A method, device and storage medium for reconstructing blood vessel images
The target image generation model generates enhanced vascular images, combined with feature image segmentation, solves the problem of difficulty in extracting vascular areas in the case of vascular obstruction, and achieves complete vascular reconstruction and clear display of blocked areas, providing better diagnostic support.
Patent Information
- Application Number
- CN202410805819.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-21
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-06-21
AI Technical Summary
The existing CTA vascular reconstruction technology cannot accurately extract the vascular area under vascular obstruction, resulting in blurred vascular structure and unable to generate complete vascular reconstruction images.
Through the first feature image corresponding to the target level, the target image generation model is used to generate an enhanced blood vessel image, and combined with the first feature image, the segmentation of the blocked blood vessel area is achieved, thereby obtaining a complete blood vessel reconstruction result.
The vascular reconstruction in the case of vascular obstruction is achieved, which can clearly display the bones and tissues surrounding the blood vessels, providing better diagnostic help.
Smart Images

Figure CN118397127B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image processing technology, and in particular to a blood vessel image reconstruction method, device and storage medium. Background Art
[0002] CTA (CT angiography) vascular reconstruction technology refers to the use of computer imaging technology to process CTA images to generate three-dimensional images of blood vessels to display the vascular structure and morphology.
[0003] In the prior art, image features in CTA images can be extracted through deep learning methods, and vascular regions can be segmented based on the obtained image features, and then vascular reconstruction images can be generated through image processing technology. However, the above method relies on the attenuation difference between the vascular segment and the surrounding tissue in the CTA image. Specifically, during CTA angiography, it is necessary to intravenously inject iodine-containing contrast agents and perform CT scans when the contrast agent concentration in the target blood vessels is optimal, so as to obtain ideal vascular enhancement images. However, some patients may have no attenuation difference between some vascular segments and surrounding tissues due to reasons such as vascular obstruction. At this time, the above vascular reconstruction technology cannot accurately extract the vascular area, and the vascular structure may be blurred in the CTA image. Summary of the invention
[0004] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.
[0005] To this end, the present invention proposes a vascular image reconstruction method, which generates a corresponding enhanced vascular image based on a first feature image corresponding to the target layer through a target image generation model, and obtains a second vascular segmentation result of the blocked vascular area based on the enhanced vascular image and the first feature image, thereby realizing vascular reconstruction of the blocked vascular area, thereby obtaining a complete vascular reconstruction result, and then obtaining the bones and tissues around the blocked blood vessels, which can provide better help to users.
[0006] Another object of the present invention is to provide a blood vessel image reconstruction device.
[0007] To achieve the above object, the present invention provides a blood vessel image reconstruction method, the method comprising:
[0008] Acquire a head and neck CTA image to be processed, and input the head and neck CTA image into a target convolutional neural network to obtain a first blood vessel segmentation result and a plurality of first feature images, wherein each CTA sequence in the head and neck CTA image corresponds to a plurality of first feature images at different levels, and the first feature image is a single-layer low-dimensional image;
[0009] Performing vascular path extraction on the first vascular segmentation result to determine the target layer where the blocked vascular region is located;
[0010] Inputting the first characteristic image corresponding to the target layer into the target image generation model to generate a corresponding enhanced blood vessel image, wherein the first characteristic image and the corresponding enhanced blood vessel image maintain a corresponding relationship at the same layer;
[0011] Obtaining a second blood vessel segmentation result of the blocked blood vessel region based on the enhanced blood vessel image and the first characteristic image;
[0012] A head and neck blood vessel reconstruction result is obtained based on the first blood vessel segmentation result and the second blood vessel segmentation result.
[0013] The blood vessel image reconstruction method of the embodiment of the present invention may also have the following additional technical features:
[0014] In one embodiment of the present invention, the extracting the blood vessel path from the first blood vessel segmentation result to determine the target layer where the blocked blood vessel region is located includes:
[0015] Performing vascular path extraction on the first vascular segmentation result to determine vascular segments whose complete paths cannot be extracted;
[0016] Determine the layer numbers at both ends of the breakpoint corresponding to the blood vessel segment;
[0017] The middle level between the level numbers at the two ends is determined as the target level where the blocked blood vessel area is located.
[0018] In one embodiment of the present invention, before inputting the first characteristic image corresponding to the target layer into the target image generation model to generate the simulated enhanced blood vessel image, the method further includes:
[0019] Determine the initial image generation model;
[0020] The initial image generation model is trained to obtain the target image generation model.
[0021] In one embodiment of the present invention, the initial image generation model includes a generator and a discriminator; the training of the initial image generation model to obtain the target image generation model includes:
[0022] Acquire a data set for training, wherein the third feature image in the data set has the same dimension as the first feature image;
[0023] Inputting the third characteristic image into the generator to obtain a simulated enhanced blood vessel image;
[0024] The discriminator discriminates the simulated enhanced blood vessel image to obtain a discrimination result;
[0025] The parameters in the generator and the discriminator are adjusted based on the discrimination result, the simulated enhanced vascular image and the real vascular image corresponding to the third feature image, and the above steps are repeated until the initial image generation model converges or the iteration reaches a preset number of times to obtain the target image generation model.
[0026] In one embodiment of the present invention, obtaining a second blood vessel segmentation result of the blocked blood vessel region based on the enhanced blood vessel image and the first feature image includes:
[0027] performing difference processing on the enhanced blood vessel image and the first characteristic image, and determining the obtained corresponding difference image as the second characteristic image corresponding to the blocked blood vessel region;
[0028] A second blood vessel segmentation result of the blocked blood vessel region is obtained based on the second feature image and the enhanced blood vessel image.
[0029] In one embodiment of the present invention, obtaining a second blood vessel segmentation result of the blocked blood vessel region based on the second feature image and the enhanced blood vessel image includes:
[0030] Inputting the second characteristic image and the enhanced blood vessel image into a blood vessel segmentation model to obtain a third characteristic image, wherein the first characteristic image and the enhanced blood vessel image maintain a corresponding relationship at the same level;
[0031] All third characteristic images corresponding to the target layer are merged to obtain a second blood vessel segmentation result of the blocked blood vessel area.
[0032] In one embodiment of the present invention, the method further comprises:
[0033] The center line of the reconstructed blocked blood vessel region in the head and neck blood vessel segmentation result is extracted, and the center line is smoothed by a B-spline curve algorithm to obtain the blocked blood vessel path.
[0034] To achieve the above object, the present invention provides a blood vessel image reconstruction device on the other hand, the device comprising:
[0035] A first processing module is used to obtain a head and neck CTA image to be processed, and input the head and neck CTA image into a target convolutional neural network to obtain a first blood vessel segmentation result and a plurality of first feature images, wherein each CTA sequence in the head and neck CTA image corresponds to a plurality of first feature images at different levels, and the first feature image is a single-layer low-dimensional image;
[0036] a determination module, configured to extract a blood vessel path from the first blood vessel segmentation result and determine a target layer where a blocked blood vessel region is located;
[0037] A second processing module is used to input the first characteristic image corresponding to the target layer into the target image generation model to generate a corresponding enhanced blood vessel image, wherein the first characteristic image and the corresponding enhanced blood vessel image maintain a corresponding relationship at the same layer;
[0038] A third processing module, configured to obtain a second blood vessel segmentation result of the blocked blood vessel region based on the enhanced blood vessel image and the first feature image;
[0039] The fourth processing module is used to obtain a head and neck blood vessel reconstruction result based on the first blood vessel segmentation result and the second blood vessel segmentation result.
[0040] The vascular image reconstruction method and device of the embodiment of the present invention generates a corresponding enhanced vascular image through a target image generation model based on the first feature image corresponding to the target layer, and obtains a second vascular segmentation result of the blocked vascular area based on the enhanced vascular image and the first feature image, thereby realizing vascular reconstruction of the blocked vascular area, thereby obtaining a complete vascular reconstruction result, and further obtaining bones and tissues around the blocked blood vessels, which can provide better help to users.
[0041] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:
[0043] Figure 1 is a flow chart of a blood vessel image reconstruction method according to an embodiment of the present invention;
[0044] Figure 2 is a structural diagram of a blood vessel image reconstruction device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0045] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0046] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0047] The following describes a blood vessel image reconstruction method and device according to an embodiment of the present invention with reference to the accompanying drawings.
[0048] Figure 1 is a flow chart of a blood vessel image reconstruction method according to an embodiment of the present invention.
[0049] like Figure 1 As shown, the method includes:
[0050] S1, obtaining a head and neck CTA image to be processed, and inputting the head and neck CTA image into a target convolutional neural network to obtain a first blood vessel segmentation result and a plurality of first feature images.
[0051] In one embodiment of the present invention, the target convolutional neural network includes an encoder and a decoder, and the decoder includes an upsampling layer and a convolutional layer. In one embodiment of the present invention, the target convolutional neural network is trained by using a single-layer 2D CTA image with head and neck blood vessel annotations as training data, and using cross entropy loss and Dice coefficient loss as loss functions.
[0052] Furthermore, in one embodiment of the present invention, after the above-mentioned head and neck CTA images to be processed are input into the target convolutional neural network, the target neural network can divide each CTA sequence in the head and neck CTA images into multiple (such as hundreds of) first feature images, and perform vascular segmentation on each first feature image, and merge the vascular segmentation results to obtain the first vascular segmentation result.
[0053] In one embodiment of the present invention, the first characteristic image may be a single-layer low-dimensional image, such as a single-layer 2D image.
[0054] S2, extracting the blood vessel path from the first blood vessel segmentation result to determine the target layer where the blocked blood vessel region is located.
[0055] In one embodiment of the present invention, after obtaining the first blood vessel segmentation result through the above steps, it is necessary to perform blood vessel path extraction on the first blood vessel segmentation result, so as to determine the target layer where the blocked blood vessel region is located.
[0056] In one embodiment of the present invention, the method of extracting the blood vessel path from the first blood vessel segmentation result and determining the target layer where the blocked blood vessel region is located may include the following steps:
[0057] S21, performing vascular path extraction on the first vascular segmentation result, and determining vascular segments whose complete paths cannot be extracted;
[0058] S22, determine the layer numbers at both ends of the vascular segment corresponding to the breakpoint;
[0059] S23, determining the middle level of the level numbers at both ends as the target level where the blocked blood vessel area is located.
[0060] For example, in one embodiment of the present invention, it is assumed that when extracting the vascular path from the first vascular segmentation result, it is determined that the image between the 200th layer and the 250th layer cannot extract a complete vascular path. At this time, it is determined that the left vertebral artery corresponding to the image between the 200th layer and the 250th layer is a blocked vascular segment, and the 200th layer and the 250th layer are the layer numbers at both ends of the break point, respectively.
[0061] In one embodiment of the present invention, the blood vessel segment from which the complete path cannot be extracted may be a blocked blood vessel region.
[0062] S3, inputting the first feature image corresponding to the target layer into the target image generation model to generate a corresponding enhanced blood vessel image.
[0063] In one embodiment of the present invention, before inputting the first characteristic image corresponding to the target layer into the target image generation model to generate the simulated enhanced blood vessel image, the method may further include the following steps:
[0064] S31, determining an initial image generation model;
[0065] S32, training the initial image generation model to obtain a target image generation model.
[0066] In one embodiment of the present invention, the initial image generation model may include a generator and a discriminator, wherein the discriminator may be a binary classification network to distinguish between real data and fake data generated by the generator.
[0067] Furthermore, in one embodiment of the present invention, the method of training the initial image generation model to obtain the target image generation model may include the following steps:
[0068] Step 1: Obtain a data set for training, wherein the third feature image in the data set has the same dimension as the first feature image;
[0069] Step 2, inputting the third feature image into the generator to obtain a simulated enhanced blood vessel image;
[0070] Step 3: The discriminator discriminates the simulated enhanced blood vessel image to obtain a discrimination result;
[0071] Step 4: Adjust the parameters in the generator and the discriminator based on the discrimination result, the simulated enhanced vascular image, and the real vascular image corresponding to the third feature image, and repeat the above steps until the initial image generation model converges or the iteration reaches a preset number of times to obtain the target image generation model.
[0072] In one embodiment of the present invention, the generator may include an encoder module and a decoder module composed of convolutional layers. Also, in one embodiment of the present invention, after the third feature map is input into the generator, a simulated enhanced blood vessel image similar to the real data distribution is generated through the encoder module and the decoder module.
[0073] Among them, in one embodiment of the present invention, in the above-mentioned training process, based on the discrimination results, the simulated enhanced vascular images and the real vascular images corresponding to the third feature images, the parameters of the generator and the discriminator are continuously optimized using a cross entropy loss function, so that the generator can generate a more difficult to distinguish simulated enhanced vascular image, and the discriminator can distinguish the real results from the simulated results more accurately.
[0074] And, in one embodiment of the present invention, after the first feature image corresponding to the target layer is input into the target image generation model, a simulated enhanced blood vessel image corresponding to the first feature image is generated by a generator in the target image generation model. In one embodiment of the present invention, the first feature image and the corresponding enhanced blood vessel image maintain a corresponding relationship at the same layer.
[0075] S4, obtaining a second blood vessel segmentation result of the blocked blood vessel region based on the enhanced blood vessel image and the first feature image.
[0076] In one embodiment of the present invention, after the enhanced blood vessel image and the first characteristic image are obtained through the above steps, a second blood vessel segmentation result of the blocked blood vessel region can be obtained based on the enhanced blood vessel image and the first characteristic image.
[0077] Specifically, in one embodiment of the present invention, the method for obtaining a second blood vessel segmentation result of a blocked blood vessel region based on the enhanced blood vessel image and the first feature image may include the following steps:
[0078] S41, performing difference processing on the enhanced blood vessel image and the first characteristic image, and determining the corresponding difference image as the second characteristic image corresponding to the blocked blood vessel area;
[0079] S42, obtaining a second blood vessel segmentation result of the blocked blood vessel region based on the second feature image and the enhanced blood vessel image.
[0080] Among them, in one embodiment of the present invention, the method for obtaining the second blood vessel segmentation result of the blocked blood vessel region based on the second characteristic image and the enhanced blood vessel image may specifically include: inputting the second characteristic image and the enhanced blood vessel image into the blood vessel segmentation model to obtain the second blood vessel segmentation result of the blocked blood vessel region. And, in one embodiment of the present invention, after the second characteristic image and the enhanced blood vessel image are input into the blood vessel segmentation model, the blood vessel segmentation model may obtain a single-layer 2-dimensional third characteristic map based on the second characteristic image and the enhanced blood vessel image, and merge the third characteristic map to obtain a 3-dimensional second blood vessel segmentation result of the blocked blood vessel region.
[0081] S5, obtaining a head and neck blood vessel reconstruction result based on the first blood vessel segmentation result and the second blood vessel segmentation result.
[0082] In one embodiment of the present invention, after obtaining the first blood vessel segmentation result and the second blood vessel segmentation result through the above steps, the first blood vessel segmentation result and the second blood vessel segmentation result can be merged to obtain a complete head and neck blood vessel reconstruction result.
[0083] Furthermore, in one embodiment of the present invention, after obtaining the head and neck vascular reconstruction results through the above steps, the center line of the reconstructed blocked blood vessel area in the head and neck vascular segmentation results can be extracted, and the center line can be smoothed by a B-spline curve algorithm to obtain the blocked blood vessel path, thereby obtaining a complete blood vessel reconstruction result, and then obtaining the bones and tissues around the blocked blood vessels, which can provide better assistance to users.
[0084] The embodiment of the present invention proposes a vascular image reconstruction method, which obtains a head and neck CTA image to be processed, inputs the head and neck CTA image into a target convolutional neural network to obtain a first vascular segmentation result and multiple first feature images, wherein each CTA sequence in the head and neck CTA image corresponds to multiple first feature images of different layers, and the first feature image is a single-layer low-dimensional image; performs vascular path extraction on the first vascular segmentation result to determine the target layer where the blocked vascular region is located; inputs the first feature image corresponding to the target layer into a target image generation model to generate a corresponding enhanced vascular image, wherein the first feature image and the corresponding enhanced vascular image maintain a corresponding relationship at the same layer; based on the enhanced vascular image and the first feature image, obtains a second vascular segmentation result of the blocked vascular region; based on the first vascular segmentation result and the second vascular segmentation result, obtains a head and neck vascular reconstruction result. Therefore, the present invention generates a corresponding enhanced vascular image through a target image generation model based on the first feature image corresponding to the target layer, and obtains a second vascular segmentation result of the blocked vascular region based on the enhanced vascular image and the first feature image, thereby realizing vascular reconstruction of the blocked vascular region, thereby obtaining a complete vascular reconstruction result, and then obtaining bones and tissues around the blocked blood vessels, which can provide better help for users.
[0085] In order to implement the above embodiment, Figure 2 As shown, this embodiment also provides a blood vessel image reconstruction device 10, which includes a first processing module 100, a determination module 200, a second processing module 300, a third processing module 400 and a fourth processing module 500;
[0086] The first processing module 100 is used to obtain a head and neck CTA image to be processed, and input the head and neck CTA image into a target convolutional neural network to obtain a first blood vessel segmentation result and a plurality of first feature images, wherein each CTA sequence in the head and neck CTA image corresponds to a plurality of first feature images at different levels, and the first feature image is a single-layer low-dimensional image;
[0087] A determination module 200 is used to extract the blood vessel path from the first blood vessel segmentation result and determine the target layer where the blocked blood vessel region is located;
[0088] The second processing module 300 is used to input the first characteristic image corresponding to the target layer into the target image generation model to generate a corresponding enhanced blood vessel image, wherein the first characteristic image and the corresponding enhanced blood vessel image maintain a corresponding relationship at the same layer;
[0089] The third processing module 400 is used to obtain a second blood vessel segmentation result of the blocked blood vessel region based on the enhanced blood vessel image and the first feature image;
[0090] The fourth processing module 500 is used to obtain a head and neck blood vessel reconstruction result based on the first blood vessel segmentation result and the second blood vessel segmentation result.
[0091] Furthermore, the determination module 200 is specifically configured to:
[0092] Performing vascular path extraction on the first vascular segmentation result to determine vascular segments whose complete paths cannot be extracted;
[0093] Determine the layer numbers at both ends of the vascular segment corresponding to the breakpoint;
[0094] The middle level between the level numbers at both ends is determined as the target level where the blocked blood vessel area is located.
[0095] Furthermore, the above device is also used for:
[0096] Determine the initial image generation model;
[0097] The initial image generation model is trained to obtain the target image generation model.
[0098] In one embodiment of the present invention, the initial image generation model includes a generator and a discriminator; the device is specifically used for:
[0099] Acquire a data set for training, wherein the third feature image in the data set has the same dimension as the first feature image;
[0100] Inputting the third feature image into the generator to obtain a simulated enhanced blood vessel image;
[0101] The discriminator discriminates the simulated enhanced blood vessel image to obtain a discrimination result;
[0102] Based on the discrimination results, the simulated enhanced vascular image and the real vascular image corresponding to the third feature image, the parameters in the generator and the discriminator are adjusted, and the above steps are repeated until the initial image generation model converges or the iteration reaches a preset number of times to obtain the target image generation model.
[0103] In one embodiment of the present invention, the third processing module 400 is specifically used for:
[0104] Performing difference processing on the enhanced blood vessel image and the first characteristic image, and determining the corresponding difference image as the second characteristic image corresponding to the blocked blood vessel area;
[0105] Based on the second characteristic image and the enhanced blood vessel image, a second blood vessel segmentation result of the blocked blood vessel area is obtained.
[0106] In one embodiment of the present invention, the third processing module 400 is further used for:
[0107] The second characteristic image and the enhanced blood vessel image are input into the blood vessel segmentation model to obtain a second blood vessel segmentation result of the blocked blood vessel area.
[0108] In one embodiment of the present invention, the above device is further used for:
[0109] The centerline of the reconstructed blocked blood vessel area in the head and neck blood vessel segmentation result is extracted, and the blocked blood vessel path is obtained by smoothing the centerline using the B-spline curve algorithm.
[0110] According to the vascular image reconstruction device of the embodiment of the present invention, a head and neck CTA image to be processed is obtained, and the head and neck CTA image is input into the target convolutional neural network to obtain a first vascular segmentation result and multiple first feature images, wherein each CTA sequence in the head and neck CTA image corresponds to multiple first feature images of different layers, and the first feature image is a single-layer low-dimensional image; the vascular path is extracted from the first vascular segmentation result to determine the target layer where the blocked vascular area is located; the first feature image corresponding to the target layer is input into the target image generation model to generate a corresponding enhanced vascular image, wherein the first feature image and the corresponding enhanced vascular image maintain a corresponding relationship at the same layer; based on the enhanced vascular image and the first feature image, a second vascular segmentation result of the blocked vascular area is obtained; based on the first vascular segmentation result and the second vascular segmentation result, a head and neck vascular reconstruction result is obtained. Therefore, the present invention generates a corresponding enhanced vascular image through the target image generation model based on the first feature image corresponding to the target layer, and obtains a second vascular segmentation result of the blocked vascular area based on the enhanced vascular image and the first feature image, thereby realizing vascular reconstruction of the blocked vascular area, thereby obtaining a complete vascular reconstruction result, and then obtaining the bones and tissues around the blocked blood vessels, which can provide better help for users.
[0111] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0112] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
Claims
1. A blood vessel image reconstruction method, characterized in that: The method comprises: Acquire a head and neck CTA image to be processed, and input the head and neck CTA image into a target convolutional neural network to obtain a first blood vessel segmentation result and a plurality of first feature images, wherein each CTA sequence in the head and neck CTA image corresponds to a plurality of first feature images at different levels, and the first feature image is a single-layer low-dimensional image; Performing vascular path extraction on the first vascular segmentation result to determine the target layer where the blocked vascular region is located, wherein the vascular path extraction is performed on the first vascular segmentation result to determine the vascular segments from which the complete path cannot be extracted; determining the layer numbers at both ends of the breakpoint corresponding to the vascular segments; and determining the middle layer between the layer numbers at both ends as the target layer where the blocked vascular region is located; Inputting the first characteristic image corresponding to the target layer into the target image generation model to generate a corresponding enhanced blood vessel image, wherein the first characteristic image and the corresponding enhanced blood vessel image maintain a corresponding relationship at the same layer; Based on the enhanced blood vessel image and the first characteristic image, a second blood vessel segmentation result of the blocked blood vessel region is obtained, wherein the enhanced blood vessel image and the first characteristic image are subjected to difference processing, and the corresponding difference image is determined as the second characteristic image corresponding to the blocked blood vessel region; the second characteristic image and the enhanced blood vessel image are input into a blood vessel segmentation model to obtain a second blood vessel segmentation result of the blocked blood vessel region; A head and neck blood vessel reconstruction result is obtained based on the first blood vessel segmentation result and the second blood vessel segmentation result.
2. The method according to claim 1, characterized in that Before inputting the first characteristic image corresponding to the target layer into the target image generation model to generate a simulated enhanced blood vessel image, the method further includes: Determine the initial image generation model; The initial image generation model is trained to obtain the target image generation model.
3. The method according to claim 2, characterized in that The initial image generation model includes a generator and a discriminator; the training of the initial image generation model to obtain the target image generation model includes: Acquire a data set for training, wherein the third feature image in the data set has the same dimension as the first feature image; Inputting the third characteristic image into the generator to obtain a simulated enhanced blood vessel image; The discriminator discriminates the simulated enhanced blood vessel image to obtain a discrimination result; The parameters in the generator and the discriminator are adjusted based on the discrimination result, the simulated enhanced vascular image and the real vascular image corresponding to the third feature image, and the above steps are repeated until the initial image generation model converges or the iteration reaches a preset number of times to obtain the target image generation model.
4. The method according to claim 1, characterized in that The method further comprises: The center line of the reconstructed blocked blood vessel region in the head and neck blood vessel segmentation result is extracted, and the center line is smoothed by a B-spline curve algorithm to obtain the blocked blood vessel path.
5. A blood vessel image reconstruction device, characterized in that: The device comprises: A first processing module is used to obtain a head and neck CTA image to be processed, and input the head and neck CTA image into a target convolutional neural network to obtain a first blood vessel segmentation result and a plurality of first feature images, wherein each CTA sequence in the head and neck CTA image corresponds to a plurality of first feature images at different levels, and the first feature image is a single-layer low-dimensional image; a determination module, configured to perform vascular path extraction on the first vascular segmentation result to determine a target layer where the blocked vascular region is located, wherein the vascular path extraction is performed on the first vascular segmentation result to determine vascular segments from which a complete path cannot be extracted; determine layer numbers at both ends of the breakpoint corresponding to the vascular segments; and determine the middle layer between the layer numbers at both ends as the target layer where the blocked vascular region is located; A second processing module is used to input the first characteristic image corresponding to the target layer into the target image generation model to generate a corresponding enhanced blood vessel image, wherein the first characteristic image and the corresponding enhanced blood vessel image maintain a corresponding relationship at the same layer; A third processing module is configured to obtain a second blood vessel segmentation result of the blocked blood vessel region based on the enhanced blood vessel image and the first characteristic image, wherein the enhanced blood vessel image and the first characteristic image are subjected to difference processing, and the corresponding difference image is determined as the second characteristic image corresponding to the blocked blood vessel region; the second characteristic image and the enhanced blood vessel image are input into a blood vessel segmentation model to obtain a second blood vessel segmentation result of the blocked blood vessel region; The fourth processing module is used to obtain a head and neck blood vessel reconstruction result based on the first blood vessel segmentation result and the second blood vessel segmentation result.
6. An electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 4.
7. A computer storage medium, wherein: The computer storage medium stores computer executable instructions; after the computer executable instructions are executed by the processor, the method according to any one of claims 1 to 4 can be implemented.
