Abdominal magnetic resonance respiration artifact eliminating system

Abstract

The invention relates to the technical field of image restoration, in particular to an abdominal magnetic resonance breathing artifact elimination system which can realize the following steps through mutual cooperation of a plurality of modules: acquiring an image to be subjected to artifact elimination, and screening out a suspected artifact area from the image to be subjected to artifact elimination; determining an artifact expression characteristic value corresponding to the suspected artifact area according to the gradient distribution condition and the gray level distribution condition in the suspected artifact area and the shape of each suspected artifact area; screening out target artifact areas, and determining an initial blurring angle and an initial blurring size corresponding to each target artifact area; correcting the initial blurring angle and the initial blurring size corresponding to each target artifact area; according to the blurring kernel of the target blurring angle and the blurring kernel of the target blurring size, artifact elimination is carried out on the target artifact area, artifact elimination of the image to be subjected to artifact elimination is achieved, and artifact area identification accuracy and artifact elimination rationality are improved.

Description

Technical Field

[0001] This invention relates to the field of image restoration technology, and more specifically to an abdominal magnetic resonance respiratory artifact elimination system. Background Technology

[0002] With the development of technology, image restoration techniques are being applied more and more widely. For example, they can be used to eliminate respiratory motion artifacts in abdominal magnetic resonance imaging (MRI) images. Currently, the common method for artifact removal is to use a blur kernel to eliminate artifact regions in the image, thereby achieving image restoration. In practice, target regions are often identified based on differences in grayscale values; therefore, artifact regions in an image can be identified based on differences in grayscale values.

[0003] However, when identifying respiratory motion artifact regions in abdominal magnetic resonance images based on different grayscale values, the following technical problems often arise: The grayscale difference between respiratory motion artifacts and some normal tissue areas is often small. Therefore, when identifying respiratory motion artifact areas in abdominal MRI images, if only the difference in grayscale values ​​is considered, it may lead to misjudgment of artifact pixels, resulting in poor accuracy of artifact area identification and consequently poor rationality of artifact removal. Summary of the Invention

[0004] To address the technical problem of poor artifact elimination due to low accuracy in artifact region identification, this invention proposes an abdominal magnetic resonance respiratory artifact elimination system.

[0005] In a first aspect, the present invention provides an abdominal magnetic resonance respiratory artifact elimination system, the system comprising: The acquisition and filtering module is used to acquire the image to be artifact-removed and filter out suspected artifact regions from the image to be artifact-removed. The artifact performance feature value determination module is used to determine the artifact performance feature value corresponding to each suspected artifact region based on the gradient distribution and grayscale distribution within each suspected artifact region, as well as the shape of each suspected artifact region. The initial parameter determination module is used to filter out the target artifact region from all suspected artifact regions based on the artifact performance characteristic value, and determine the initial blur angle and initial blur size corresponding to each target artifact region. The parameter correction module is used to correct the initial blur angle and initial blur size of each target artifact region based on the pre-acquired historical abdominal magnetic resonance image with the same respiratory stage as the image to be eliminated, so as to obtain the target blur angle and target blur size of each target artifact region. The artifact removal module is used to remove artifacts from the target artifact region based on the blur kernel of the target blur angle and target blur size, thus realizing artifact removal of the image to be removed.

[0006] In conjunction with the first aspect above, in one possible implementation, the step of filtering out suspected artifact regions from the image to be artifact-removed includes: Based on the gradient value and grayscale distribution of each pixel in the image to be artifact-removed, the initial value of the blur artifact feature corresponding to each pixel in the image to be artifact-removed is determined. If the initial value of the blur artifact feature corresponding to a pixel is greater than the preset blur artifact threshold, then the pixel is identified as a suspected artifact edge pixel. Based on the suspected artifact edge pixels in the image to be artifact-removed, a suspected artifact region is constructed.

[0007] In conjunction with the first aspect above, in one possible implementation, constructing the initial value of the blur artifact feature corresponding to each pixel in the image to be removed, based on the gradient value and grayscale distribution of each pixel in the image to be removed, includes: Any pixel in the image to be artifact removed is designated as a marker pixel. The gradient value corresponding to the marker pixel and the entropy value of the gray values ​​of all pixels in the preset neighborhood of the marker pixel are used to determine the initial value of the blur artifact feature corresponding to the marker pixel.

[0008] In conjunction with the first aspect above, in one possible implementation, determining the artifact representation feature value corresponding to each suspected artifact region based on the gradient distribution and grayscale distribution within each suspected artifact region, as well as the shape of each suspected artifact region, includes: Based on the gradient values ​​and grayscale values ​​of all edge pixels in each suspected artifact region, determine the dynamic edge blur value corresponding to each suspected artifact region. Based on the minimum bounding rectangle of each suspected artifact region and the gradient direction of all pixels within each suspected artifact region, determine the shape feature value corresponding to each suspected artifact region. The product between the edge blur dynamic value and the shape feature value corresponding to each suspected artifact region is normalized to obtain the artifact performance feature value corresponding to each suspected artifact region.

[0009] In conjunction with the first aspect above, in one possible implementation, determining the dynamic edge blur value corresponding to each suspected artifact region based on the gradient values ​​and grayscale values ​​corresponding to all edge pixels on each suspected artifact region includes: Any suspected artifact region in the image to be artifact removed is identified as a marked suspected artifact region, and the absolute value of the difference between the gradient values ​​corresponding to every two edge pixels in the marked suspected artifact region is identified as the target gradient difference, thus obtaining the target gradient difference set corresponding to the marked suspected artifact region. The entropy value of the grayscale value of all pixels in the preset neighborhood corresponding to each edge pixel in the marked suspected artifact region is determined as the grayscale representative entropy value of each edge pixel in the marked suspected artifact region. The edge blur dynamic value corresponding to the marked suspected artifact region is determined based on the largest target gradient difference in the target gradient difference set corresponding to the marked suspected artifact region, the average gradient value of all edge pixels in the marked suspected artifact region, and the average gray-level entropy value of all edge pixels in the marked suspected artifact region.

[0010] In conjunction with the first aspect above, in one possible implementation, determining the shape feature value corresponding to each suspected artifact region based on the minimum bounding rectangle of each suspected artifact region and the gradient direction corresponding to all pixels within each suspected artifact region includes: The angle corresponding to the gradient direction of each pixel in each suspected artifact region is determined as the gradient direction angle of each pixel in each suspected artifact region. Based on the aspect ratio of the minimum bounding rectangle of each suspected artifact region and the standard deviation of the gradient direction angles of all pixels within each suspected artifact region, the shape feature value corresponding to each suspected artifact region is determined.

[0011] In conjunction with the first aspect above, in one possible implementation, the step of filtering out the target artifact region from all suspected artifact regions based on artifact performance feature values ​​includes: If the artifact performance characteristic value corresponding to the suspected artifact region is greater than the preset artifact performance threshold, then the suspected artifact region is determined as the target artifact region.

[0012] In conjunction with the first aspect above, in one possible implementation, determining the initial blur angle and initial blur size corresponding to each target artifact region includes: Perform a Fast Fourier Transform on each target artifact region to obtain the Fourier spectrum corresponding to each target artifact region; By using Radon transform, the direction of the stripes in the Fourier spectrum corresponding to each target artifact region is detected, and the initial blur angle and initial blur size corresponding to each target artifact region are obtained.

[0013] In conjunction with the first aspect above, in one possible implementation, the step of correcting the initial blur angle and initial blur size corresponding to each target artifact region based on a pre-acquired historical abdominal MRI image of the same respiratory stage as the image to be artifact-removed, to obtain the target blur angle and target blur size corresponding to each target artifact region, includes: Any target artifact region in the image to be artifact-removed is determined as the artifact region to be determined, and each historical abdominal MRI image with the same respiratory stage as the image to be artifact-removed is determined as the reference image. Identify historical artifact regions from each reference image and determine the initial blur angle and initial blur size for each historical artifact region; Based on the initial blur angle and initial blur size corresponding to each historical artifact region in each reference image, and the initial blur angle and initial blur size corresponding to the undetermined artifact region, the target blur angle and target blur size corresponding to the undetermined artifact region are determined.

[0014] In conjunction with the first aspect above, in one possible implementation, determining the target blur angle and target blur size corresponding to the undetermined artifact region based on the initial blur angle and initial blur size corresponding to each historical artifact region in each reference image, and the initial blur angle and initial blur size corresponding to the undetermined artifact region, includes: Based on the respiratory standard degree of each reference image to which it belongs, the mean and standard deviation of the initial blur angles corresponding to all historical artifact regions in each reference image, and the initial blur angles corresponding to the undetermined artifact regions, the angle reference weights corresponding to each reference image are determined. Based on the respiratory standard degree of each reference image to which it belongs, the mean and standard deviation of the initial blur size corresponding to all historical artifact regions in each reference image, and the initial blur size corresponding to the undetermined artifact region, the size reference weight corresponding to each reference image is determined. The target blur angle corresponding to the undetermined artifact region is determined based on the average of the angle reference weights corresponding to each reference image and the initial blur angles corresponding to all historical artifact regions within them, as well as the initial blur angles corresponding to the undetermined artifact region. The target blur size of the undetermined artifact region is determined based on the average of the size reference weights corresponding to each reference image and the initial blur size of all historical artifact regions within it, as well as the initial blur size of the undetermined artifact region.

[0015] Secondly, the present invention provides a method for eliminating abdominal magnetic resonance respiratory artifacts using an abdominal magnetic resonance respiratory artifact elimination system, the method comprising: Obtain the image to be artifact-removed, and filter out suspected artifact regions from the image; Based on the gradient distribution and grayscale distribution within each suspected artifact region, as well as the shape of each suspected artifact region, determine the artifact performance feature value corresponding to each suspected artifact region. Based on the artifact performance characteristic values, target artifact regions are selected from all suspected artifact regions, and the initial blur angle and initial blur size corresponding to each target artifact region are determined. Based on the pre-acquired historical abdominal magnetic resonance images of the same respiratory stage as the image to be artifact-removed, the initial blur angle and initial blur size corresponding to each target artifact region are corrected to obtain the target blur angle and target blur size corresponding to each target artifact region. Based on the blur kernel of the target blur angle and target blur size, artifact removal is performed on the target artifact region, thus realizing artifact removal of the image to be removed.

[0016] Thirdly, a server is provided, including a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, causing the device to perform the aforementioned method for eliminating respiratory artifacts on abdominal magnetic resonance imaging.

[0017] Fourthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to perform the aforementioned method for eliminating respiratory artifacts on abdominal magnetic resonance imaging.

[0018] Fifthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the aforementioned method for eliminating respiratory artifacts on abdominal magnetic resonance imaging.

[0019] The present invention has the following beneficial effects: This invention discloses an abdominal magnetic resonance imaging (MRI) respiratory artifact removal system. It achieves the identification of target artifact regions and the removal of artifacts from the image to be removed, solving the technical problem of poor artifact removal efficiency due to poor accuracy in artifact region identification, and improving both the accuracy of artifact region identification and the rationality of artifact removal. Specifically, this invention analyzes the shape of suspected artifact regions and their gradient and grayscale distributions, quantifying the artifact performance characteristic values ​​corresponding to each suspected artifact region. This achieves the identification of target artifact regions and improves the accuracy of artifact region identification, thereby enhancing the rationality of artifact removal. Secondly, by analyzing historical abdominal MRI images with the same respiratory stage as the image to be removed, the initial blur angle and initial blur size corresponding to each target artifact region are corrected. This improves the rationality of the blur angle and blur size settings to a certain extent, further enhancing the rationality of artifact removal. Attached Figure Description

[0020] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the structure of an abdominal magnetic resonance respiratory artifact elimination system according to the present invention; Figure 2 This is a flowchart of a method for eliminating respiratory artifacts on abdominal magnetic resonance imaging according to the present invention; Figure 3 This is a schematic diagram of the structure of a computer device according to the present invention. Detailed Implementation

[0022] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the technical solution proposed according to the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0023] Unless otherwise defined, 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.

[0024] In practice, the severity of respiratory artifacts in different tissue regions of abdominal MRI images is often directly related to the amplitude of regional motion, and the amplitude of motion in different regions often varies. For example, the top of the liver, which is close to the diaphragm, often has a larger amplitude of motion, while the kidney, which is far from the diaphragm, often has a smaller amplitude of motion. Therefore, it is necessary to divide the image into different artifact regions and adaptively set the blur direction and blur size for different respiratory artifact regions caused by respiratory motion. Then, the local blur kernel is estimated, and dynamic non-uniform deblurring is performed on different tissue regions that produce different amounts of blur to effectively eliminate respiratory artifacts.

[0025] refer to Figure 1 A schematic diagram of an abdominal magnetic resonance respiratory artifact elimination system according to the present invention is shown. The abdominal magnetic resonance respiratory artifact elimination system includes: The acquisition and filtering module 101 is used to acquire the image to be removed from the artifacts and filter out the suspected artifact regions from the image to be removed from the artifacts.

[0026] The image to be artifact-removed can be an abdominal magnetic resonance imaging (MRI) image from which artifact removal is to be performed.

[0027] As an example, the acquisition and filtering module 101 implements the following steps: The first step is to obtain the image to be free of artifacts.

[0028] For example, magnetic resonance imaging (MRI) technology can be used to acquire abdominal MRI images of a patient to be examined, which can then be used as images to eliminate artifacts.

[0029] The second step is to determine the initial value of the blur artifact feature corresponding to each pixel in the image to be artifact-removed, based on the gradient value and grayscale distribution of each pixel in the image to be artifact-removed.

[0030] For example, any pixel in the image to be removed can be designated as a marker pixel, and the gradient value corresponding to the marker pixel and the entropy value of the gray values ​​of all pixels in the preset neighborhood corresponding to the marker pixel can be used to determine the initial value of the blur artifact feature corresponding to the marker pixel.

[0031] The preset neighborhood can be a pre-set neighborhood, which can be a 3×3 neighborhood.

[0032] For example, the formula for determining the initial value of the blur artifact feature corresponding to the marked pixel can be: ; in, AIt is the initial value of the blur artifact feature corresponding to the marked pixel. It is a normalization function. p It is the entropy value of the grayscale values ​​of all pixels in the preset neighborhood corresponding to the marked pixel. It is an exponential function with the natural constant as its base. D It is the gradient value corresponding to the marked pixel.

[0033] It should be noted that, in reality, the blurred areas caused by breathing artifacts often appear in MRI images as having low edge gradients, and due to gray-level disorder, they often have high gray-level entropy within the region. Therefore, when A A larger value often indicates that the marked pixel is more likely to be a blurry artifact edge point.

[0034] The third step is to determine the pixel as a suspected artifact edge pixel if the initial value of the blur artifact feature corresponding to the pixel is greater than the preset blur artifact threshold.

[0035] The preset blur artifact threshold can be a pre-set threshold, which can be manually set based on the actual scene. For example, the preset blur artifact threshold can be 0.5.

[0036] It should be noted that suspected artifact edge pixels can be the coarsely screened artifact edge pixels.

[0037] The fourth step is to construct the suspected artifact region based on the suspected artifact edge pixels in the image to be removed.

[0038] Among them, the suspected artifact region can be the area enclosed by the edge pixels of the suspected artifact.

[0039] The artifact performance feature value determination module 102 is used to determine the artifact performance feature value corresponding to each suspected artifact region based on the gradient distribution and grayscale distribution in each suspected artifact region, as well as the shape of each suspected artifact region.

[0040] As an example, determining the artifact performance characteristic value corresponding to each suspected artifact region may include the following steps: The first step, determining the dynamic edge blur value for each suspected artifact region based on the gradient and grayscale values ​​of all edge pixels in each suspected artifact region, may include the following sub-steps: The first sub-step involves identifying any suspected artifact region in the image to be artifact-removed as a marked suspected artifact region, and determining the absolute value of the difference between the gradient values ​​of every two edge pixels in the marked suspected artifact region as the target gradient difference, thereby obtaining the target gradient difference set corresponding to the marked suspected artifact region.

[0041] The second sub-step is to determine the entropy value of the grayscale value of all pixels in the preset neighborhood corresponding to each edge pixel in the marked suspected artifact region as the grayscale representative entropy value of each edge pixel in the marked suspected artifact region.

[0042] The third sub-step involves determining the dynamic edge blur value corresponding to the marked suspected artifact region based on the largest target gradient difference in the target gradient difference set corresponding to the marked suspected artifact region, the average gradient value of all edge pixels in the marked suspected artifact region, and the average gray-scale entropy value of all edge pixels in the marked suspected artifact region.

[0043] For example, the formula for determining the dynamic value of the edge blur corresponding to the marked suspected artifact region can be: ; in, B It is the dynamic value of the edge blur corresponding to the suspected artifact region. S It is the average grayscale value representing the entropy of all edge pixels on the suspected artifact region. It is an exponential function with the natural constant as its base. H It is the largest target gradient difference in the set of target gradient differences corresponding to the suspected artifact region. h It is the mean of the gradient values ​​corresponding to all edge pixels on the suspected artifact region.

[0044] It should be noted that, in reality, within the physical context of MRI breathing artifacts, the "blurred dynamics" of the region's edges are often essentially the degree of edge signal diffusion caused by respiratory motion. The edge gradient values ​​of areas with artifacts are often relatively lower than those of clear, normal areas. The degree to which the same region is affected by breathing is often consistent, and the grayscale distribution of normal edges is often regular, while the pixel grayscale distribution in areas with artifacts often becomes disordered. Therefore, when... B The larger the value, the more obvious the artifacts may be at the boundary of the suspected artifact region, and the more likely the suspected artifact region is to be a real artifact region.

[0045] The second step, determining the shape feature value corresponding to each suspected artifact region based on the minimum bounding rectangle of each suspected artifact region and the gradient direction corresponding to all pixels within each suspected artifact region, may include the following sub-steps: The first sub-step is to determine the gradient direction angle corresponding to each pixel in each suspected artifact region.

[0046] The second sub-step involves determining the shape feature value corresponding to each suspected artifact region based on the aspect ratio of the minimum bounding rectangle of each suspected artifact region and the standard deviation of the gradient direction angles corresponding to all pixels within each suspected artifact region.

[0047] For example, the formula for determining the shape feature value corresponding to a suspected artifact region can be: ; in, It is the first in the image to be artifact-removed i The shape feature value corresponding to the suspected artifact region. i It is the sequence number of the suspected artifact region in the image to be artifact-removed. It is the first in the image to be artifact-removed i The aspect ratio of the smallest bounding rectangle of a suspected artifact region. It is an exponential function with the natural constant as its base. It is the first in the image to be artifact-removed i The standard deviation of the gradient direction angles of all pixels within a suspected artifact region.

[0048] It should be noted that in reality, the essence of respiratory artifacts is often continuous blurring caused by the periodic movement of organs. Therefore, they should typically appear as stripes extending along the direction of movement. Furthermore, since respiratory artifacts are produced by periodic movement, the blurring direction within the same artifact area should be highly consistent, while the direction of the normal tissue edges is often more random. The larger the value, the more likely it is to indicate the first i The more likely a suspected artifact region is to appear as a long strip, the more likely it is to appear as a long strip. The smaller the size, the more likely it is to indicate the first i The more consistent the blur direction within a suspected artifact region, the better. Therefore, when The larger the value, the more likely it is to indicate the first i The more likely a region is to be a real artifact, the more likely it is to be a region that appears to be an artifact.

[0049] The third step is to normalize the product between the edge blur dynamic value and the shape feature value corresponding to each suspected artifact region to obtain the artifact performance feature value corresponding to each suspected artifact region.

[0050] It should be noted that the larger the artifact characteristic value corresponding to the suspected artifact region, the more likely the suspected artifact region is to be a real artifact region.

[0051] The initial parameter determination module 103 is used to filter out the target artifact region from all suspected artifact regions based on the artifact performance characteristic value, and determine the initial blur angle and initial blur size corresponding to each target artifact region.

[0052] As an example, the initial parameter determination module 103 specifically implements the following steps: The first step is to determine the suspected artifact region as the target artifact region if the artifact performance feature value corresponding to the suspected artifact region is greater than the preset artifact performance threshold.

[0053] The preset artifact rendering threshold can be a pre-set threshold, which can be manually set based on the actual scene. For example, the preset artifact rendering threshold can be 0.7.

[0054] The second step is to perform a fast Fourier transform on each target artifact region to obtain the Fourier spectrum corresponding to each target artifact region.

[0055] The third step involves using Radon transform to detect the direction of the stripes in the Fourier spectrum corresponding to each target artifact region, thereby obtaining the initial blur angle and initial blur size for each target artifact region.

[0056] The initial blur angle and initial blur size can be obtained through Radon transform.

[0057] It should be noted that for each target artifact region, the blur kernel can be estimated by analyzing the image patch of that region. Since breathing artifacts in MRI images are linear motion blurs, their mathematical model is the convolution of the original sharp image and the blur kernel. According to the convolution theorem of Fourier transform, the spectrum of the sharp image and the spectrum of the blur kernel are multiplied. The Fourier spectrum of a rectangular blur kernel often exhibits parallel zero-value fringes, physically corresponding to the suppressed frequency components in the blur direction. For each target artifact region, frequency domain analysis algorithms can be used on its image patch to analyze its Fourier spectrum. Blur often results in parallel zero-value fringes in the spectrum, with the fringe direction perpendicular to the blur direction. Horizontal vibration blur often results in vertical zero-value fringes in the spectrum. The fringe spacing is inversely proportional to the blur size; the longer the blur size (the greater the displacement during exposure time), the wider the fringe spacing (corresponding to the suppression of lower frequency components). The direction of these fringes can be detected using Radon transform, allowing for precise calculation of the blur angle and blur size of the region.

[0058] The parameter correction module 104 is used to correct the initial blur angle and initial blur size of each target artifact region based on the historical abdominal magnetic resonance image with the same respiratory stage as the image to be artifact eliminated, obtained in advance, so as to obtain the target blur angle and target blur size of each target artifact region.

[0059] In practice, to facilitate the division of respiratory phases, patients can be instructed to use the 448 breathing method during abdominal MRI image acquisition. The respiratory cycle of the 448 breathing method can be divided into multiple respiratory phases. The number of respiratory phases can be preset, such as 3. For example, if the number of respiratory phases is 3, there can be 3 respiratory phases: inhalation, breath-holding, and exhalation. The inhalation phase is typically the phase where the patient takes a deep breath through their nose or mouth. The breath-holding phase is typically the phase where the patient stops breathing after taking a deep breath. The exhalation phase is typically the phase where the patient exhales the inhaled air. Historical abdominal MRI images can be those acquired during the same respiratory phase using MRI technology before the acquisition of images to be acquired for artifact removal.

[0060] As an example, correcting the initial blur angle and initial blur size for each target artifact region to obtain the target blur angle and target blur size for each target artifact region may include the following steps: The first step is to identify any target artifact region in the image to be artifacted as the undetermined artifact region, and to identify each historical abdominal MRI image with the same respiratory stage as the image to be artifacted as the reference image.

[0061] The second step is to identify historical artifact regions from each reference image and determine the initial blur angle and initial blur size corresponding to each historical artifact region.

[0062] It should be noted that the method for identifying historical artifact regions can be the same as the method for identifying target artifact regions, and will not be repeated here. The methods for obtaining the initial blur angle and initial blur size corresponding to historical artifact regions can be the same as the methods for obtaining the initial blur angle and initial blur size corresponding to target artifact regions, and will not be repeated here.

[0063] The third step is to determine the target blur angle and target blur size corresponding to the aforementioned undetermined artifact regions based on the initial blur angle and initial blur size corresponding to each historical artifact region in each reference image, as well as the initial blur angle and initial blur size corresponding to the aforementioned undetermined artifact regions.

[0064] In practice, the smaller the difference between the blur parameters of the region to be identified and the blur parameters of the reference image, the more valid the reference basis of the reference image, which originates from the same respiratory movement pattern, tends to be. Furthermore, the more stable the patient's breathing (without directional abnormal movements such as coughing or shallow, rapid breathing) during the acquisition of the reference image, the more accurately its mainstream direction reflects the normal movement direction of that respiratory phase, thus making the determination of the blur kernel parameters of the region to be identified more reliable. A higher degree of respiratory standardization in the respiratory cycle to which the reference image belongs often indicates a greater similarity between the respiratory conditions of the reference image and the respiratory conditions of the image to be eliminated, thus suggesting a higher reference value for the reference image.

[0065] For example, determining the target blur angle and target blur size corresponding to the above-mentioned undetermined artifact region may include the following sub-steps: The first sub-step involves determining the angle reference weight for each reference image based on the respiratory standard degree of the respiratory cycle to which each reference image belongs, the mean and standard deviation of the initial blur angles corresponding to all historical artifact regions in each reference image, and the initial blur angles corresponding to the aforementioned undetermined artifact regions.

[0066] The respiratory standard of the reference image's respiratory cycle can be characterized by the similarity between the respiratory signals of the patient to be tested within the reference image's respiratory cycle and the respiratory signals of the patient to be tested within the same respiratory cycle as the image to be artifact-removed. The similarity between different respiratory signals can be obtained using the Dynamic Time Warping (DTW) algorithm.

[0067] For example, the formula for determining the angle reference weight corresponding to the reference image can be: ; ; in, It is the first j Angle reference weights corresponding to each reference image. j It is the reference image number. It is the first j Initial weights for the angles corresponding to each reference image. G It is the sum of the initial angle weights corresponding to all reference images. It is an absolute value function. It is the initial blur angle corresponding to the undetermined artifact region. It is the first j The mean of the initial blur angles corresponding to all historical artifact regions in the reference image can characterize the first... j The mainstream blur angle of a reference image. It is the first jThe standard deviation of the initial blur angles corresponding to all historical artifact regions in a reference image. It is a pre-set factor greater than 0, mainly used to prevent the denominator from being 0, and it can be 0.0001. It is the first j The standard degree of breathing for the respiratory cycle to which each reference image belongs.

[0068] It should be noted that when The smaller the value, the closer the initial blur angle corresponding to the undetermined artifact region is to the first value. j The prevailing blur angle of the first reference image often indicates the first... j The more reference images there are, the more angularly relevant the region of artifact to be determined. The smaller the size, the more likely it is to indicate the first j The more stable the patient's breathing is during the acquisition of the reference image, the more accurately its mainstream direction tends to reflect the normal direction of movement during the respiratory phase, and the more reliable the determination of the fuzzy kernel parameters of the aforementioned undetermined artifact region becomes. The larger the value, the more likely it is to indicate the first j The more similar the breathing pattern of the reference image to the breathing pattern of the image to be artifact-removed during its breathing cycle, the better the indication of the first reference image's breathing pattern. j The more reference images there are, the more valuable they are. Therefore, when The larger the value, the more likely it is to indicate the first j The higher the angular reference value of the first reference image for the region of artifact to be determined, the more likely it is that the first reference image is more relevant. j The mainstream blur angle of a reference image can, to a certain extent, represent the blur angle of the undetermined artifact region.

[0069] The second sub-step involves determining the size reference weight for each reference image based on the respiratory standard degree of each pre-acquired respiratory cycle, the mean and standard deviation of the initial blur size corresponding to all historical artifact regions in each reference image, and the initial blur size corresponding to the aforementioned undetermined artifact regions.

[0070] For example, the formula for determining the size reference weight corresponding to the reference image can be: ; ; in, It is the first j The size reference weights corresponding to each reference image. j It is the reference image number. It is the first j Initial weights for the dimensions of each reference image. It is the sum of the initial weights of the dimensions corresponding to all reference images. It is an absolute value function. d It is the initial blur size corresponding to the undetermined artifact region. It is the first j The mean of the initial blur size corresponding to all historical artifact regions in the reference image can characterize the first... j The mainstream blur size of a reference image. It is the first j The standard deviation of the initial blur size corresponding to all historical artifact regions in a reference image. It is a pre-set factor greater than 0, mainly used to prevent the denominator from being 0, and it can be 0.0001. It is the first j The standard degree of breathing for the respiratory cycle to which each reference image belongs.

[0071] It should be noted that when The larger the value, the more likely it is to indicate the first j The more relevant the reference image is to the size of the artifact region to be determined, the more likely it is that the first reference image is more relevant. j The mainstream blur size of a reference image can, to a certain extent, represent the blur size of the undetermined artifact region.

[0072] The third sub-step involves determining the target blur angle corresponding to the undetermined artifact region based on the average of the angle reference weights corresponding to each reference image and the initial blur angles corresponding to all historical artifact regions within them, as well as the initial blur angles corresponding to the undetermined artifact regions.

[0073] For example, the formula for determining the target blur angle corresponding to the undetermined artifact region can be: ; in, It is the target blur angle corresponding to the undetermined artifact region. and . yes The weight can be 0.7. yes The weight can be 0.3. It is the initial blur angle corresponding to the undetermined artifact region. N This is the number of reference images. j It is the reference image number. It is the first j Angle reference weights corresponding to each reference image. It is the first j The mean of the initial blur angles corresponding to all historical artifact regions in a reference image.

[0074] The fourth sub-step involves determining the target blur size corresponding to the undetermined artifact region based on the average of the size reference weights corresponding to each reference image and the initial blur size corresponding to all historical artifact regions within them, as well as the initial blur size corresponding to the undetermined artifact region.

[0075] For example, the formula for determining the target blur size corresponding to the undetermined artifact region can be: ; in, Q It is the target blur size corresponding to the undetermined artifact region. and . yes d The weight can be 0.7. yes The weight can be 0.3. d It is the initial blur size corresponding to the undetermined artifact region. N This is the number of reference images. j It is the reference image number. It is the first j The size reference weights corresponding to each reference image. It is the first j The mean of the initial blur size corresponding to all historical artifact regions in a reference image.

[0076] The artifact removal module 105 is used to remove artifacts from the target artifact region based on the blur kernel of the target blur angle and the target blur size, thereby realizing artifact removal of the image to be removed.

[0077] As an example, any target artifact region in the image to be artifacted can be defined as the undetermined artifact region. The target blur angle and target blur size corresponding to the undetermined artifact region can be defined as the undetermined blur angle and undetermined blur size, respectively. The undetermined blur angle and undetermined blur size can be used as the blur angle and blur size of the blur kernel, and the artifact removal of the undetermined artifact region can be achieved through the blur kernel.

[0078] Optionally, after obtaining the blur direction and size of the artifact region, a dedicated blur kernel can be constructed for different regions based on the characteristics of the 448 breathing method scene. Specifically, a rectangular linear motion blur kernel with the same blur direction and size is generated for the artifact region. For example, a pixel in the vertically downward direction during exhalation with a blur size of 5 can correspond to a 1×5 pixel kernel. Artifact-free regions can use a 1×1 pixel spread function to form a "region-kernel" mapping. Then, the Wiener deconvolution algorithm is used to perform deconvolution on each region, processing each sub-region independently before Gaussian smoothing the edges to avoid stitching artifacts. Finally, image quality is verified to achieve accurate elimination of breathing artifacts while preserving the details required for diagnosis, making it suitable for clinical applications. Since breathing is a continuous motion, the blur kernel between adjacent frames should change smoothly. The smoothing intensity can be adjusted based on the rate of change of the breathing signal, i.e., the difference in breathing signals at adjacent time points. If breathing changes slowly, the blur kernels of adjacent frames should be similar; if breathing changes drastically, larger differences are allowed.

[0079] refer to Figure 2 Based on the same inventive concept as the above-described method embodiments, the present invention provides a method for eliminating respiratory artifacts on abdominal magnetic resonance imaging, comprising the following steps: Step 201: Obtain the image to be artifact-removed and filter out suspected artifact regions from the image to be artifact-removed.

[0080] Step 202: Based on the gradient distribution and grayscale distribution within each suspected artifact region, as well as the shape of each suspected artifact region, determine the artifact performance feature value corresponding to each suspected artifact region.

[0081] Step 203: Based on the artifact performance feature values, select the target artifact region from all suspected artifact regions, and determine the initial blur angle and initial blur size corresponding to each target artifact region.

[0082] Step 204: Based on the historical abdominal MRI images of the same respiratory stage as the image to be artifact-removed, which were obtained in advance, the initial blur angle and initial blur size corresponding to each target artifact region are corrected to obtain the target blur angle and target blur size corresponding to each target artifact region.

[0083] Step 205: Based on the blur kernel of the target blur angle and target blur size, artifact removal is performed on the target artifact region, thus realizing artifact removal of the image to be removed.

[0084] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. For example, as shown... Figure 3As shown, the computer device 300 includes: a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302, wherein when the processor 302 executes the computer program 303, the computer device can perform the aforementioned method for eliminating abdominal magnetic resonance respiratory artifacts.

[0085] Based on the same inventive concept as the above-described method embodiments, the present invention provides a server, including a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, causing the device to perform the above-described method for eliminating respiratory artifacts on abdominal magnetic resonance imaging.

[0086] Based on the same inventive concept as the above-described method embodiments, the present invention provides a computer program product comprising: computer program code, which, when run on a computer, causes the computer to execute the above-described method for eliminating respiratory artifacts on abdominal magnetic resonance imaging.

[0087] Based on the same inventive concept as the above-described method embodiments, the present invention provides a computer-readable storage medium storing computer program code, which, when executed on a computer, causes the computer to perform the above-described method for eliminating respiratory artifacts on abdominal magnetic resonance imaging.

[0088] In summary, this invention quantifies the artifact performance characteristics of each suspected artifact region by analyzing its shape, gradient distribution, and grayscale distribution. This enables the identification of target artifact regions and improves the accuracy of artifact region identification, thereby enhancing the rationality of artifact removal. Secondly, by analyzing historical abdominal MRI images with the same respiratory stage as the image to be removed, the initial blur angle and initial blur size for each target artifact region are corrected. This improves the rationality of the blur angle and blur size settings to a certain extent, further enhancing the rationality of artifact removal.

[0089] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A system for eliminating respiratory artifacts on abdominal magnetic resonance imaging, characterized in that, The system includes: The acquisition and filtering module is used to acquire the image to be artifact-removed and filter out suspected artifact regions from the image to be artifact-removed. The artifact performance feature value determination module is used to determine the artifact performance feature value corresponding to each suspected artifact region based on the gradient distribution and grayscale distribution within each suspected artifact region, as well as the shape of each suspected artifact region. The initial parameter determination module is used to filter out the target artifact region from all suspected artifact regions based on the artifact performance characteristic value, and determine the initial blur angle and initial blur size corresponding to each target artifact region. The parameter correction module is used to correct the initial blur angle and initial blur size of each target artifact region based on the pre-acquired historical abdominal magnetic resonance image with the same respiratory stage as the image to be eliminated, so as to obtain the target blur angle and target blur size of each target artifact region. The artifact removal module is used to remove artifacts in the target artifact region based on the blur kernel of the target blur angle and target blur size, thus realizing the artifact removal of the image to be removed. The step of determining the artifact performance feature value corresponding to each suspected artifact region based on the gradient distribution and grayscale distribution within each suspected artifact region, as well as the shape of each suspected artifact region, includes: determining the edge blur dynamic value corresponding to each suspected artifact region based on the gradient values ​​and grayscale values ​​corresponding to all edge pixels in each suspected artifact region; determining the shape feature value corresponding to each suspected artifact region based on the minimum bounding rectangle of each suspected artifact region and the gradient direction corresponding to all pixels in each suspected artifact region; and normalizing the product between the edge blur dynamic value and the shape feature value corresponding to each suspected artifact region to obtain the artifact performance feature value corresponding to each suspected artifact region.

2. The abdominal magnetic resonance respiratory artifact elimination system according to claim 1, characterized in that, The step of filtering out suspected artifact regions from the image to be artifact-removed includes: Based on the gradient value and grayscale distribution of each pixel in the image to be artifact-removed, the initial value of the blur artifact feature corresponding to each pixel in the image to be artifact-removed is determined. If the initial value of the blur artifact feature corresponding to a pixel is greater than the preset blur artifact threshold, then the pixel is identified as a suspected artifact edge pixel. Based on the suspected artifact edge pixels in the image to be artifact-removed, a suspected artifact region is constructed.

3. The abdominal magnetic resonance respiratory artifact elimination system according to claim 2, characterized in that, The step of constructing initial values ​​for blur artifact features corresponding to each pixel in the image to be removed, based on the gradient value and grayscale distribution of each pixel, includes: Any pixel in the image to be artifact removed is designated as a marker pixel. The gradient value corresponding to the marker pixel and the entropy value of the gray values ​​of all pixels in the preset neighborhood of the marker pixel are used to determine the initial value of the blur artifact feature corresponding to the marker pixel.

4. The abdominal magnetic resonance respiratory artifact elimination system according to claim 1, characterized in that, The step of determining the dynamic edge blur value corresponding to each suspected artifact region based on the gradient value and grayscale value corresponding to all edge pixels on each suspected artifact region includes: Any suspected artifact region in the image to be artifact removed is identified as a marked suspected artifact region, and the absolute value of the difference between the gradient values ​​corresponding to every two edge pixels in the marked suspected artifact region is identified as the target gradient difference, thus obtaining the target gradient difference set corresponding to the marked suspected artifact region. The entropy value of the grayscale value of all pixels in the preset neighborhood corresponding to each edge pixel in the marked suspected artifact region is determined as the grayscale representative entropy value of each edge pixel in the marked suspected artifact region. The edge blur dynamic value corresponding to the marked suspected artifact region is determined based on the largest target gradient difference in the target gradient difference set corresponding to the marked suspected artifact region, the average gradient value of all edge pixels in the marked suspected artifact region, and the average gray-level entropy value of all edge pixels in the marked suspected artifact region.

5. The abdominal magnetic resonance respiratory artifact elimination system according to claim 1, characterized in that, The step of determining the shape feature value corresponding to each suspected artifact region based on the minimum bounding rectangle of each suspected artifact region and the gradient direction corresponding to all pixels within each suspected artifact region includes: The angle corresponding to the gradient direction of each pixel in each suspected artifact region is determined as the gradient direction angle of each pixel in each suspected artifact region. Based on the aspect ratio of the minimum bounding rectangle of each suspected artifact region and the standard deviation of the gradient direction angles of all pixels within each suspected artifact region, the shape feature value corresponding to each suspected artifact region is determined.

6. The abdominal magnetic resonance respiratory artifact elimination system according to claim 1, characterized in that, The step of filtering out the target artifact region from all suspected artifact regions based on artifact performance characteristic values ​​includes: If the artifact performance characteristic value corresponding to the suspected artifact region is greater than the preset artifact performance threshold, then the suspected artifact region is determined as the target artifact region.

7. The abdominal magnetic resonance respiratory artifact elimination system according to claim 1, characterized in that, Determining the initial blur angle and initial blur size corresponding to each target artifact region includes: Perform a Fast Fourier Transform on each target artifact region to obtain the Fourier spectrum corresponding to each target artifact region; By using Radon transform, the direction of the stripes in the Fourier spectrum corresponding to each target artifact region is detected, and the initial blur angle and initial blur size corresponding to each target artifact region are obtained.

8. The abdominal magnetic resonance respiratory artifact elimination system according to claim 1, characterized in that, The step involves correcting the initial blur angle and initial blur size of each target artifact region based on a pre-acquired historical abdominal MRI image corresponding to the same respiratory stage as the image to be artifact-removed, to obtain the target blur angle and target blur size for each target artifact region, including: Any target artifact region in the image to be artifact-removed is determined as the artifact region to be determined, and each historical abdominal MRI image with the same respiratory stage as the image to be artifact-removed is determined as the reference image. Identify historical artifact regions from each reference image and determine the initial blur angle and initial blur size for each historical artifact region; Based on the initial blur angle and initial blur size corresponding to each historical artifact region in each reference image, and the initial blur angle and initial blur size corresponding to the undetermined artifact region, the target blur angle and target blur size corresponding to the undetermined artifact region are determined.

9. The abdominal magnetic resonance respiratory artifact elimination system according to claim 8, characterized in that, The step of determining the target blur angle and target blur size corresponding to the undetermined artifact region based on the initial blur angle and initial blur size corresponding to each historical artifact region in each reference image, and the initial blur angle and initial blur size corresponding to the undetermined artifact region, includes: Based on the respiratory standard degree of each reference image to which it belongs, the mean and standard deviation of the initial blur angles corresponding to all historical artifact regions in each reference image, and the initial blur angles corresponding to the undetermined artifact regions, the angle reference weights corresponding to each reference image are determined. Based on the respiratory standard degree of each reference image to which it belongs, the mean and standard deviation of the initial blur size corresponding to all historical artifact regions in each reference image, and the initial blur size corresponding to the undetermined artifact region, the size reference weight corresponding to each reference image is determined. The target blur angle corresponding to the undetermined artifact region is determined based on the average of the angle reference weights corresponding to each reference image and the initial blur angles corresponding to all historical artifact regions within them, as well as the initial blur angles corresponding to the undetermined artifact region. The target blur size of the undetermined artifact region is determined based on the average of the size reference weights corresponding to each reference image and the initial blur size of all historical artifact regions within it, as well as the initial blur size of the undetermined artifact region.

Images