Aortic dissection detection method and system

Abstract

The application discloses an aortic dissection detection method and system, belongs to the technical field of medical image processing, adopts a first segmentation network to segment an aorta region in each layer slice of a CTA image, obtains a slice image of the aorta region in each layer slice, and further segments an inner valve region in the aorta region slice image by using a second segmentation network; the inner valve region is accurately detected step by step in a step-by-step segmentation mode; and in the process, after the slice image of the aorta region in each layer slice is obtained, the formed slice sequence is filtered by using a sliding window and local continuity to remove the slice images incorrectly divided in the slice sequence, so that the accurate segmentation of the inner valve region in the second stage is ensured; and the inner valve region in the CTA image is accurately segmented, and the aortic dissection can be accurately detected.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of medical image processing, and more particularly relates to an aortic dissection detection method and system. BACKGROUND

[0002] Aortic dissection (AD) is a life-threatening aortic disease. The aorta contains three layers of intima, media and adventitia, and is divided into three parts: ascending aorta, aortic arch and descending aorta. The intimal layer ruptures, causing blood to seep into the middle layer, resulting in aortic dissection. The mortality rate of untreated acute aortic dissection patients increases by 1%-2% per hour shortly after the onset of symptoms. Artificial detection is prone to miss some not obvious lesions, and is affected by experience, fatigue and other factors. Therefore, a rapid and accurate intelligent detection method is crucial to shorten the preoperative time.

[0003] Computed tomography angiography (CTA) is an effective means for detecting aortic dissection. The intimal flap is the most significant imaging feature of aortic dissection, and the aortic dissection class under DeBakey typing can be obtained by the existence area and number of the intimal flap. However, due to the problems of calcification, artifacts, low contrast and other imaging features, it is difficult to accurately segment the intimal flap in the aorta, and thus it is difficult to accurately detect the aortic dissection. SUMMARY

[0004] In view of the above defects or improvement needs of the prior art, the present application provides an aortic dissection detection method and system to solve the technical problem that the prior art cannot accurately detect aortic dissection.

[0005] To achieve the above purpose, in a first aspect, the present application provides an aortic dissection detection method, comprising:

[0006] The first segmentation network is used to segment the to-be-detected CTA image to obtain the slice images of the aortic region in each layer slice of the stacked CTA image and the corresponding aortic class; the slice images of each aortic region are sorted according to the order of the corresponding slice layer in the to-be-detected CTA image along the stacking direction to form a slice sequence;

[0007] A sliding window is used to slide on the slice sequence to segment the slice sequence into slice subsequences of different aortic categories, and slice images that do not belong to the aortic category in each slice subsequence are removed and reassembled to obtain a new slice sequence. When the sliding window is at different positions, the number of slice images of different aortic categories contained in the sliding window is counted. When the number exceeds a preset threshold and the aortic category begins to change, the changed aortic category is used as the current category, and the first slice image of the current category in the current sliding window is separated from the previous slice image to segment the slice sequence.

[0008] The second segmentation network is used to segment the inner valve region in each slice image in the new slice sequence, and the number of slice images containing the inner valve region obtained by segmentation under different aortic categories is counted respectively, thereby obtaining the DeBakey classification of aortic dissection in the CTA image to be detected.

[0009] Further preferably, the aorta categories include: aortic arch, ascending aorta and descending aorta;

[0010] Count the number of slice images containing the inner valve area obtained by segmentation under the aortic arch, ascending aorta and descending aorta categories respectively 、 and , and then calculate the ascending and descending aortic dissection scores;

[0011] Among them, ascending and descending aortic dissection scores 、 They are:

[0012]

[0013]

[0014] is the preset threshold.

[0015] Further preferably, when the ascending and descending aortic dissection scores are both less than or equal to a preset score, the CTA image to be detected is determined to be normal;

[0016] When the ascending and descending aortic dissection scores are both greater than the preset scores, the aortic dissection of the CTA image to be detected is determined to be type I according to the DeBakey classification;

[0017] When the ascending aortic dissection score is greater than or equal to the preset score, and the descending aortic dissection score is less than or equal to the preset score, the aortic dissection in the CTA image to be detected is determined to be DeBakey type II dissection;

[0018] When the ascending aortic dissection score is less than a preset score and the descending aortic dissection score is greater than a preset score, the aortic dissection of the CTA image to be detected is determined to be a DeBakey type III dissection.

[0019] Further preferably, the first segmentation network comprises: a cascaded encoding module, an expansion module and a decoding module;

[0020] The encoding module includes M cascaded residual modules; M ≥ 2; the input end of each residual module is also connected to its input end through a residual;

[0021] The dilation module consists of one or more cascaded dilated convolutional layers.

[0022] Further preferably, the first segmentation network is a pre-trained segmentation network, and its training process includes:

[0023] Inputting the first training set into the first segmentation network, and training the first segmentation network by minimizing the first loss function;

[0024] The first training set includes CTA images of healthy individuals and patients with different types of aortic dissection according to the DeBakey classification. The labels carried by each slice of the CTA image include the actual segmentation results of the aortic region in the slice image and the corresponding aortic category.

[0025] The first loss function includes: the difference loss between the aorta region segmentation result of each slice in the CTA image in the first training set and the corresponding true segmentation result.

[0026] Further preferably, the second segmentation network includes: a feature extraction module, a semantic extractor, a spatial feature extraction module, an attention module, a feature fusion module and a segmentation head;

[0027] The feature extraction module is used to extract the features of the slice image step by step to obtain N feature images with decreasing scales from large to small. ; N ≥ 3;

[0028] The semantic extractor is used to They are spliced ​​together and semantic extraction is performed based on the channel attention mechanism and the spatial attention mechanism to obtain global semantic encoding information; the global semantic encoding information includes: The corresponding semantic encoding information ;

[0029] The spatial feature extraction module is used to extract the spatial features of the slice image f ;

[0030] The feature fusion module includes N semantic injection units;

[0031] No. j A semantic injection unit is used to Perform upsampling to make it consistent with After the same size as Based on the attention mechanism, we can get j Decoding features ; ;

[0032] The N-1th semantic injection unit is used to Perform upsampling to make it consistent with After the same size as The features extracted by the attention module are fused based on the attention mechanism to obtain the N-1th decoding feature ;

[0033] The Nth semantic injection unit is used to Perform upsampling to make it consistent with f After the same size as f Fusion based on the attention mechanism to obtain the Nth decoding feature ;

[0034] The split header is used based on The spliced ​​features are used to segment the inner valve area in the slice image.

[0035] Further preferably, the semantic extractor includes: a cascaded channel attention module and a spatial attention module.

[0036] Further preferably, the attention module includes: one or more cascaded channel attention modules.

[0037] Further preferably, the spatial feature extraction module includes: multiple cascaded CNN modules and a spatial attention module connected to the output end of the last-level CNN module.

[0038] Further preferably, the second segmentation network is a pre-trained segmentation network, and its training process includes:

[0039] Inputting the second training set into the second segmentation network, and training the second segmentation network by minimizing the second loss function;

[0040] The second training set includes: slice images of aorta regions of different aorta categories; the labels carried by each slice image include: the true segmentation result and the true boundary line image of the inner valve region in the slice image;

[0041] During the training process, the second segmentation network also includes a boundary line detection module for Extracting the boundary line of the inner valve region in the slice image to obtain a boundary line image of the inner valve region in the slice image;

[0042] The second loss function includes: the difference loss between the inner valve area segmentation result of the slice image in the second training set obtained by the second segmentation network and the corresponding true segmentation result, and the difference loss between the boundary line image obtained by the boundary line detection module and the corresponding true boundary line image.

[0043] Further preferably, the second training set further includes: slice images with shadows at the edge of the inner valve region.

[0044] In a second aspect, the present invention provides an aortic dissection detection system, comprising: a memory and a processor, wherein the memory stores a computer program, and the processor executes the method provided in the first aspect of the present invention when executing the computer program.

[0045] In a third aspect, the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is executed by a processor, the device where the storage medium is located is controlled to execute the method provided in the first aspect of the present invention.

[0046] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0047] 1. The present invention provides a method for detecting aortic dissection, which uses a first segmentation network to segment the aortic region in each layer of slices of a CTA image to obtain slice images of the aortic region in each layer of slices, and then uses a second segmentation network to further segment the inner valve region in the slice images of the aortic region; through a step-by-step segmentation method, the inner valve region is accurately detected step by step; and in this process, after the slice images of the aortic region in each layer of slices are obtained, the slice sequence is further filtered through a sliding window and local continuity to remove erroneously divided slice images in the slice sequence, thereby providing a guarantee for further accurate segmentation of the inner valve region in the second stage; the present invention can accurately detect aortic dissection by accurately segmenting the inner valve region in the CTA image.

[0048] 2. Furthermore, in the aortic dissection detection method provided by the present invention, the first segmentation network includes a cascaded encoding module, a dilation module and a decoding module; the encoding module includes multiple cascaded residual modules; the input end of each residual module is also connected to its input end through a residual; the dilation module includes one or more cascaded hole convolution layers, thereby effectively expanding the network's receptive field, improving the ability to capture global information, and being able to more accurately locate the aortic area.

[0049] 3. Furthermore, in the aortic dissection detection method provided by the present invention, the second segmentation network includes: a feature extraction module, a semantic extractor, a spatial feature extraction module, an attention module, a feature fusion module and a segmentation head; while ensuring the second segmentation network's ability to extract and summarize the semantic features of the inner valve, the spatial information expression capability of the second segmentation network is enriched, and the inner valve area can be located more accurately.

[0050] 4. Furthermore, in the aortic dissection detection method provided by the present invention, during the training process of the second segmentation network, the second segmentation network further includes a boundary line detection module. The boundary lines of the inner valve area in the slice image are extracted to achieve early guidance of shallow features, and a difference loss is constructed between the boundary line image extracted by the boundary line detection module and the corresponding real boundary line image to supervise the spatial features, which can better take into account the spatial and semantic information, so that the second segmentation network can more accurately locate the inner valve area. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flow chart of a method for detecting aortic dissection provided in an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of the aortic dissection detection process provided by an embodiment of the present invention;

[0053] Figure 3 A schematic diagram of the structure of a first segmentation network provided in an embodiment of the present invention;

[0054] Figure 4 A schematic diagram of the structure of a second segmentation network provided in an embodiment of the present invention;

[0055] Figure 5 A schematic diagram of an aorta region slice image and an inner valve region segmentation result of the aorta region slice image provided by an embodiment of the present invention;

[0056] Figure 6 A schematic diagram of the structure of the semantic extractor LSSE provided in an embodiment of the present invention;

[0057] Figure 7 A schematic diagram of the structure of the spatial feature extraction module LSPath provided in an embodiment of the present invention;

[0058] Figure 8 Schematic diagram of the process of acquiring the boundary line image and the real boundary line image provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0060] In order to achieve the above objectives, firstly, Figure 1 As shown, the present invention provides a method for detecting aortic dissection, comprising:

[0061] 1) First stage segmentation:

[0062] The first segmentation network is used to segment the CTA image to be tested, obtaining slice images of the aorta region and the corresponding aorta categories in each stacked slice layer of the CTA image. The slice images of each aorta region are sorted according to the order of their corresponding slice layers along the stacking direction in the CTA image to form a slice sequence. The aorta categories include the aortic arch, ascending aorta, and descending aorta. Each slice layer of the CTA image corresponds to a scan segment.

[0063] In an optional embodiment, the specific detection process is as follows Figure 2 As shown in the figure, to provide more detailed surgical advice, the aortic diameter is measured in the slice images of the aortic region. Specifically, based on the aortic segmentation results, the aortic diameter is measured. For the ascending aorta, the average diameter from the aortic root to the aortic arch is calculated; for the descending aorta, the diameter is measured from the aortic arch to the abdominal aorta. To verify the accuracy of the diameter measurement, the physician can manually measure the aortic diameter of the selected vessel segment every five scan slices.

[0064] Preferably, in an optional embodiment, the CTA image to be examined is preprocessed before being input into the first segmentation network. Preprocessing operations may include one or more of denoising and image enhancement. Specifically, in an optional embodiment, the CTA image to be examined is denoised using the SRAD algorithm, which reduces speckle anisotropic diffusion, and the resolution of the CTA image to be examined is adjusted to 256×256. In an optional embodiment, considering that different doses of contrast enhancement agents may be used during patient examinations, the brightness of the CTA image to be examined is adjusted to half its original brightness for data enhancement, and the image is rotated 90° counterclockwise.

[0065] It should be noted that the first segmentation network can be a segmentation network such as SegNet, Mask-RCNN, DeepLab, etc. Preferably, in an optional embodiment, the first segmentation network includes: a cascaded encoding module, an expansion module, a decoding module and a mapping layer;

[0066] The encoding module includes M cascaded residual modules; M ≥ 2; the input end of each residual module is further connected to its input end through a residual; preferably, each residual module includes multiple cascaded convolutional layers;

[0067] The dilation module consists of one or more cascaded dilated convolutional layers;

[0068] The mapping layer maps the decoded feature map of the slice output by the decoding module to the aorta region pixel by pixel, and obtains the aorta region segmentation result of the slice. Generate the segmentation result. This function calculates the probability of each pixel in the decoded feature map being the aorta region or the non-aorta region. When the probability of the aorta region is large, the pixel is divided into the aorta region. Otherwise, the pixel is divided into the non-aorta region.

[0069] In one embodiment, a first segmentation network having 21 hidden layers is provided, and its structural diagram is shown as follows: Figure 3 As shown; in this embodiment, the introduction of residual blocks can effectively suppress the gradient vanishing phenomenon, and the addition of void convolution at the bottleneck of the first CNN can improve the global information capture capability of the convolutional neural network so as to accurately locate the aorta; the use of dropout and rectified linear units (ReLU) can improve the convergence speed.

[0070] In an optional implementation manner, the first segmentation network is a pre-trained segmentation network, and its training process includes:

[0071] Inputting the first training set into the first segmentation network, and training the first segmentation network by minimizing the first loss function;

[0072] The first training set includes CTA images of healthy individuals and patients with different types of aortic dissection according to the DeBakey classification. The labels carried by each slice of the CTA image include the actual segmentation results of the aortic region in the slice image and the corresponding aortic category.

[0073] The first loss function includes: the difference loss between the aorta region segmentation result of each slice in the CTA image in the first training set and the corresponding true segmentation result.

[0074] It should be noted that the loss function for measuring the difference between the aorta region segmentation results of each slice in the CTA image of the first training set and the corresponding true segmentation results can be a variety of loss functions, such as Dice loss function, cross entropy loss function, Focal loss function, Jaccard loss function, etc. Among them, the Dice loss function is:

[0075]

[0076] Among them, X and Y represent the predicted result and the true label respectively. The calculation process does not consider the background pixels.

[0077] Furthermore, in an optional embodiment, experienced physicians used the medical image processing software 3DSlicer to perform pixel-level annotation of the aorta region in each slice of the CTA image, obtaining corresponding labels. Before inputting the CTA images in the first training set into the first segmentation network, they were also preprocessed using the same method used for the CTA images to be tested.

[0078] 2) Filter the incorrectly divided slice images in the slice sequence by sliding window and local continuity:

[0079] A sliding window is used to slide on the slice sequence to segment the slice sequence into slice subsequences of different aortic categories, and slice images that do not belong to the aortic category in each slice subsequence are removed and reassembled to obtain a new slice sequence; wherein, when the sliding window is at different positions, the number of slice images of different aortic categories contained in the sliding window is counted respectively, and when the aortic category begins to change after the number exceeds a preset number threshold (in an optional embodiment, the value is 10), the changed aortic category is used as the current category, and the first slice image of the current category in the current sliding window is separated from the previous slice image to segment the slice sequence.

[0080] In one optional embodiment, the sliding window is five image slices long, with a step size of two image slices. Specifically, a sliding window of five slices is used to detect the boundaries of different aortic scan segments, dividing the entire CTA volume into three segments: the aortic arch, the descending aorta, and the ascending and descending aorta. Each slice is classified into one of these three segments based on the number and shape of the aortic regions detected within the slice. If a slice's category does not match the scan segment it belongs to, the slice is discarded.

[0081] In one optional embodiment, the slice images of each aortic region are sorted in order of their corresponding slice layers' positions along the stacking direction of the CTA image to be examined to form a slice sequence. For example, in a top-down stacking direction (i.e., the scanning direction of the CTA image), the aorta categories of the first and last image slices in the slice sequence are the aortic arch and descending aorta, respectively.

[0082] When the sliding window is at different positions, the number of image slices of different aorta categories contained in the sliding window is obtained respectively; in the initial sliding stage, the number of image slices of the aortic arch category exceeds a preset threshold; after multiple sliding, when the number of image slices of the ascending aorta category exceeds the preset threshold, the first image slice of the ascending aorta category in the current sliding window is separated from the previous image slice; when the number of image slices of the descending aorta category exceeds the preset threshold, the first image slice of the descending aorta category in the current sliding window is separated from the previous image slice.

[0083] 3) Second stage segmentation:

[0084] The second segmentation network is used to segment the inner valve region in each slice image in the new slice sequence, and the number of slice images containing the inner valve region obtained by segmentation under different aortic categories is counted respectively, thereby obtaining the DeBakey classification of aortic dissection in the CTA image to be detected.

[0085] In an optional embodiment, the number of slice images containing the inner valve region obtained by segmentation in the aortic arch, ascending aorta and descending aorta categories is counted respectively. 、 and , and then calculate the ascending and descending aortic dissection scores;

[0086] Among them, ascending and descending aortic dissection scores 、 They are:

[0087]

[0088]

[0089] is a preset threshold, and in an optional implementation manner, the value is 10.

[0090] In an optional embodiment, when the ascending and descending aortic dissection scores are both less than or equal to a preset score, the CTA image to be tested is determined to be normal;

[0091] When the ascending and descending aortic dissection scores are both greater than the preset scores, the aortic dissection of the CTA image to be detected is determined to be type I according to the DeBakey classification;

[0092] When the ascending aortic dissection score is greater than or equal to the preset score, and the descending aortic dissection score is less than or equal to the preset score, the aortic dissection in the CTA image to be detected is determined to be DeBakey type II dissection;

[0093] When the ascending aortic dissection score is less than a preset score and the descending aortic dissection score is greater than a preset score, the aortic dissection of the CTA image to be detected is determined to be a DeBakey type III dissection.

[0094] Furthermore, surgical recommendations can be given based on the classification results. For patients with type I dissection, ascending aorta / aortic arch replacement surgery and antegrade TEVAR are recommended simultaneously. For patients with type II dissection, ascending aorta / aortic arch replacement surgery is recommended, while for patients with type III dissection, only TEVAR is recommended, which can effectively improve the diagnostic efficiency of aortic dissection.

[0095] It should be noted that the second segmentation network can be a segmentation network such as SegNet, Mask-RCNN, DeepLab, etc.

[0096] Preferably, in an alternative embodiment, as Figure 4 As shown, the second segmentation network includes: feature extraction module, semantic extractor LSSE, spatial feature extraction module LSPath, attention module, feature fusion module and segmentation head;

[0097] The feature extraction module includes N cascaded feature extraction units; the feature extraction module is used to extract the features of the slice image step by step to obtain N feature images with decreasing scales from large to small. ; N ≥ 3; preferably, N is 4;

[0098] The semantic extractor LSSE is used to They are spliced ​​together and semantic extraction is performed based on the channel attention mechanism and the spatial attention mechanism to obtain global semantic encoding information; the global semantic encoding information includes: The corresponding semantic encoding information ;

[0099] The spatial feature extraction module LSPath is used to extract the spatial features of slice images f ;

[0100] The feature fusion module includes N semantic injection units;

[0101] No. j A semantic injection unit is used to Perform upsampling to make it consistent with After the same size as Based on the attention mechanism, we can get j Decoding features ; ;

[0102] The N-1th semantic injection unit is used to Perform upsampling to make it consistent with After the same size as The features extracted by the attention module are fused based on the attention mechanism to obtain the N-1th decoding feature ;

[0103] The Nth semantic injection unit is used to Perform upsampling to make it consistent with f After the same size as f Fusion based on the attention mechanism to obtain the Nth decoding feature ;

[0104] The split header is used based on The spliced ​​features are used to segment the inner valve area in the slice image.

[0105] like Figure 5 Shown are the aorta region slice image and the inner valve region segmentation results of the aorta region slice image.

[0106] In an optional implementation, the semantic injection unit includes an upsampling unit and a SIM module; the upsampling unit is used to perform an upsampling operation; and the SIM module is used to implement a fusion operation based on an attention mechanism.

[0107] In an optional embodiment, the semantic extractor LSSE includes: a cascaded channel attention module and a spatial attention module. Figure 6 As shown, in an optional embodiment, the semantic extractor LSSE includes a multi-layer cascade of channel attention modules and a multi-layer cascade of spatial attention modules; wherein the output end of the last-level channel attention module is connected to the input end of the first-level spatial attention module.

[0108] In one optional embodiment, the attention module includes: one or more cascaded channel attention modules. In another optional embodiment, the attention module includes: one or more cascaded spatial attention modules.

[0109] In an optional embodiment, the spatial feature extraction module LSPath comprises a plurality of cascaded CNN modules and a spatial attention module connected to the output end of the last CNN module; wherein the plurality of cascaded CNN modules are used to extract low-dimensional detailed spatial features of the slice image.

[0110] Introducing additional network layers to obtain more feature information inevitably increases the overall computational load of the network, reducing the speed of network inference. Considering the above problems, the embodiment designs a lightweight spatial feature extraction module LSPath to obtain additional spatial information, and the structure is as shown in Figure 7

[0111] Generally speaking, small convolution kernels focus on details, and large convolution kernels focus on context information. Considering that the purpose of the spatial feature extraction module LSPath is to extract spatial information, 3x3 convolution layers are used in the first three layers of the spatial feature extraction module LSPath to process the features. In addition, the number of channels often represents multi-dimensional abstract features, and a low number of channels corresponds to more detailed information. Therefore, the number of channels of the first three layers of the spatial feature extraction module LSPath is designed to be 16, 32, and 32, respectively. Small convolution kernels and low channel numbers also reduce the burden introduced by the spatial feature extraction module LSPath on network inference. And in the first two layers of convolution, a convolution with a step of 2 is used to realize fast downsampling of the feature map, further reducing the computational load of the network. The last layer of convolution of the LSPath is designed as a 1x1 convolution kernel in this embodiment, which is used to fuse the channel information of the first three layers of features, and at the same time, the 32-channel features are upgraded to 64 channels to match the input requirements of the subsequent feature fusion module. After convolution, two cascaded spatial refinement modules are connected to enhance the spatial feature representation of the path, and finally a feature map with a size of 1 / 4 of the original feature map is obtained, which is used to replace the features output by the first stage feature extraction unit (stage1) in the original feature extraction backbone network as the input of the feature fusion module. Figure 1

[0112] In an optional embodiment, the second segmentation network is a pre-trained segmentation network, and the training process thereof comprises:

[0113] The second training set is input into the second segmentation network, and the second segmentation network is trained by minimizing the second loss function;

[0114] The second training set comprises: aortic region slice images of different aortic categories; and the label carried by each slice image comprises: a real segmentation result of an inner valve region in the slice image and a real boundary line image.

[0115] ​​During the training process, the second segmentation network also includes a boundary line detection module FlapHead, which is used to detect the boundary line based on the Extracting the boundary line of the inner valve region in the slice image to obtain a boundary line image of the inner valve region in the slice image;

[0116] The second loss function includes: the difference loss between the inner valve area segmentation result of the slice image in the second training set obtained by the second segmentation network and the corresponding true segmentation result, and the difference loss between the boundary line image obtained by the boundary line detection module and the corresponding true boundary line image.

[0117] It should be noted that the aorta region slice images in the second training set can be directly obtained by manual clipping, or can be obtained by extracting the aorta region in each layer of the CTA image slices through the first segmentation network of the present invention, which is not limited here.

[0118] It should be noted that the boundary line detection module FlapHead is a neural network, which can be CNN, YOLO, SSD, etc. Preferably, in an optional embodiment, the boundary line detection module FlapHead includes a cascaded first convolution layer, a BN layer, an activation layer (preferably using ReLu) and a second convolution layer; in this embodiment, the second convolution layer is 1 1 convolutional layer.

[0119] In an optional embodiment, the second training set further includes slice images with shadows at the edge of the inner valve region.

[0120] In an optional embodiment, when obtaining the labels carried by the slice image, the actual segmentation result of the inner valve region in the slice image is represented by a mask, which can be obtained by manual annotation or by performing preliminary inner valve detection on each slice image based on the Hessian matrix (HM) method. The supervision information of the inner valve boundary (the actual boundary line image) is obtained by the mask of the inner valve region; Figure 8 As shown, when extracting the inner valve boundary, two types of labels are included, namely, inner valve and non-inner valve elements. The contour information is obtained by moving the Laplace operator on the mask annotation with a variety of different moving steps to obtain edge contour maps of different scales. After the edge contour maps of different scales are transformed into images of the same size as the mask, they are averaged to obtain an average edge contour map. The pixels in the average edge contour map with values ​​greater than the preset contour point threshold are divided into the boundary area to obtain the inner valve contour map. The inner valve contour map is expanded to obtain the real boundary line image. Specifically, the moving step sizes are set to 1, 2, and 4, respectively, so that the size is consistent with the original image, and the size is the same as the original image. Figure 1 / 2 and 1 / 4 of the size of the edge contour map, in order to obtain contour information of different scales. The size is the original Figure 1 The edge contour images of size 1 / 4 and 1 / 2 are upsampled to the same size as the original image, and the mean of the three contour images is calculated. The pixels with a mean value greater than the preset contour point threshold (0.1) are recorded as contour points (points in the boundary area) to obtain the inner valve contour image. Considering that the feature map itself only has the original image, the edge contour image of the inner valve is upsampled to the same size as the original image. Figure 1 The resolution of the image is / 8, and obtaining accurate contours directly from shallow features is too demanding. Therefore, this paper uses the expansion operation to thicken the inner valve contour map to obtain its soft contour map, so as to avoid overfitting of the network caused by too strong constraints. The formula of the expansion operation is:

[0121]

[0122] In the above formula, is the size of the rectangular dilation operator, Represents the inner valve contour, is the contour image after dilation, which is used as the real boundary line image.

[0123] Considering that the number of boundary pixels is far less than that of non-boundary pixels, boundary prediction is a class-imbalanced binary classification problem. Therefore, in this embodiment, binary cross entropy and Dice loss are used to jointly optimize boundary learning. The expression of this joint loss is:

[0124]

[0125] in, , They represent the boundary line image obtained by the boundary line detection module FlapHead and the real boundary line image respectively. and Represent binary cross entropy loss and Dice loss respectively, and their expressions are:

[0126]

[0127]

[0128] Where, H and W They represent the height and width of the boundary line image obtained by the boundary line detection module.

[0129] Binary cross-entropy loss is used for binary classification problems. For inner valve prediction, the inner valve region image can be divided into two categories: foreground and background elements, corresponding to two label values ​​of 0 or 1. Elements in the inner valve region image are probability values ​​between 0 and 1. Binary cross-entropy loss measures model performance by comparing the probability distribution of the network output with the distribution of the actual labels. Because the predicted labels are evaluated using the log function, it is more sensitive to large deviations between the predicted value and the actual value, which can accelerate network iteration.

[0130] However, the binary cross entropy loss usually calculates the loss of each pixel independently without considering their spatial connection, and finally takes the average of the loss of each pixel, which makes it difficult to take into account errors in details. In addition, this feature also reduces its performance in class imbalance problems such as inner valve boundary prediction, because the network tends to predict pixels as categories with a larger number, resulting in vague results for other categories. From this perspective, additional losses need to be added to alleviate this problem. Compared with the binary cross entropy loss, the Dice loss measures the accuracy of the network by measuring the overlap between the boundary line image obtained by the boundary line detection module FlapHead and the true boundary line image. In addition, the Dice loss only considers foreground pixels during the calculation process, which has a better effect on class imbalance problems such as boundary line extraction and evaluation. Therefore, the embodiment combines it with the binary cross entropy loss to make up for its shortcomings and further improve the effect of the network.

[0131] In addition, since the output results of the boundary line detection module FlapHead are not re-input into the backbone network, they are only used in the training phase to guide the direction of feature extraction of the backbone network, and do not increase the overhead of network inference.

[0132] Two branches were added to the original network structure. The LSPath branch of the feature extraction module takes the slice image as its starting point, directly extracting high-resolution spatial information from the slice image and outputting it to the subsequent feature fusion module. The FlapHead branch of the boundary detection module uses the 1 / 8-scale feature map output by the second-stage feature extraction unit (stage 2) in the feature extraction backbone as its starting point and outputs the result to the loss function calculation module to guide shallow feature extraction.

[0133] The final loss function of the network is as follows, The network loss calculated by the feature output of the FlapHead guidance head of the boundary line detection module is represented by the weight . Represents the network loss calculated from the inner valve region segmentation results and some semantic features.

[0134]

[0135] The expression is:

[0136]

[0137] It is caused by focal loss and feature separation loss Composition, where is the weight of feature separation loss.

[0138] focal loss The expression is:

[0139]

[0140] in, Represents the number of predicted categories, which is equivalent to the number of channels output by the network. The current element calculated by the network belongs to c The probability of the class category; with the help of the product term , focal loss Reduce the loss calculated from high-confidence predictions, so that the network pays more attention to difficult-to-recognize examples, ξ is the adjustment factor that controls the slope of the scheme.

[0141] Feature separation loss Different features are separated early by calculating the feature differences of elements belonging to different categories. The expression is shown in the following formula:

[0142]

[0143] in, In the expression and Represents the number of elements of different categories, represents the number of channels of the feature map, and Represents feature maps m Previous n The eigenvalues ​​of different categories of elements on the channels, is the mean of the channel eigenvalues. With this formula, The difference between the features at different category element positions in the feature map is calculated. Based on the formula The feature difference between the 1 / 16 feature map obtained by the feature extraction backbone at the corresponding positions of the two types of labels 0 and 1 is calculated and used as part of the loss function to encourage the network to distinguish them early.

[0144] It should be noted that the above processes are all executed in a computer or other program execution module.

[0145] In a second aspect, the present invention provides an aortic dissection detection system, comprising: a memory and a processor, wherein the memory stores a computer program, and the processor executes the method provided in the first aspect of the present invention when executing the computer program.

[0146] The related technical solutions are the same as the method provided in the first aspect of the present invention and will not be described in detail here.

[0147] In a third aspect, the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is executed by a processor, the device where the storage medium is located is controlled to execute the method provided in the first aspect of the present invention.

[0148] The related technical solutions are the same as the method provided in the first aspect of the present invention and will not be described in detail here.

[0149] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for detecting aortic dissection, characterized in that: include: The first segmentation network is used to segment the CTA image to be detected, and the slice images of the aorta region in each stacked slice in the CTA image and the corresponding aorta category are obtained; Slice images of each aorta region are sorted according to the order of their corresponding slice layers along the stacking direction in the CTA image to be detected to form a slice sequence; A sliding window is used to slide on the slice sequence to segment the slice sequence into slice subsequences of different aortic categories, and slice images that do not belong to the aortic category in each slice subsequence are removed and reassembled to obtain a new slice sequence; wherein, when the sliding window is at different positions, the number of slice images of different aortic categories contained in the sliding window is counted respectively, and when the aortic category begins to change after the number exceeds a preset number threshold, the changed aortic category is used as the current category, and the first slice image of the current category in the current sliding window is segmented from the previous slice image to segment the slice sequence; Using a second segmentation network to segment the inner valve region in each slice image in the new slice sequence, and counting the number of slice images containing the inner valve region obtained under different aorta categories, thereby obtaining the DeBakey classification of aortic dissection in the CTA image to be detected; The first segmentation network includes: a cascaded encoding module, a dilation module, a decoding module and a mapping layer; The encoding module includes M cascaded residual modules; M ≥ 2; the input end of each residual module is also connected to its input end through a residual; the expansion module includes one or more cascaded hollow convolution layers; the mapping layer maps the decoded feature map of the slice output by the decoding module to the aorta region pixel by pixel, thereby obtaining the aorta region segmentation result of the slice; The second segmentation network includes: a feature extraction module, a semantic extractor, a spatial feature extraction module, an attention module, a feature fusion module and a segmentation head; The feature extraction module is used to extract the features of the slice image step by step to obtain N feature images with decreasing scales from large to small. ; N ≥ 3; The semantic extractor includes a cascaded channel attention module and a spatial attention module for Splice them together and perform semantic extraction based on the channel attention mechanism and the spatial attention mechanism to obtain global semantic encoding information; the global semantic encoding information includes: The corresponding semantic encoding information ; The spatial feature extraction module is used to extract the spatial features of the slice image f ; The feature fusion module includes N semantic injection units; No. j A semantic injection unit is used to Perform upsampling to make it consistent with After the same size as Based on the attention mechanism, we can get j Decoding features ; ; The N-1th semantic injection unit is used to Perform upsampling to make it consistent with After the same size as The features extracted by the attention module are fused based on the attention mechanism to obtain the N-1th decoding feature ; The Nth semantic injection unit is used to Perform upsampling to make it consistent with f After the same size as f Fusion based on the attention mechanism to obtain the Nth decoding feature ; The segmentation header is used based on The spliced ​​features are used to segment the inner valve area in the slice image.

2. The aortic dissection detection method according to claim 1, characterized in that: The aorta categories include: aortic arch, ascending aorta, and descending aorta; Count the number of slice images containing the inner valve area obtained by segmentation under the aortic arch, ascending aorta and descending aorta categories respectively 、 and , and then calculate the ascending and descending aortic dissection scores; Among them, ascending and descending aortic dissection scores 、 They are: is the preset threshold.

3. The aortic dissection detection method according to claim 1, characterized in that: The first segmentation network is a pre-trained segmentation network, and its training process includes: Inputting a first training set into the first segmentation network, and training the first segmentation network by minimizing a first loss function; The first training set includes: CTA images of normal subjects and CTA images of patients with different types of aortic dissection according to the DeBakey classification; wherein the labels carried by each slice of the CTA image include: the true segmentation result of the aorta region in the slice image and the corresponding aorta category; The first loss function includes: a difference loss between the aorta region segmentation result of each slice in the CTA image in the first training set and the corresponding true segmentation result.

4. The method for detecting aortic dissection according to any one of claims 1 to 3, wherein: The spatial feature extraction module includes: multiple cascaded CNN modules and a spatial attention module connected to the output end of the last-level CNN module.

5. The aortic dissection detection method according to any one of claims 1 to 3, characterized in that: The second segmentation network is a pre-trained segmentation network, and its training process includes: Inputting the second training set into the second segmentation network, and training the second segmentation network by minimizing the second loss function; The second training set includes: slice images of aorta regions of different aorta categories; the labels carried by each slice image include: the true segmentation result and the true boundary line image of the inner valve region in the slice image; During the training process, the second segmentation network also includes a boundary line detection module for Extracting the boundary line of the inner valve region in the slice image to obtain a boundary line image of the inner valve region in the slice image; The second loss function includes: the difference loss between the inner valve area segmentation result of the slice image in the second training set obtained by the second segmentation network and the corresponding true segmentation result, and the difference loss between the boundary line image obtained by the boundary line detection module and the corresponding true boundary line image.

6. An aortic dissection detection system, characterized in that: include: A memory and a processor, wherein the memory stores a computer program, and the processor executes the method according to any one of claims 1 to 5 when executing the computer program.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed by a processor, the device where the storage medium is located is controlled to execute the method according to any one of claims 1 to 5.

Images