Method for accurately locating pulmonary segment boundary by adaptive watershed growth

Abstract

The present invention relates to the field of medical image processing technology, and more specifically, to a method for accurately locating lung segment boundaries using adaptive watershed growth, comprising: acquiring lung CT image data; performing image preprocessing based on the lung CT image data to obtain a preprocessed lung image; determining a main point set of a pulmonary segment artery according to the preprocessed lung image; determining an initial seed point based on the main point set using a sub-point intelligent arrangement strategy; performing point competition-based adaptive watershed growth based on the initial seed point to obtain a preliminary positioning result of the lung segment boundary; performing boundary conflict detection and processing based on the preliminary positioning result of the lung segment boundary to obtain an optimized lung segment boundary; and outputting the optimized lung segment boundary as a final precise positioning result of the lung segment boundary. By dynamically adjusting growth parameters and introducing a point competition mechanism, the method can more accurately capture the complex morphology of the lung segment boundary and significantly improve the accuracy and robustness of segmentation.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and more particularly to a method for accurately locating lung segment boundaries using adaptive watershed growth. Background Art

[0002] Precise localization of lung segment boundaries is crucial for lung disease diagnosis, surgical planning, and radiotherapy. Advances in medical imaging technology, particularly the widespread use of computed tomography (CT), have greatly improved the visualization of lung anatomy. However, due to the complexity of lung structure and individual variability, accurately localizing lung segment boundaries remains a challenging task.

[0003] Traditional methods for locating lung segment boundaries rely primarily on manual or semi-automatic segmentation techniques. These methods are not only time-consuming and labor-intensive but also susceptible to subjective operator judgment, resulting in poor consistency and reproducibility of results. In recent years, with the advancement of computer vision and artificial intelligence technologies, several automated lung segmentation methods have been proposed. These methods can be broadly categorized as those based on anatomical knowledge, those based on image features, and those based on deep learning.

[0004] Anatomical knowledge-based methods typically use anatomical landmarks such as lobar fissures and vascular trees to locate pulmonary segment boundaries. While these methods offer the advantage of good interpretability, their accuracy is often limited by the precision of anatomical landmark identification. In particular, in certain pathological conditions, such as those with emphysema or pulmonary fibrosis, anatomical landmarks may become obscured or distorted, rendering these methods ineffective.

[0005] Image feature-based methods, such as region growing and level set methods, perform segmentation by analyzing image features such as grayscale and texture. These methods perform well when dealing with normal lung structures, but lack robustness when dealing with diseased areas or when dealing with large individual differences. Furthermore, these methods often require manual configuration of numerous parameters, which significantly impacts the results, limiting their widespread clinical application.

[0006] In recent years, deep learning-based methods, particularly convolutional neural networks (CNNs), have made significant progress in the field of medical image segmentation. These methods are able to automatically learn image features and even surpass the performance of human experts in certain tasks. However, deep learning methods also face several challenges. First, they typically require large amounts of precisely labeled training data, which is expensive and time-consuming in the medical field. Second, deep learning models are often black-box and lack interpretability, which may raise ethical and legal issues in medical applications. Finally, the generalization ability of these methods in handling edge cases and rare cases still needs to be improved.

[0007] Existing technologies, both traditional and deep learning approaches, face a common challenge: balancing accuracy, efficiency, and adaptability. This is particularly true when dealing with diverse patients, different scanning devices, and various pathological conditions. Furthermore, most methods lack effective mechanisms for integrating expert knowledge with automated algorithms, failing to fully leverage the experience of clinical experts to continuously improve system performance. Summary of the Invention

[0008] This invention aims to address the aforementioned technical issues by providing a method for accurately locating lung segment boundaries using adaptive watershed growing. This method innovatively combines image processing, machine learning, and deep learning techniques to achieve high-precision, high-efficiency, and highly adaptable localization of lung segment boundaries.

[0009] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0010] The adaptive watershed growing method for accurate lung segment boundary positioning includes:

[0011] The acquisition steps include:

[0012] Acquire lung CT image data;

[0013] Processing steps include:

[0014] Performing image preprocessing based on the lung CT image data to obtain a preprocessed lung image;

[0015] determining a set of main points of the pulmonary segmental artery according to the preprocessed lung image;

[0016] Based on the set of backbone points, an initial seed point is determined by adopting a sub-point intelligent arrangement strategy;

[0017] Based on the initial seed points, adaptive watershed growth based on point competition is performed to obtain a preliminary positioning result of the lung segment boundary;

[0018] Based on the preliminary positioning result of the lung segment boundary, boundary conflict detection and processing are performed to obtain an optimized lung segment boundary;

[0019] Output steps include:

[0020] The optimized lung segment boundary is output as the final lung segment boundary precise positioning result.

[0021] Preferably, the image preprocessing specifically includes:

[0022] performing bilateral filtering on the lung CT image data to obtain a denoised lung image;

[0023] Blood vessels are extracted based on the denoised lung image to obtain a lung blood vessel mask.

[0024] Preferably, the sub-point intelligent arrangement strategy specifically includes:

[0025] Based on the anatomical structure of the pulmonary artery in the human body, determine the origin and branching points of the aorta;

[0026] Determine the anatomical landmarks based on the connection points of the main pulmonary artery and the left and right pulmonary arteries with the left and right hilums;

[0027] The position of the initial seed point is determined based on the aorta starting point, branch point and anatomical landmark point.

[0028] Preferably, the adaptive watershed growth based on point competition specifically includes:

[0029] Constructing a point group based on the initial seed point;

[0030] Using Euclidean distance as a similarity parameter, unsupervised clustering is performed on the point group;

[0031] According to the clustering results, determine the watershed corresponding to each class;

[0032] Based on the watershed, adaptive seed growing is performed, which includes:

[0033] Define the point group growth rule based on the distance between seed points, including adaptive step size and distance threshold parameters;

[0034] Set the adaptive convergence criterion based on the average distance between the class points and the seed points;

[0035] By comparing the competition scores of each point, boundaries are drawn between adjacent point classes.

[0036] Preferably, the boundary conflict detection and processing specifically includes:

[0037] Calculate the seed competition score between the boundary area point class and the adjacent point class area;

[0038] identifying boundary conflict regions between a plurality of watersheds based on the seed competition scores;

[0039] For detected boundary conflicts, the narrowest channel between two seed points is found in the conflict area;

[0040] According to the narrowest channel, the point cluster in the conflict area is divided into two parts to resolve the boundary conflict.

[0041] As an option, it also includes:

[0042] Based on the deep learning model, the optimized lung segment boundaries are further optimized, wherein:

[0043] A convolutional neural network model is used to compare the model-predicted boundaries with the standard results;

[0044] The boundary correction model is optimized through the loss function to obtain the final accurate positioning result of the lung segment boundary.

[0045] Preferably, the adaptive seed growth further comprises:

[0046] Anisotropic tensor fields are used to simulate the directional constraints of local expansion;

[0047] Based on the anisotropic tensor field, the control point group grows along the blood vessel direction.

[0048] As an advantage, it also includes:

[0049] Performing lung tumor analysis based on the precise positioning results of the lung segment boundaries;

[0050] The lung segment boundary is used as the reference boundary for tumor staging;

[0051] Combined with imaging omics analysis tools, it provides tumor pathological feature analysis information.

[0052] Preferably, the output step further comprises:

[0053] Visually displaying the precise positioning result of the lung segment boundary;

[0054] A three-dimensional model including the lung segment boundaries is generated.

[0055] As an advantage, it also includes:

[0056] Setting a user interaction interface to allow doctors to fine-tune the precise positioning result of the lung segment boundary;

[0057] Based on the fine-tuning results, the training dataset of the deep learning model is updated.

[0058] The method of the present invention has the following significant technical effects:

[0059] First, the adaptive watershed growing model proposed in this paper can effectively adapt to the anatomical differences and various pathological conditions of different patients. By dynamically adjusting the growing parameters and introducing a point competition mechanism, this method can more accurately capture the complex morphology of lung segment boundaries, significantly improving the accuracy and robustness of segmentation.

[0060] Secondly, this invention cleverly combines the advantages of traditional image processing techniques and deep learning methods. Initial segmentation based on image features provides a good starting point, while the introduction of a deep learning model further optimizes the segmentation results. This hybrid approach not only improves the algorithm's performance but also enhances its interpretability, making it more likely to gain the trust and acceptance of clinical experts.

[0061] Furthermore, the boundary conflict detection and resolution mechanism introduced in this invention effectively resolves the issue of blurred boundaries between multiple lung segments. This innovation significantly improves the consistency and reliability of segmentation results, providing more accurate anatomical information for subsequent clinical applications such as surgical planning and radiotherapy.

[0062] Furthermore, the user interface and continuous learning mechanism designed in this invention establish a collaborative relationship between clinical experts and the AI system. This not only allows experts to fine-tune the results based on their clinical experience, but also feeds these corrections back into the model, enabling continuous optimization of system performance. This human-machine collaboration significantly improves the system's practicality and reliability.

[0063] Finally, the method of this invention has broad application prospects. In addition to precisely localizing lung segment boundaries, this method can also be extended to multiple fields, including lung tumor analysis, surgical planning, and radiotherapy. By generating precise three-dimensional models and combining them with radiomics analysis, this invention provides powerful technical support for personalized medicine.

[0064] In general, the adaptive watershed growth method for precise positioning of lung segment boundaries provided by the present invention innovatively solves the problem of balancing accuracy, efficiency, and adaptability, providing a more reliable and accurate tool for the diagnosis and treatment of lung diseases, and is expected to play an important role in improving medical quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 The figure is an overall flow chart of the method of the present invention.

[0066] Figure 2 This is a flow chart of the adaptive watershed growth based on point competition of the present invention.

[0067] Figure 3 Flowchart of boundary conflict detection and processing of the present invention.

[0068] Figure 4 Flowchart for deep learning model optimization of the present invention. DETAILED DESCRIPTION

[0069] like Figure 1-4As shown, the present invention provides a method for accurately locating lung segment boundaries using adaptive watershed growing. This method achieves high-precision localization of lung segment boundaries by innovatively combining image processing, machine learning, and deep learning techniques. Specific embodiments of the present invention are described in detail below.

[0070] The method for accurately locating lung segment boundaries using adaptive watershed growth provided by the present invention comprises the following steps:

[0071] First, in the acquisition step, the method acquires lung CT image data. This CT image data is typically acquired by a CT scanner at a medical institution and contains detailed anatomical information about the patient's lungs. Preferably, the CT image resolution should be no less than 512x512 pixels, with a slice thickness no greater than 1 mm, to ensure adequate image quality and detail.

[0072] Next, in the processing step, the method of the present invention first performs image preprocessing based on the acquired lung CT image data to obtain a preprocessed lung image. Image preprocessing is a key step that can effectively improve the accuracy and efficiency of subsequent processing. In one embodiment of the present invention, image preprocessing can include operations such as noise removal, contrast enhancement, and image normalization. For example, a Gaussian filter can be used to remove high-frequency noise in the image, while a histogram equalization method can be used to enhance image contrast.

[0073] After preprocessing is completed, this method determines the main point set of the pulmonary segmental artery based on the preprocessed lung image. This step is the basis for subsequent precise positioning. Preferably, a threshold-based segmentation method combined with morphological operations can be used to extract the pulmonary vascular structure, and then a skeletonization algorithm can be used to obtain the centerline of the blood vessel to determine the main point set. For example, the CT value threshold can be set between -600 and 400 HU. This range can usually effectively segment the pulmonary vascular structure.

[0074] After determining the set of main points, the method of the present invention uses a sub-point intelligent placement strategy based on this set of main points to determine the initial seed points. The purpose of this step is to provide a good starting point for the subsequent watershed growing algorithm. In one specific embodiment, the sub-point intelligent placement strategy can determine the seed points based on the spatial distribution of the main points and the local vascular morphological characteristics. For example, seed points can be placed at the midpoint between the main points, while taking into account the bifurcation of the blood vessels and adding additional seed points at the bifurcation.

[0075] Next, based on the determined initial seed points, this method performs adaptive watershed growth based on point competition to obtain preliminary localization results for lung segment boundaries. This is one of the core steps of the present invention, which achieves adaptive growth of lung segment boundaries by simulating the competitive relationship between points. Specifically, the following formula can be used to calculate the competition intensity between points:

[0076] ,

[0077] in, Indicates a point and The competition intensity between them, I(p) represents the image intensity at point p, is an adjustable parameter used to control the decay rate of competition intensity. Can be set between 10% and 20% of the image intensity range.

[0078] After obtaining preliminary segment boundary localization results, the method of the present invention performs boundary conflict detection and resolution based on these results to obtain optimized segment boundaries. This step aims to address potential boundary overlap or gaps between different segments. In one embodiment, morphological operations can be used to detect and resolve boundary conflicts. For example, dilation and erosion operations can be performed on the preliminary localization results, and the results before and after the operations can be compared to identify areas of potential conflict.

[0079] Finally, in the output step, the method outputs the optimized lung segment boundaries as the final lung segment boundary precise positioning result. This result can be presented in various forms such as binary image, contour point set, or 3D model to meet different application requirements.

[0080] The method of the present invention may also include other optimization steps. For example, a deep learning model can be introduced to further optimize the boundary localization results. In this case, a classic image segmentation network such as U-Net can be used, taking the preliminary localization results as input and outputting a more refined boundary. Training such a network generally requires a large amount of annotated data; it is recommended to use at least 1,000 expert-annotated lung segment CT images for training.

[0081] In summary, the adaptive watershed growing method for precise lung segment boundary localization provided by this invention achieves high-precision localization of lung segment boundaries by integrating image processing, machine learning, and deep learning techniques. This method boasts strong adaptability, high accuracy, and excellent robustness, and can provide important technical support for the diagnosis and treatment of lung diseases.

[0082] The method of the present invention, when performing adaptive watershed growth based on point competition, specifically includes the following steps:

[0083] First, a point cluster is constructed based on the initial seed points. This step provides basic data for subsequent clustering and watershed growth. Preferably, the point cluster can be constructed using the k-nearest neighbor (k-NN) algorithm, which selects the k nearest points around each seed point. In one embodiment of the present invention, the k value can be set between 50 and 100, which generally strikes a good balance between computational efficiency and coverage.

[0084] Next, this method uses Euclidean distance as the similarity parameter to perform unsupervised clustering on the constructed point group. The purpose of this step is to classify similar points into the same watershed. Preferably, the K-means clustering algorithm can be used. The advantage of the K-means algorithm is that it is simple, efficient, and easy to implement. In practical applications, the number of clusters K can be determined based on the number of initial seed points, and can usually be set to 1.2 to 1.5 times the number of seed points to allow a certain degree of over-segmentation to facilitate subsequent boundary optimization.

[0085] Based on the clustering results, this method determines the watershed corresponding to each class. This step converts the clustering results into preliminary lung segment boundaries. In a preferred embodiment of the present invention, the following formula can be used to define the watershed:

[0086] ,

[0087] in, represents the i-th watershed, p represents any point in the image, and represent the i-th and j-th cluster centers respectively, represents the Euclidean distance.

[0088] Based on the determined watershed, the method of the present invention performs adaptive seed growth. This is one of the core steps of the present invention, which achieves adaptive segmentation of lung segments of different shapes and sizes by dynamically adjusting growth parameters. Specifically, adaptive seed growth includes the following key points:

[0089] First, define the point group growth rule based on the distance between seed points, including the adaptive step size and distance threshold parameters. Preferably, the adaptive step size can be calculated using the following formula:

[0090] ,

[0091] in, represents the step size of the i-th iteration, is the initial step size (can be set from 1 to 3 pixels), is the distance from the current point to the nearest seed point, is the maximum distance between all points and their nearest seed point. This formula can make the area close to the seed point grow slower, while the area far from the seed point grows faster, thus adapting to lung segments of different sizes.

[0092] Secondly, the method sets an adaptive convergence criterion based on the average distance between the class points and the seed points. The purpose of this step is to obtain appropriate segmentation results in lung segments of different sizes. In one embodiment of the present invention, the following formula can be used to determine whether convergence has occurred:

[0093] ,

[0094] Where N is the number of points in the current class, is the i-th point in the class, c is the center point of the class, is a small positive number representing the convergence threshold. Can be set from 1 to 3 pixels.

[0095] Finally, this method divides the boundaries between adjacent point classes by comparing the competition scores of each point. The purpose of this step is to obtain more accurate lung segment boundaries. The competition score can be calculated using the following formula:

[0096] ,

[0097] Where S(p) is the competition score of point p, N(p) is the neighborhood of p, w(p,q) is the weight between p and q (which can be defined based on distance or image intensity), is the indicator function, when q belongs to the same region as p 1 if yes, 0 otherwise.

[0098] The method of the present invention, when performing boundary conflict detection and processing, specifically includes the following steps:

[0099] First, the seed competition score between the boundary region point class and the adjacent point class region is calculated. The purpose of this step is to quantify the competition relationship between different regions. In a preferred embodiment of the present invention, the competition score can be calculated using the following formula:

[0100] ,

[0101] in, Indicates area and The competition score between represents the boundary point set of two regions, and Represents point p to region and degree of belonging.

[0102] Next, the method identifies the boundary conflict areas between multiple watersheds based on the calculated seed competition scores. Preferably, a threshold T can be set. >T, the area is considered and There are boundary conflicts between them. The choice of threshold T can be determined by experiment and can usually be set to 1.5 to 2 times the average competition score.

[0103] For detected boundary conflicts, the method of the present invention searches for the narrowest path between the two seed points in the conflicting region. The purpose of this step is to find the most reasonable boundary demarcation location. In one embodiment of the present invention, the Dijkstra algorithm can be used to find the narrowest path, using the image gradient as the edge weight.

[0104] Finally, this method divides the point clusters in the conflicting area into two parts based on the narrowest channel found, thereby resolving boundary conflicts. This step can effectively eliminate boundary overlap or gap problems between different lung segments and improve segmentation accuracy.

[0105] In a preferred embodiment of the present invention, a deep learning model can be introduced to further optimize the optimized lung segment boundaries. Specifically, a convolutional neural network model can be used to compare the model-predicted boundaries with the standard results. Classic image segmentation network structures such as U-Net or DeepLab can be used here. The network input can be a fusion of the original CT image and the preliminary segmentation results, and the output is the optimized segmentation mask.

[0106] To train such a deep learning model, a large number of lung segment CT images with expert annotations are required. Preferably, the training set should be no less than 1,000 cases to ensure the generalization ability of the model. During training, the Dice coefficient can be used as a loss function, which effectively balances the contributions of objects of different sizes. The loss function can be expressed as:

[0107] ,

[0108] Among them, X is the predicted segmentation result, Y is the true label, Indicates the size of the collection.

[0109] By optimizing the boundary correction model using a loss function, this method ultimately achieves precise lung segment boundary localization. This result not only considers local image features but also incorporates the deep learning model's understanding of global semantic information, achieving high accuracy.

[0110] In another embodiment of the present invention, adaptive seed growth further includes using an anisotropic tensor field to simulate the directional constraints of local expansion. This method can better adapt to the complex structure of pulmonary blood vessels. Specifically, the anisotropic tensor field can be constructed using the Hessian matrix:

[0111] ,

[0112] in, 、 、 etc. are the second-order partial derivatives of the image in the corresponding direction. By analyzing the eigenvalues and eigenvectors of the Hessian matrix, the main directions of the local structure can be obtained. Based on the constructed anisotropic tensor field, this method controls the growth of the point group along the blood vessel direction. This can be achieved by modifying the adaptive step size formula mentioned above:

[0113] ,

[0114] in, is the unit vector of the current growth direction, is the diffusion tensor derived from the Hessian matrix, is a parameter that controls the degree of anisotropy. Can be set between 0.5 and 2.

[0115] By introducing anisotropic tensor fields, the method of the present invention can better adapt to the complex anatomical structure of the lungs and improve the accuracy of segmentation, especially in small blood vessels and bifurcations. This is of great significance for subsequent clinical diagnosis and treatment planning. The method of the present invention can not only accurately locate the boundaries of lung segments, but can also be further applied to lung tumor analysis. In a preferred embodiment of the present invention, lung tumor analysis can be performed based on the obtained precise positioning results of the lung segment boundaries. This analysis method uses the lung segment boundaries as the reference boundaries for tumor staging, and combines imaging genomics analysis tools to provide comprehensive tumor pathology feature analysis information.

[0116] Specifically, this method first uses precisely located lung segment boundaries as anatomical reference to determine the precise location and extent of the tumor. Preferably, deep learning-based object detection algorithms, such as Faster R-CNN or YOLO, can be used to automatically detect and localize lung tumors. These algorithms can incorporate lung segment boundary information as additional feature input during training, thereby improving detection accuracy.

[0117] After determining the location of the tumor, the method of the present invention combines radiomics analysis tools to extract multi-dimensional features of the tumor. These features may include but are not limited to:

[0118] 1. Morphological characteristics: such as tumor volume, surface area, maximum diameter, sphericity, etc.

[0119] 2. Density characteristics: such as average CT value, CT value distribution histogram, etc.

[0120] 3. Texture features: such as gray-level co-occurrence matrix (GLCM) features, run-length matrix (RLRM) features, etc.

[0121] 4. Wavelet features: multi-scale features extracted through wavelet transform.

[0122] Preferably, feature selection methods such as principal component analysis (PCA) or LASSO can be used to select the most discriminative feature subset from a large number of features. This can effectively reduce feature redundancy and improve the efficiency and accuracy of subsequent analysis.

[0123] Based on the extracted features, this method can construct a machine learning model to predict tumor malignancy, stage, and potential prognosis. In one embodiment of the present invention, a prediction model can be constructed using algorithms such as random forests or support vector machines (SVMs). These models take the extracted radiomics features as input and output information such as the tumor's TNM stage and malignancy score.

[0124] In order to improve the reliability of the prediction, the method of the present invention can also integrate clinical information, such as the patient's age, gender, smoking history, etc. This information can be input into the prediction model as additional features to obtain more comprehensive and accurate analysis results.

[0125] Through the above steps, the method of the present invention not only accurately locates lung segment boundaries but also provides important reference information for diagnosis and treatment decisions of lung tumors. This method, combining anatomical localization and radiomics analysis, is expected to significantly improve the accuracy of lung cancer diagnosis and the effectiveness of personalized treatment.

[0126] In another embodiment of the present invention, the output step further includes visualizing the precise location of the lung segment boundaries and generating a three-dimensional model containing the lung segment boundaries. This visualization and three-dimensional modeling not only intuitively demonstrate the segmentation results but also provide important reference for doctors' diagnosis and surgical planning.

[0127] Specifically, the visualization display can use multi-planar reconstruction (MPR) technology to simultaneously display the lung segment boundaries in the axial, coronal, and sagittal planes. Preferably, different colors can be used to identify different lung segments to enhance the visual effect. For example, the following color scheme can be used:

[0128] Right upper lung segment: red;

[0129] Right middle lung segment: green;

[0130] Right lower lung segment: blue;

[0131] Left upper lung segment: yellow;

[0132] Left lower lung segment: purple;

[0133] The display interface can provide interactive adjustment functions, allowing users to adjust the window width and window position to suit different observation needs. In addition, measurement tools can be provided to enable doctors to accurately measure the size and distance of the area of interest.

[0134] To generate a 3D model, the method of the present invention preferably employs surface reconstruction techniques. First, a surface point cloud is extracted for each lung segment based on the segmentation results. Then, an initial triangular mesh model is generated using algorithms such as Marching Cubes. To reduce model complexity and smooth the surface, mesh simplification and smoothing algorithms, such as Laplacian smoothing, can be further applied.

[0135] The generated 3D model should support rotation, scaling, and translation, allowing users to observe the morphology of the lung segments from any angle. Preferably, a semi-transparent display function should also be provided to make internal structures visible. Furthermore, other relevant information, such as the location and size of the tumor, and the orientation of major blood vessels and bronchi, can be overlaid on the 3D model.

[0136] Through this visualization and 3D modeling approach, the present invention not only provides accurate lung segment boundary localization results, but also presents these results in an intuitive and interactive manner, greatly improving the usability and clinical value of the results.

[0137] In another embodiment of the present invention, the method further includes providing a user interface that allows doctors to fine-tune the precise localization of lung segment boundaries and, based on the fine-tuning results, update the deep learning model's training dataset. This interactive feedback mechanism can continuously improve system performance, making it better adapted to the image characteristics of different hospitals and equipment.

[0138] The user interface preferably adopts an intuitive graphical design and mainly includes the following functions:

[0139] 1. Boundary Editing: Allows doctors to modify lung segment boundaries directly on the image using a mouse or stylus. Tools such as brushes and erasers are provided to make the editing process more convenient.

[0140] 2. Marking tools: Provides a series of marking tools, such as arrows, text annotations, etc., allowing doctors to mark special anatomical structures or lesion areas.

[0141] 3. Comparative display: The original segmentation results and the manually corrected results are displayed simultaneously to facilitate comparison and evaluation by doctors.

[0142] 4. Undo / Redo: Provides multi-level undo and redo functions to make the editing process more flexible.

[0143] 5. Automatic optimization: Based on the doctor's local modifications, the system can automatically optimize the segmentation results of adjacent areas, reducing the workload of manual operations.

[0144] Once the doctor completes fine-tuning, the system records the original segmentation results, the manual correction results, and the corresponding original image data. This data is used to update the training dataset of the deep learning model. Preferably, online learning or incremental learning methods can be used to enable the model to continuously adapt to new data.

[0145] The process of updating the training dataset can be done as follows:

[0146] 1. Data enhancement: Perform appropriate data enhancement on the results of manual correction, such as rotation, scaling, adding noise, etc., to increase data diversity.

[0147] 2. Sample weight adjustment: Assign higher weights to newly added samples so that the model pays more attention to these new, expert-corrected samples.

[0148] 3. Model fine-tuning: Use the updated dataset to fine-tune the existing model. A smaller learning rate can be used to maintain the stability of the model.

[0149] 4. Validation and deployment: Validate the fine-tuned model to ensure improved performance before deploying it in real-world applications.

[0150] Through this interactive feedback and continuous learning mechanism, the method of the present invention can continuously improve its accuracy and robustness, better adapting to various situations in clinical practice. This not only improves the practicality of the system but also establishes a positive collaborative relationship between doctors and AI systems, which is conducive to the promotion and application of AI-assisted diagnosis technology.

[0151] The above description is only a preferred specific embodiment of the present invention; however, the protection scope of the present invention is not limited thereto; any person familiar with the art who makes equivalent replacements or changes based on the scheme and improved concepts of the present invention within the scope disclosed by the present invention shall be covered by the protection scope of the present invention.

Claims

1. Adaptive watershed growing method for accurate positioning of lung segment boundaries, characterized by: include: The acquisition steps include: Acquire lung CT image data; Processing steps include: Performing image preprocessing based on the lung CT image data to obtain a preprocessed lung image; determining a set of main points of the pulmonary segmental artery according to the preprocessed lung image; Based on the set of backbone points, an initial seed point is determined by adopting a sub-point intelligent arrangement strategy; Based on the initial seed points, adaptive watershed growth based on point competition is performed to obtain a preliminary positioning result of the lung segment boundary; Based on the preliminary positioning result of the lung segment boundary, boundary conflict detection and processing are performed to obtain an optimized lung segment boundary; Output steps include: Outputting the optimized lung segment boundary as the final lung segment boundary precise positioning result; The adaptive watershed growth based on point competition specifically includes: Constructing a point group based on the initial seed point; Using Euclidean distance as a similarity parameter, unsupervised clustering is performed on the point group; According to the clustering results, determine the watershed corresponding to each class; Based on the watershed, adaptive seed growing is performed, which includes: Define the point group growth rule based on the distance between seed points, including adaptive step size and distance threshold parameters; Set the adaptive convergence criterion based on the average distance between the class points and the seed points; By comparing the competitive scores of pixels to be classified in the image, boundaries are drawn between adjacent point classes.

2. The method according to claim 1, characterized in that The image preprocessing specifically includes: performing bilateral filtering on the lung CT image data to obtain a denoised lung image; Blood vessels are extracted based on the denoised lung image to obtain a lung blood vessel mask.

3. The method according to claim 1, characterized in that The sub-point intelligent layout strategy specifically includes: Based on the anatomical structure of the pulmonary artery in the human body, determine the origin and branching points of the aorta; Determine the anatomical landmarks based on the connection points of the main pulmonary artery and the left and right pulmonary arteries with the left and right hilums; The position of the initial seed point is determined based on the aorta starting point, branch point and anatomical landmark point.

4. The method according to claim 1, wherein The boundary conflict detection and processing specifically includes: Calculate the seed competition score between the boundary area point class and the adjacent point class area; identifying boundary conflict regions between a plurality of watersheds based on the seed competition scores; For detected boundary conflicts, the narrowest channel between two seed points is found in the conflict area; According to the narrowest channel, the point cluster in the conflict area is divided into two parts to resolve the boundary conflict.

5. The method according to claim 1, wherein Also includes: Based on the deep learning model, the optimized lung segment boundaries are further optimized, wherein: A convolutional neural network model is used to predict lung segment boundaries; Comparing the boundaries predicted by the convolutional neural network model with standard results; The convolutional neural network model is optimized by the loss function to obtain the final accurate positioning result of the lung segment boundary.

6. The method according to claim 1, characterized in that The adaptive seed growth further comprises: Anisotropic tensor fields are used to simulate the directional constraints of local expansion; Based on the anisotropic tensor field, the control point group grows along the blood vessel direction.

7. The method according to claim 1, characterized in that Also includes: Performing lung tumor analysis based on the precise positioning result of the lung segment boundaries; The lung segment boundary is used as the reference boundary for tumor staging; Combined with imaging omics analysis tools, it provides tumor pathological feature analysis information.

8. The method according to claim 1, characterized in that The output step further includes: Visually displaying the precise positioning result of the lung segment boundary; A three-dimensional model including the lung segment boundaries is generated.

9. The method according to claim 1, characterized in that Also includes: Setting a user interaction interface to allow doctors to fine-tune the precise positioning result of the lung segment boundary; Based on the fine-tuning results, the training dataset of the deep learning model is updated.

Images