Aorta segmentation method and device based on plain-scan CT (Computed Tomography) image

Abstract

The invention provides an aorta segmentation method and device based on a plain-scan CT image. The aorta segmentation method comprises the following steps: S1, preprocessing the plain-scan CT image; s2, carrying out spine skeleton segmentation; s3, acquiring an aorta distribution area; s4, performing enhancement processing on the aorta distribution area; and S5, performing joint segmentation based on multiple algorithms on the aorta distribution region after enhancement processing. According to the method, in the plain-scan CT image, the aorta distribution area is determined based on the spine skeleton, the data size needing to be processed is reduced, joint segmentation based on multiple algorithms is carried out on the aorta distribution area, the segmentation accuracy is improved through the segmentation models of different structures, and faster and more accurate aorta segmentation is achieved while the application range is expanded.

Description

Technical Field

[0001] The present invention relates to the field of medical imaging technology, and in particular to an aortic segmentation method and device based on plain CT images. Background Art

[0002] The aorta is the thickest artery in the human body. It originates from the left ventricle of the heart, arches upward, right, and then downward, and then runs downward along the spine, branching into many smaller arteries in the chest and abdomen. The aorta is the main conduit for delivering blood to all parts of the body. Currently, aortic segmentation and analysis are mainly based on CT angiography. However, this imaging method is costly, time-consuming, has a large amount of data, and has low diagnostic efficiency. Some patients cannot undergo CT angiography due to physical reasons, and a large number of first- and second-level hospitals do not have the medical equipment and conditions for CT angiography, so they can only analyze based on plain CT image data. Summary of the Invention

[0003] In order to solve the technical problems of high cost and many limitations in the scope of application of aortic segmentation and analysis based on CT angiography in the prior art, the aortic segmentation method based on plain CT images proposed by the present invention includes the following steps:

[0004] S1. Preprocess the plain CT image;

[0005] S2. Perform spinal bone segmentation;

[0006] S3. Obtain the aortic distribution area;

[0007] S4. Perform enhancement processing on the aortic distribution area;

[0008] S5. Perform combined segmentation based on multiple algorithms on the enhanced aortic distribution area.

[0009] Preferably, in S1, the preprocessing includes obtaining the original plain CT image data, performing pixel resolution normalization, and performing noise suppression processing through bilateral filtering.

[0010] Preferably, in S2, the specific steps of spinal bone segmentation include:

[0011] S21. Perform gray histogram analysis on the preprocessed plain CT image, determine the gray boundary point between bone tissue and soft tissue according to the peak and valley characteristics of the histogram, and determine the bone tissue area;

[0012] S22. Perform edge detection processing on the bone tissue area to obtain bone tissue contour data, and perform connected component labeling processing on the bone tissue contour data to obtain bone tissue area data;

[0013] S23. Determine the target area where the spinal bones are located according to the distribution characteristics of the spinal bones;

[0014] S24. Perform boundary enhancement processing, region growing segmentation processing, and morphological smoothing processing on the target area in sequence to obtain the final spinal bone segmentation data.

[0015] Preferably, in S3, extend a first set distance to the left and right respectively to form the aortic distribution area.

[0016] Preferably, in S3, the determination process of the first set distance is to obtain a pre-set extension ratio and use the product of the extension ratio and the width of the current plain CT image as the first set distance.

[0017] Preferably, in S3, the extension ratio is set to 15%.

[0018] Preferably, in S4, the specific process of the enhancement processing is to convert the image data of the aortic distribution area from the spatial domain to the frequency domain through two-dimensional fast Fourier transform to obtain spectral components, perform high-pass filtering on the spectral components using a Butterworth high-pass filter, convert the high-frequency image data from the frequency domain to the spatial domain through two-dimensional inverse fast Fourier transform, perform edge detection on the image data of the aortic distribution area based on the Canny operator, perform gray-scale adjustment based on the result of the edge detection, and perform histogram equalization on the image data after gray-scale adjustment.

[0019] Preferably, in S4, the specific process of performing gray-scale adjustment based on the result of the edge detection is to increase the gray scale of the edge area by 10% and decrease the gray scale of the non-edge area by 10% to enhance the contrast.

[0020] Preferably, in S5, the specific process of the joint segmentation is as follows:

[0021] S51. Copy the image information of the aortic distribution area to obtain the first image information and the second image information;

[0022] S52. Use the TransUNet model to perform aortic segmentation on the first image information to obtain the first segmentation data;

[0023] S53. Use the TransFuse model to perform aortic segmentation on the second image information to obtain the second segmentation data;

[0024] S54. Register the first segmentation data and the second segmentation data, take the overlapping part as the final segmentation data, and match the final segmentation data to the image information of the original aortic distribution area.

[0025] The aortic segmentation device based on non-contrast CT images proposed by the present invention includes a processor, a memory, and a communication module. The communication module is used to receive non-contrast CT image data. The memory stores instructions executable by the processor, and the processor can implement the above-mentioned aortic segmentation method by executing the instructions.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] In non-contrast CT images, the aortic distribution area is determined based on the spinal skeleton, reducing the amount of data to be processed. For the aortic distribution area, joint segmentation based on multiple algorithms is performed, and the segmentation accuracy is improved through segmentation models with different structures, achieving faster and more accurate aortic segmentation while expanding the applicable range. Description of the Drawings

[0028] Figure 1 It is a flowchart of the aortic segmentation method based on non-contrast CT images of the present invention. Detailed Embodiments

[0029] The technical solutions of the present invention will be described in detail below in conjunction with the drawings and specific embodiments.

[0030] As Figure 1 shown, the aortic segmentation method based on non-contrast CT images proposed by the present invention includes the following steps:

[0031] S1. Preprocess the non-contrast CT image. The preprocessing includes obtaining the original non-contrast CT image data, performing pixel resolution standardization, adjusting the resolution to 1mm×1mm, and performing noise suppression processing through bilateral filtering.

[0032] S2. Perform spinal skeleton segmentation. The specific steps of spinal skeleton segmentation include:

[0033] S21. Perform gray histogram analysis on the preprocessed non-contrast CT image. According to the peak-valley characteristics of the histogram, determine the gray boundary point between bone tissue and soft tissue, and determine the bone tissue area.

[0034] S22. Perform edge detection processing on the bone tissue area to obtain bone tissue contour data, and perform connected component labeling processing on the bone tissue contour data to obtain bone tissue area data.

[0035] S23. Determine the target area where the spinal skeleton is located according to the distribution characteristics of the spinal skeleton.

[0036] S24. Perform boundary enhancement processing, region growing segmentation processing, and morphological smoothing processing on the target area in sequence to obtain the final spinal skeleton segmentation data.

[0037] S3. Obtain the aortic distribution area. Considering that the aorta is distributed near the spinal bones, take the area where the spinal bones are located as the center and extend a first set distance to the left and right respectively to form the aortic distribution area. The determination process of the first set distance is to obtain a pre-set extension ratio and take the product of the extension ratio and the width of the current plain CT image as the first set distance. To ensure that the possible areas of the aorta can be covered, the extension ratio is set to 15%.

[0038] S4. Perform enhancement processing on the aortic distribution area. The specific process of the enhancement processing is to convert the image data of the aortic distribution area from the spatial domain to the frequency domain through two-dimensional fast Fourier transform to obtain spectral components, perform high-pass filtering on the spectral components using a Butterworth high-pass filter, convert the high-frequency image data from the frequency domain to the spatial domain through two-dimensional inverse fast Fourier transform, perform edge detection on the image data of the aortic distribution area based on the Canny operator, perform gray-scale adjustment based on the result of the edge detection, and perform histogram equalization on the image data after gray-scale adjustment. The specific process of performing gray-scale adjustment based on the result of the edge detection is to increase the gray scale of the edge area by 10% and decrease the gray scale of the non-edge area by 10% to enhance the contrast. Through the enhancement processing, the gap in contrast and resolution between the plain CT image and CT angiography can be compensated, and the accuracy of subsequent recognition can be improved.

[0039] S5. Perform joint segmentation on the aortic distribution area after enhancement processing based on multiple algorithms. The specific process of the joint segmentation is as follows:

[0040] S51. Copy the image information of the aortic distribution area to obtain the first image information and the second image information.

[0041] S52. Use the TransUNet model to perform aortic segmentation on the first image information to obtain the first segmentation data. The TransUNet model combines Transformer and U-Net. The global self-attention mechanism of Transformer can effectively obtain global information and make up for the defect that U-Net cannot well model long-distance dependence relationships.

[0042] S53. Use the TransFuse model to perform aortic segmentation on the second image information to obtain the second segmentation data. The TransFuse model combines Transformers and CNN in a parallel manner, and both global dependence and shallow spatial details can be effectively captured in a more straightforward way.

[0043] S54. Register the first segmentation data and the second segmentation data, take the overlapping part as the final segmentation data, and match the final segmentation data to the image information of the original aortic distribution area.

[0044] The aortic segmentation device based on non-contrast CT images includes a processor, a memory, and a communication module. The communication module is used to receive non-contrast CT image data. The memory stores instructions executable by the processor, and the processor can implement the above-mentioned aortic segmentation method by executing the instructions.

[0045] The above-disclosed are only the preferred embodiments of the present invention. Of course, the scope of the rights of the present invention cannot be limited thereby. It should be pointed out that for those skilled in the art of this technology, any equivalent changes made to the present invention without departing from the design structure and principle of the present invention are regarded as the protection scope of the present invention.

Claims

1. A method for segmenting the aorta based on plain scan CT images, characterized in that ,The aorta segmentation method comprises the following steps: S1. Preprocessing the plain scan CT image; S2, segment the spine bones; S3, obtain the aorta distribution area; S4, enhance the aorta distribution area; S5. Perform joint segmentation based on multiple algorithms on the enhanced aorta distribution area.

2. The aorta segmentation method according to claim 1, characterized in that: In S1, the preprocessing includes acquiring original plain scan CT image data, normalizing pixel resolution, and performing noise suppression processing through bilateral filtering.

3. The aorta segmentation method according to claim 1, characterized in that: In S2, the specific steps of spine skeleton segmentation include: S21, performing grayscale histogram analysis on the preprocessed plain scan CT image, determining the grayscale demarcation point between bone tissue and soft tissue according to the peak and valley characteristics of the histogram, and determining the bone tissue area; S22, performing edge detection processing on the bone tissue region to obtain bone tissue contour data, and performing connected domain labeling processing on the bone tissue contour data to obtain bone tissue region data; S23, determining the target area where the spinal bones are located according to the distribution characteristics of the spinal bones; S24, performing boundary enhancement processing, region growing segmentation processing, and morphological smoothing processing on the target region in sequence to obtain the final spinal bone segmentation data.

4. The aorta segmentation method according to claim 1, characterized in that: In the above-mentioned S3, the first set distance is extended to the left and to the right respectively to form the aorta distribution area.

5. The aorta segmentation method according to claim 4, characterized in that: In S3, the process of determining the first set distance is to obtain a pre-set extension ratio, and use the product of the extension ratio and the width of the current plain scan CT image as the first set distance.

6. The aorta segmentation method according to claim 5, characterized in that: In the above-mentioned S3, the stretching ratio is set to 15%.

7. The aorta segmentation method according to claim 1, characterized in that: In S4, the specific process of the enhancement processing is to convert the image data of the aorta distribution area from the spatial domain to the frequency domain through a two-dimensional fast Fourier transform to obtain spectral components, use a Butterworth high-pass filter to high-pass filter the spectral components, convert the high-frequency image data from the frequency domain to the spatial domain through a two-dimensional inverse fast Fourier transform, perform edge detection on the image data of the aorta distribution area based on the Canny operator, perform grayscale adjustment based on the result of edge detection, and perform histogram equalization on the image data after the grayscale adjustment.

8. The aorta segmentation method according to claim 7, characterized in that: In S4, the specific process of grayscale adjustment based on the edge detection result is to increase the grayscale of the edge area by 10% and reduce the grayscale of the non-edge area by 10% to enhance the contrast.

9. The aorta segmentation method according to claim 1, characterized in that: In S5, the specific process of joint segmentation is as follows: S51, copying the image information of the aorta distribution area to obtain first image information and second image information; S52, performing aorta segmentation using a TransUNet model on the first image information to obtain first segmentation data; S53, performing aorta segmentation using the TransFuse model on the second image information to obtain second segmentation data; S54 , registering the first segmented data and the second segmented data, taking the overlapping parts as the final segmented data, and matching the final segmented data to the original image information of the aorta distribution area.

10. An aorta segmentation device based on plain scan CT images, characterized in that The aorta segmentation device includes a processor, a memory and a communication module. The communication module is used to receive plain scan CT image data. The memory stores processor executable instructions. The processor can implement the aorta segmentation method as described in any one of claims 1-9 by executing the instructions.

Images