CT liver segmentation and three-dimensional reconstruction method based on image enhancement

Abstract

The invention relates to the technical field of medical image processing and computer-aided diagnosis, in particular to a CT liver segmentation and three-dimensional reconstruction method combined with image enhancement. The method comprises the following steps: 1) establishing a plain-scan CT database, and collecting enough human body plain-scan CT image data; 2) screening and cleaning data in the plain-scan CT database; 3) performing image enhancement on CT plain scanning data according to a region of interest (ROI); 4) performing semantic segmentation on the enhanced image, and automatically labeling the liver in the plain-scan CT image by a model; and 5) performing three-dimensional reconstruction and smoothing processing on the segmented result image data to generate a simulated three-dimensional model, which is helpful for a doctor to more clearly master the relative position of the human anatomical structure, improves the safety and treatment effect of an operation, and facilitates the communication between the doctor and a patient at the same time.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical image processing and three-dimensional reconstruction, and specifically relates to a CT liver segmentation and three-dimensional reconstruction method based on image enhancement, which is widely used in scenarios such as preoperative planning, intraoperative navigation and liver disease diagnosis. Background Art

[0002] With the development of medical imaging technology, CT (computed tomography) has become an important tool for the diagnosis and treatment of liver diseases. However, due to the low contrast between the liver and surrounding tissues (such as blood vessels, fat, etc.) in CT images, traditional liver segmentation methods have certain limitations in accuracy and robustness. Existing technologies mostly rely on manual annotation or semi-automatic segmentation, which is time-consuming and easily affected by subjective factors, and it is difficult to meet clinical needs. In addition, three-dimensional reconstruction technology is particularly important in applications such as surgical planning and liver volume measurement, but accurate segmentation of the liver is a prerequisite for obtaining a high-quality three-dimensional model. To solve these problems, there is an urgent need for a CT liver segmentation method combined with image enhancement to improve segmentation accuracy and achieve efficient three-dimensional reconstruction to better assist clinical diagnosis and treatment. Summary of the invention

[0003] In response to the challenges mentioned in the background technology section, the present invention provides an innovative method, a CT liver segmentation and three-dimensional reconstruction method based on image enhancement. This method improves the recognition of organs such as the liver by performing image enhancement processing on plain scan CT images, thereby significantly improving the accuracy of organ segmentation and ensuring that the generated three-dimensional reconstruction model is closer to the real organ. This not only helps to accurately locate and evaluate lesions, but also provides a reliable basis for preoperative planning and treatment. At the same time, the enhanced three-dimensional visualization model also makes it easier for patients to better understand their own condition, and has strong practicality and broad application prospects.

[0004] The CT liver segmentation and three-dimensional reconstruction method based on image enhancement of the present invention adopts the following technical solutions:

[0005] The CT liver segmentation and three-dimensional reconstruction method based on image enhancement includes the following steps:

[0006] 1) Establish a plain scan CT database and collect a sufficient amount of human plain scan CT image data;

[0007] 2) Screening and cleaning of data in the plain scan CT database;

[0008] 3) Image enhancement of CT plain scan data based on the region of interest (ROI);

[0009] 4) Using a semantic segmentation model on the enhanced image to automatically annotate the liver in the plain scan CT image;

[0010] 5) Perform three-dimensional reconstruction and smoothing on the image data resulting from the segmentation process to generate a simulated three-dimensional model.

[0011] Preferably, in step 3), image enhancement is performed on the CT scan data according to the region of interest (ROI), including an image normalization operation, using maximum and minimum normalization, namely:

[0012]

[0013] Preferably, in step 3), image enhancement is performed on the CT scan data according to the region of interest (ROI), and the pixels of the high-intensity bone structure are suppressed using an image threshold. The suppression formula is

[0014] Right now:

[0015]

[0016] Preferably, in step 3), image enhancement is performed on the CT plain scan data according to the region of interest (ROI), including an operation of adaptive histogram equalization (CLAHE) contrast enhancement, namely:

[0017]

[0018] Preferably, the image enhancement of the CT plain scan data according to the region of interest (ROI) in step 3) includes the operation of adjusting the window width and window level of the CT plain scan data, that is:

[0019]

[0020] Preferably, in step 3), image enhancement is performed on the CT plain scan data according to the region of interest (ROI), including performing a Gaussian mixture model enhancement image contrast operation on the image, namely:

[0021]

[0022] Preferably, in step 3), image enhancement is performed on the CT scan data according to the region of interest (ROI), and Gaussian smoothing is performed on the image, that is:

[0023]

[0024] Preferably, in step 4), a semantic segmentation model automatically marks the liver in the enhanced image, and a loss function combining Dice loss and cross-entropy loss (CE) is used to optimize the segmentation effect of the liver. The formula of the combined loss function can be expressed as:

[0025]

[0026] Substitute (21)(22) into (23), and (0<α<1)

[0027] L=α·L CE +(1-α)·L Dice (twenty three);

[0028] Preferably, in step 4), a semantic segmentation model for automatically annotating the liver in the enhanced image uses a Leaky ReLU activation function instead of a traditional ReLU activation function. The model uses the Leaky ReLU formula to optimize the segmentation effect of the liver, and the formula is expressed as:

[0029]

[0030] Preferably, in step 4), a semantic segmentation model for automatically identifying the liver in the enhanced image uses instance normalization instead of the widely used batch normalization as its normalization method. The formula of Instance Normalization is:

[0031]

[0032] Preferably, in step 5), for isosurface extraction, the Marching Cubes algorithm is used to extract isosurfaces from floating-point image data, and the input three-dimensional scalar field data is used to generate corresponding polygonal meshes. The Marching Cubes algorithm calculates the intersection of the isosurface and the cube edge by linear interpolation, thereby determining the shape of the three-dimensional isosurface. The linear interpolation formula is as follows:

[0033]

[0034] Preferably, in step 5), the extracted polygonal mesh is processed by a smoothing filter (Smooth PolyDataFilter). The filter uses a Laplacian smoothing algorithm to iteratively adjust the position of each vertex, reduce noise and irregularities, and improve the smoothness and accuracy of the mesh. The vertex position update formula of Laplacian smoothing is as follows:

[0035]

[0036] Preferably, in step 5), in order to ensure the lighting effect and normal consistency of the polygonal mesh, the normal vector of each vertex is calculated, and the direction of the normal vector is automatically adjusted (AutoOrientNormals). The calculation formula of the normal vector is as follows:

[0037]

[0038] The beneficial effects of the present invention are:

[0039] 1. Improve the accuracy of liver and surrounding tissue segmentation and reduce errors: By performing image enhancement processing on CT images, the present invention can highlight the boundary features of the liver and surrounding tissues and significantly improve the accuracy of the segmentation algorithm. Compared with traditional methods, it reduces the segmentation error caused by low tissue contrast, which is especially suitable for cases with complex liver morphology or lesions, laying a reliable foundation for subsequent three-dimensional reconstruction.

[0040] 2. It has significant teaching significance and improves the effect of medical education: With the help of the generated high-precision three-dimensional liver model, the present invention provides a powerful tool for medical teaching. Students and clinicians can intuitively understand the anatomical structure of the liver and its pathological changes through three-dimensional visualization. Compared with traditional 2D images, the 3D model shows more comprehensive anatomical details, which helps medical students master complex anatomical knowledge and provides doctors with teaching scenarios for simulated surgery and case discussions, thereby improving the effect of medical education and training.

[0041] 3. It helps to accurately locate the lesion and supports preoperative planning and treatment decision-making: With the help of enhanced image data for segmentation, the present invention can generate a more accurate three-dimensional liver model. This technology plays an important role in preoperative planning, assisting doctors in designing the best surgical path and plan, reducing surgical risks, and improving treatment effects, especially in liver tumor surgery.

[0042] 4. Improve patients’ understanding of their condition and enhance doctor-patient communication: Through the generated high-quality 3D model, doctors can more intuitively show patients the condition of the liver and lesions, helping them to clearly understand their condition. This visual display method improves the efficiency of doctor-patient communication, enhances patients’ sense of trust, helps patients better participate in treatment decisions, and relieves anxiety about surgery and treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings involved in the embodiments or the prior art description are briefly introduced. It is obvious that the drawings in the following description are only some embodiments of the present invention, and ordinary technicians in this field can still obtain other drawings based on these drawings without creative work.

[0044] Figure 1 It is a flow chart of the CT liver segmentation and three-dimensional reconstruction method based on image enhancement of the present invention;

[0045] Figure 2 The plain scan CT original image of the present invention;

[0046] Figure 3 It is a plain scan CT image after image enhancement of the present invention;

[0047] Figure 4 It is the plain scan CT liver semantic segmentation map of the present invention;

[0048] Figure 5 A schematic diagram of liver modeling according to the present invention; DETAILED DESCRIPTION

[0049] The following will provide a clear and complete description of the technical solution of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described are only a part of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments that can be obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention. An embodiment of the CT liver segmentation and three-dimensional reconstruction method based on image enhancement of the present invention, such as Figure 1 As shown, including:

[0050] See also Figure 1 The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement shown in the figure comprises the following steps:

[0051] 1) Establish a plain scan CT database and collect a sufficient amount of human plain scan CT image data;

[0052] 2) Screening and cleaning of data in the plain scan CT database;

[0053] 3) Image enhancement of CT plain scan data based on the region of interest (ROI);

[0054] 4) Using a semantic segmentation model on the enhanced image to automatically annotate the liver in the plain scan CT image;

[0055] 5) Perform three-dimensional reconstruction and smoothing on the image data resulting from the segmentation process to generate a simulated three-dimensional model.

[0056] Furthermore, in step 3), the CT scan data is subjected to image normalization according to the region of interest (ROI), and the maximum and minimum normalization is adopted, that is:

[0057]

[0058] Among them I norm (x, y) is the normalized intensity value after standardization, I(x, y) is the original intensity value of the pixel with coordinates (x, y) in the CT image, and I min is the minimum intensity value in the image, I max is the maximum intensity value in the image.

[0059] Furthermore, in step 3), the CT plain scan data is thresholded to suppress the pixels of high-intensity bone structures according to the region of interest (ROI), and the suppression formula is:

[0060]

[0061] Among them I suppressed (x, y) is the intensity value after suppression, I(x, y) is the original intensity value of the pixel with coordinates (x, y) in the CT image, and T is the threshold for distinguishing between bones (high intensity) and soft tissues (low intensity).

[0062] Furthermore, in step 3), the CT plain scan data is subjected to an adaptive histogram equalization (CLAHE) operation according to the region of interest (ROI), namely:

[0063]

[0064] Where I is the original image, I i,j To divide the original image into several non-overlapping sub-images, i and j represent the row and column indices of the sub-image, m×n is set to the size of the original image, and a×b is set to the size of each sub-image.

[0065]

[0066] Among them, H i,j (k) is sub-graph I i,j Sub-grayscale histogram, k represents the grayscale level (0-255), δ is the Kroneckerdelta function, when I(x,y)=k, δ=1, otherwise δ=0.

[0067]

[0068] ClipLimit is the set contrast limit threshold, and the part exceeding ClipLimit is evenly distributed to other gray levels.

[0069]

[0070] in is the enhanced grayscale value, which is obtained by looking up the table.

[0071]

[0072] Among them I enhanced is the enhanced original image combined from all sub-images, For sub-graph.

[0073] Further, in step 3), the window width and window level of the CT scan data are adjusted according to the region of interest (ROI), that is:

[0074]

[0075] Among them, Display Value is the displayed grayscale value, which usually ranges from 0 to 255, CT Value is the original CT value of each pixel in the image, Window Level (WL) is the window level, that is, the center point of the window width, Window Width (WW) is the window width, that is, the CT value range around the window level, Scale Factor and Offset are used to ensure that all grayscale values ​​are in the range of 0 to 255. The scaling factor and offset.

[0076] Further, in step 3), a Gaussian mixture model operation is performed on the CT scan data according to the region of interest (ROI), namely:

[0077]

[0078] Where G is the gradient magnitude of the pixel and θ is the direction of each pixel.

[0079]

[0080] where π k is the mixing coefficient of the kth Gaussian component, N(x i |μ k ,∑k) is the probability density function of the kth Gaussian component.

[0081]

[0082] where μ k is the mean of the kth Gaussian component, ∑k is the covariance matrix, γ ik is the probability of the kth component and N is the total number of data points.

[0083] Further, in step 3), Gaussian smoothing operation is performed on the CT scan data according to the region of interest (ROI), that is:

[0084]

[0085] Among them, the Gaussian kernel G(x,y) is a two-dimensional normal distribution function, σ is the standard deviation, and (x,y) is the coordinate of the kernel.

[0086]

[0087] Among them, i, j are the indices of the Gaussian kernel matrix, and k is the center index of the kernel (i.e. ).

[0088]

[0089] Among them, Ismooth (x,y) is the smoothed image, G(i,j) is the weight of the Gaussian kernel, I(xi,yj) is the pixel value of the original image, and k is the size of the kernel.

[0090] Furthermore, in step 4), a semantic segmentation model automatically marks the liver in the enhanced image, and a loss function combining Dice loss and cross-entropy loss (CE) is used to optimize the segmentation effect of the liver. The formula of the combined loss function can be expressed as:

[0091]

[0092] Among them, X is the predicted segmentation result, Y is the actual segmentation result, |X∩Y| is the size of the intersection of the predicted result and the actual result, and |X| and |Y| are the total areas of the predicted result and the actual result, respectively.

[0093]

[0094] Where M is the number of categories, y o,c is the one-hot encoding of the true label, p o,c is the probability that the model predicts class c.

[0095] Substitute (21)(22) into (23)

[0096] L=α·L CE +(1-α)·L Dice (twenty three);

[0097] Among them, L is the total loss, L CE is the cross entropy loss, which measures the difference between the model’s predicted probability distribution and the true label, L Dice is the Dice loss, which measures the similarity between the predicted segmentation result and the true segmentation result. α is a weight parameter used to balance the two losses, usually between 0 and 1.

[0098] Furthermore, in step 4), the semantic segmentation model uses LeakyReLU as the activation function to replace the traditional ReLU activation function. The formula of LeakyReLU is:

[0099]

[0100] Among them, x is the input value, α is a small positive number (usually between 0.01 and 0.1) used to control the gradient of negative input. When x>0, the function is the same as ReLU and outputs x; when x≤0, the function outputs αx, so that even if the input is negative, the neuron can still receive a certain gradient signal, thus avoiding complete inactivation.

[0101] Furthermore, in step 4), a semantic segmentation model for automatically identifying the liver in the enhanced image uses instance normalization instead of the widely used batch normalization as its normalization method. The formula of Instance Normalization is:

[0102]

[0103] Where x is a single sample of the input feature map, μ is the mean of x, and σ 2 is the standard deviation of x, and ε is a small constant used to prevent division by zero.

[0104] Further, in step 5), for isosurface extraction, the Marching Cubes algorithm is used to extract isosurfaces from floating-point image data, and the input three-dimensional scalar field data is converted into a corresponding polygonal mesh. The Marching Cubes algorithm calculates the intersection of the isosurface and the cube edge by linear interpolation, thereby determining the shape of the three-dimensional isosurface. The linear interpolation formula is as follows:

[0105]

[0106] Among them, p1 and p2 are the two vertices of the cube edge, f1 and f2 are their scalar values, and C is the set isosurface threshold.

[0107] Further, in step 5), the extracted polygon mesh is processed by a smoothing filter (SmoothPolyDataFilter). The filter uses the Laplacian Smoothing algorithm to iteratively adjust the position of each vertex, reduce noise and irregularities, and improve the smoothness and accuracy of the mesh. The vertex position update formula of Laplacian smoothing is as follows:

[0108]

[0109] Among them, p i is the current vertex, N(i) is the set of vertices adjacent to it, and λ is the relaxation factor.

[0110] Furthermore, in step 5), in order to ensure the lighting effect and normal consistency of the polygonal mesh, the normal vector of each vertex is calculated, and the direction of the normal vector is automatically adjusted (AutoOrientNormals). The calculation formula of the normal vector is as follows:

[0111]

[0112] Among them, v1, v2, and v3 are the three vertices that constitute the triangular surface.

[0113] After executing the above process on the plain scan image, the following Figure 2-5 effect.

[0114] Advantages of the present invention:

[0115] 1) High segmentation accuracy: Through image enhancement processing, the contrast between the liver and surrounding tissues is improved, which significantly improves the accuracy of liver segmentation and reduces segmentation errors.

[0116] 2) 3D model authenticity: The generated 3D reconstructed model is closer to the real organ, can retain detailed features, and helps to accurately display the organ structure and lesion location.

[0117] 3) Preoperative planning support: Three-dimensional reconstruction helps to accurately locate lesions, assist doctors in designing the best surgical plan, reduce surgical risks, and improve treatment outcomes.

[0118] 4) Enhance doctor-patient communication: Through intuitive 3D model display, enhance patients’ understanding of their condition, improve the efficiency of doctor-patient communication, and enhance patients’ confidence in participating in treatment.

[0119] 5) High teaching value: The 3D reconstructed model provides an intuitive learning tool for medical education, helping students and young doctors better understand the complex anatomical structure and pathological characteristics of the liver.

[0120] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope defined in the claims.

Claims

1. A CT liver segmentation and three-dimensional reconstruction method based on image enhancement, characterized in that: The following steps are involved: 1) Establish a plain scan CT database and collect a sufficient amount of human plain scan CT image data; 2) Screening and cleaning of data in the plain scan CT database; 3) Image enhancement of CT plain scan data based on the region of interest (ROI); 4) Using a semantic segmentation model on the enhanced image to automatically annotate the liver in the plain scan CT image; 5) Perform three-dimensional reconstruction and smoothing on the image data resulting from the segmentation process to generate a simulated three-dimensional model.

2. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 3), image enhancement is performed on the CT scan data according to the region of interest (ROI), including image normalization operation, namely: Among them I norm (x, y) is the normalized intensity value after standardization, I(x, y) is the original intensity value of the pixel with coordinates (x, y) in the CT image, and I min is the minimum intensity value in the image, I max is the maximum intensity value in the image.

3. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 3), the CT scan data is enhanced according to the region of interest (ROI), and the high-intensity bone structure is suppressed using the image threshold, that is: Among them I suppressed (x, y) is the intensity value after suppression, I(x, y) is the original intensity value of the pixel with coordinates (x, y) in the CT image, and T is the threshold for distinguishing between bones (high intensity) and soft tissues (low intensity).

4. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 3), image enhancement is performed on the CT scan data according to the region of interest (ROI), including the operation of adaptive histogram equalization (CLAHE) contrast enhancement, namely: Where I is the original image, I i,j To divide the original image into several non-overlapping sub-images, i and j represent the row and column indices of the sub-image, m×n is set to the size of the original image, and a×b is set to the size of each sub-image. Among them, H i,j (k) is sub-graph I i,j Sub-grayscale histogram, k represents the grayscale level (0-255), δ is the Kronecker delta function, when I(x, y) = k, δ = 1, otherwise δ = 0. ClipLimit is the set contrast limit threshold, and the part exceeding ClipLimit is evenly distributed to other gray levels. in is the enhanced grayscale value, which is obtained by looking up the table. Among them I enhanced is the enhanced original image combined from all sub-images, For sub-graph.

5. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 3), image enhancement is performed on the CT scan data according to the region of interest (ROI), including the operation of adjusting the window width and window level of the CT scan data, namely: Among them, Display Value is the displayed grayscale value, which usually ranges from 0 to 255, CT Value is the original CT value of each pixel in the image, Window Level (WL) is the window level, that is, the center point of the window width, Window Width (WW) is the window width, that is, the CT value range around the window level, Scale Factor and Offset are used to ensure that all grayscale values ​​are in the range of 0 to 255. The scaling factor and offset.

6. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 3), the CT scan data is enhanced according to the region of interest (ROI), including performing a Gaussian mixture model to enhance the image contrast, namely: Where G is the gradient magnitude of the pixel and θ is the direction of each pixel. where π k is the mixing coefficient of the kth Gaussian component, N(x i |μ k ,∑k) is the probability density function of the kth Gaussian component. where μ k is the mean of the kth Gaussian component, Σk is the covariance matrix, γ ik is the probability of the kth component and N is the total number of data points.

7. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 3), the CT scan data is enhanced according to the region of interest (ROI), and the image is subjected to Gaussian smoothing operation, namely: Among them, the Gaussian kernel G(x,y) is a two-dimensional normal distribution function, σ is the standard deviation, and (x,y) is the coordinate of the kernel. Among them, i, j are the indices of the Gaussian kernel matrix, and k is the center index of the kernel (i.e. ). Among them, I smooth (x,y) is the smoothed image, G(i,j) is the weight of the Gaussian kernel, I(xi,yj) is the pixel value of the original image, and k is the size of the kernel.

8. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 4), the semantic segmentation model automatically marks the liver in the enhanced image, and uses a loss function that combines Dice loss and cross-entropy loss (CE) to optimize the segmentation effect of the liver. The formula of the combined loss function can be expressed as: Where X is the predicted segmentation result, Y is the actual segmentation result, |X∩Y| is the size of the intersection of the predicted result and the actual result, and |X| and |Y| are the total areas of the predicted result and the actual result, respectively. Where M is the number of categories, y o,c is the one-hot encoding of the true label, p o,c is the probability that the model predicts class c. Substitute (21)(22) into (23) L=α·L CE +(1-a)·L Dice (23).

9. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 4), the semantic segmentation model uses LeakyReLU as the activation function to replace the traditional ReLU activation function. The formula of LeakyReLU is: Among them, x is the input value, α is a small positive number (usually between 0.01 and 0.1) used to control the gradient of negative input. When x>0, the function is the same as ReLU and outputs x; when x≤0, the function outputs αx, so that even if the input is negative, the neuron can still receive a certain gradient signal, thus avoiding complete inactivation.

10. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 4), a semantic segmentation model for automatically identifying the liver in the enhanced image uses instance normalization instead of the widely used batch normalization as its normalization method. The formula of Instance Normalization is: Where x is a single sample of the input feature map, μ is the mean of x, and σ 2 is the standard deviation of x, and ε is a small constant used to prevent division by zero.

11. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 5), for isosurface extraction, the Marching Cubes algorithm is used to extract isosurfaces from floating-point image data, and the input three-dimensional scalar field data is converted into a corresponding polygonal mesh. The Marching Cubes algorithm calculates the intersection of the isosurface and the cube edge through linear interpolation to determine the shape of the three-dimensional isosurface. The linear interpolation formula is as follows: Among them, p1 and p2 are the two vertices of the cube edge, f1 and f2 are their scalar values, and C is the set isosurface threshold.

12. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 5), the extracted polygon mesh is processed by a smoothing filter (SmoothPolyDataFilter). The filter uses the Laplacian smoothing algorithm to iteratively adjust the position of each vertex, reduce noise and irregularities, and improve the smoothness and accuracy of the mesh. The vertex position update formula of Laplacian smoothing is as follows: Among them, p i is the current vertex, N(i) is the set of vertices adjacent to it, and λ is the relaxation factor.

13. The method for CT liver segmentation and three-dimensional reconstruction based on image enhancement according to claim 1, characterized in that: In step 5), in order to ensure the lighting effect and normal consistency of the polygonal mesh, the normal vector of each vertex is calculated and the direction of the normal vector is automatically adjusted (AutoOrientNormals). The calculation formula of the normal vector is as follows: Among them, v1, v2, and v3 are the three vertices that constitute the triangular surface.

Images