A medical image segmentation and three-dimensional reconstruction method and system based on region growing

Abstract

This invention discloses a method and system for medical image segmentation and 3D reconstruction based on region growing. The method includes the following steps: S1, acquiring the original medical image data of the target object and preprocessing it to generate a preprocessed 2D image sequence; S2, selecting seed points on the preprocessed 2D image sequence; S3, segmenting based on the seed points to generate segmented regions; S4, optimizing the boundaries of the segmented regions to generate an optimized binary mask; S5, generating the final binary mask; S6, completing the 3D model reconstruction; S7, spatially registering and fusing the reconstructed 3D model. The semi-automated segmentation process of this invention reduces the subjective differences of manual delineation, achieving over 90% consistency in reconstruction results for the same image across different operators, providing a reproducible technical basis for multi-center studies and efficacy evaluation.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, specifically relating to a method and system for medical image segmentation and three-dimensional reconstruction based on region growing. Background Technology

[0002] Three-dimensional reconstruction technology for medical images has been widely used in clinical diagnosis, surgical planning, and doctor-patient communication. Existing three-dimensional reconstruction methods mainly include the following steps: acquiring CT or MRI two-dimensional image sequences, extracting the target tissue region through threshold segmentation or manual delineation, and generating a three-dimensional surface model.

[0003] Existing technologies mainly include: (1) traditional threshold segmentation methods, such as fixed threshold method, Otsu method, and histogram-based bimodal method; (2) region growing methods, such as ITK's ConfidenceConnectedImageFilter, which uses a fixed multiplier and gray mean and standard deviation to calculate the confidence interval; (3) graph cut methods, such as the GrowCut algorithm, which mainly segment based on the gray similarity of neighboring pixels.

[0004] However, existing technologies have the following shortcomings: 1) Limited segmentation accuracy: Traditional thresholding methods are ineffective at segmenting complex areas such as the junction of soft tissue and bone, and the edges of lesions, often resulting in blurred boundaries or oversegmentation, leading to discrepancies between the reconstructed model and the actual anatomical structure. Existing region growing algorithms (such as ConfidenceConnected) use globally fixed multiplier coefficients, which cannot dynamically adjust the growing threshold based on local image features (especially gradient information at edges), making them prone to leakage or undersegmentation at tissue junctions.

[0005] 2) The segmented area is not editable: After the routine segmentation is completed, if the doctor finds that a certain area is not segmented accurately, the entire segmentation operation usually needs to be repeated. It is impossible to make fine local adjustments to specific areas, resulting in low processing efficiency.

[0006] 3) Difficulty in displaying multiple tissues together: Current technology makes it difficult to accurately register and display multiple types of tissues such as bones, blood vessels, and soft tissues in the same coordinate system, which is not conducive to doctors' comprehensive understanding of anatomical relationships. Summary of the Invention

[0007] To address the above problems, this invention proposes a method and system for medical image segmentation and 3D reconstruction based on region growing.

[0008] The technical solution of this invention is: a medical image segmentation and three-dimensional reconstruction method based on region growing, comprising the following steps: S1. Obtain the original medical image data of the target object, perform preprocessing, and generate a preprocessed two-dimensional image sequence. S2. Select seed points on the preprocessed two-dimensional image sequence; S3. Based on the seed points, perform segmentation to generate segmented regions; S4. Optimize the boundaries of the segmented region to generate an optimized binary mask; S5. Perform editable processing on the optimized binary mask to generate the final binary mask; S6. Based on the final binary mask, complete the 3D model reconstruction; S7. Perform spatial registration and fusion on the reconstructed 3D model.

[0009] Furthermore, in S1, preprocessing includes noise reduction, contrast enhancement, and normalization. Noise denoising employs a bilateral filtering method. The expressions for the spatial domain kernel and the gray-level kernel of the bilateral filter are as follows: ; ; in, Spatial weights for the spatial domain kernel The grayscale weights of the grayscale kernel. For pixels coordinate, For pixels coordinate, Grayscale value For spatial scale parameters, This refers to the grayscale scale parameter; Contrast enhancement employs an adaptive histogram equalization method to enhance the contrast of raw medical image data.

[0010] Furthermore, in S2, seed points are selected automatically or manually; Automatic selection specifically involves: identifying the main peak interval, and selecting the geometric center or centroid as the seed point within the connected region that satisfies the main peak interval.

[0011] Furthermore, S3 includes the following sub-steps: S31. Add seed points to the queue to be processed, initialize the grown region as a set of seed points, and calculate the mean gray value and standard gray value of the seed point set. S32. Extract the first pixel of the queue to be processed, and traverse the neighboring pixels of the first pixel. S33. Calculate the adaptive coefficients based on the local gradients of neighboring pixels; S34, satisfy the adaptive coefficients The neighboring pixels are added to the seed point set, and the gray mean and gray standard deviation of the seed point set are updated. This represents the average gray level of the currently grown area. The gray standard deviation of the currently grown area. For the adaptive coefficients of neighboring pixels, Grayscale value; S35. Repeat S32-S34 until the queue to be processed is empty or the number of voxels in the grown region reaches the preset upper limit, and generate the segmented region.

[0012] Furthermore, in S33, the adaptive coefficient The expression is: ; in, For local gradients, As the benchmark adjustment coefficient, The gradient sensitivity coefficient is... This is the gradient reference value.

[0013] Furthermore, in S4, morphological closing operations and edge correction are used to complete boundary optimization.

[0014] Furthermore, S5 includes the following sub-steps: S51, Receive editing instructions for the optimized binary mask; S52. Based on the editing instructions, determine the three-dimensional sub-region to be edited; S53. Perform editing operations on the three-dimensional sub-region to be edited, and propagate the editing operation results of the previous layer to the adjacent layers. The editing operations specifically include: boundary adjustment, region merging, and region segmentation. S52. Perform boundary optimization on the 3D sub-region after editing to generate the final binary mask.

[0015] Furthermore, in S6, the moving cube method is used to extract isosurfaces, generate triangular meshes, and determine smoothing parameters based on the grayscale characteristics of the final binary mask to perform smoothing and complete the reconstruction of the three-dimensional model.

[0016] Based on the above methods, this invention also proposes a medical image segmentation and three-dimensional reconstruction system based on region growing, including an image preprocessing module, a seed point selection module, a region growing segmentation module, a boundary optimization module, an editable processing module, a three-dimensional reconstruction module, and a multi-tissue fusion display module; The image preprocessing module is used to acquire the raw medical image data of the target object, perform preprocessing, and generate a preprocessed two-dimensional image sequence. The seed point selection module is used to select seed points on the preprocessed two-dimensional image sequence; The region growing and segmentation module is used to generate segmented regions; The boundary optimization module is used to optimize the boundaries of the segmented region and generate an optimized binary mask; The editable processing module is used to perform editable processing on the optimized binary mask to generate the final binary mask; The 3D reconstruction module is used to complete the reconstruction of 3D models; The multi-organism fusion display module is used for spatial registration and fusion of the reconstructed 3D model.

[0017] The beneficial effects of this invention are: (1) Significantly improved segmentation accuracy: This invention introduces an adaptive region growth algorithm, which dynamically adjusts the growth threshold based on local grayscale features. Compared with traditional fixed threshold segmentation methods, the segmentation accuracy in complex areas such as soft tissue-bone junctions and lesion edges is significantly improved. Clinical tests show that the average error between the reconstructed 3D model and the real anatomical structure can be controlled within 0.5 mm, meeting the requirements for high-precision surgical planning.

[0018] (2) High editability of segmented regions: The region editing module of this invention allows doctors to make fine adjustments to specific segmented regions without affecting other regions. When a doctor finds that the segmentation of a certain level is inaccurate, the boundary can be corrected directly at that level, and the system will automatically propagate the correction to adjacent levels, avoiding the repetition of all segmentation operations. Clinical simulation tests show that compared with the traditional method that requires complete re-segmentation, the editing operation efficiency of this invention is improved by more than 50%.

[0019] (3) Multi-tissue joint display and interaction: The multi-tissue fusion display module of the present invention realizes the synchronous segmentation and joint three-dimensional display of bones, blood vessels, soft tissues and lesions, allowing doctors to clearly observe the spatial adjacency between different tissues. Combined with interactive functions such as dissection and transparency, it helps to understand complex anatomical structures more comprehensively and provides intuitive basis for precise tumor localization and surgical approach planning.

[0020] (4) Standardized operation process: The semi-automated segmentation process of the present invention reduces the subjective differences of manual delineation, and makes the consistency of reconstruction results of the same image among different operators reach more than 90%, providing a reproducible technical basis for multi-center studies and efficacy evaluation. Attached Figure Description

[0021] Figure 1 This is a flowchart of a region-growing-based medical image segmentation and 3D reconstruction method. Figure 2 This is a schematic diagram of a region-growing-based medical image segmentation and 3D reconstruction system.

[0022] Figure 3Adaptive coefficients With local gradient The functional relationship graph; Figure 4 For traditional fixed coefficient Adaptive coefficients of the present invention A comparative diagram showing the segmentation effect at the junction of complex tissues. Detailed Implementation

[0023] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0024] like Figure 1 As shown, this invention provides a method for medical image segmentation and 3D reconstruction based on region growing, comprising the following steps: S1. Obtain the original medical image data of the target object, perform preprocessing, and generate a preprocessed two-dimensional image sequence. S2. Select seed points on the preprocessed two-dimensional image sequence; S3. Based on the seed points, perform segmentation to generate segmented regions; S4. Optimize the boundaries of the segmented region to generate an optimized binary mask; S5. Perform editable processing on the optimized binary mask to generate the final binary mask; S6. Based on the final binary mask, complete the 3D model reconstruction; S7. Perform spatial registration and fusion on the reconstructed 3D model.

[0025] This invention aims to solve the following technical problems: 1) The primary issue: How to improve the accuracy of medical image segmentation, especially the segmentation accuracy in complex areas such as the junction of soft tissue and bone and the edge of lesions.

[0026] 2) Secondary issue: How to achieve local editability of segmented regions, allowing doctors to make fine adjustments to specific regions while preserving the segmentation results of other regions.

[0027] 3) The third question: How to achieve simultaneous segmentation and joint three-dimensional display of multiple tissue structures to improve doctors' cognitive efficiency of spatial relationships of anatomical structures.

[0028] The main objective of this invention is to provide a three-dimensional reconstruction method that can significantly improve segmentation accuracy and reconstruction quality, while also possessing region editability, thereby providing more reliable imaging evidence for precise clinical diagnosis and treatment.

[0029] In this embodiment of the invention, S1 includes preprocessing, including noise reduction, contrast enhancement, and normalization. Noise denoising employs a bilateral filtering method. The expressions for the spatial domain kernel and the gray-level kernel of the bilateral filter are as follows: ; ; in, Spatial weights for the spatial domain kernel The grayscale weights of the grayscale kernel. For pixels coordinate, For pixels coordinate, Grayscale value For spatial scale parameters, This refers to the grayscale scale parameter; Contrast enhancement employs an adaptive histogram equalization method to enhance the contrast of raw medical image data.

[0030] The process involves acquiring raw medical image data (including CT or MRI) of the target object, denoising, normalizing, and grayscale stretching the images to generate a preprocessed two-dimensional image sequence. Specifically, a bilateral filtering algorithm is used to remove noise while preserving edge information, and adaptive histogram equalization is employed to enhance image contrast.

[0031] Normalization: Linearly maps the grayscale values ​​of an image to a standard range (such as 0-255 or 0-4095) to facilitate subsequent unified processing.

[0032] In this embodiment of the invention, in S2, seed points are selected automatically or manually; Automatic selection specifically involves: identifying the main peak interval, and selecting the geometric center or centroid as the seed point within the connected region that satisfies the main peak interval.

[0033] On the preprocessed 2D image, initial seed points are automatically or manually selected for the target tissue (bone, blood vessel, soft tissue, or lesion) to be segmented. The automatic selection method uses a clustering method (such as K-means) to pre-classify the image and select representative points of the target category as seeds.

[0034] Manual selection allows doctors to click to select seed points at key levels (such as the most prominent lesion layer or the main vascular layer), and supports multiple seed points to cover multi-connected regions.

[0035] In this embodiment of the invention, S3 includes the following sub-steps: S31. Add seed points to the queue to be processed, initialize the grown region as a set of seed points, and calculate the mean gray value and standard gray value of the seed point set. S32. Extract the first pixel of the queue to be processed, and traverse the neighboring pixels of the first pixel. S33. Calculate the adaptive coefficients based on the local gradients of neighboring pixels; S34, satisfy the adaptive coefficients The neighboring pixels are added to the seed point set, and the gray mean and gray standard deviation of the seed point set are updated. This represents the average gray level of the currently grown area. The gray standard deviation of the currently grown area. For the adaptive coefficients of neighboring pixels, Grayscale value; S35. Repeat S32-S34 until the queue to be processed is empty or the number of voxels in the grown region reaches the preset upper limit, and generate the segmented region.

[0036] Based on the gray-level statistical features and gradient information of the local region, the similarity threshold between the current candidate pixel and the seed region is dynamically calculated.

[0037] threshold and The calculation formula is as follows: ; ; in: This represents the average gray level of the currently grown area. The gray standard deviation of the currently grown area. The adaptive adjustment coefficient is dynamically calculated based on the magnitude of the local gradient at the candidate pixel.

[0038] Gradient calculation: Taking 3D images as an example, candidate pixels gradient magnitude at The following formula can be used for calculation: ; The partial derivatives can be approximated by the central difference or by the difference after convolution with a Gaussian kernel.

[0039] Technical effect: When When the area is large (edge ​​region), Decrease, growth criteria The stricter regulations effectively suppress cross-boundary growth; when When the area is small (uniform), Approaching It allows for relatively loose growth, ensuring complete coverage of the target area.

[0040] The adaptive adjustment coefficient is dynamically adjusted based on the magnitude of the gradient in the local region; the larger the gradient (i.e., the edge region), the better. The smaller the value, the stricter the growth criteria, preventing excessive growth from crossing tissue boundaries.

[0041] Optional Functional form (as a parallel technical solution): Exponential form: The attenuation coefficient; Piecewise linear form: when hour ;when hour ,and .

[0042] In this embodiment of the invention, in S33, the adaptive coefficient The expression is: ; in, For local gradients, As the benchmark adjustment coefficient, The gradient sensitivity coefficient is... This is the gradient reference value.

[0043] User-configurable, typical value 2.0 to 3.5, corresponding to the relaxation threshold in regions with zero or uniform gradient; Typical values ​​are 0.01–0.5, controlling the gradient pair. The intensity of the impact; The Sobel operator, central difference, or Gaussian gradient can be used for calculation. Used for normalization.

[0044] In this embodiment of the invention, in S4, morphological closing operation and edge correction are used to complete boundary optimization.

[0045] Morphological closing operation: Using structuring elements (such as spheres or cubes with a radius of 1 to 3 pixels) to perform a closing operation on a binary mask to fill small holes.

[0046] Edge correction: Canny edge detection or gradient magnitude thresholding is used to extract real edges on the original grayscale image, and the segmentation boundary is moved closer to the detected edge to correct jagged edges.

[0047] In this embodiment of the invention, S5 includes the following sub-steps: S51, Receive editing instructions for the optimized binary mask; S52. Based on the editing instructions, determine the three-dimensional sub-region to be edited; S53. Perform editing operations on the three-dimensional sub-region to be edited, and propagate the editing operation results of the previous layer to the adjacent layers. The editing operations specifically include: boundary adjustment, region merging, and region segmentation. S52. Perform boundary optimization on the 3D sub-region after editing to generate the final binary mask.

[0048] Local editing range determination: Based on the area selected, clicked or drawn by the user, determine the three-dimensional sub-region to be edited (such as a columnar or spherical region based on a certain layer).

[0049] Within the sub-region to be edited, users are allowed to perform (1) boundary adjustment: manually add or erase; (2) region merging: merge adjacent sub-regions of the same organization; (3) region splitting: split the current sub-region into multiple sub-regions.

[0050] Edit propagation: When a user modifies the boundary at a certain level, the system propagates the correction result to adjacent levels based on the continuity between levels (such as through interpolation or contour matching), without the need for re-segmenting the entire map.

[0051] Non-editable region protection: Editing operations only apply to the local region specified by the user, and the segmentation results of other regions remain unchanged.

[0052] In this embodiment of the invention, in step S6, the isosurface is extracted using the moving cube method to generate a triangular mesh, and the smoothing parameters are determined based on the grayscale features of the final binary mask to perform smoothing and complete the reconstruction of the three-dimensional model.

[0053] Linked smoothing parameters: Different smoothing parameters are set according to the grayscale characteristics of the segmented region -- a smaller smoothing intensity is used for high gradient edge regions (such as bone-soft tissue junction) to maintain sharpness, and a larger smoothing intensity is used for uniform regions to eliminate step artifacts.

[0054] Smoothing can be achieved through Laplace filtering or Taubin filtering. The number of smoothing iterations or the intensity of smoothing are inversely proportional to the local gray-level variance or gradient magnitude.

[0055] Based on the above methods, such as Figure 2 As shown, the present invention also proposes a medical image segmentation and three-dimensional reconstruction system based on region growing, including an image preprocessing module, a seed point selection module, a region growing segmentation module, a boundary optimization module, an editable processing module, a three-dimensional reconstruction module, and a multi-tissue fusion display module; The image preprocessing module is used to acquire the raw medical image data of the target object, perform preprocessing, and generate a preprocessed two-dimensional image sequence. The seed point selection module is used to select seed points on the preprocessed two-dimensional image sequence; The region growing and segmentation module is used to generate segmented regions; The boundary optimization module is used to optimize the boundaries of the segmented region and generate an optimized binary mask; The editable processing module is used to perform editable processing on the optimized binary mask to generate the final binary mask; The 3D reconstruction module is used to complete the reconstruction of 3D models; The multi-organism fusion display module is used for spatial registration and fusion of the reconstructed 3D model.

[0056] The generated 3D models of various tissues are spatially registered and fused for display in the same coordinate system.

[0057] Specific implementation: All tissue models share the same image coordinate system, requiring no additional registration; Supports setting different colors, transparency, and rendering modes (surface / volume drawing) for different organizations; It provides interactive operations such as rotation, scaling, sectioning (sectioning in any plane), and transparency switching; The cutting planes can be updated in real time, and multiple cutting planes can be displayed simultaneously.

[0058] like Figure 3 As shown, these are the adaptive coefficients. With local gradient The functional relationship.

[0059] like Figure 4 As shown, the adaptive coefficient region growing method of the present invention has a significantly better segmentation effect at the junction of complex tissues compared with the traditional fixed coefficient region growing method: Figure 4 The image on the left is the original medical image, which shows the bones (high grayscale area), soft tissue (medium grayscale area), and the anatomical structure at the junction of the two. The arrow points to a typical tissue junction area.

[0060] Figure 4 The upper right side shows the segmentation results using the traditional fixed-coefficient region growing method. It can be seen that in homogeneous tissue regions (such as inside bone or soft tissue), the traditional method can achieve relatively complete segmentation; however, at tissue boundaries (indicated by the arrow), the traditional method exhibits a significant "overgrowth" phenomenon—the segmented region crosses the real tissue boundary and infiltrates adjacent tissue, resulting in a jagged or spiky outline at the segmentation boundary, leading to a decrease in segmentation accuracy. This is because the traditional method uses a globally fixed multiplier coefficient (α is a constant), which cannot dynamically adjust the growth threshold in edge regions, making it prone to leakage when local grayscale differences are small.

[0061] Figure 4The lower right side shows the segmentation results using the adaptive coefficient region growth method of this invention. It can be seen that the method of this invention achieves precise boundary locking at tissue boundaries. Because the adaptive coefficient α automatically decreases in the edge region (when the local gradient G is large) according to the formula α = α0·G_ref / (G_ref + k·G), the growth criterion... The stricter gradient effectively inhibits cross-boundary growth. The segmentation boundary basically matches the real tissue boundary, with smooth and complete edges, and no obvious leakage or undersegmentation. Within the homogeneous tissue (when the local gradient G is small), α approaches α0, maintaining relaxed growth conditions and ensuring complete coverage of the target area.

[0062] Combination Figure 3 The curve showing the functional relationship between the adaptive coefficient α and the local gradient G indicates that when G is large (tissue edge), α decreases rapidly, and the growth threshold tightens; when G is small (uniform region), α approaches α0, and the growth threshold loosens. This is the key mechanism by which the segmentation accuracy of this invention is superior to traditional methods.

[0063] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for medical image segmentation and 3D reconstruction based on region growing, characterized in that, Includes the following steps: S1. Obtain the original medical image data of the target object, perform preprocessing, and generate a preprocessed two-dimensional image sequence. S2. Select seed points on the preprocessed two-dimensional image sequence; S3. Based on the seed points, perform segmentation to generate segmented regions; S4. Optimize the boundaries of the segmented region to generate an optimized binary mask; S5. Perform editable processing on the optimized binary mask to generate the final binary mask; S6. Based on the final binary mask, complete the 3D model reconstruction; S7. Perform spatial registration and fusion on the reconstructed 3D model.

2. The medical image segmentation and three-dimensional reconstruction method based on region growing according to claim 1, characterized in that, In S1, the preprocessing includes noise reduction, contrast enhancement, and normalization. The noise reduction employs a bilateral filtering method, and the expressions for the spatial domain kernel and the gray-level kernel of the bilateral filter are as follows: ； ； in, Spatial weights for the spatial domain kernel The grayscale weights of the grayscale kernel. For pixels coordinate, For pixels coordinate, Grayscale value For spatial scale parameters, This refers to the grayscale scale parameter; The contrast enhancement employs an adaptive histogram equalization method to enhance the contrast of the original medical image data.

3. The medical image segmentation and 3D reconstruction method based on region growing according to claim 1, characterized in that, In S2, seed points are selected automatically or manually. The automatic selection specifically involves: identifying the main peak interval, and selecting the geometric center or centroid as a seed point within the connected region that satisfies the main peak interval.

4. The medical image segmentation and three-dimensional reconstruction method based on region growing according to claim 1, characterized in that, S3 includes the following sub-steps: S31. Add seed points to the queue to be processed, initialize the grown region as a set of seed points, and calculate the mean gray value and standard gray value of the seed point set. S32. Extract the first pixel of the queue to be processed, and traverse the neighboring pixels of the first pixel. S33. Calculate the adaptive coefficients based on the local gradients of neighboring pixels; S34, satisfy the adaptive coefficients The neighboring pixels are added to the seed point set, and the gray mean and gray standard deviation of the seed point set are updated. This represents the average gray level of the currently grown area. The gray standard deviation of the currently grown area. For the adaptive coefficients of neighboring pixels, Grayscale value; S35. Repeat S32-S34 until the queue to be processed is empty or the number of voxels in the grown region reaches the preset upper limit, and generate the segmented region.

5. The medical image segmentation and three-dimensional reconstruction method based on region growing according to claim 4, characterized in that, In S33, the adaptive coefficient The expression is: ； in, For local gradients, As the benchmark adjustment coefficient, This is the gradient sensitivity coefficient. This is the gradient reference value.

6. The medical image segmentation and three-dimensional reconstruction method based on region growing according to claim 1, characterized in that, In S4, morphological closing operations and edge correction are used to complete boundary optimization.

7. The medical image segmentation and three-dimensional reconstruction method based on region growing according to claim 1, characterized in that, S5 includes the following sub-steps: S51, Receive editing instructions for the optimized binary mask; S52. Based on the editing instructions, determine the three-dimensional sub-region to be edited; S53. Perform editing operations on the three-dimensional sub-region to be edited, and propagate the editing operation results of the previous layer to the adjacent layers. The editing operations specifically include: boundary adjustment, region merging, and region segmentation. S52. Perform boundary optimization on the 3D sub-region after editing to generate the final binary mask.

8. The medical image segmentation and three-dimensional reconstruction method based on region growing according to claim 1, characterized in that, In step S6, the isosurface is extracted using the moving cube method to generate a triangular mesh, and the smoothing parameters are determined based on the grayscale features of the final binary mask to perform smoothing and complete the reconstruction of the three-dimensional model.

9. A medical image segmentation and 3D reconstruction system based on region growing, characterized in that, It includes an image preprocessing module, a seed point selection module, a region growing and segmentation module, a boundary optimization module, an editable processing module, a 3D reconstruction module, and a multi-tissue fusion display module; The image preprocessing module is used to acquire the original medical image data of the target object, perform preprocessing, and generate a preprocessed two-dimensional image sequence. The seed point selection module is used to select seed points on the preprocessed two-dimensional image sequence; The region growth and segmentation module is used to generate segmented regions; The boundary optimization module is used to optimize the boundaries of the segmented region and generate an optimized binary mask. The editable processing module is used to perform editable processing on the optimized binary mask to generate the final binary mask; The 3D reconstruction module is used to complete the 3D model reconstruction. The multi-organism fusion display module is used to perform spatial registration and fusion of the reconstructed 3D model.

Images