CN105976384A - Human body thoracic and abdominal cavity CT image aorta segmentation method based on GVF Snake model - Google Patents

Abstract

The invention discloses a GVF Snake model-based method for aortic segmentation of CT images of the human thoracoabdominal cavity, which mainly overcomes the shortcomings of traditional manual segmentation and semi-automatic segmentation with a large workload and long time consumption. At the same time, the invention has good repeatability and avoids artificial Uncertainty caused by segmentation; the implementation process is: (1) read the CT image and perform image preprocessing; (2) set the initial contour of the GVF Snake model on the image obtained after preprocessing; (3) obtain The edge image of the image obtained after preprocessing; (4) obtain the gradient vector flow GVF as the external energy field through the diffusion equation based on the obtained edge image; (5) establish the internal energy model to maintain the smoothness of the contour; (6) use The energy function E is constructed by the internal energy and the external energy, and the minimum value of the energy E is obtained through iterative operation, and finally the contour reaches the target boundary; the present invention has important application value in the field of diagnosis and treatment of human thoracoabdominal aortic dissection.

Description

Aortic Segmentation Method Based on GVF Snake Model in Human Thoracic and Abdominal CT Images

technical field

The invention belongs to the technical field of medical image processing, and relates to a GVF Snake model-based aorta segmentation method in a CT image of a human thoracoabdominal cavity, which can be used for three-dimensional reconstruction of the aorta in a human thoracoabdominal cavity.

Background technique

For patients with cardiovascular and cerebrovascular diseases, thoracoabdominal aortic dissection (AoD) is much more life-threatening than cerebral infarction, myocardial infarction and malignant tumors. The dissection is peeled off, resulting in the appearance of true lumen and false lumen in the aorta in the human thoracoabdominal cavity. When the blood flows in the true and false lumen, the blood vessel wall is squeezed, and at the same time, the blood enters the middle layer of the aorta through the intima to form a hematoma. Patients do not fully understand the condition, so the mortality rate within 48 hours after the onset of the condition can be as high as 36% to 71%, which poses a great threat to human life; currently, the main treatment for aortic dissection is endovascular isolation. As for the chief surgeon, it is necessary to obtain as much information as possible about the spatial relationship between the local anatomy and adjacency of the lesion during the operation, so as to achieve accurate diagnosis, operation and postoperative evaluation, especially for the location and extent of thoracoabdominal aortic dissection. It is one of the main factors that determine the indications of surgery and the success of the operation; therefore, it is very necessary to develop a medical image processing system to quickly and accurately extract the thoracoabdominal aorta and its dissection It is of great significance to improve the success rate of the operation; because each patient has a large number of CT tomographic images, the workload of manual segmentation is very large. For the three-dimensional image This is especially true for operations; another problem is that there is uncertainty in manual segmentation, and there are great differences in the segmentation results of different medical experts, and even the segmentation results of the same image by the same expert at different times and in different states are also different. Not a small difference; if the CT image can be automatically segmented, then the problems associated with the manual segmentation method will be solved; currently the most clinically used 3D reconstruction based on CT tomographic images, except for a few tissues with obvious contrast between the Housfield value and the surrounding , such as bones, lungs, and blood vessels after angiography, other tissues and organs cannot be automatically segmented; even for these tissues, due to the influence of scanning layer thickness and volume effect, the contours of automatic computer segmentation are often unsatisfactory to medical researchers; Therefore, it is currently a difficult point to extract the aorta accurately and quickly from complex and irregular tissues and organs.

Contents of the invention

The outstanding advantage of this method is that it can realize the automatic, accurate and rapid extraction of the aorta from the complex human thoracic and abdominal cavity, avoiding the problems of inaccurate manual segmentation and huge workload; the technical solution adopted in the present invention is a A method for aortic segmentation of human thoracoabdominal CT images based on the GVF Snake model, comprising the following steps:

(1) Read the CT image and perform image preprocessing;

(2) carry out the initialization profile setting of GVF Snake model on the image after processing in step (1);

(3) to obtain its edge image to the image obtained after processing in step (1);

(4) seek the gradient vector flow GVF as the external energy field by the diffusion equation based on the edge image obtained in the step (3);

(5) Establish an internal energy model to maintain the smoothness of the contour;

(6) Construct the energy function E based on steps (4) and (5), and obtain the minimum value of the energy E through iterative operations, and finally make the contour reach the target boundary;

In step (1), due to the complexity of the internal structure of the human chest and abdominal cavity, and the overall brightness of the CT image is dark, and the contrast between various tissues and organs is low, it is necessary to properly adjust the brightness of the read-in CT image; In addition, if the subsequent calculation is performed on the entire CT image, the amount of calculation will be very large, which will reduce the calculation efficiency, and other tissues and organs will also interfere with the automatic segmentation of the aorta. Therefore, an appropriate size region of interest should be selected for Subsequent operation processing;

In step (2), since this GVF Snake model is more flexible in initializing the position of the contour than the traditional Snake model, the initial contour is set inside, outside or intersecting with the target boundary, and finally converges to the aorta correctly. on the border of

In step (3), define f(x, y) as the edge image of the image I(x, y) obtained after preprocessing; the edge image f(x, y) has a larger value near the target boundary nature;

In step (4), the gradient vector field is defined as: V(x, y) = (u(x, y), v(x, y)), then the gradient vector flow is the vector field that minimizes the following energy functional :

In step (5), the elastic energy and the bending energy are collectively referred to as internal energy, which is used to control the elastic deformation of the contour line and play a role in maintaining the continuity and smoothness of the contour;

In step (6), the contour curve is "actively" deformed and displaced under the joint constraints of the internal force from the model itself and the external force from outside the model; the internal force constrains the shape characteristics of the contour curve, and the external force guides the behavior of the curve; through Multiple iterative operations continuously update the position of the model contour curve, and finally make the contour reach the target boundary; compared with the existing manual and semi-automatic methods for segmenting the aorta, the present invention has the following advantages:

1. The present invention avoids the disadvantages of traditional manual segmentation and semi-automatic segmentation, such as large workload and long time consumption, and realizes the rapid and automatic segmentation of the aorta from the complicated human chest and abdominal cavity;

2. The present invention has good repeatability and avoids the uncertainty caused by manual segmentation. Since the segmentation results of different medical experts are very different, even the same expert can segment the same image at different times and in different states. There are also not small differences in the results; because this invention is fully automatic segmentation, there is no human intervention, so the segmentation results will avoid the occurrence of this uncertainty and improve the segmentation accuracy.

Description of drawings

Fig. 1 is the algorithm flowchart of the present invention;

Fig. 2 is the CT image of input and the graph I (x, y) that obtains after preprocessing; (a) is the original CT image of input; (b) is the graph I (x, y) that obtains after preprocessing;

Fig. 3 is the situation of setting the initial profile of the GVF Snake model on I(x, y);

Fig. 4 is the edge image of I (x, y);

Figure 5 is the distribution of the gradient vector flow GVF;

Figure 6 shows the final evolution result of the active contour model based on GVF Snake and the corresponding aorta segmentation result; (a) is the final evolution result of the contour; (b) the aorta segmentation result.

detailed description

The algorithm flow chart of the present invention is as shown in Figure 1, at first read CT image, carry out image preprocessing; Then carry out the initial contour setting of GVF Snake model on the image obtained after preprocessing; Then ask for the image obtained after preprocessing Based on the obtained edge image, the gradient vector flow GVF is obtained as the external energy field through the diffusion equation; then the internal energy model is established to maintain the smoothness of the contour; finally, the energy function E is constructed by using the internal energy and external energy, through iteration Calculate the minimum value of energy E, and finally make the contour reach the target boundary. The specific implementation process of the technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings.

1. Read CT images and perform image preprocessing

Due to the complexity of the internal structure of the human thoracic and abdominal cavity, the overall brightness of the CT image is dark, and the contrast between various tissues and organs is low, it is necessary to properly adjust the brightness of the read-in CT image. The input original CT image is as follows: As shown in Figure 2(a); firstly, according to the prior knowledge, the gray scale range of the useful information in the CT image is known, and then it is mapped to the entire gray scale space, thereby improving the brightness of the entire image; The amount of calculation will be very large, and at the same time, in order to eliminate the interference of other tissues and organs on the aorta segmentation, an appropriate size region of interest should be selected for subsequent calculation and processing. Here, the size of 128*128 is used. As the region of interest, and the aorta can be completely displayed in this region; the final image obtained after the above preprocessing operations is shown in Figure 2(b);

2. Perform the initial contour setting of the GVF Snake model on the image obtained after preprocessing

The traditional Snake model is very sensitive to the position of the initial contour, requiring that the position of the initial contour must be as close as possible to the target boundary, otherwise a wrong convergence result will be obtained; at this point, the GVF Snake model is more flexible, and its initial contour can be If it is set inside, outside or intersecting with the target boundary, it can eventually converge to the boundary of the aorta correctly; when setting the initial contour, first select at least 4 points, and at the same time, take points strictly in a clockwise direction , and then insert several new points between the pre-selected points, so as to describe the initialized contour more completely and make the contour smoother; the initial contour setting of the GVF Snake model is shown in Figure 3;

3. Find the edge image of the image obtained after preprocessing

Since the GVF force field is obtained by the negative gradient vector of the edge image through the diffusion equation, in order to obtain the GVF force field, it is necessary to define f(x, y) as the preprocessed image I(x, y) Edge image; the edge image f(x, y) has the property of taking a larger value near the target boundary, and it can be defined as any of the following four forms:

$\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; \&dtri;</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}$

$\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; \&dtri;</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; G</span>}}_{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}$

f(x,y)=-I(x,y)

f(x,y)=-(G _σ (x,y)*I(x,y))

It can be considered that the edge image f(x, y) is obtained from the image I(x, y), which is very close to the real edge of the image, and the obtained edge image is shown in Figure 4;

4. Calculate the gradient vector flow GVF as the external energy field through the diffusion equation based on the edge image

The gradient vector flow is a static image force that does not change over time; the gradient vector field is defined as: V(x, y) = (u(x, y), v(x, y) ), then the gradient vector flow is the vector field that minimizes the following energy functional:

$\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \&Integral;</span><span\; class="notranslate">\; \&Integral;</span>\mathrm{<span\; class="notranslate">\; \&mu;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; \&dtri;</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; \&dtri;</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}$

Among them, the first item on the right end is the smoothing item, and the second item is the data item; μ is the weighting coefficient, and its value needs to be set according to the noise of the image. The greater the noise, the corresponding increase in the value of μ; then using the variational method, Use the following Euler equations to solve the minimization process of the energy functional:

$\mathrm{<span\; class="notranslate">\; \&mu;</span>}{<span\; class="notranslate">\; \&dtri;</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}$

$\mathrm{<span\; class="notranslate">\; \&mu;</span>}{<span\; class="notranslate">\; \&dtri;</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}$

in Represents the Laplacian operator, f _x , f _y , is the first-order derivative and second-order derivative of the edge graph with respect to x and y, and the GVF field can be obtained by using the above formula to solve the process, as shown in Figure 5;

5. Build the internal energy model to maintain the smoothness of the contour

Elastic energy and bending energy are collectively referred to as internal energy, which is used to control the elastic deformation of the contour and maintain the continuity and smoothness of the contour; in the process of minimizing the energy function E, the elastic energy quickly compresses the contour into a A smooth circle, the bending energy drives the contour line into a smooth curve; the Snake model can be expressed as a parametric curve defined on s∈(0, 1), that is, X(s)=(x(s), y(s)), Then the internal energy function can be defined as:

${\mathrm{<span\; class="notranslate">\; E.</span>}}_{\mathrm{<span\; class="notranslate">\; int</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; \&Integral;</span>}_{\mathrm{<span\; class="notranslate">\; 0</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&alpha;</span>}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&beta;</span>}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}$

Among them, α is the continuous constraint coefficient applied to two adjacent points on the contour curve, and its function is to adjust the stretching force of the contour; β controls the stiffness of the contour; their values are related to the image noise distribution, the greater the noise, the greater the value of α and β The value should also be larger so that the contour curve can cross the local minimum position caused by noise; at the same time, the values of α and β determine the performance of the contour convergence; since α controls the first-order derivative vector mode component of the contour curve, The larger α is, the faster the contour shrinks; while β controls the second-order derivative vector mode component of the contour curve, the larger β is, the smoother the contour is; therefore, by choosing the values of α and β reasonably, the contour can converge to the image A more reasonable position in

6. Construct the energy function E, and obtain the minimum value of the energy E through iterative operations, and finally make the contour reach the target boundary. The energy function E should contain both internal energy and external energy, defined as:

$\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; \&Integral;</span>}_{\mathrm{<span\; class="notranslate">\; 0</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&alpha;</span>}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&beta;</span>}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; E.</span>}}_{\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}$

When the active contour is approaching the target boundary, its essence is the process of finding the minimum value of the energy function E; from the variational principle, the necessary condition for the energy functional to take the minimum value is that it satisfies the following after transformation Euler's equation:

${\mathrm{<span\; class="notranslate">\; \&alpha;X</span>}}^{<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&beta;X</span>}}^{<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; \&dtri;</span>{\mathrm{<span\; class="notranslate">\; E.</span>}}_{\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}$

In order to solve the above-mentioned Euler equation, X(s) can be regarded as a function of time t and independent variable s, that is, X(s)=X(s, t); at this time, the solution of the above-mentioned Euler equation can be obtained by solving The numerical solution of the following formula gives:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&alpha;X</span>}}^{<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&beta;X</span>}}^{<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; \&dtri;</span>{\mathrm{<span\; class="notranslate">\; E.</span>}}_{\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; t</span>}}$

in, X _t (s, t) is the partial derivative of X (s, t) with respect to time t. When the equation solution X (s, t) is stable, X _t (s, t) = 0, which is convenient for obtaining the solution of the Euler equation ; Solve the equation through discretization and continuous iterative calculation, and finally make the energy E reach the minimum value, and the contour reaches the target boundary accordingly.

Claims (1)

1. A human thoracic and abdominal cavity CT image aorta segmentation method based on a GVF Snake model comprises the following steps:

(1) reading a CT image and carrying out image preprocessing;

(2) carrying out initialization contour setting of a GVF Snake model on the image processed in the step (1);

(3) solving an edge image of the image obtained after the processing in the step (1);

(4) solving gradient vector flow GVF as an external energy field through a diffusion equation based on the edge image obtained in the step (3);

(5) establishing an internal energy model for keeping the smoothness of the contour;

(6) constructing an energy function E based on the step (4) and the step (5), and solving a minimum value of the energy E through iterative operation to finally enable the contour to reach a target boundary;

in the step (1), due to the complexity of the internal structure of the human body thoracic cavity and abdominal cavity, the overall brightness of the CT image is dark, and the contrast between tissues and organs is low, the brightness of the read CT image needs to be properly adjusted; in addition, if the whole CT image is subjected to subsequent operation, the calculated amount is very large, the operation efficiency is reduced, and other tissues and organs also generate certain interference on the automatic segmentation of the aorta, so that an interested area with a proper size is selected for subsequent operation processing;

in the step (2), because the GVF Snake model has higher flexibility on the outline initialization position than that of the traditional Snake model, the initial outline is arranged inside or outside the target boundary or is intersected with the target boundary, and finally the initial outline can be correctly converged on the boundary of the aorta;

in the step (3), f (x, y) is defined as an edge image of the image I (x, y) obtained after preprocessing; the edge image f (x, y) has the property of being relatively large near the target boundary;

in step (4), the gradient vector field is defined as: v (x, y) ═ u (x, y), V (x, y)), then the gradient vector flow is a vector field that minimizes the energy functional:

in the step (5), the elastic energy and the bending energy are called internal energy together and are used for controlling the elastic deformation of the contour line, so that the continuity and the smoothness of the contour line are kept;

in the step (6), the contour curve is actively deformed and displaced under the common constraint of the internal force from the model and the external force from the outside of the model; wherein the internal force constrains the shape characteristics of the profile curve and the external force guides the behavior of the curve; and continuously updating the position of the model contour curve through multiple iterative operations, and finally enabling the contour to reach the target boundary.