lungs; Performing gray-scale normalization processing on the lung image; A processing step, including: Based on the lung image after gray-scale normalization processing, performing pulmonary segment artery segmentation; According to the connection relationship between the pulmonary segment arteries, determining a lobe division map of 4 lung regions in the left lung and 3 lung regions in the right lung; Based on the lobe division map, obtaining a lobe division table; According to the lobe division map, determining the left and right attribution of each branch artery to form a lobe marking matrix; Inside each lobe, for each branch artery, using this branch artery as the main trunk, performing region growing to obtain the corresponding connected region; Taking the connected region as the initial segmentation result of the lung segment corresponding to this branch artery; According to the connection situation between the lung segments and the lobe division table, performing merging and optimization processing on adjacent lung segments to obtain an optimized lung segment segmentation result; For any lobe, determining its innermost lung segment boundary and outermost lung segment boundary; Obtaining the innermost boundary of this lobe according to the innermost lung segment boundary, and obtaining the outermost boundary of this lobe according to the outermost lung segment boundary; Determining the gray-scale segmentation threshold corresponding to the outermost boundary and the gray-scale segmentation threshold corresponding to the innermost boundary; According to the gray-scale segmentation threshold corresponding to the outermost boundary, determining the average gray scale of the region between the outermost boundary and the innermost boundary of this lobe, and determining the gray-scale gradient range of the region between the outermost boundary and the innermost boundary of this lobe according to the average gray scale; According to the gray-scale segmentation threshold corresponding to the innermost boundary, determining the region after threshold segmentation, determining the curve parameter of the boundary line of this region according to the length of the boundary line of this region, and determining the smooth coefficient of the lobe according to the curve parameter; An output step, including: According to the gray-scale gradient range and the smooth coefficient, performing final lobe boundary segmentation to obtain a lung segment boundary positioning result.

2. The lung segment boundary localization method according to claim 1, wherein The lung segment segmentation and optimization processing in the processing step specifically include: Initializing boundary parameters, including the maximum threshold, minimum threshold, step factor, maximum number of iterations, segmentation threshold of the outermost boundary, area of the outermost boundary, segmentation threshold of the innermost boundary, area of the innermost boundary, and smooth coefficient; Taking the sum and average of the current maximum threshold and minimum threshold as the threshold parameter used in this iteration, and performing threshold segmentation on the image; Judging whether there are unsegmented lung segments. If so, calculating the boundary lengths of all unsegmented lung segments and obtaining the average value of the boundary lengths; If the average value is less than the minimum value of all boundary lengths in the region after threshold segmentation and the number of iterations in this iteration is less than the maximum number of iterations, modifying the minimum threshold according to the step factor and returning to execute the threshold segmentation step; Taking the current minimum threshold as the segmentation threshold of the innermost boundary; Judging whether there are lung segments that do not reach the minimum area threshold. If so, returning to execute the threshold segmentation step; Taking the current minimum threshold as the segmentation threshold of the outermost boundary; Judge whether the area of the outermost boundary is less than the minimum area threshold. If so, modify the segmentation threshold of the outermost boundary to the segmentation threshold of the innermost boundary, and modify the segmentation threshold of the innermost boundary to the segmentation threshold of the outermost boundary in the previous iteration; Judge whether the number of final iterations is equal to the maximum number of iterations. If so, output the obtained boundary parameters to get the lung segment segmentation result.

3. The lung segment boundary localization method according to claim 1, characterized in that, The lung segment merging and optimization process in the above processing steps specifically includes: Initialize the boundary parameters, including the maximum threshold, minimum threshold, step factor, maximum number of iterations, segmentation threshold of the outermost boundary, area of the outermost boundary, segmentation threshold of the innermost boundary, area of the innermost boundary, and smoothing coefficient; Take the sum and average of the current maximum threshold and minimum threshold as the threshold parameter used in this iteration, and perform threshold segmentation processing on the image; Judge whether there is an unsegmented lung lobe area in the threshold segmentation result. If so, calculate the average gray value of all connected regions and determine the segmentation threshold of the outermost boundary; Determine the final image through the weight mixing algorithm based on the gradient magnitude and curvature magnitude; Judge whether the segmentation threshold of the outermost boundary is greater than the current minimum threshold. If so, calculate the average length of all current connected regions; If the average length is less than the area of the outermost boundary and the current number of iterations is less than the maximum number of iterations, update the minimum threshold and return to execute the threshold segmentation processing step; Calculate the average length of all connected regions. If the average length is greater than the area of the outermost boundary, update the segmentation threshold of the outermost boundary and return to execute the threshold segmentation processing step; Save the current minimum threshold; Output the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary as the final result of the lung segment merging and optimization process.

4. The lung segment boundary localization method according to claim 1, wherein The specific process of determining the segmentation threshold of the innermost boundary and the segmentation threshold of the outermost boundary in the above processing steps includes: Determine the gray value of the outermost boundary in the lung image after the gray normalization process, and use the current gray value of the outermost boundary as the segmentation threshold of the outermost boundary; Perform reverse gray value sorting. If the average value of the gray values of the three voxel points with the largest gray values in the current lung lobe is less than the segmentation threshold of the innermost boundary, determine the average gray value through the gray value of the outermost boundary and the segmentation threshold of the innermost boundary in the lung image after the gray normalization process, and then obtain the gray range centered on the average gray value; Judge whether the average gray value of the lung lobe area is within the gray range. If so, judge whether the difference between the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary is less than 5; If the difference is less than 5, the segmentation threshold of the innermost boundary is the current gray value of the outermost boundary, otherwise the segmentation threshold of the innermost boundary is the sum of the current gray value of the outermost boundary and 5; If the difference between the maximum value and the minimum value of the gray values of all voxel points in the lung lobe area exceeds 5, determine the gray range by subtracting the gray value of the outermost boundary from the average value of the gray values of 50% of the voxel points with the largest gray values in the lung lobe; Determine the gray range by subtracting the segmentation threshold of the innermost boundary from the maximum value of the gray values of all current voxel points in the lung lobe; Determine the smoothing coefficient based on the segmentation threshold of the current outermost boundary and the segmentation threshold of the current innermost boundary.

5. The lung segment boundary localization method according to claim 1, wherein The output step specifically includes: Segment the lung image after the grayscale normalization process using the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary to obtain two segmented regions. Segment the lung image after the grayscale normalization process using the segmentation threshold of the innermost boundary and the segmentation threshold of the outermost boundary. Perform hole filling on the segmentation result of the previous step to obtain the grayscale gradient range. Segment all voxel points between the segmentation threshold of the outermost boundary and the segmentation threshold of the innermost boundary, and update the grayscale gradient range with the intensity and position of the voxel points. Perform maximum threshold segmentation on the lung image after the grayscale normalization process, and then perform a closing operation on the maximum threshold segmentation result. Perform curvature amplitude analysis on the closing result. Gaussian filtering is used for the curvature amplitude analysis. Then calculate the ratio of the curvature amplitude value of all voxel points to the average curvature amplitude value, and determine the smoothing coefficient based on this ratio. Obtain the final grayscale segmentation threshold based on the grayscale gradient range and the smoothing coefficient, and then segment the lung image after the grayscale normalization process using this grayscale segmentation threshold. Perform region division using local grayscale value averaging to obtain the final segmentation result.

6. The lung segment boundary localization method according to claim 1, characterized in that, In the output step, use a machine learning-based model to verify the correctness of the boundaries marked by the adaptive watershed.

7. The lung segment boundary localization method according to claim 1, wherein In the processing step, determine the lung lobe division map and the lung lobe marking matrix based on the blood vessel position information.

8. The lung segment boundary localization method according to claim 1, wherein, In the processing step and the output step, for each segmentation step, eliminate the noise of small structures through morphological filtering and morphological closing operations.

9. The method for positioning the pulmonary segment boundary according to claim 8, characterized in that, In the processing step, select a point from the outermost boundary region, calculate its Euclidean distance and Manhattan distance to the innermost boundary, and use the difference between the Euclidean distance and the Manhattan distance as the smoothing coefficient.

10. The lung segment boundary localization method according to claim 1, wherein The method further includes: Perform three-dimensional registration on the lung lobe image to obtain the three-dimensionally registered lung lobe image. Based on the three-dimensionally registered lung lobe image, calculate the characteristic parameters of the lung lobe and lung segment surfaces, and complete the lung segment segmentation work of the lung lobe image according to the characteristic parameters to obtain the lung segment segmentation result. Extract the lung segment boundaries of the lung segment segmentation result to obtain a boundary curve containing multiple boundary convex points and boundary concave points. Through the convex hull algorithm, perform convex hull stitching operations on the boundary curves with the same direction to obtain a closed curve. Obtain the cavity region enclosed by the closed curve, set boundary convex points according to the volume of the cavity region, and set lung segment boundary concave points. Based on the calculation criterion of the watershed parameters with the set region growth threshold in the watershed algorithm as the adaptive rule, obtain the optimal lung segment boundary so that the boundary of the lung segment segmentation region corresponding to the optimal watershed threshold coincides with the lung segment boundary.

Images