A method, device and equipment for extracting an aortic dissection membrane

Abstract

The application discloses an aortic dissection membrane extraction method, device and equipment. After a target slice image in a CTA image is acquired, an aortic blood vessel region in the target slice image is recognized. Eigenvalues of a two-dimensional Hessian matrix of each pixel point in the aortic blood vessel region are analyzed, and an intima region is extracted. Pixel points of the intima region are set as a first pixel value, and pixel points of other regions are set as 0, so that a first binary image is generated. The first binary image is eroded, the pixel value of the pixel points in the region near the aortic membrane in the first binary image is removed by erosion, and a second binary image is generated. Pixel points with the first pixel value in the second binary image are clustered into two categories, and a connected region with an area greater than a first threshold in the cluster is determined as an aortic dissection membrane region, so that the aortic dissection membrane region can be correctly extracted for one or two aortic dissection membrane regions in the aortic region. The aortic dissection membrane region in the CTA image is accurately extracted.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical image processing, in particular to an aortic dissection membrane extraction method, device and equipment. BACKGROUND

[0002] Aortic dissection refers to the blood in the aorta entering the aortic media from the aortic intimal tear, causing the media to separate, and expanding along the long axis of the aorta to form a true lumen and a false lumen separation pattern of the aortic wall. A CTA (CT angiography) image is an image obtained after CT scanning of the human body blood vessels after injection of contrast agent. The blood vessels in the CTA image are clearly visible, and the aorta and main branches can be clearly displayed, which can help doctors understand the blood supply of the patient. Through the CTA image, the doctor can determine whether the patient has an aortic dissection.

[0003] There is currently a need to automatically determine whether the CTA image contains an aortic dissection membrane and to locate the region of the aortic dissection membrane. However, there is currently little research on aortic dissection membrane extraction, and it is not possible to accurately extract the aortic dissection membrane in the CTA image. SUMMARY

[0004] Therefore, the embodiments of the present application provide an aortic dissection membrane extraction method, device and equipment to accurately extract the aortic dissection membrane in the CTA image.

[0005] To solve the above problems, the technical solutions provided by the embodiments of the present application are as follows:

[0006] An aortic dissection membrane extraction method, the method comprising:

[0007] Obtaining a target slice image in a CTA image, the target slice image being any slice image in the CTA image;

[0008] Identifying an aortic vessel region in the target slice image;

[0009] Determining an intimal region from the aortic vessel region according to the eigenvalues of the two-dimensional Hessian matrix of each pixel point in the aortic vessel region;

[0010] Setting the pixel points of the intimal region to a first pixel value and the pixel points of other regions to 0 to generate a first binary image;

[0011] Performing erosion processing on the first binary image to generate a second binary image;

[0012] The pixels in the second binary image whose pixel value is the first pixel value are subjected to binary clustering, and the connected regions with an area greater than the first threshold in the clusters are determined to be the aortic dissection membrane region.

[0013] In one possible implementation, determining the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region includes:

[0014] Calculate the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels; the target pixels are each pixel in the aortic vascular region.

[0015] The predicted value of the target pixel is calculated using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel, and the predicted value of the target pixel is inversely proportional to the gray value of the target pixel;

[0016] Pixels whose predicted values ​​are greater than the second threshold are identified as inner membrane regions.

[0017] In one possible implementation, the eigenvalues ​​of the two-dimensional Hessian matrix include a first eigenvalue and a second eigenvalue, wherein the first eigenvalue is the element value on the main diagonal of the first row of the two-dimensional Hessian matrix, and the second eigenvalue is the element value on the main diagonal of the second row of the two-dimensional Hessian matrix.

[0018] The step of calculating the predicted value of the target pixel using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel includes:

[0019] When the second feature value of the target pixel is less than or equal to 0, the predicted value of the target pixel is determined to be 0.

[0020] When the second feature value of the target pixel is greater than 0, the ratio of the absolute value of the first feature value of the target pixel to the absolute value of the second feature value of the target pixel is calculated to obtain the first parameter. The square root of the sum of the squares of the first feature value and the second feature value of the target pixel is calculated to obtain the second parameter. The predicted value of the target pixel is calculated based on the first parameter and the second parameter.

[0021] In one possible implementation, calculating the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel includes:

[0022] The eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels at different scales are calculated for the aortic vascular region.

[0023] In one possible implementation, calculating the predicted value of the target pixel using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel includes:

[0024] Using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels at different scales, the predicted values ​​of the target pixels at different scales are calculated respectively.

[0025] The maximum value among the predicted values ​​of the target pixel corresponding to the different scales is selected as the predicted value of the target pixel.

[0026] In one possible implementation, the step of performing binary clustering on pixels in the second binary image whose pixel value is the first pixel value, and determining the connected regions with an area greater than a first threshold in the clustered clusters as aortic dissection regions, includes:

[0027] Perform binary clustering on the pixels in the second binary image whose pixel value is the first pixel value, and perform region growing on the pixels in the clusters obtained by clustering to obtain multiple connected regions;

[0028] Calculate the area of ​​each of the connected regions, and determine the connected regions with an area greater than a first threshold as aortic dissection membrane regions.

[0029] In one possible implementation, after eroding the first binary image to generate a second binary image, the method further includes:

[0030] Determine whether the number of pixels with the first pixel value in the second binary image is greater than a third threshold;

[0031] If so, perform binary classification clustering on the second binary image, and determine the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection region.

[0032] If not, determine that the target slice image does not contain the aortic dissection region.

[0033] A method for assembling an aortic dissection membrane, the device comprising:

[0034] The acquisition unit is used to acquire a target slice image in a CTA image, wherein the target slice image is any slice image in the CTA image;

[0035] The identification unit is used to identify the aortic vessel region in the target slice image;

[0036] The first determining unit is configured to determine the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region.

[0037] The first generation unit is used to set the pixels in the inner membrane region to a first pixel value and set the pixels in other regions to 0 to generate a first binary image.

[0038] The second generation unit is used to perform erosion processing on the first binary image to generate a second binary image;

[0039] The second determining unit is used to perform binary classification and clustering on the pixels in the second binary image whose pixel value is the first pixel value, and to determine the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection membrane region.

[0040] An aortic dissection membrane extraction device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aortic dissection membrane extraction method as described above.

[0041] A computer-readable storage medium storing instructions that, when executed on a terminal device, cause the terminal device to perform the aortic dissection membrane extraction method described above.

[0042] Therefore, the embodiments of this application have the following beneficial effects:

[0043] This embodiment of the application, after acquiring a target slice image from a CTA image, identifies the aortic vessel region within the target slice image. By analyzing the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vessel region, the intima region, i.e., the aortic dissection membrane region and the region near the aortic membrane, can be extracted. Then, pixels in the intima region are set to a first pixel value, while pixels in other regions are set to 0, generating a first binary image. Erosion is applied to the first binary image to remove the pixel values ​​of pixels in the region near the aortic membrane, generating a second binary image. Finally, pixels in the second binary image with the first pixel value are clustered into two classes. Connected regions with an area greater than a first threshold in the clusters are identified as aortic dissection membrane regions. This ensures that one or two aortic dissection membrane regions can be correctly extracted from the aortic region. Thus, accurate extraction of the aortic dissection membrane region from the CTA image is achieved. Attached Figure Description

[0044] Figure 1 A schematic diagram illustrating an exemplary application scenario provided in this application embodiment;

[0045] Figure 2 A flowchart illustrating a method for extracting aortic dissection membrane as provided in an embodiment of this application;

[0046] Figure 3 A flowchart of another method for extracting aortic dissection membrane provided in an embodiment of this application;

[0047] Figure 4 This is a schematic diagram of an aortic dissection membrane extraction device provided in an embodiment of this application. Detailed Implementation

[0048] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the embodiments of this application will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0049] To facilitate understanding and explanation of the technical solutions provided in the embodiments of this application, the background technology of this application will be described first.

[0050] Aortic dissection occurs when blood from within the aorta tears at the intima, causing separation of the media and extending along the long axis of the aorta, creating a separation of the true and false lumens in the aortic wall. CTA images are obtained by performing a CT scan after contrast agents have been injected into the blood vessels. CTA images show clearer visualization of blood vessels, clearly displaying arteries and major branches, allowing doctors to determine if a patient has aortic dissection.

[0051] Currently, there is an urgent need for computer-assisted diagnosis by automatically determining whether aortic dissection is present in aortic CTA images and identifying the true and false lumens. Extraction of the aortic dissection membrane is a crucial and challenging part of this process.

[0052] Based on this, embodiments of this application provide a method, apparatus, and device for extracting aortic dissection membrane. After acquiring a target slice image from a CTA image, the aortic vessel region in the target slice image is identified. The feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vessel region are analyzed to extract the intima region, i.e., the aortic dissection membrane region and the region near the aortic membrane. Then, the pixels in the intima region are set to a first pixel value, and the pixels in other regions are set to 0, generating a first binary image. The first binary image is then eroded to remove the pixel values ​​of the pixels near the aortic membrane, generating a second binary image. Finally, the pixels in the second binary image with the first pixel value are clustered into two classes. Connected regions with an area greater than a first threshold in the clusters are identified as aortic dissection membrane regions. This ensures that one or two aortic dissection membrane regions can be correctly extracted from the aortic region, thereby achieving accurate extraction of the aortic dissection membrane region from the CTA image.

[0053] To facilitate understanding of the method for extracting aortic dissection membrane provided in the embodiments of this application, the following is combined with... Figure 1 The example scenario is shown below. See also... Figure 1 As shown in the figure, this figure is a schematic diagram of an exemplary application scenario provided in the embodiments of this application.

[0054] In practical applications, CTA images of the aortic region are first acquired. These CTA images consist of multiple slice images. Aortic dissection may or may not be present in the CTA images. Any slice image from the CTA images is used as the target slice image for illustration.

[0055] The aortic vessel region in the target slice image is identified. This region can include the intima (darker gray area in the image) and the blood region (lighter gray area). Then, the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vessel region are analyzed to extract the intima region, which can include the aortic dissection membrane region and the area near the aortic membrane. The aortic membrane is the adventitia of the aorta, and the aortic dissection membrane region is the area where the aortic dissection membrane exists after aortic dissection occurs.

[0056] Then, by setting the pixels in the intima region to the first pixel value and setting the pixels in other regions to 0, a first binary image can be generated. Erosion is then applied to the first binary image to remove the pixel values ​​of the pixels in the region near the aortic membrane, generating a second binary image.

[0057] The manifestation of aortic dissection can be either a single connected region or two connected regions within the aortic vascular region (the figure shows the case of two connected regions within the aortic vascular region). Therefore, pixels in the second binary image with a pixel value equal to the first pixel value are clustered into two classes. Connected regions within these clusters with an area greater than a first threshold are identified as aortic dissection regions. This allows for the correct extraction of aortic dissection regions in both cases, thus achieving accurate extraction of aortic dissection regions from CTA images.

[0058] Those skilled in the art will understand that Figure 1 The schematic diagram shown is merely one example in which embodiments of this application can be implemented. The scope of application of the embodiments of this application is not limited by any aspect of this framework.

[0059] To facilitate understanding of this application, the following description, in conjunction with the accompanying drawings, illustrates a method for extracting aortic dissection membrane according to an embodiment of this application.

[0060] See Figure 2As shown, this figure is a flowchart of a method for extracting aortic dissection membrane according to an embodiment of this application. Figure 2 As shown, the method may include S201-S206:

[0061] S201: Obtain the target slice image from the CTA image. The target slice image is any slice image in the CTA image.

[0062] Acquire CTA images of the aortic region, which consist of multiple slice images. Use any slice image from the CTA images as the target slice image and read its pixel values. The matrix corresponding to the pixel values ​​of the target slice image is as follows. Where a mn This represents the pixel value of the pixel in the m-th row and n-th column of the target slice image. Since the slice images included in the CTA image are all grayscale images, the pixel value of each pixel in the target slice image is a grayscale value, which can reflect the CT value corresponding to each pixel.

[0063] S202: Identify the aortic vessel region in the target slice image.

[0064] By preprocessing the target slice image using methods such as convolutional neural networks, the aortic vessel region can be extracted from the target slice image.

[0065] In practical applications, the segmentation mask for the aortic vessel region in the target slice image can first be obtained: Where b mn This indicates whether the pixel in the m-th row and n-th column is a mask for the aortic blood vessel region. If it is, the mask is 1; otherwise, the mask is 0.

[0066] Then, the matrix corresponding to the pixel values ​​of the target slice image is multiplied by the segmentation mask of the aortic vessel region in the target slice image, i.e., Image*Mask. Aorta This yields the corresponding matrix of the image, including the aortic vascular region. Where f mn This represents the pixel value of the pixel in the m-th row and n-th column. If the pixel belongs to the aortic blood vessel region, the pixel value is greater than 0; otherwise, it is 0.

[0067] S203: Determine the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region.

[0068] In CTA images, due to the presence of contrast agents in the blood, the blood portion of the aortic vessel region has a higher pixel value, while the intima region has a lower pixel value. Therefore, the intima region of the aortic vessel region can be extracted based on an algorithm using two-dimensional Hessian matrix eigenvalue analysis.

[0069] The Hessian matrix is ​​a square matrix composed of the second-order partial derivatives of a multivariable function. For a two-dimensional image f(x,y), the two-dimensional Hessian matrix at points x,y is defined as follows: Where I xx The convolution of the second-order Gaussian derivative with the image f(x,y) can be used to obtain I. xy and I yy I xx and I yy These are the eigenvalues ​​of a two-dimensional Hessian matrix. These two eigenvalues ​​can represent the degree of image variation in the tangent and normal directions at each pixel.

[0070] Since the pixel values ​​of the blood portion and the intima portion in the aortic vascular region differ significantly, the eigenvalues ​​of the two-dimensional Hessian matrix can be used to determine the intima portion from the aortic vascular region.

[0071] In one possible implementation, this application provides a specific method for determining the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region. Please refer to A1-A3 or B1-B4 below for details.

[0072] S204: Set the pixels in the inner membrane region to the first pixel value, and set the pixels in other regions to 0 to generate the first binary image.

[0073] After determining the inner membrane region, the pixels in the inner membrane region can be set as the first pixel value. For example, the pixels in the inner membrane region can be set to 1, and the pixels in other regions can be set to 0 to generate a first binary image.

[0074] In practical applications, the matrix corresponding to the first binary image can be obtained. Where c mn This represents the pixel value of the pixel in the m-th row and n-th column of the first binary image. If the pixel belongs to the inner membrane region, then the pixel value is the first pixel value; otherwise, it is 0.

[0075] S205: Perform erosion processing on the first binary image to generate a second binary image.

[0076] In morphological erosion, the erosion process primarily targets the highlighted areas (pixel-valued portions) of a binary image, eliminating boundary points and causing the boundaries to shrink inward. The first binary image primarily shows the intima region, including the aortic dissection membrane region and the area near the aortic wall. The area near the aortic wall mainly includes the aortic wall itself and any misidentified areas nearby. The erosion process uses a pre-defined convolution kernel, such as a 3x3 kernel, to remove the aortic wall portion located at the edge of the intima region in the first binary image, leaving the misidentified areas near the aortic wall and the aortic dissection membrane region, thus generating the second binary image.

[0077] In practical applications, the corresponding matrix of the second binary image can be obtained. Where d mn This represents the pixel value of the pixel in the m-th row and n-th column of the second binary image. The pixel value of each pixel is either the first pixel value or 0.

[0078] S206: Perform binary clustering on the pixels in the second binary image whose pixel value is the first pixel value, and determine the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection membrane region.

[0079] The manifestations of aortic dissection are diverse. If the tear is in the middle of the dissection, the aortic dissection area is distributed at both ends of the aortic vessel area, forming two connected areas. If the tear starts from one side of the dissection, the aortic dissection area is distributed at one end of the aortic vessel area, forming a connected area.

[0080] Clustering pixels in the second binary image whose pixel value is the first pixel value into two classes is primarily to satisfy the two conditions mentioned above. Ideally, for two connected regions distributed at both ends of the aortic region, the center of each cluster corresponds to the centroid of the two connected regions. For a connected region distributed at one end of the aorta, the center of at least one cluster corresponds to the centroid of this connected region.

[0081] Clustering can be performed using the k-means clustering algorithm, also known as k-means clustering. "k" indicates that there are k clusters, and "means" represents the mean of the data values ​​in each cluster as the centroid. The algorithm's core idea is to iteratively find a way to partition the data into k clusters that minimizes the loss function. The loss function can be defined as the sum of the squared errors of each sample's distance from its cluster centroid.

[0082] After clustering into two clusters, each cluster may contain misidentified regions near the aortic wall and aortic dissection regions. The area A of each connected region in the cluster is calculated. A threshold S is set; if the area of ​​a connected region is greater than or equal to the threshold S, it is considered an aortic dissection region. Misidentified regions near the aortic wall are multiple small areas, considered noise points. By applying the area threshold, these noise points are removed, preserving the true aortic dissection regions.

[0083] If the area of ​​all connected regions in a cluster is less than the first threshold, it means that the target slice image does not include the aortic dissection membrane region.

[0084] To improve the accuracy of connected regions, region growing can be performed on image pixels. In some possible implementations, S206 performs binary clustering on pixels in the second binary image whose pixel value is the first pixel value, and determines the connected regions with an area greater than a first threshold in the resulting clusters as aortic dissection regions. Specific implementations of this can include:

[0085] Perform binary clustering on the pixels in the second binary image whose pixel value is the first pixel value, and then perform region growing on the pixels in the clusters to obtain multiple connected regions.

[0086] Calculate the area of ​​each connected region, and determine the connected regions with an area greater than the first threshold as aortic dissection membrane regions.

[0087] In this process, seed points are automatically found within each cluster, and a region growing algorithm is used to identify multiple connected regions. The area A of each connected region is calculated. A first threshold S is set; if the area of ​​a connected region is greater than or equal to the first threshold S, it is considered to be an aortic dissection region. This application does not limit the value of the first threshold in its embodiments.

[0088] For a pixel p in the second binary image whose pixel value is the first pixel value, A(p) represents the area of ​​the connected region corresponding to this pixel p, and G(p) is the mask value of this pixel p.

[0089]

[0090] The region where G(p) is 1 is the aortic dissection region.

[0091] Based on the above steps S201-S206, this embodiment of the application, after acquiring the target slice image in the CTA image, identifies the aortic vessel region in the target slice image. By analyzing the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vessel region, the intima region, i.e., the aortic dissection membrane region and the region near the aortic membrane, can be extracted. Then, the pixels in the intima region are set to a first pixel value, and the pixels in other regions are set to 0, generating a first binary image. Erosion is performed on the first binary image to remove the pixel values ​​of the pixels near the aortic membrane, generating a second binary image. Finally, the pixels in the second binary image with the first pixel value are clustered into two classes. The connected regions in the clusters with an area greater than a first threshold are identified as aortic dissection membrane regions. This ensures that one or two aortic dissection membrane regions can be correctly extracted from the aortic region. Thus, accurate extraction of the aortic dissection membrane region from the CTA image is achieved.

[0092] In one possible implementation, S203 determines the specific implementation of the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region, which may include A1-A3:

[0093] A1: Calculate the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel; the target pixel is each pixel in the aortic vascular region.

[0094] For a two-dimensional image f(x,y), the two-dimensional Hessian matrix at points x,y is: The eigenvalues ​​of the two-dimensional Hessian matrix include a first eigenvalue and a second eigenvalue. The first eigenvalue λ1 is the element value on the main diagonal of the first row of the two-dimensional Hessian matrix (i.e., I). xx The second eigenvalue λ2 is the element value on the main diagonal of the second row of the two-dimensional Hessian matrix (i.e., I). yy ).

[0095] A2: Calculate the predicted value of the target pixel using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel. The predicted value of the target pixel is inversely proportional to the gray value of the target pixel.

[0096] Since the pixel values ​​in the inner membrane region are low, λ1, representing the tangent direction of the pixel, has an absolute value close to 0. λ2, representing the normal direction of the pixel, has an absolute value much greater than 0, and satisfies |λ1|≤|λ2|. Using the eigenvalues ​​of the target pixel's two-dimensional Hessian matrix, a predicted value can be obtained. A larger predicted value indicates a lower pixel value, making the target pixel more likely to be within the inner membrane region.

[0097] In one possible implementation, the predicted value of the target pixel is calculated using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel, including:

[0098] When the second feature value of the target pixel is less than or equal to 0, the predicted value of the target pixel is determined to be 0.

[0099] When the second feature value of the target pixel is greater than 0, calculate the ratio of the absolute value of the first feature value of the target pixel to the absolute value of the second feature value of the target pixel to obtain the first parameter. Calculate the square root of the sum of the squares of the first feature value and the second feature value of the target pixel to obtain the second parameter. Calculate the predicted value of the target pixel based on the first parameter and the second parameter.

[0100] That is, for each target pixel p in the aortic region, the predicted value S(p) of the target pixel is calculated. The calculation method is given by the following formula:

[0101]

[0102] Where: the first parameter Second parameter

[0103] In the inner membrane region, λ1 represents the tangent direction of a pixel, with an absolute value close to 0. λ2 represents the normal direction of a pixel, with an absolute value much greater than 0. Therefore, R... f It is the ratio of the absolute value of λ1 to the absolute value of λ2, and the corresponding value is closer to 0, so It is closer to 1. R n It is the square root of the sum of the squares of λ1 and λ2, and the corresponding value is much greater than 0, so Closer to 0 It is closer to 1. and Multiplying them together makes the predicted value closer to 1.

[0104] The parameters α and β control the computational sensitivity. α and β can be set according to the actual situation; for example, experience shows that setting α to 1 and β to 10 results in better recognition. Using the above method, the predicted values ​​of all pixels in the aortic vessel region can be calculated.

[0105] A3: Pixels with predicted values ​​greater than the second threshold are identified as inner membrane regions.

[0106] The above formula shows that the predicted value of a pixel in the inner membrane region should be closer to 1. Therefore, a second threshold is set. If the predicted value is greater than the second threshold, it means that the corresponding pixel belongs to the inner membrane region. This application does not limit the value of the second threshold in its embodiments.

[0107] In practical applications, a second threshold T is set, and the predicted value S(p) of each pixel p in the aortic vascular region is thresholded to obtain I(p), which is the mask value of pixel p.

[0108]

[0109] The region where I(p) is 1 is the endometrial region.

[0110] In this embodiment, the intima region can be determined by analyzing the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region.

[0111] To more accurately determine the intima region, in some possible implementations, S203 determines the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region. This specific implementation may also include B1-B4:

[0112] B1: Calculate the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels at different scales for the aortic vascular region.

[0113] When calculating the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel, the eigenvalues ​​of the two-dimensional Hessian matrix for the target pixel at different scales are calculated first.

[0114] B2: Calculate the predicted values ​​of the target pixels at different scales using the eigenvalues ​​of the two-dimensional Hessian matrix corresponding to the target pixels at different scales.

[0115] B3: Select the maximum value among the predicted values ​​of the target pixel corresponding to different scales as the predicted value of the target pixel.

[0116] The specific implementation of calculating the predicted value of the target pixel using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel includes steps B2-B3 above. The predicted value of the target pixel corresponding to different scales is S(σ,p), where σ represents the scale. Then, the maximum value in S(σ,p) is selected as the predicted value of the target pixel, i.e., S(p) = max S(σ,p). The method of calculating S(σ,p) is the same as the method of calculating S(p) in step A2, and will not be repeated here.

[0117] B4: Pixels with predicted values ​​greater than the second threshold are identified as inner membrane regions.

[0118] In practical applications, a second threshold T is set, and the predicted value S(p) of each pixel p in the aortic vascular region is thresholded to obtain I(p), which is the mask value of pixel p.

[0119]

[0120] The region where I(p) is 1 is the endometrial region.

[0121] In this embodiment, since the thickness of the vascular intima varies, the predicted values ​​of the target pixels corresponding to different scales are calculated, so that images including intima of different thicknesses can be processed. The predicted values ​​of the target pixels determined in the end are closer to the actual situation, thereby achieving more accurate determination of the intima region.

[0122] See Figure 3 The diagram shows a flowchart of another method for extracting aortic dissection membrane provided in an embodiment of this application.

[0123] Since the target slice image may not include the aortic dissection region, in order to improve efficiency, based on the above... Figure 2 In a corresponding embodiment, after performing erosion processing on the first binary image in S205 to generate the second binary image, the method may further include:

[0124] S301: Determine whether the number of pixels of the first pixel value in the second binary image is greater than the third threshold; if yes, execute the above S206 to perform binary classification clustering on the second binary image, and determine the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection membrane region; if no, execute S302.

[0125] S302: Determine that the target slice image does not contain the aortic dissection region.

[0126] After eroding the first binary image, the aortic wall region is removed. If the target slice image does not include the aortic dissection membrane region, the second binary image will only include a small number of pixels with the first pixel value. At this time, it can be directly determined that the target slice image does not contain the aortic dissection membrane region, without having to perform the clustering process in S206, thus improving the execution efficiency.

[0127] In this embodiment, S201-S206 can be referred to in the above embodiment, and will not be repeated here.

[0128] Based on the method embodiment provided above for extracting aortic dissection membrane, this application embodiment also provides an aortic dissection membrane extraction device, which will be described below with reference to the accompanying drawings.

[0129] SeeFigure 4 As shown in the figure, this is a schematic diagram of the structure of an aortic dissection membrane extraction device provided in an embodiment of this application. Figure 4 As shown, the aortic dissection membrane extraction device includes:

[0130] Acquisition unit 401 is used to acquire a target slice image in a CTA image, wherein the target slice image is any slice image in the CTA image;

[0131] The identification unit 402 is used to identify the aortic vessel region in the target slice image;

[0132] The first determining unit 403 is used to determine the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region.

[0133] The first generation unit 404 is used to set the pixels in the inner membrane region to the first pixel value and set the pixels in other regions to 0 to generate a first binary image.

[0134] The second generation unit 405 is used to perform erosion processing on the first binary image to generate a second binary image;

[0135] The second determining unit 406 is used to perform binary classification and clustering on the pixels in the second binary image whose pixel value is the first pixel value, and to determine the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection membrane region.

[0136] In one possible implementation, the first determining unit includes:

[0137] The first computational subunit is used to calculate the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel; the target pixel is each pixel in the aortic vascular region.

[0138] The second calculation subunit is used to calculate the predicted value of the target pixel using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel, wherein the predicted value of the target pixel is inversely proportional to the gray value of the target pixel.

[0139] The first determining subunit is used to determine the pixels whose predicted value is greater than the second threshold as the inner membrane region.

[0140] In one possible implementation, the eigenvalues ​​of the two-dimensional Hessian matrix include a first eigenvalue and a second eigenvalue, wherein the first eigenvalue is the element value on the main diagonal of the first row of the two-dimensional Hessian matrix, and the second eigenvalue is the element value on the main diagonal of the second row of the two-dimensional Hessian matrix.

[0141] The second calculation subunit is specifically used for:

[0142] When the second feature value of the target pixel is less than or equal to 0, the predicted value of the target pixel is determined to be 0.

[0143] When the second feature value of the target pixel is greater than 0, the ratio of the absolute value of the first feature value of the target pixel to the absolute value of the second feature value of the target pixel is calculated to obtain the first parameter. The square root of the sum of the squares of the first feature value and the second feature value of the target pixel is calculated to obtain the second parameter. The predicted value of the target pixel is calculated based on the first parameter and the second parameter.

[0144] In one possible implementation, the first computational subunit is specifically used for:

[0145] The eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels at different scales are calculated for the aortic vascular region.

[0146] In one possible implementation, the second computational subunit is specifically used for:

[0147] Using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels at different scales, the predicted values ​​of the target pixels at different scales are calculated respectively.

[0148] The maximum value among the predicted values ​​of the target pixel corresponding to the different scales is selected as the predicted value of the target pixel.

[0149] In one possible implementation, the second determining unit includes:

[0150] The clustering subunit is used to perform binary classification and clustering on pixels in the second binary image whose pixel value is the first pixel value, and to perform region growing on the pixels in the clusters obtained by clustering to obtain multiple connected regions.

[0151] The second determining subunit is used to calculate the area of ​​each of the connected regions and determine the connected regions with an area greater than a first threshold as aortic dissection membrane regions.

[0152] In one possible implementation, the device further includes:

[0153] The judgment unit is used to determine whether the number of pixels of the first pixel value in the second binary image is greater than a third threshold.

[0154] If the judgment result of the judgment unit is yes, the triggering unit triggers the second determining unit to perform binary classification and clustering on the second binary image, and determines the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection membrane region.

[0155] The third determining unit is used to determine that the target slice image does not contain the aortic dissection membrane region if the judgment result of the judgment unit is negative.

[0156] In addition, this application embodiment also provides an aortic dissection membrane extraction device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aortic dissection membrane extraction method as described in any of the above claims.

[0157] In addition, embodiments of this application also provide a computer-readable storage medium storing instructions that, when executed on a terminal device, cause the terminal device to perform the aortic dissection membrane extraction method as described in any of the preceding claims.

[0158] This embodiment of the application, after acquiring a target slice image from a CTA image, identifies the aortic vessel region within the target slice image. By analyzing the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vessel region, the intima region, i.e., the aortic dissection membrane region and the region near the aortic membrane, can be extracted. Then, pixels in the intima region are set to a first pixel value, while pixels in other regions are set to 0, generating a first binary image. Erosion is applied to the first binary image to remove the pixel values ​​of pixels in the region near the aortic membrane, generating a second binary image. Finally, pixels in the second binary image with the first pixel value are clustered into two classes. Connected regions with an area greater than a first threshold in the clusters are identified as aortic dissection membrane regions. This ensures that one or two aortic dissection membrane regions can be correctly extracted from the aortic region. Thus, accurate extraction of the aortic dissection membrane region from the CTA image is achieved.

[0159] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems or apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple, and relevant parts can be referred to the method section.

[0160] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0161] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0162] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0163] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for extracting aortic dissection membrane, characterized in that, The method includes: Obtain a target slice image from a CTA image, wherein the target slice image is any slice image of the CTA image; Identify the aortic vessel region in the target slice image; The intima region is determined from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region. The pixels in the inner membrane region are set to the first pixel value, and the pixels in other regions are set to 0 to generate a first binary image; The first binary image is subjected to erosion processing to generate a second binary image; The pixels in the second binary image whose pixel value is the first pixel value are subjected to binary clustering, and the connected regions with an area greater than the first threshold in the clusters are determined to be the aortic dissection membrane region.

2. The method according to claim 1, characterized in that, Determining the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region includes: Calculate the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels; the target pixels are each pixel in the aortic vascular region. The predicted value of the target pixel is calculated using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel, and the predicted value of the target pixel is inversely proportional to the gray value of the target pixel; Pixels whose predicted values ​​are greater than the second threshold are identified as inner membrane regions.

3. The method according to claim 2, characterized in that, The eigenvalues ​​of the two-dimensional Hessian matrix include a first eigenvalue and a second eigenvalue. The first eigenvalue is the element value on the main diagonal of the first row of the two-dimensional Hessian matrix, and the second eigenvalue is the element value on the main diagonal of the second row of the two-dimensional Hessian matrix. The step of calculating the predicted value of the target pixel using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel includes: When the second feature value of the target pixel is less than or equal to 0, the predicted value of the target pixel is determined to be 0. When the second feature value of the target pixel is greater than 0, the ratio of the absolute value of the first feature value of the target pixel to the absolute value of the second feature value of the target pixel is calculated to obtain the first parameter. The square root of the sum of the squares of the first feature value and the second feature value of the target pixel is calculated to obtain the second parameter. The predicted value of the target pixel is calculated based on the first parameter and the second parameter.

4. The method according to claim 2, characterized in that, The calculation of the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel includes: The eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels at different scales are calculated for the aortic vascular region.

5. The method according to claim 4, characterized in that, The step of calculating the predicted value of the target pixel using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixel includes: Using the eigenvalues ​​of the two-dimensional Hessian matrix of the target pixels at different scales, the predicted values ​​of the target pixels at different scales are calculated respectively. The maximum value among the predicted values ​​of the target pixel corresponding to the different scales is selected as the predicted value of the target pixel.

6. The method according to claim 1, characterized in that, The step of performing binary clustering on pixels in the second binary image whose pixel value is the first pixel value, and determining the connected regions with an area greater than a first threshold in the clusters obtained from the clustering as aortic dissection regions, includes: Perform binary clustering on the pixels in the second binary image whose pixel value is the first pixel value, and perform region growing on the pixels in the clusters obtained by clustering to obtain multiple connected regions; Calculate the area of ​​each of the connected regions, and determine the connected regions with an area greater than a first threshold as aortic dissection membrane regions.

7. The method according to claim 1, characterized in that, After performing erosion processing on the first binary image to generate a second binary image, the method further includes: Determine whether the number of pixels with the first pixel value in the second binary image is greater than a third threshold; If so, perform binary classification clustering on the second binary image, and determine the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection region. If not, determine that the target slice image does not contain the aortic dissection region.

8. A method for assembling an aortic dissection membrane, characterized in that, The device includes: The acquisition unit is used to acquire a target slice image in a CTA image, wherein the target slice image is any slice image in the CTA image; The identification unit is used to identify the aortic vessel region in the target slice image; The first determining unit is configured to determine the intima region from the aortic vascular region based on the feature values ​​of the two-dimensional Hessian matrix of each pixel in the aortic vascular region. The first generation unit is used to set the pixels in the inner membrane region to a first pixel value and set the pixels in other regions to 0 to generate a first binary image. The second generation unit is used to perform erosion processing on the first binary image to generate a second binary image; The second determining unit is used to perform binary classification and clustering on the pixels in the second binary image whose pixel value is the first pixel value, and to determine the connected regions with an area greater than the first threshold in the clusters obtained by clustering as the aortic dissection membrane region.

9. An extraction device for aortic dissection membrane, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method for extracting aortic dissection membrane as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a terminal device, cause the terminal device to perform the method for extracting aortic dissection membrane as described in any one of claims 1-7.

Images