Image enhancement method, device, module, system and medium for cardiac interventional operation
Through a semantic-driven dual-pixel image defocusing defuzzing network, combined with the CLIP model, the problems of low image clarity and high computational complexity in cardiac interventional surgery are solved, and high-quality and real-time image defuzzing is achieved to adapt to the heart. Complex scenarios of interventional surgery.
Patent Information
- Application Number
- CN202411209527.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has problems such as motion blur and defocus blur in X-ray image processing in cardiac interventional surgery, resulting in low image clarity, high computational complexity, insufficient robustness, and difficult to meet the real-time requirements.
The semantic-driven double-pixel image defocusing defuzzing network is used to estimate the fuzzy map through unsupervised learning, and combined with the semantic understanding ability of the CLIP model and the geometric information of the dual-pixel image, image stitching, feature extraction, defuzzing and fusion are carried out. Generate high-quality coronary angiography images.
Effectively remove defocus blur of X-ray images in cardiac interventional surgery, improve image clarity, simplify algorithm processes, meet real-time requirements, have strong robustness, and adapt to interference factors such as heartbeat and respiratory movement.
Smart Images

Figure CN119090772A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of medical image processing, and in particular to a real-time image enhancement method, device, system and computer-readable storage medium for cardiac intervention surgery. Background Art
[0002] Cardiac intervention is an important means of treating cardiovascular diseases such as coronary heart disease and arrhythmia. During the operation, doctors need to rely on X-ray angiography images to observe the internal conditions of blood vessels and guide the operation of devices such as catheters and stents. However, due to the influence of factors such as heart beating and respiratory movement, X-ray angiography images often have problems such as motion blur and defocus blur, which affect doctors' observation of lesions and the accuracy of surgical operations.
[0003] At present, some image deblurring algorithms have been applied to cardiac interventional surgery image enhancement, but most of these methods have the following problems:
[0004] (1) The blur kernel needs to be acquired in advance: Traditional deconvolution methods usually require estimating the blur kernel of the image, which is difficult to obtain accurately in practical applications, especially for complex dynamic scenes such as cardiac interventional surgery.
[0005] (2) Reliance on complex image registration: Some methods eliminate motion blur by registering multiple frames of images, but image registration itself is a complex problem and can easily introduce new errors.
[0006] (3) High computational complexity: Although many advanced deblurring algorithms have good effects, they are computationally intensive and time-consuming, making it difficult to meet the real-time requirements during surgery.
[0007] (4) Lack of specificity: General image deblurring algorithms do not take into account the characteristics of cardiac interventional surgery images, such as vascular structure and instruments, and the deblurring effect is often not ideal.
[0008] (5) Lack of robustness: Existing methods are not adaptable enough to situations such as different degrees of blur and images of different patients, making it difficult to maintain stable results in practical applications.
[0009] Therefore, a new method that can achieve real-time, high-quality deblurring of X-ray images during cardiac interventional surgery is urgently needed to improve the safety and accuracy of the surgery. Summary of the invention
[0010] The embodiment of the present application provides an image enhancement method for cardiac interventional surgery, which can effectively remove defocus blur in cardiac interventional surgery images and improve image clarity, so as to assist doctors in performing accurate diagnosis and surgical operations.
[0011] In a first aspect, an embodiment of the present application provides an image enhancement method for cardiac intervention surgery, the method comprising the following steps:
[0012] Obtain two images of the left and right views of the coronary angiography;
[0013] Splicing the two images of the left and right perspectives after preprocessing, extracting a feature map of the spliced image, and generating a blur map based on the feature map;
[0014] Deblurring the blurred image;
[0015] The deblurred left and right view images are fused to generate the final coronary angiography image.
[0016] Preferably, the two images of left and right perspectives of coronary angiography are X-ray images of two perspectives obtained by a dual-source X-ray machine performing angiography on the patient's coronary arteries from two directions, left anterior oblique and right anterior oblique, respectively.
[0017] Preferably, the preprocessing and splicing of the two images of the left and right perspectives, extracting a feature map of the spliced image, and generating a blur map based on the feature map specifically includes:
[0018] Image preprocessing: normalizing the two images of the left and right perspectives, scaling the pixel values to the range of [0, 1];
[0019] Image stitching: Flip the preprocessed left-view image vertically, and then stitch it to the right side of the preprocessed right-view image to form a stitched image;
[0020] Feature map extraction: based on the spliced image, extract its semantic features to form a feature map;
[0021] Fuzzy image generation: Using the extracted feature map, a fuzzy image is generated through a convolutional neural network.
[0022] Preferably, the step of fusing the deblurred left and right view images to generate a final coronary angiography image specifically includes:
[0023] Image alignment: Accurately align two images from left and right perspectives;
[0024] Weight calculation: Based on the blur map, the fusion weight is calculated for each pixel position, where the area with lower blur is given a higher fusion weight;
[0025] Weighted average: Use the calculated fusion weights to perform pixel-level weighted average of the left and right view images to obtain the fused image.
[0026] Furthermore, the step of fusing the left and right view images after deblurring to generate a final coronary angiography image further includes:
[0027] Post-processing: Edge enhancement and contrast adjustment are performed on the fused image to further improve image quality.
[0028] In a second aspect, an embodiment of the present application provides an image enhancement device for cardiac intervention surgery, comprising:
[0029] An image acquisition module, used for acquiring two images of left and right perspectives of coronary angiography;
[0030] A blur map estimation module is used to splice the two images of the left and right perspectives after preprocessing, extract a feature map of the spliced image, and generate a blur map based on the feature map;
[0031] Defuzzification network module: used for defuzzifying the fuzzy graph;
[0032] Image fusion module: used to fuse the left and right view images after deblurring to generate the final coronary angiography image.
[0033] Preferably, the deblurring network module specifically includes:
[0034] The encoder is used to gradually reduce the spatial size of the feature map while increasing the number of channels to extract multi-scale features of the image;
[0035] The decoder is used to gradually restore the spatial size of the feature map and reconstruct a clear image;
[0036] A jump connector connects the encoder and the decoder.
[0037] In a third aspect, an embodiment of the present application provides an image enhancement module for cardiac interventional surgery, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned image enhancement method for cardiac interventional surgery when executing the program.
[0038] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned image enhancement method for cardiac interventional surgery.
[0039] In a fifth aspect, an embodiment of the present application provides an image enhancement system for cardiac intervention surgery, comprising:
[0040] A dual-pixel image acquisition module, including a dual-source X-ray machine for simultaneously capturing left and right perspective coronary angiography images;
[0041] The above-mentioned image enhancement module for cardiac intervention surgery is connected to the dual-pixel image acquisition module and is used to enhance the coronary angiography images of the left and right viewing angles;
[0042] The display module is connected to the image enhancement module for cardiac intervention surgery and is used to display the processed image on the screen in real time.
[0043] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:
[0044] 1. Dual-pixel processing: The present invention adopts a semantically driven dual-pixel image defocus deblurring network, which can effectively remove the defocus blur of X-ray angiography images in cardiac interventional surgery and improve the clarity of the image, so as to assist doctors in accurate diagnosis and surgical operations, thereby improving the safety and accuracy of the surgery;
[0045] 2. Unsupervised learning: The present invention uses an unsupervised learning method to estimate the blur map, without the need to obtain the blur kernel in advance or perform complex image registration, which simplifies the algorithm process and improves practicality;
[0046] 3. High-precision deblurring: The present invention combines the semantic understanding capability of the CLIP model with the geometric information of the dual-pixel image to estimate the blur map more accurately, thereby achieving a more precise deblurring effect.
[0047] 4. Strong real-time performance: The present invention adopts a lightweight defuzzification network, which can meet the real-time requirements of cardiac interventional surgery.
[0048] 5. Good robustness: The present invention has strong robustness against interference factors such as heart beat and respiratory movement, and can stably improve image clarity. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0050] Figure 1 This is a flow chart of the image enhancement method for cardiac intervention surgery provided in Example 1 of the present application;
[0051] Figure 2 This is a performance comparison diagram of the image enhancement method for cardiac interventional surgery provided in Example 1 of the present application and the traditional method (Baseline Method) in image denoising;
[0052] Figure 3 This is a structural block diagram of the image enhancement device for cardiac intervention surgery provided in Example 2 of the present application;
[0053] Figure 4 This is a structural block diagram of the fuzzy image estimation module in the second embodiment of the present application;
[0054] Figure 5 This is a structural block diagram of the defuzzification network module in the second embodiment of the present application;
[0055] Figure 6 This is a structural block diagram of the image fusion module in the second embodiment of the present application;
[0056] Figure 7 This is a structural block diagram of the image enhancement module for cardiac intervention surgery provided in Example 3 of the present application;
[0057] Figure 8 This is a structural block diagram of the image enhancement system for cardiac interventional surgery provided in Example 5 of the present application.
[0058] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0059] The embodiments of the present application provide an image enhancement method, device, module, system and computer-readable storage medium for cardiac interventional surgery, which can effectively remove defocus blur in cardiac interventional surgery images and improve image clarity, so as to assist doctors in performing accurate diagnosis and surgical operations.
[0060] In order to better understand the above technical solution, exemplary embodiments will be described in detail here, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are only examples of devices consistent with some aspects of the present application as detailed in the attached claims.
[0061] Embodiment 1
[0062] Figure 1 This is a flow chart of the image enhancement method for cardiac intervention surgery provided in Example 1 of the present application. The image enhancement method for cardiac intervention surgery comprises the following steps:
[0063] Step S1: acquiring two images of left and right perspectives of coronary angiography;
[0064] Specifically, a dual-source X-ray machine is used to simultaneously angiogram the patient's coronary arteries from the left anterior oblique (LAO) and right anterior oblique (RAO) directions to obtain X-ray images from two viewing angles. These two images have the same imaging time, but due to different viewing angles, they contain different spatial information of the coronary arteries.
[0065] Step S2: splicing the two images of the left and right perspectives after preprocessing, extracting a feature map of the spliced image, and generating a blur map based on the feature map.
[0066] The specific steps are as follows:
[0067] Step S21: Image preprocessing
[0068] The two images of the left and right perspectives are normalized to scale the pixel values to the range of [0, 1].
[0069] Step S22: Image stitching
[0070] The preprocessed left-view image is flipped vertically and then spliced to the right side of the preprocessed right-view image to form a spliced image with an aspect ratio of 2:1.
[0071] Step S23: Feature map extraction
[0072] Based on the spliced image, its semantic features are extracted to form a feature map.
[0073] In a preferred embodiment, the stitched image is input into a pre-trained CLIP model, and feature maps are extracted from the middle layer of the CLIP model, wherein the feature maps contain multi-scale information of the stitched image.
[0074] Among them, the CLIP (Contrastive Language-Image Pre-training) model is a multimodal pre-trained neural network model released by OpenAI in 2021, which is mainly used to learn the association between images and text. The CLIP model is pre-trained on a large-scale image-text pair dataset through contrastive learning, enabling the model to understand and match image content with text descriptions.
[0075] The architecture of the CLIP model consists of two main parts: an image encoder and a text encoder. The image encoder can be a convolutional neural network, such as ResNet, or a Vision Transformer (ViT), which is used to convert images into feature vectors; the text encoder is usually a Transformer model, which is used to convert text into feature vectors. The CLIP model is trained through contrastive learning, using a symmetric contrastive loss function, aiming to make the feature vectors of matching image-text pairs as close as possible, and the feature vectors of mismatched pairs as far apart as possible.
[0076] In this embodiment, high-level semantic features of the spliced image are extracted through the CLIP model.
[0077] Step S24: Fuzzy image generation
[0078] Using the extracted feature map, a small convolutional neural network is used to generate a blur map. The generated blur map reflects the blur degree of different areas in the image.
[0079] In a preferred embodiment, the convolutional neural network includes 3-4 convolutional layers and a ReLU activation function, and the last layer uses a Sigmoid activation function to map the output to a range of [0, 1].
[0080] Step S3: Deblurring the blurred image.
[0081] In a preferred embodiment, the blur image is deblurred using a deblurring network module, which adopts an encoder-decoder structure and introduces a blur priority attention block (BPA) to guide the deblurring network module to pay more attention to the areas with severe blurring.
[0082] The deblurring network module specifically includes:
[0083] The encoder includes a plurality of downsampling convolution blocks, each of which is composed of a convolution layer, a batch normalization layer, and a LeakyReLU activation function. The encoder gradually reduces the spatial size of the feature map while increasing the number of channels to extract multi-scale features of the image.
[0084] The decoder includes a plurality of upsampling convolution blocks, each of which is composed of a deconvolution layer, a batch normalization layer, and a ReLU activation function. The decoder gradually restores the spatial size of the feature map and reconstructs a clear image.
[0085] A skip connector connects the encoder and decoder to preserve high-resolution detail information.
[0086] Furthermore, a blur priority attention block (BPA) module is introduced at each scale of the encoder and decoder. The BPA module receives the feature map of the current scale and the blur map of the corresponding scale as input. The BPA module first fuses the blur map with the feature map, and then generates a spatial attention map and a channel attention map. These attention maps are used to modulate the feature map so that the deblurring network module pays more attention to the areas with severe blur.
[0087] Furthermore, the loss functions of both the encoder and the decoder use a combination of multiple loss functions, including:
[0088] L1 loss: measures the pixel-level difference between the reconstructed image and the real clear image.
[0089] Perceptual loss: Use the pre-trained VGG network to extract features and calculate the difference between the reconstructed image and the real image in the feature space.
[0090] Adversarial loss: Introduce a discriminator network to improve the authenticity of generated images through adversarial training.
[0091] Structural Similarity Loss: Use SSIM (Structural Similarity Index) as part of the loss function to preserve the structural information of the image.
[0092] Step S4: Fusing the left and right view images after deblurring to generate a final clear image of coronary angiography.
[0093] The specific steps are as follows:
[0094] Step S41: Image alignment
[0095] Since there may be a slight position deviation between the left and right view images, an image registration algorithm is first used to accurately align the two images.
[0096] Step S42: Weight calculation
[0097] Based on the blur map, a fusion weight is calculated for each pixel location. Areas with less blur are given higher weights.
[0098] Step S43: Weighted average
[0099] Using the calculated weights, the left and right view images are weighted averaged at the pixel level to obtain a fused image.
[0100] Step S44: Post-processing
[0101] The fused image is edge enhanced and contrast adjusted to further improve image quality.
[0102] In a specific clinical trial, we compared the performance of the traditional method (Baseline Method) and the proposed method for cardiac intervention surgery in image denoising. Figure 2 shown.
[0103] Figure 2 In the figure, the horizontal axis represents Epoch, which is the number of iterations of the model. The vertical axis represents PSNR (peak signal-to-noise ratio), which is the signal-to-noise ratio of the image after denoising. Figure 2 It can be seen that the image enhancement method for cardiac interventional surgery provided by this embodiment is always better than the traditional method in terms of PSNR, indicating that the image enhancement method for cardiac interventional surgery provided by this embodiment can more effectively remove image noise and obtain higher quality images. With the increase of Epoch, the PSNR value of the image enhancement method for cardiac interventional surgery provided by this embodiment continues to increase, while the PSNR value of the traditional method tends to be stable. This shows that the image enhancement method for cardiac interventional surgery provided by this embodiment has better convergence and stability.
[0104] In summary, the image enhancement method for cardiac interventional surgery provided in this embodiment has better performance and effect in image denoising.
[0105] Embodiment 2
[0106] Based on the same inventive concept as the image enhancement method for cardiac intervention surgery in the aforementioned embodiment 1, this embodiment also provides an image enhancement device for cardiac intervention surgery, such as Figure 3 As shown, the device comprises:
[0107] An image acquisition module 10 is used to acquire two images of left and right perspectives of coronary angiography;
[0108] A blur map estimation module 20 is used to splice the two images of the left and right perspectives after preprocessing, extract a feature map of the spliced image, and generate a blur map based on the feature map;
[0109] Defuzzification network module 30: used for defuzzifying the fuzzy graph;
[0110] Image fusion module 40: used for fusing the left and right view images after deblurring to generate a final coronary angiography image.
[0111] Specifically, the image acquisition module 10 acquires two X-ray images of the patient's coronary arteries from two viewing angles, the left anterior oblique (LAO) and the right anterior oblique (RAO). The two images have the same imaging time, but contain different spatial information of the coronary arteries due to different viewing angles.
[0112] Combination Figure 4 , the fuzzy map estimation module 20 comprises:
[0113] Image preprocessing module 201: used for normalizing the two images of the left and right perspectives, and scaling the pixel values to the range of [0, 1];
[0114] Image stitching module 202: used to flip the pre-processed left-view image in the vertical direction, and then stitch it on the right side of the pre-processed right-view image to form a stitched image with an aspect ratio of 2:1;
[0115] Feature map extraction module 203: used to extract semantic features based on the spliced image to form a feature map.
[0116] The fuzzy map generation module 204 is used to generate a fuzzy map by using the extracted feature map through a small convolutional neural network.
[0117] Combination Figure 5 , the deblurring network module 30 specifically includes:
[0118] The encoder 301 includes a plurality of downsampling convolution blocks, each of which is composed of a convolution layer, a batch normalization layer, and a LeakyReLU activation function. The encoder gradually reduces the spatial size of the feature map while increasing the number of channels to extract multi-scale features of the image.
[0119] Decoder 303, the decoder includes multiple upsampling convolution blocks, each of which is composed of a deconvolution layer, a batch normalization layer and a ReLU activation function. The decoder gradually restores the spatial size of the feature map and reconstructs a clear image.
[0120] A skip connector 302 connects the encoder and the decoder to preserve high-resolution detail information.
[0121] Furthermore, a blur priority attention block (BPA) module is introduced at each scale of the encoder and decoder. The BPA module receives the feature map of the current scale and the blur map of the corresponding scale as input. The BPA module first fuses the blur map with the feature map, and then generates a spatial attention map and a channel attention map. These attention maps are used to modulate the feature map so that the deblurring network module pays more attention to the areas with severe blur.
[0122] Furthermore, the loss functions of both the encoder and the decoder use a combination of multiple loss functions, including:
[0123] L1 loss: measures the pixel-level difference between the reconstructed image and the real clear image.
[0124] Perceptual loss: Use the pre-trained VGG network to extract features and calculate the difference between the reconstructed image and the real image in the feature space.
[0125] Adversarial loss: Introduce a discriminator network to improve the authenticity of generated images through adversarial training.
[0126] Structural Similarity Loss: Use SSIM (Structural Similarity Index) as part of the loss function to preserve the structural information of the image.
[0127] Combination Figure 6 , the image fusion module 40 includes:
[0128] The image alignment module 401 is used to accurately align two images using an image registration algorithm.
[0129] The weight calculation module 402 is used to calculate the fusion weight for each pixel position based on the blur map. A higher weight is assigned to a region with a lower blur level.
[0130] The weighted averaging module 403 is used to perform pixel-level weighted averaging on the left and right view images using the calculated weights to obtain a fused image.
[0131] The post-processing module 404 is used to perform edge enhancement and contrast adjustment on the fused image to further improve the image quality.
[0132] Embodiment 3
[0133] Based on the same inventive concept as the image enhancement method for cardiac intervention surgery in the aforementioned embodiment 1, this embodiment further provides an image enhancement module for cardiac intervention surgery, such as Figure 7 As shown, a computer program is stored thereon, and when the program is executed by a processor, the steps of the image enhancement method for cardiac intervention surgery described in Example 1 are implemented.
[0134] Among them, Figure 7 In the embodiment of the present invention, a bus architecture (represented by bus 500) is shown, which may include any number of interconnected buses and bridges, and bus 500 links various circuits including one or more processors represented by processor 502 and memory represented by memory 504. Bus 500 may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are not further described herein. Bus interface 506 provides an interface between bus 500 and receiver 501 and transmitter 503. Receiver 501 and transmitter 503 may be the same element, namely a transceiver, which provides a unit for communicating with various other devices over a transmission medium.
[0135] The processor 502 is responsible for managing the bus 500 and general processing, while the memory 504 may be used to store data used by the processor 502 when performing operations.
[0136] Embodiment 4
[0137] Based on the same inventive concept as the image enhancement method for cardiac interventional surgery in the aforementioned embodiment 1, this embodiment also provides a computer-readable storage medium on which a computer program is stored, which, when executed by a processor, implements the steps of the image enhancement method for cardiac interventional surgery described in embodiment 1.
[0138] Embodiment 5
[0139] Based on the same inventive concept as the image enhancement method for cardiac intervention surgery in the aforementioned embodiment 1, this embodiment also provides an image enhancement system for cardiac intervention surgery, such as Figure 8 As shown, the image enhancement system for cardiac intervention surgery includes:
[0140] The dual-pixel image acquisition module 60 includes a dual-source X-ray machine for simultaneously capturing coronary angiography images of left and right viewing angles, and an image acquisition device for acquiring the coronary angiography images of left and right viewing angles;
[0141] The real-time image enhancement module 61 is connected to the image acquisition device and is used to enhance the coronary angiography images of the left and right viewing angles. The real-time image enhancement module is specifically as described in the third embodiment.
[0142] The display module 62 is connected to the real-time image enhancement module 61 and is used to display the processed clear image on the screen in real time for doctors to view.
[0143] In a preferred embodiment, the image enhancement system for cardiac intervention surgery further includes: a storage module connected to the real-time image enhancement module 61, and used to store the original image, intermediate processing results and the final clear image.
[0144] In a preferred embodiment, the image enhancement system for cardiac interventional surgery further includes: a human-computer interaction module connected to the display module 62, for providing a graphical user interface to allow doctors to adjust processing parameters, such as deblurring strength, fusion weight, etc.
[0145] In a preferred embodiment, the image enhancement system for cardiac intervention surgery further includes: a GPU acceleration module connected to the real-time image enhancement module 61, and used for performing parallel computing using the GPU to improve the real-time performance of the system.
[0146] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0147] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0148] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0149] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0150] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0151] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. A method for image enhancement for cardiac intervention surgery, the method comprising the following steps: Obtain two images of the left and right views of the coronary angiography; Splicing the two images of the left and right perspectives after preprocessing, extracting a feature map of the spliced image, and generating a blur map based on the feature map; Deblurring the blurred image; The deblurred left and right view images are fused to generate the final coronary angiography image.
2. The image enhancement method for cardiac intervention surgery according to claim 1, characterized in that: The two images of the left and right perspectives of the coronary angiography are X-ray images of two perspectives obtained by a dual-source X-ray machine performing angiography on the patient's coronary artery from two directions, left anterior oblique and right anterior oblique, respectively.
3. The image enhancement method for cardiac intervention surgery according to claim 1, characterized in that: The preprocessing and splicing of the two images of the left and right perspectives, extracting a feature map of the spliced image, and generating a blur map based on the feature map specifically includes: Image preprocessing: normalizing the two images of the left and right perspectives, scaling the pixel values to the range of [0, 1]; Image stitching: Flip the preprocessed left-view image vertically, and then stitch it to the right side of the preprocessed right-view image to form a stitched image; Feature map extraction: based on the spliced image, extract its semantic features to form a feature map; Fuzzy image generation: Using the extracted feature map, a fuzzy image is generated through a convolutional neural network.
4. The image enhancement method for cardiac intervention surgery according to claim 1, characterized in that: The step of fusing the left and right view images after deblurring to generate a final coronary angiography image specifically includes: Image alignment: Accurately align two images from left and right perspectives; Weight calculation: Based on the blur map, the fusion weight is calculated for each pixel position, where the area with lower blur is given a higher fusion weight; Weighted average: Use the calculated fusion weights to perform pixel-level weighted average of the left and right view images to obtain the fused image.
5. The image enhancement method for cardiac intervention surgery according to claim 4, characterized in that: The step of fusing the left and right view images after deblurring to generate a final coronary angiography image further includes: Post-processing: Perform edge enhancement and contrast adjustment on the fused image to further improve image quality.
6. An image enhancement device for cardiac intervention surgery, characterized in that: include: An image acquisition module, used for acquiring two images of left and right perspectives of coronary angiography; A blur map estimation module is used to splice the two images of the left and right perspectives after preprocessing, extract a feature map of the spliced image, and generate a blur map based on the feature map; Defuzzification network module: used for defuzzifying the fuzzy graph; Image fusion module: used to fuse the left and right view images after deblurring to generate the final coronary angiography image.
7. The image enhancement device for cardiac interventional surgery according to claim 6, characterized in that: The deblurring network module specifically includes: The encoder is used to gradually reduce the spatial size of the feature map while increasing the number of channels to extract multi-scale features of the image; The decoder is used to gradually restore the spatial size of the feature map and reconstruct a clear image; A jump connector connects the encoder and the decoder.
8. An image enhancement module for cardiac intervention surgery, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the image enhancement method for cardiac intervention surgery as claimed in claim 1 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the image enhancement method for cardiac intervention surgery as claimed in claim 1 are implemented.
10. An image enhancement system for cardiac intervention surgery, characterized in that: include: A dual-pixel image acquisition module, including a dual-source X-ray machine for simultaneously capturing left and right perspective coronary angiography images; The image enhancement module for cardiac interventional surgery according to claim 8, connected to the dual-pixel image acquisition module, is used to enhance the coronary angiography images of the left and right viewing angles; The display module is connected to the image enhancement module for cardiac intervention surgery and is used to display the processed image on the screen in real time.
