An intravascular ultrasound image repair method and system

Abstract

The application discloses an intravascular ultrasound image repairing method and system. The method comprises the following steps: analyzing an original intravascular ultrasound image file to obtain an intravascular ultrasound image sequence and other non-image parameters; performing image preprocessing on the intravascular ultrasound image sequence to obtain an intravascular ultrasound image sequence to be repaired; performing initialization repairing on the intravascular ultrasound image sequence to be repaired to obtain an initial image repairing sequence; performing space-time modeling on the initial image repairing sequence to obtain a space-time consistency image repairing sequence, that is, a refined image repairing sequence; synthesizing a final image repairing sequence, converting the final image repairing sequence into an original format intravascular ultrasound image file, and archiving the final image repairing sequence; and the method can repair a guide wire artifact existing in the intravascular ultrasound image and provide a complete intravascular wall image of a coronary artery.

Description

Technical Field

[0001] This invention relates to the field of intravascular ultrasound medical imaging technology, and in particular to a method and system for intravascular ultrasound image restoration. Background Technology

[0002] The statements in this section merely refer to the background art related to this invention and do not necessarily constitute prior art.

[0003] With the rapid development of medical imaging technology and digital imaging, intravascular ultrasound (IVUS) imaging technology is frequently combined with coronary angiography for the clinical diagnosis of coronary artery disease and has become the gold standard for diagnosing coronary atherosclerosis. This technology combines invasive catheter technology with non-invasive ultrasound technology, utilizing a transducer within the catheter to continuously transmit ultrasound pulses. A computer imaging system then generates high-resolution images of the inner wall of the coronary arteries based on the ultrasound echoes. However, current intravascular ultrasound acquisition instruments primarily rely on two types of transducers for data acquisition: mechanical rotating transducers and phased array transducers. Unlike phased array transducers, mechanical rotating transducers are affected by the guidewire located outside the catheter, producing guidewire artifacts and causing localized loss of imaging information.

[0004] Meanwhile, thanks to the rapid development of artificial intelligence, digital image restoration technology has received increasing attention. Current image restoration techniques mainly focus on the restoration of natural color images, while restoration techniques for medical images, especially ultrasound images, are rarely heard of. The presence of guidewire artifacts in intravascular ultrasound images not only damages the integrity of the image and interferes with clinical diagnosis, but also poses significant challenges to subsequent image processing tasks such as segmentation of the arterial wall and intima, as well as the calculation of other relevant physiological parameters. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for repairing intravascular ultrasound images, which overcomes the deficiencies in medical ultrasound image repair. It can repair guide wire artifacts in intravascular ultrasound images, providing a complete image of the inner wall of the coronary artery, and has potential value for clinical diagnosis.

[0006] The first aspect of the present invention provides a method for repairing intravascular ultrasound images.

[0007] A method for intravascular ultrasound image restoration includes:

[0008] The original intravascular ultrasound image file is parsed to obtain the intravascular ultrasound image sequence and other non-image parameters;

[0009] The intravascular ultrasound image sequence is preprocessed to obtain the intravascular ultrasound image sequence to be repaired.

[0010] The intravascular ultrasound image sequence to be repaired is initialized and repaired to obtain an initial image repair sequence;

[0011] Spatiotemporal modeling is performed on the initial image restoration sequence to obtain a spatiotemporally consistent image restoration sequence, i.e., a refined image restoration sequence;

[0012] The final image restoration sequence is synthesized, converted into an intravascular ultrasound image file in its original format, and archived.

[0013] A second aspect of the present invention provides an intravascular ultrasound image restoration system.

[0014] An intravascular ultrasound image restoration system, comprising:

[0015] The image parsing module is configured to parse the raw intravascular ultrasound image file to obtain the intravascular ultrasound image sequence and other non-image parameters;

[0016] The image preprocessing module is configured to perform image preprocessing on the intravascular ultrasound image sequence to obtain the intravascular ultrasound image sequence to be repaired.

[0017] The initial repair module is configured to initialize and repair the intravascular ultrasound image sequence to be repaired, thereby obtaining an initial image repair sequence;

[0018] The fine-grained restoration module is configured to perform spatiotemporal modeling on the initial image restoration sequence to obtain a spatiotemporally consistent image restoration sequence, i.e., a fine-grained image restoration sequence.

[0019] The image synthesis module synthesizes the final image restoration sequence, converts it into an intravascular ultrasound image file in its original format, and archives it.

[0020] A third aspect of the present invention provides a computer-readable storage medium.

[0021] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of an intravascular ultrasound image restoration method as described in the first aspect above.

[0022] A fourth aspect of the present invention provides a computer device.

[0023] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of an intravascular ultrasound image restoration method as described in the first aspect above.

[0024] Compared with the prior art, the beneficial effects of the present invention are:

[0025] This invention addresses the shortcomings of existing technologies in medical ultrasound image restoration. This technology can repair guidewire artifacts in intravascular ultrasound images, providing a complete image of the coronary artery wall, which has potential value for clinical diagnosis, and in particular, provides important reference value for subsequent image processing tasks.

[0026] The advantages of additional aspects of the invention will be set forth in part in the description which follows, or may be learned by practice of the invention. Attached Figure Description

[0027] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0028] Figure 1 This is a flowchart of the intravascular ultrasound image restoration method in an embodiment of the present invention;

[0029] Figure 2 This is a schematic diagram of the intravascular ultrasound image restoration device in an embodiment of the present invention;

[0030] Figure 3(a) shows an intravascular ultrasound image of the blood vessel to be repaired in an embodiment of the present invention;

[0031] Figure 3(b) shows the repaired intravascular ultrasound image in an embodiment of the present invention. Detailed Implementation

[0032] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0033] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0034] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0035] All data acquisition in this embodiment is carried out in accordance with laws and regulations and with user consent, and the data is used legally.

[0036] Example 1

[0037] like Figure 1 As shown, this embodiment provides a method for intravascular ultrasound image restoration, including:

[0038] Step S1: Parse the original intravascular ultrasound image file to obtain the intravascular ultrasound image sequence and other non-image parameters;

[0039] Step S2: Perform image preprocessing on the intravascular ultrasound image sequence to obtain the intravascular ultrasound image sequence to be repaired;

[0040] Step S3: Initialize and repair the intravascular ultrasound image sequence to be repaired to obtain an initial image repair sequence;

[0041] Step S4: Perform spatiotemporal modeling on the initial image restoration sequence to obtain a spatiotemporally consistent image restoration sequence, i.e., a refined image restoration sequence;

[0042] Step S5: Synthesize the final image restoration sequence, convert it into an intravascular ultrasound image file in its original format, and archive it.

[0043] In step S1 of this embodiment, the original intravascular ultrasound image file is parsed, as follows:

[0044] After acquiring image files using an intravascular ultrasound instrument, the raw intravascular ultrasound image files are obtained and parsed using the Python library function Pydicom. The raw DICOM format intravascular ultrasound image file contains an intravascular ultrasound image sequence reflecting information about the tissue structure of the coronary artery wall, as well as other non-image parameters, such as acquisition instrument information, patient information, and other diagnostic parameters.

[0045] Therefore, the original DICOM format intravascular ultrasound image file is parsed into ultrasound image sequences and other non-image parameters, such as acquisition instrument information, patient information, and other diagnostic parameters.

[0046] In step S2 of this embodiment, the intravascular ultrasound image sequence is preprocessed.

[0047] Step S21: Adjust the grayscale value range of the intravascular ultrasound image sequence;

[0048] Step S22: Annotate each frame of the intravascular ultrasound image sequence to generate a mask and determine the known pixel area and the damaged area;

[0049] Step S23: Obtain the intravascular ultrasound image sequence of the vessel to be repaired.

[0050] Because intravascular ultrasound images are affected by guidewire artifacts, they cannot display a complete image of the inner wall of the artery. Each frame of the image can be divided into known pixel areas and damaged areas. The image preprocessing stage needs to determine the known pixel areas and damaged areas of the image.

[0051] Specifically, in step S21, the image grayscale values ​​are scaled to the range [0.0, 1.0] using Min-MaxNormalization.

[0052]

[0053] Where x′ is the original image matrix, x max x min These are the maximum and minimum pixel values, respectively, and x is the normalized image matrix with an image size of 512x512.

[0054] In step S22, during image preprocessing, a mask represented by 0s and 1s is generated for each image, where 0s represent damaged areas and 1s represent known pixel areas. The mask is multiplied by the normalized image matrix to obtain the image sequence to be repaired.

[0055] During network training, the 0 values ​​of the mask are marked in the effective pixel area of ​​the image to simulate the damaged area. After the network is trained, the mask is marked according to the actual damaged and undamaged areas during the inference stage.

[0056] In step S3 of this embodiment, an initial image restoration sequence is obtained through an initial restoration network.

[0057] In this embodiment, given that the shape and size of the guide wire artifact region in the image to be repaired are not regular, and that ordinary convolution operation has the characteristic of weight sharing, it cannot effectively distinguish between known regions and damaged regions in the image when extracting feature maps.

[0058] Gated convolutional units can effectively solve this problem by using a dynamically learnable method to extract features, and can effectively handle known and damaged areas in images.

[0059] Specifically, each image in the image sequence to be repaired is passed through the initial repair network one by one to obtain the initial repair image sequence, thereby completing the initial repair of the damaged areas of the image and obtaining the initial image repair sequence.

[0060] Specifically, the initial repair network consists of a series of gated convolutional units, where the basic operation principle of gated convolution is as follows:

[0061] The image to be repaired or the feature map output from the previous gated convolutional unit is used as input, and a gated convolutional filter W is employed. g Perform filtering gate value G x，y , where (x, y) is the position of the pixel coordinates;

[0062] On the other hand, a feature filter W is used. f Features are extracted from it to obtain feature map F. x，y The final feature map output x，y By the gate value G x，y and feature map F x，y Multiply them to get the result.

[0063] The operation of gated convolution can be expressed mathematically as follows:

[0064] G x，y =∑∑W g F x，y

[0065] F x，y =∑∑W f F x,y

[0066] Output x，y =σ(G x，y )φ(F x，y )

[0067] Where σ(·) represents the sigmoid function, after which the gate value G can be reduced. x，y Limited to the range of 0 to 1, the activation function φ(·) in this embodiment is the LeakyReLU function.

[0068] After the image-by-image restoration in this stage, the image sequence to be restored has a rough restoration effect. That is, only the restoration of a single image is considered, while the jitter and inconsistency in the entire image sequence are ignored. Therefore, a second stage of restoration is carried out on the basis of the initial image restoration sequence. By performing spatiotemporal modeling on the entire image sequence, a restoration sequence that is consistent in both spatial and temporal dimensions is obtained.

[0069] In step S4 of this embodiment, spatiotemporal modeling is performed on the initial image restoration sequence to obtain a spatiotemporally consistent image restoration sequence, i.e., a refined image restoration sequence.

[0070] The initial image restoration sequence only focuses on the restoration of a single frame image without considering the issue of temporal consistency. The spatiotemporal modeling network takes the initial image restoration sequence as input, fully explores the redundant information between adjacent images, further improves the consistency of the image sequence in both spatial and temporal dimensions, reduces blurring effects, and improves the accuracy of the restored content.

[0071] In this embodiment of the invention, spatiotemporal modeling is performed on the initial image restoration sequence to obtain a spatiotemporally consistent image restoration sequence, i.e., a refined image restoration sequence. The specific process is as follows:

[0072] The initial image restoration sequence is input as a whole into the trained spatiotemporal modeling network;

[0073] The spatiotemporal modeling network is based on bidirectional convolutional long short-term memory units and extracts relevant information between images in two directions: from front to back and from back to front.

[0074] Furthermore, the repair effect is further optimized through an iterative error control feedback mechanism to obtain spatiotemporal consistency repair results.

[0075] The spatiotemporal modeling network uses an eight-layer bidirectional convolutional long short-term memory (ConvLSTM) network as its backbone, ultimately outputting a refined image restoration sequence. The convolutional long short-term memory network can fully utilize the relevant information between image sequences in the time dimension, making the restoration effect of the image sequence more consistent in both time and space dimensions. The bidirectional convolutional long short-term memory network can simultaneously mine potential hidden relationships within the input image sequence from both the forward and reverse directions.

[0076] A single convolutional short-term memory network can be represented by the following formula:

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083] σ(*) represents the sigmoid activation function, X t and H t-1 Let i be the input value at time t and the hidden state value at time t-1, respectively. t ,f t ,o tThese represent the output values ​​of the input gate, forget gate, and output gate, respectively, and tanh(*) denotes the hyperbolic tangent function. The symbol represents a convolution operation, and ° represents a pixel-level multiplication. W xf W xc Wxo, WhfWhc, and Who are all learnable parameters.

[0084] Specifically, the entire network parameter update step only exists during the training phase. After the network is trained, this phase is skipped and the next step is performed directly.

[0085] Specifically, the embodiments of the present invention employ reconstruction loss function and adversarial loss function during the entire network training process. In this embodiment, the dataset was obtained through animal experiments. During the entire network training process, a pair of training datasets are used, namely the image sequence to be repaired and the real image sequence. This is done by marking the 0 values ​​of the mask in the effective pixel area of ​​the image to simulate the damaged area. The mask and the original image sequence are multiplied pixel by pixel to obtain the image sequence to be repaired along with the mask, which is then input into the network. The original image sequence is used as the real image sequence.

[0086] After the entire network is trained, the mask corresponding to the original image sequence is labeled according to the actual damaged area. The original image sequence and the corresponding mask are multiplied pixel by pixel and then input into the trained network along with the mask.

[0087] The parameter updates during network training are divided into two stages. First, the initial inpainting network parameters are updated. In this stage, a reconstruction loss function is used to measure the difference between the real image sequence and the initial inpainted image sequence. The Adam optimizer is used to update the parameters of the initial inpainting network. The learning rate of Adam is l. r The initial parameter is 1e-4, and the β value is 0.9. After the initial network parameters have converged, the second stage of parameter update is performed, that is, the parameters of the initial repair network are fixed, and the parameters of the spatiotemporal modeling network are updated, using the same optimizer and its hyperparameter settings.

[0088] The reconstruction loss function uses the L1 norm metric and consists of two parts: one part calculates the reconstruction loss function for known pixel regions, and the other part calculates the reconstruction loss function for artifact-damaged regions. See formula

[0089]

[0090]

[0091]

[0092] in The reconstruction loss value representing the damaged area, Let λ be the reconstruction loss value for the known pixel region. i ,λ v These are the weighting coefficients, where m represents the mask. y represents the initial image restoration sequence or the refined image restoration sequence, y represents the real image sequence, and ⊙ represents the Hadamard product.

[0093] The formula for the adversarial loss function used is:

[0094]

[0095]

[0096] in For the discriminator to resist the loss value, Let D represent the generator adversarial loss value, y represent the real image sequence, x represent the image sequence to be repaired, and G represent the generator. The discriminator uses five layers of conventional three-dimensional convolutional operation units.

[0097] In step S5 of this embodiment, the final image restoration sequence is synthesized, converted into an intravascular ultrasound image file in its original format, and archived. Specifically:

[0098] The known pixel regions of the intravascular ultrasound image sequence to be repaired and the damaged regions in the refined image repair sequence are stitched together to synthesize the final image repair sequence.

[0099] This can be expressed by a mathematical formula as follows:

[0100]

[0101] Where y out This represents the final image restoration sequence, where m⊙x represents the selected known pixel region of the image sequence to be restored. This indicates the damaged area selected for the refined image repair sequence.

[0102] The final image restoration sequence and other non-image parameters from the analysis are combined to create the original DICOM format intravascular ultrasound image data.

[0103] And archive them.

[0104] Specifically, the final image restoration sequence is obtained by stitching together the known pixel regions of the intravascular ultrasound image sequence to be restored and the damaged regions in the refined image restoration sequence.

[0105] As shown in Figures 3(a) and 3(b), the repair effect of the trained network on intravascular ultrasound images in this embodiment is demonstrated. Figure 3(a) is the intravascular ultrasound image to be repaired, and Figure 3(b) is the repaired intravascular ultrasound image. The present invention can effectively repair guide wire artifacts in intravascular ultrasound images.

[0106] The raw format of intravascular ultrasound image data includes both the final image restoration sequence and other non-image parameters obtained during the analysis. Raw format intravascular ultrasound image data facilitates interaction with Picture Archiving and Communication System (PACS).

[0107] Example 2

[0108] This embodiment provides an intravascular ultrasound image restoration system, including:

[0109] The image parsing module is configured to parse the raw intravascular ultrasound image file to obtain the intravascular ultrasound image sequence and other non-image parameters;

[0110] The image preprocessing module is configured to perform image preprocessing on the intravascular ultrasound image sequence to obtain the intravascular ultrasound image sequence to be repaired.

[0111] The initial repair module is configured to initialize and repair the intravascular ultrasound image sequence to be repaired, thereby obtaining an initial image repair sequence;

[0112] The fine-grained restoration module is configured to perform spatiotemporal modeling on the initial image restoration sequence to obtain a spatiotemporally consistent image restoration sequence, i.e., a fine-grained image restoration sequence.

[0113] The image synthesis module synthesizes the final image restoration sequence, converts it into an intravascular ultrasound image file in its original format, and archives it.

[0114] It should be noted that the image analysis module, image preprocessing module, initial restoration module, refined restoration module, and image synthesis module mentioned above correspond to steps S1 to S5 in Embodiment 1. The examples and application scenarios implemented by these modules and their corresponding steps are the same, but they are not limited to the content disclosed in Embodiment 1. It should be noted that these modules, as part of the system, can be executed in a computer system such as a set of computer-executable instructions.

[0115] The descriptions of each embodiment in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0116] The proposed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative, and the division of modules described above is only a logical functional division. In actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed.

[0117] Example 3

[0118] like Figure 2 As shown, this embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the intravascular ultrasound image repair method described in Embodiment 1 above.

[0119] It also includes: communication modules, displays;

[0120] The communication module is used to communicate with intravascular ultrasound instruments and hospital PACS systems wirelessly or via wired means to transmit intravascular ultrasound image data.

[0121] The memory is used to store the image data that needs to be saved, as well as the model program code and related code configuration library;

[0122] The monitor is used to display intravascular ultrasound images before and after repair.

[0123] It should be understood that in this embodiment, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0124] Memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of memory may also include non-volatile random access memory. For example, memory may also store information about the device type.

[0125] In the implementation process, each step of the above method can be completed by the integrated logic circuits in the processor hardware or by software instructions.

[0126] The method in Embodiment 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The software modules can reside in readily available storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not provided here.

[0127] Example 4

[0128] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the intravascular ultrasound image restoration method as described in Embodiment 1 above.

[0129] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0130] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0131] Those skilled in the art will recognize that the units and algorithm steps described in connection with the various examples of this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0132] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intravascular ultrasound image repair method, characterized by, The application relates to a method for repairing an intravascular ultrasound image sequence, and a device thereof. The method comprises the following steps: analyzing an original intravascular ultrasound image file to obtain an intravascular ultrasound image sequence and other non-image parameters; The intravascular ultrasound image sequence to be repaired is initialized and repaired to obtain an initial image repair sequence. Specifically, an initial repair network is used to repair each frame of the intravascular ultrasound image sequence to be repaired, completing the initial repair of the damaged areas of the image; thus obtaining the initial image repair sequence. The initial repair network consists of several gated convolutional units, wherein the gated convolution operation process is as follows: the intravascular ultrasound image to be repaired or the feature map output by the previous gated convolutional unit is used as input, and a gated convolutional filter is applied. Perform filtering and gating Where (x, y) represents the pixel coordinates; on the other hand, a feature filter is used. Features are extracted from it to obtain a feature map. The final feature map By gating value and feature map Multiplying, we obtain: After image-by-image initialization and repair, the intravascular ultrasound image sequence to be repaired has a rough repair effect. That is, only the repair of a single image is considered, while the jitter and inconsistency in the whole image sequence are ignored. Therefore, a second stage of repair is carried out on the basis of the initial image repair sequence. By performing spatiotemporal modeling on the whole image sequence, a repair sequence that is consistent in both spatial and temporal dimensions is obtained. performing image preprocessing on the intravascular ultrasound image sequence to obtain an intravascular ultrasound image sequence to be repaired; spatiotemporally modeling an initial image repair sequence to obtain a spatiotemporally consistent image repair sequence, that is, a refined image repair sequence, specifically: the initial image repair sequence is input into a trained spatiotemporal modeling network as a whole to obtain the spatiotemporally consistent image repair sequence; the spatiotemporal modeling network is a convolutional long short-term memory network, which extracts relevant information between images in two directions, from front to back and from back to front, and further optimizes the repair effect through an iterative error control feedback mechanism to obtain a spatiotemporally consistent repair sequence result; 2. The intravascular ultrasound image repair method of claim 1, wherein, synthesizing a final image repair sequence, converting the final image repair sequence into an original format intravascular ultrasound image file, and archiving the final image repair sequence, specifically: performing image splicing on a known pixel region of the intravascular ultrasound image sequence to be repaired and a damaged region in the refined image repair sequence; synthesizing a final image repair sequence, combining the final image repair sequence and other non-image parameters into an original format intravascular ultrasound image file, and archiving the final image repair sequence. The analyzing of the original intravascular ultrasound image file to obtain the intravascular ultrasound image sequence and the other non-image parameters comprises the following steps: analyzing an original medical digital imaging and communication (DICOM) format intravascular ultrasound image file into an ultrasound image sequence and other non-image parameters, 3. The method for intravascular ultrasound image restoration as described in claim 1, characterized in that, The other non-image parameters include acquisition instrument information, patient information and other diagnosis-related parameters. The image preprocessing of the intravascular ultrasound image sequence to obtain the intravascular ultrasound image sequence to be repaired comprises the following steps: adjusting a gray value range of the intravascular ultrasound image sequence; determining a known pixel region and a damaged region of the intravascular ultrasound image sequence; 4. An intravascular ultrasound image repair system, comprising: obtaining the intravascular ultrasound image sequence to be repaired. The device comprises: an image analysis module configured to analyze an original intravascular ultrasound image file to obtain an intravascular ultrasound image sequence and other non-image parameters; The initial repair module is configured to initialize and repair the intravascular ultrasound image sequence to be repaired, obtaining an initial image repair sequence. Specifically, it uses an initial repair network to repair each frame of the intravascular ultrasound image sequence to be repaired, completing the initial repair of the damaged areas of the image; thus obtaining the initial image repair sequence. The initial repair network consists of several gated convolutional units, wherein the gated convolution operation process is as follows: the intravascular ultrasound image to be repaired or the feature map output by the previous gated convolutional unit is used as input, and a gated convolutional filter is applied. Perform filtering and gating Where (x, y) represents the pixel coordinates; on the other hand, a feature filter is used. Features are extracted from it to obtain a feature map. The final feature map By gating value and feature map Multiplying, we obtain: After image-by-image initialization and repair, the intravascular ultrasound image sequence to be repaired has a rough repair effect. That is, only the repair of a single image is considered, while the jitter and inconsistency in the whole image sequence are ignored. Therefore, a second stage of repair is carried out on the basis of the initial image repair sequence. By performing spatiotemporal modeling on the whole image sequence, a repair sequence that is consistent in both spatial and temporal dimensions is obtained. an image preprocessing module configured to perform image preprocessing on the intravascular ultrasound image sequence to obtain an intravascular ultrasound image sequence to be repaired; a refined repair module configured to spatiotemporally model an initial image repair sequence to obtain a spatiotemporally consistent image repair sequence, that is, a refined image repair sequence, specifically: the initial image repair sequence is input into a trained spatiotemporal modeling network as a whole to obtain the spatiotemporally consistent image repair sequence; the spatiotemporal modeling network is a convolutional long short-term memory network, which extracts relevant information between images in two directions, from front to back and from back to front, and further optimizes the repair effect through an iterative error control feedback mechanism to obtain a spatiotemporally consistent repair sequence result; An image synthesis module synthesizes the final image repair sequence, converts the intravascular ultrasound image file in the original format, and archives, specifically: image splicing is performed on the known pixel area of the intravascular ultrasound image sequence to be repaired and the damaged area in the refined image repair sequence; the final image repair sequence is synthesized, the final image repair sequence and other non-image parameters are synthesized into an intravascular ultrasound image file in the original format; and archiving is performed.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps in the intravascular ultrasound image repair method of any one of claims 1-3.

6. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps in the intravascular ultrasound image repair method of any one of claims 1-3.

Images