Self-navigating cardiac magnetic resonance imaging method and system based on adaptive sector sampling

Abstract

The application provides a self-navigation cardiac magnetic resonance imaging method and system based on adaptive sector sampling, comprising: establishing an ideal square kyz plane, a real rectangular kyz plane and an affine transformation relationship between the two; dividing the real rectangular kyz plane into n concentric ring regions; forming radial spoke ideal trajectories rotating along a golden angle in the ideal square kyz plane; mapping the radial spoke ideal trajectories to the real rectangular kyz plane to generate basic trajectories; establishing a sector search window in each spoke near angle of the basic trajectories, forming n sector ring intersection regions with the concentric ring regions, optimizing the positions of sampling points in each sector ring to obtain optimized sampling trajectories; and performing scanning according to the adjusted optimized sampling trajectories, estimating a respiratory motion field after binning the collected k-space data and reconstructing a three-dimensional image. The application can reduce the k-space sampling repetition rate, improve the sampling efficiency and form better imaging quality in a shorter imaging time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of magnetic resonance imaging, in particular to a three-dimensional self-navigation cardiac magnetic resonance imaging method and system based on k-space adaptive variable-density sector sampling. BACKGROUND

[0002] Three-dimensional cardiac magnetic resonance imaging technology is a technology for collecting three-dimensional images of the heart using magnetic resonance scanning, which can be applied to coronary artery imaging, left atrial delayed gadolinium enhancement imaging (LGE), and three-dimensional movie imaging. The main challenge of three-dimensional cardiac magnetic resonance imaging is to handle the heartbeat and respiratory motion of the heart. If these two motions are not handled in imaging, the cardiac magnetic resonance image will be blurred. For heartbeat motion, coronary artery imaging and left atrial delayed gadolinium enhancement imaging use electrocardiogram triggering for navigation, and three-dimensional movie imaging reconstructs images of different heart conditions separately. For respiratory motion, existing methods mostly use respiratory belts or diaphragm navigation technology for processing. These two methods use an additional respiratory belt signal or one-dimensional magnetic resonance signal of the diaphragm position for navigation, and only collect magnetic resonance signals at the end of expiration, thereby avoiding blurring of the magnetic resonance image due to respiratory motion.

[0003] However, the existing respiratory belt navigation and diaphragm navigation methods have the following problems: 1) These methods only use k-space data at the end of expiration, and the data acquisition efficiency is as low as less than 50%; 2) Since the data acquisition efficiency is related to the length of time when the subject is at the end of expiration, this leads to an uncertain length of time required for scanning, which is not conducive to clinical examination; 3) The respiratory belt signal and one-dimensional diaphragm signal cannot directly reflect the true respiratory motion, and using this signal for navigation may have errors in the classification of respiratory states.

[0004] Self-navigating cardiac magnetic resonance imaging (MRI) effectively avoids the above problems by continuously acquiring magnetic resonance k-space data and retrospectively reconstructing with motion compensation. The main trajectories used in three-dimensional self-navigating cardiac MRI scans include: G-CASPR trajectory, three-dimensional radial trajectory and stack-of-star trajectory. The G-CASPR trajectory (DOI: 10.1002 / jmri.24602) is a three-dimensional pseudo-spiral trajectory, and its generation principle is: fully sampling in the kx direction, dividing multiple concentric elliptical rings with equal area in the kyz plane, and numbering each ring in polar coordinates. Each trajectory line takes the points with increasing numbers in each ring in turn to obtain a spiral-like trajectory, and each two trajectory lines rotate at an approximate golden angle. The disadvantages of the G-CASPR trajectory are: 1) there are limitations on the shape and area of the imaging region, it is only used for scanning in the kyz plane elliptical field of view, the ring shape cannot be changed as needed, and the number of points in the imaging region must be adjusted to the product of the number of sampling spiral lines and the number of points in each spiral line, which cannot be flexibly adjusted as needed; 2) there is no different sampling strategy for the low-frequency and high-frequency regions of k-space, so the same motion state of k-space at the same position may be repeatedly sampled, and the redundancy is high. The three-dimensional radial trajectory and the stack-of-star trajectory belong to non-Cartesian trajectories, and their advantages are that the sampling efficiency is high and the acquisition speed is fast. Their disadvantages are: 1) when non-Cartesian trajectories are applied to fast imaging, frequent gradient switching is required, which causes serious eddy current effect, resulting in more eddy current artifacts in the reconstructed image; 2) the trajectory does not directly fill the k-space region, and additional steps are required for correction, and the reconstruction process is complex; 3) when applied to free-breathing sequences, the trajectory is unevenly distributed in each breathing state, making image reconstruction difficult.

[0005] To accurately reconstruct the motion-compensated cardiac magnetic resonance image, it is necessary to ensure the uniformity of k-space sampling in each motion state to remove motion artifacts in the reconstructed magnetic resonance image. Due to the inconsistent breathing rhythm and amplitude of different subjects, the process of dividing k-space sampling points into different breathing states obtained by free-breathing sequence acquisition is random. If traditional three-dimensional radial, spiral trajectory and other trajectories are used in free-breathing magnetic resonance imaging sequence scanning, the distribution of k-space trajectory lines in each breathing state during motion compensation will be completely random, resulting in uneven distribution of k-space sampling data in each breathing state, causing serious image quality artifacts, which are difficult to remove by image reconstruction algorithm. Therefore, a k-space sampling trajectory is needed to uniformly sample k-space data in each breathing state under free-breathing state.

[0006] In the magnetic resonance imaging sequence of free breathing acquisition, since the breathing and heartbeat processes are continuous, the k-space low frequency signals of adjacent motion states have similarity. In the design of the trajectory, the similarity of the low frequency signals of adjacent motion states can be utilized to estimate the unacquired low frequency k-space by using the low frequency k-space of adjacent motion states. In order to fully utilize the similarity of the signals of adjacent motion states, it is required that the k-space trajectories acquired by adjacent motion states have as large difference as possible, so that the estimated k-space data is uniformly distributed in the k-space with the acquired k-space data. Therefore, a k-space sampling trajectory is required, which makes the k-space sampling masks of adjacent breathing states have large difference and uniform distribution under the condition of free breathing. SUMMARY

[0007] In view of the defects in the prior art, the purpose of the present application is to provide a self-navigation cardiac magnetic resonance imaging method and system based on adaptive sector sampling.

[0008] According to the self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling provided by the present application, the method comprises the following steps:

[0009] Step S1: establishing an ideal square kyz plane, a real rectangular kyz plane and an affine transformation relationship between the two planes;

[0010] Step S2: dividing the real rectangular kyz plane into n concentric ring-shaped regions based on a preset area ratio, wherein n is the number of kyz plane points collected per heartbeat of the trajectory;

[0011] Step S3: forming a radial spoke ideal trajectory rotating along a golden angle in the ideal square kyz plane;

[0012] Step S4: mapping the ideal trajectory into the real rectangular kyz plane by affine transformation to generate a basic trajectory;

[0013] Step S5: establishing a sector search window within an angle around each spoke of the basic trajectory to form n sector-ring intersection regions with the concentric ring-shaped regions, and optimizing the positions of the sampling points in each sector ring based on the principle of historical sampling frequency and low frequency region near the k-space center to obtain an optimized sampling trajectory;

[0014] Step S6: performing scanning according to the adjusted optimized sampling trajectory, performing binning on the acquired k-space data by a self-gated binning algorithm, estimating a breathing motion field based on the binned data, and reconstructing a three-dimensional image by a motion compensation image reconstruction algorithm.

[0015] Preferably, the step S1 comprises the following sub-steps:

[0016] Step S1.1: Establish an ideal square kyz plane with an aspect ratio of 1:1;

[0017] Step S1.2: Establish a real rectangular kyz plane, the number of long side points of the real rectangular kyz plane is equal to the actual sampling point number in the ky direction, and the number of wide side points is equal to the actual sampling point number in the kz direction;

[0018] Step S1.3: Establish an affine transformation relationship between the real rectangular kyz plane and the ideal square kyz plane, for the polar coordinate angle in the ideal square kyz plane , after affine transformation, it has:

[0019]

[0020] , wherein, is the polar coordinate angle corresponding to the real rectangular kyz plane. ny and nz represent the actual sampling point number in the ky and kz directions of the real rectangular kyz plane respectively.

[0021] Preferably, the step S2 comprises the following steps:

[0022] Step S2.1: Set a concentric ring shape, including but not limited to an elliptical ring, a rectangular ring;

[0023] Step S2.2: Set the number of each heartbeat collection points n;

[0024] Step S2.3: Set a concentric ring area proportion function, by adjusting the area function of the concentric ring, to reduce the sampling redundancy of the high frequency region of k space in different motion phase, improve the sampling rate of the low frequency region of k space, and realize adaptive sampling of low frequency region high density and high frequency region low density;

[0025] Step S2.4: According to the area proportion function obtained by the concentric ring and the size of the real rectangular kyz plane, calculate the radius of each ring, and divide the real rectangular kyz plane into n variable density concentric rings with specified size, number and shape.

[0026] Preferably, the step S3 comprises the following sub-steps:

[0027] Step S3.1: In the ideal square kyz plane, each heartbeat collects a spoke from the k space center to the surrounding along a fixed direction;

[0028] Step S3.2: Rotate the current track spoke by a golden angle in the clockwise or counterclockwise direction to obtain the next track spoke;

[0029] Repeat steps S3.1 and S3.2 until the track meets the sampling rate requirement, and record the polar coordinate angle corresponding to the kth ideal track line in the ideal square kyz plane as .

[0030] Preferably, the step S4 comprises the following sub-steps:

[0031] Step S4.1: for each ideal trajectory spoke in the ideal square kyz plane, calculate the real rectangular kyz plane's basis trajectory spoke angle according to the polar coordinate angle, using the affine transformation obtained in step S1, for the kth ideal trajectory spoke, the corresponding polar coordinate angle of the ideal trajectory spoke in the ideal square kyz plane is , and:

[0032]

[0033] wherein, is the corresponding polar coordinate angle of the kth basis trajectory spoke in the real rectangular kyz plane;

[0034] Step S4.2: according to the polar coordinate angle , obtain the kth basis trajectory spoke in the real rectangular kyz plane;

[0035] Repeat steps S4.1 and S4.2 until all ideal trajectories are mapped to the basis trajectories in the real rectangular kyz plane.

[0036] Preferably, the step S5 comprises:

[0037] Step S5.1: for each spoke of the basis trajectory, establish a sector search window within angle in its polar coordinate representation, and the n sector-ring intersection areas are formed by the n variable-density concentric annular areas, and the intersection area of the kth search window and the ith variable-density concentric annular area is denoted as the sector-ring search window area :

[0038]

[0039] wherein, represents the coordinate point in the polar coordinate representation, is the polar radius, is the polar angle, is the specified angle tolerance window half-width, i.e. the maximum absolute value of the acceptable deviation angle, represents the corresponding polar coordinate angle of the kth basis trajectory spoke in the real rectangular kyz plane, represents the ith variable-density concentric annular area;

[0040] Step S5.2: grid the basis trajectory to the Cartesian coordinate system, for the sector-ring search window area , obtain the ith candidate point set of the kth spoke , i.e.:

[0041]

[0042] wherein, denotes taking a set of discrete points in the Cartesian coordinate system within the region;

[0043] Step S5.3: confirming and optimizing the sampling point positions of the target trajectory, combining the historical visit times and the search rule, dynamically selecting the optimal sampling points from the candidate point set ;

[0044] Step S5.4: for the sequence that needs continuous data acquisition including three-dimensional cardiac cine imaging and is prone to eddy current, using the peripheral-center-peripheral alternating sampling sequence to recombine adjacent trajectory lines, the sampling points of the kth recombined trajectory line are:

[0045]

[0046] wherein, n is the total number of concentric rings, denotes the sampling point of the kth optimized trajectory on the ith variable-density concentric ring.

[0047] Preferably, the search rule is determined according to the trajectory application scenario, including preferentially selecting the point with the least cumulative sampling times before, preferentially selecting the point close to the central region for the low-frequency region near the k-space center , and preferentially selecting the point close to the central region for the high-frequency region outside the k-space periphery , and randomly selecting a single sampling point from the point set with the least sampling times.

[0048] Preferably, the step S6 comprises the following sub-steps:

[0049] Step S6.1: using the optimized sampling trajectory scanning to collect magnetic resonance k-space data;

[0050] Step S6.2: analyzing the k-space center signal principal components to obtain each principal component containing motion information;

[0051] Step S6.3: performing frequency analysis on each extracted principal component to obtain self-navigated respiratory and cardiac curves;

[0052] Step S6.4: after binning processing of the respiratory and cardiac curves, obtaining the under-sampled k-space data of each respiratory / cardiac dimension bin;

[0053] ​Step S6.5: using the k-space data of the respiratory dimension binning, performing coil sensitivity estimation and respiratory binning image reconstruction by using a parallel imaging reconstruction method to obtain images of each respiratory state; using an image registration method, specifying the end-expiratory phase as the target respiratory state, and registering the images of the remaining respiratory states to the target respiratory state to obtain the respiratory displacement field of the remaining respiratory state images registered to the target respiratory state;

[0054] Step S6.6: using the estimated respiratory displacement field, using a motion-compensated image reconstruction method to perform motion-compensated image reconstruction on the undersampled k-space data of each respiratory / heart dimension binning to obtain a three-dimensional cardiac magnetic resonance image with respiratory and heart motion artifacts removed.

[0055] Preferably, the analysis of the k-space center signal principal component includes:

[0056] Performing inverse Fourier transform on the full-sampling k-space center line of each trajectory acquisition to obtain a one-dimensional projection image of the current layer block along the frequency encoding direction, and the projection image reflects the motion state of the current layer block image along the frequency encoding direction;

[0057] Arranging the projection images obtained at each time along the time dimension to obtain a dynamic motion image of the projection image of the current layer block along the frequency encoding direction over time;

[0058] Performing principal component analysis on the obtained one-dimensional frequency encoding direction projection-time image to obtain each principal component containing motion information.

[0059] Preferably, step S6.4 includes removing abnormal data with a respiratory curve amplitude exceeding a preset threshold of 5% according to the self-navigation respiratory curve, equally dividing the respiratory curve according to the curve height, and dividing the k-space data corresponding to the respiratory curve points into corresponding respiratory bins according to the height of the points on the respiratory curve to complete the k-space data binning in the respiratory dimension.

[0060] According to the self-navigation heart curve, recording the interval between every two maximum values, and recording the interval distance between each point and the previous maximum value point, and dividing the k-space data corresponding to each heart curve point into corresponding heart phase bins according to the distance to complete the k-space data binning in the heart dimension.

[0061] After extracting the self-navigation respiratory curve and heart curve using the k-space center signal, the collected k-space data is simultaneously subjected to respiratory and heart binning, that is, the undersampled k-space data of each respiratory / heart dimension binning is obtained.

[0062] According to the self-navigation cardiac magnetic resonance imaging system based on adaptive sector sampling provided by the application, comprising:

[0063] Module M1: establish ideal square kyz plane, real rectangular kyz plane and affine transformation relationship between the two;

[0064] Module M2: divide the real rectangular kyz plane into n concentric ring-shaped areas based on a preset area ratio, wherein n is the number of kyz plane points collected for each heartbeat of the trajectory;

[0065] Module M3: form a radial spoke ideal trajectory rotating along a golden angle in the ideal square kyz plane;

[0066] Module M4: map the ideal trajectory into the real rectangular kyz plane through affine transformation to generate a basic trajectory;

[0067] Module M5: establish a sector search window within an angle around each spoke of the basic trajectory Module M5: establish a sector search window within an angle around each spoke of the basic trajectory

[0068] Module M6: scan according to the adjusted optimized sampling trajectory, perform binning on the collected k-space data through a self-gated binning algorithm, estimate a respiratory motion field based on the binned data, and reconstruct a three-dimensional image through a motion-compensated image reconstruction algorithm.

[0069] Compared with the prior art, the present application has the following beneficial effects:

[0070] Compared with the prior art, the present application has the following beneficial effects: BRIEF DESCRIPTION OF DRAWINGS

[0071] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments, read in conjunction with the accompanying drawings:

[0072] Figure 1 is a working method flowchart of the present application;

[0073] Figure 2 is a trajectory generation principle diagram of the k-space adaptive variable-density sector sampling trajectory generation method of the present application suitable for three-dimensional self-navigation cardiac magnetic resonance imaging;

[0074] Figure 3A schematic diagram for the application of the present application to free-breathing three-dimensional cardiac coronary imaging;

[0075] Figure 4 A comparison diagram for three-dimensional cardiac imaging reconstructed by the present application and the comparative method G-CASPR at the same acceleration rate. DETAILED DESCRIPTION

[0076] The present application will be described in detail below with specific embodiments. The following examples will help those skilled in the art to further understand the present application, but do not limit the present application in any form. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of changes and improvements can be made. These all belong to the protection scope of the present application.

[0077] Example 1

[0078] According to the self-navigation cardiac magnetic resonance imaging method based on adaptive fan sampling provided by the present application, as shown in the figure, it comprises: Figure 1

[0079] Step S1: Establish an ideal square kyz plane, a real rectangular kyz plane and an affine transformation relationship between the two, wherein the length and width of the real rectangular kyz plane are determined by the actual sampling point numbers ny and nz in the ky and kz directions. The step S1 comprises the following sub-steps:

[0080] Step S1.1: Establish an ideal square kyz plane with an aspect ratio of 1:1.

[0081] Step S1.2: Establish a real rectangular kyz plane, the number of long side points of the real rectangular kyz plane is equal to the actual sampling point number in the ky direction, and the number of wide side points is equal to the actual sampling point number in the kz direction.

[0082] Step S1.3: Establish an affine transformation relationship between the real rectangular kyz plane and the ideal square kyz plane, for the polar coordinate angle , after affine transformation, it has:

[0083]

[0084] , wherein is the corresponding polar coordinate angle of the real rectangular kyz plane. ny and nz represent the actual sampling point numbers in the ky and kz directions of the real rectangular kyz plane, respectively.

[0085] Step S2: Divide the real rectangular kyz plane into n concentric annular regions based on a preset area ratio, wherein n is the kyz plane point number collected by each heartbeat of the trajectory. The step S2 comprises the following sub-steps: ​

[0086] Step S2.1: Set the concentric ring shape, including but not limited to an elliptical ring, a rectangular ring.

[0087] Step S2.2: Set the number n of each heartbeat acquisition point.

[0088] Step S2.3: Set the concentric ring area ratio function, adjust the area function of the concentric ring, reduce the sampling redundancy of the high-frequency region of the k-space in different motion phases, and improve the sampling rate of the low-frequency region of the k-space, to realize adaptive sampling of high density in the low-frequency region and low density in the high-frequency region.

[0089] Step S2.4: According to the area ratio function obtained by the concentric ring and the real rectangular kyz plane size, calculate the radius of each ring, and divide the real rectangular kyz plane into n variable-density concentric rings of specified size, number and shape.

[0090] Step S3: Form a radial spoke ideal trajectory rotating along a golden angle in the ideal square kyz plane. The step S3 includes the following sub-steps:

[0091] Step S3.1: In the ideal square kyz plane, each heartbeat acquires one spoke from the k-space center to the surrounding along a fixed direction.

[0092] Step S3.2: Rotate the current trajectory spoke by a golden angle in a clockwise or counterclockwise direction to obtain the next trajectory spoke.

[0093] Repeat steps S3.1 and S3.2 until the trajectory meets the sampling rate requirement, and record the corresponding polar coordinate angle of the kth ideal trajectory line in the ideal square kyz plane as .

[0094] Step S4: Map the ideal trajectory to the real rectangular kyz plane by affine transformation to generate a base trajectory. The step S4 includes the following sub-steps:

[0095] Step S4.1: For each ideal trajectory spoke in the ideal square kyz plane, calculate the spoke angle of the base trajectory in the real rectangular kyz plane using the affine transformation obtained in step S1 according to the polar coordinate angle, and for the kth ideal trajectory spoke, the corresponding polar coordinate angle in the ideal square kyz plane is , and

[0096]

[0097] wherein, is the corresponding polar coordinate angle of the kth base trajectory spoke in the real rectangular kyz plane.

[0098] Step S4.2: According to the polar coordinate angle This yields the k-th basic trajectory spoke within the true rectangular kyz plane.

[0099] Repeat steps S4.1 and S4.2 until all ideal trajectories are mapped to the base trajectories in the real rectangular kyz plane.

[0100] Step S5: Near each spoke of the base trajectory A fan-shaped search window is established within the angle, forming n fan-ring intersection regions with the concentric annular region. Within each fan-ring, the position of the sampling point is optimized based on the historical sampling count and the principle of prioritizing low-frequency regions near the center of k space, thus obtaining the optimized sampling trajectory. Step S5 includes:

[0101] Step S5.1: For each spoke of the basic trajectory, in its vicinity in polar coordinates... A fan-shaped search window is established within the angle, forming n fan-ring intersection regions with n concentric ring regions of varying density. The intersection region of the k-th search window and the ith concentric ring of varying density is denoted as the fan-ring search window region. :

[0102]

[0103] in, This represents the coordinates of a point in polar coordinates. Polar radius, Polar angle, The tolerance window half-width for a specified angle, i.e., the maximum absolute value of the acceptable deviation angle. This represents the polar coordinate angle of the k-th basic trajectory spoke within the real rectangular kyz plane. This represents the i-th concentric ring region with varying density. In polar coordinates, it represents the polar radius of a point within the target sector search window region. The formula expresses: in polar coordinates, if the point in polar coordinates... Satisfies: ① The polar coordinate angle of the point lies within the angle tolerance window around the target angle (corresponding to the formula). ); ② This point is located within the i-th variable-density concentric ring region (corresponding to the formula in the formula). If the value is 0, then the point is accepted into the fan-ring search window area.

[0104] Step S5.2: Grid the basic trajectory to a Cartesian coordinate system, targeting the fan-ring search window region. This yields the i-th candidate point set for the k-th spoke. ,Right now:

[0105]

[0106] in, Indicates taking A set of discrete points in the region in a Cartesian coordinate system.

[0107] Step S5.3: Confirm and optimize the sampling point positions of the target trajectory, dynamically select the optimal sampling points from the candidate point set according to the historical access times and the search rule of giving priority to the low-frequency region near the k-space center . The specific search rule is determined according to the application scenario of the trajectory, including but not limited to:

[0108] (1) Preferentially select the point with the least cumulative sampling times before.

[0109] (2) For the low-frequency region near the center of the k-space , preferentially select the point close to the center region.

[0110] (3) For the high-frequency region outside the k-space , randomly select a single sampling point from the set of points with the least sampling times.

[0111] Step S5.4: For sequences that need to continuously collect data and are prone to eddy currents, including three-dimensional cardiac cine imaging, adopt an alternating sampling sequence of "periphery-center-periphery" to recombine adjacent trajectory lines, so as to reduce the gradient switching distance and thus reduce the eddy current of fast sampling. The sampling points of the kth trajectory line after recombination are:

[0112]

[0113] wherein n is the total number of concentric rings, represents the sampling point of the kth optimized trajectory on the ith variable-density concentric ring.

[0114] Step S6: Scan according to the adjusted optimized sampling trajectory, perform binning on the collected k-space data through a self-gated binning algorithm, estimate the respiratory motion field based on the binned data, and reconstruct a three-dimensional image through a motion-compensated image reconstruction algorithm. The scanning and motion-compensated image reconstruction in the step S6 include the following sub-steps:

[0115] Step S6.1: Scan using the optimized sampling trajectory to collect magnetic resonance k-space data.

[0116] ​Step S6.2: k-space center signal principal component analysis, inverse Fourier transform is performed on the full-sampling k-space center line collected by each trajectory to obtain a one-dimensional projection image of the layer block along the frequency encoding direction, which reflects the motion state of the current layer block image along the frequency encoding direction; the projection image obtained continuously at each time is arranged along the time dimension, thereby obtaining a dynamic motion image of the projection image of the layer block along the frequency encoding direction over time; principal component analysis is performed on the obtained one-dimensional frequency encoding direction projection-time image to obtain principal components containing motion information.

[0117] Step S6.3: motion phase extraction, frequency analysis is performed on the extracted principal components, and principal components conforming to the breathing frequency, heartbeat frequency, etc. are extracted for filtering according to the breathing frequency, heartbeat frequency, etc. to obtain self-navigation breathing curves and heartbeat curves, etc.

[0118] Step S6.4: original data breathing / heartbeat motion binning, according to the self-navigation breathing curve, abnormal data with a breathing curve amplitude exceeding a preset threshold of 5% is removed, and the breathing curve is equally divided into m bins according to the curve height, where m is the number of divided breathing bins; according to the height of the point on each breathing curve, the k-space data corresponding to the breathing curve point is divided into the corresponding breathing bin to complete the k-space data binning in the breathing dimension; according to the self-navigation heartbeat curve, the interval between each two maximum values is recorded, and the interval distance between each point and the previous maximum value point (i.e. the number of corresponding trajectory lines collected) is recorded starting from the first maximum value point, and the k-space data corresponding to each heartbeat curve point is divided into the corresponding heartbeat phase bin according to the distance to complete the k-space data binning in the heartbeat dimension; at the same time, after the self-navigation breathing curve and heartbeat curve are extracted using the k-space center signal, the collected k-space data is subjected to breathing and heartbeat binning, thereby obtaining undersampling k-space data in each breathing / heartbeat dimension bin.

[0119] Step S6.5: breathing displacement field estimation, using the k-space data in the breathing dimension bin, coil sensitivity estimation and image reconstruction of the breathing bin are performed by using a traditional parallel imaging reconstruction method to obtain images of each breathing state, which do not distinguish the heartbeat state; by using an image registration method, the end-expiratory phase is specified as the target breathing state, and the images of the remaining breathing states are registered to the target breathing state to obtain the breathing displacement field of the registration of the images of the remaining breathing states to the target breathing state.

[0120] Step S6.6: using the estimated breathing displacement field, motion-compensated image reconstruction is performed on the undersampling k-space data in each breathing / heartbeat dimension bin by using a motion-compensated image reconstruction method including but not limited to MoCo-SENSE and MoCo-LLR-SENSE, thereby obtaining a three-dimensional cardiac magnetic resonance image in which the breathing and heartbeat artifacts are removed.

[0121] Embodiment 2

[0122] An embodiment of the application applied to 3D cardiac coronary imaging will be described in detail below with reference to the accompanying drawings. Figure 1 The flowchart of the embodiment of the application is as follows, Figure 2 The principle diagram of trajectory generation of the application is as follows. Specifically, the embodiment includes the following steps:

[0123] Step S1: establishing an ideal square kyz plane, a real rectangular kyz plane and an affine transformation relationship between the two, wherein the length and width of the real rectangular kyz plane are determined by the actual sampling point numbers ny and nz in the ky and kz directions, including the following steps:

[0124] Step S1.1: establishing an ideal square kyz plane with an aspect ratio of 1:1.

[0125] Step S1.2: establishing a real rectangular kyz plane with the number of long side points equal to the actual sampling point number 266 in the ky direction and the number of wide side points equal to the actual sampling point number 88 in the kz direction.

[0126] Step S1.3: establishing an affine transformation relationship between the real rectangular kyz plane and the ideal square kyz plane, for the polar coordinate angle in the ideal square kyz plane, after affine transformation, there is:

[0127]

[0128] wherein is the corresponding polar coordinate angle of the real rectangular kyz plane.

[0129] Step S2: dividing the real rectangular kyz plane into n concentric ring regions based on a preset area ratio, wherein n is the number of kyz plane points collected per heartbeat of the trajectory, including the following steps:

[0130] Step S2.1: setting the concentric ring shape as an elliptical ring.

[0131] Step S2.2: setting the number of points collected per heartbeat n according to the subject's heart rate and the length of time when the heart coronary is not moving, and n can take values including: 20, 24, 29, 34.

[0132] Step S2.3: setting a concentric ring area ratio function, designing the center ring area as 2000 / n and the outermost ring area as 34769 / n, and distributing the areas of the intermediate rings according to an arithmetic sequence.

[0133] Step S2.4: calculating the radii of the rings according to the area ratio function of the concentric rings and the size of the real rectangular kyz plane, and dividing the real rectangular kyz plane into n variable-density concentric rings with specified size, number and shape.

[0134] Step S3: Forming the radial spoke ideal trajectory rotating along the golden angle in the ideal square kyz plane, comprising the following steps:

[0135] Step S3.1: In the ideal square kyz plane, each heartbeat collects one spoke from the k-space center to the periphery along a fixed direction.

[0136] Step S3.2: Rotating the current trajectory spoke in the clockwise direction by a golden angle of 137.51°, which is in radian measure, to obtain the next trajectory spoke. The angle of the golden angle is obtained by golden section of , which can be expressed in radian measure as , that is, . The angle is also equal to in value. The golden angle is slightly larger than the commonly used golden angle ( ), which can also ensure uniform sampling when the number of sampling points is relatively small, and conforms to the characteristics of highly undersampled k-space data in each respiratory state for coronary artery imaging, and good results have been achieved in actual implementation.

[0137] Step S3.3: Repeat steps S3.1 and S3.2 until the trajectory can cover each variable density concentric ring after affine transformation and optimization in subsequent steps, and the polar coordinate angle corresponding to the kth ideal trajectory line in the ideal square kyz plane is denoted as .

[0138] Step S4: Mapping the ideal trajectory to the real rectangular kyz plane through affine transformation to generate the base trajectory, comprising the following steps:

[0139] Step S4.1: For each ideal trajectory spoke in the ideal square kyz plane, calculate the spoke angle of the base trajectory in the real rectangular kyz plane using the affine transformation obtained in step S1 according to the polar coordinate angle. For the kth ideal trajectory spoke, the polar coordinate angle corresponding to it in the ideal square kyz plane is , and

[0140]

[0141] where is the polar coordinate angle corresponding to the kth base trajectory spoke in the real rectangular kyz plane.

[0142] Step S4.2: Obtain the kth base trajectory spoke in the real rectangular kyz plane according to the polar coordinate angle .

[0143] Step S4.3: Repeat step S4.1, S4.2 until all ideal trajectories are mapped to the base trajectories in the real rectangular kyz plane.

[0144] Step S5: For each spoke of the base trajectory, a sector search window is established around it, intersecting with the n concentric annular regions to form n sector-annular intersection regions, and in each sector-annular region, the position of the sampling point is optimized based on the principle of historical sampling frequency and low-frequency region near the k-space center, to obtain an optimized sampling trajectory, including the following steps:

[0145] Step S5.1: For each spoke of the base trajectory, a sector search window is established around it in the polar coordinate representation within an angle, intersecting with the n variable-density concentric annular regions to form n sector-annular intersection regions, and the intersection region of the kth search window and the ith variable-density concentric annular region is denoted as the sector-annular search window region

[0146]

[0147] wherein, represents the coordinate point in polar coordinate representation, is the polar radius, is the polar angle, is the specified angle tolerance window half-width, i.e., the maximum absolute value of the acceptable deviation angle, represents the corresponding polar coordinate angle of the kth base trajectory spoke in the real rectangular kyz plane, represents the ith variable-density concentric annular region.

[0148] Step S5.2: Grid the base trajectory to the Cartesian coordinate system, and for the sector-annular search window region , obtain the ith candidate point set of the kth spoke , i.e.:

[0149]

[0150] wherein represents taking the set of discrete points in the Cartesian coordinate system within the region.

[0151] Step S5.3: Confirm and optimize the sampling point position of the target trajectory, dynamically select the optimal sampling point from the candidate point set combining the search rules of historical access frequency and low-frequency region near the k-space center, and the specific search rules are determined according to the trajectory application scenario, including:

[0152] (1) Preferentially select the point with the least cumulative sampling frequency before. ​​​

[0153] (2) For the low-frequency region near the center of k-space , prefer to select points near the center region.

[0154] (3) For the high-frequency region of the outer periphery of k-space , randomly select a single sampling point from the set of points with the least number of samples.

[0155] Step S6: According to the adjusted optimized sampling trajectory, scanning is performed, the k-space data collected is binned through the self-gated binning algorithm, the respiratory motion field is estimated based on the binned data, and the three-dimensional cardiac coronary magnetic resonance image is reconstructed through the motion compensated image reconstruction algorithm, including the following steps:

[0156] Step S6.1: Using the optimized sampling trajectory, a gradient echo sequence triggered by electrocardiogram gating is used to collect three-dimensional cardiac coronary magnetic resonance k-space data.

[0157] Step S6.2: k-space center signal principal component analysis, inverse Fourier transform is performed on the full-sampling k-space center line collected by each trajectory to obtain a one-dimensional projection image of the layer block along the frequency encoding direction. The projection image reflects the motion state of the current layer block image along the frequency encoding direction. The projection image obtained continuously at each time is arranged along the time dimension, i.e. the dynamic motion image of the projection image of the layer block along the frequency encoding direction is obtained. The one-dimensional frequency encoding direction projection-time image obtained is subjected to principal component analysis to obtain principal components containing motion information.

[0158] Step S6.3: Respiratory motion phase extraction, frequency analysis is performed on the extracted principal components, and the principal components conforming to the respiratory frequency are extracted, filtered according to the respiratory frequency, and the self-navigation respiratory curve is obtained.

[0159] Step S6.4: Raw data respiratory motion binning, according to the self-navigation respiratory curve, abnormal data with an amplitude exceeding a preset threshold of 5% is removed, the respiratory curve is equally divided according to the curve height, and m is the number of respiratory bins divided; according to the height of the points on the respiratory curve, the k-space data corresponding to the points on the respiratory curve is divided into corresponding respiratory bins (Bin), the k-space data is binned in the respiratory dimension, and undersampled k-space data of each respiratory dimension bin is obtained;

[0160] Step S6.5: Respiratory displacement field estimation, using the respiratory dimension binned k-space data, using sensitivity encoding (SENSE) method to perform coil sensitivity estimation and respiratory binned image reconstruction to obtain images of each respiratory state, which does not distinguish the cardiac state; using image registration method, specifying the end-expiratory phase as the target respiratory state, and registering the images of the remaining respiratory states to the target respiratory state to obtain the respiratory displacement field of the remaining respiratory state images registered to the target respiratory state.

[0161] Step S6.6: Using the estimated respiratory displacement field, using motion compensated image reconstruction method MoCo-SENSE to perform motion compensated image reconstruction on each respiratory / cardiac dimension binned undersampled k-space data to obtain three-dimensional cardiac magnetic resonance images with respiratory artifacts removed, and using two-point Dixon water-fat separation algorithm to obtain three-dimensional cardiac coronary magnetic resonance images.

[0162] Figure 3 The results of this embodiment after water-fat separation are shown in the schematic diagram. The first column is the three-dimensional cardiac coronary magnetic resonance imaging coronal image and transverse image generated using the trajectory acquisition method, and the second column is the left anterior descending coronary artery and right coronary artery obtained after surface reconstruction using the three-dimensional image, wherein LAD represents the left anterior descending coronary artery, and RCA represents the right coronary artery.

[0163] Figure 4 The results of this embodiment and the comparative method G-CASPR are compared under the same acceleration undersampling condition. When the acceleration rate is set to 3.2 times, the first column is the three-dimensional cardiac magnetic resonance anti-phase coronal image and sampling trajectory mask generated using the 3.2 times accelerated trajectory acquisition method proposed in the present application, and the second column is the three-dimensional cardiac magnetic resonance anti-phase coronal image and sampling trajectory mask generated using the G-CASPR method. Compared with the comparative method G-CASPR, the trajectory generated using the method proposed in the present application has fewer reconstruction image artifacts and better image quality.

[0164] For free-breathing three-dimensional cardiac magnetic resonance imaging, given any shape field of view and any resolution condition, based on the present application, the sampling trajectory of k-space can be calculated by the above steps. Compared with other free-breathing three-dimensional cardiac magnetic resonance imaging trajectories, the present application increases the applicability under the condition of any rectangular field of view and any resolution, reduces the k-space sampling repetition rate and improves the sampling efficiency through secondary search and optimization of each sampling point of the basic trajectory, forms better imaging quality in a shorter imaging time, and has better practical value in clinical free-breathing three-dimensional cardiac magnetic resonance imaging.

[0165] Embodiment 3

[0166] The application also provides a three-dimensional self-navigation cardiac magnetic resonance imaging system, which can be realized by performing the flow steps of the three-dimensional self-navigation cardiac magnetic resonance imaging method, i.e., the three-dimensional self-navigation cardiac magnetic resonance imaging method can be understood by those skilled in the art as a preferred embodiment of the three-dimensional self-navigation cardiac magnetic resonance imaging system.

[0167] According to the application, a self-navigation cardiac magnetic resonance imaging system based on adaptive sector sampling is provided, comprising:

[0168] Module M1: establish an ideal square kyz plane, a real rectangular kyz plane and an affine transformation relationship therebetween. The module M1 comprises the following sub-modules: Module M1.1: establish an ideal square kyz plane with an aspect ratio of 1:1. Module M1.2: establish a real rectangular kyz plane, the number of long side points of the real rectangular kyz plane is equal to the actual sampling point number in the ky direction, and the number of wide side points is equal to the actual sampling point number in the kz direction. Module M1.3: establish an affine transformation relationship between the real rectangular kyz plane and the ideal square kyz plane, for the polar coordinate angle in the ideal square kyz plane , after affine transformation, there is: , wherein, is the polar coordinate angle corresponding to the real rectangular kyz plane. ny and nz respectively represent the actual sampling point number in the ky and kz directions of the real rectangular kyz plane.

[0169] Module M2: divide the real rectangular kyz plane into n concentric annular regions based on a preset area ratio, wherein n is the kyz plane point number collected by each heartbeat of the trajectory.

[0170] Module M3: form a radial spoke ideal trajectory rotating along the golden angle in the ideal square kyz plane. The module M3 comprises the following sub-modules: Module M3.1: in the ideal square kyz plane, each heartbeat collects one spoke from the k-space center to the periphery along a fixed direction. Module M3.2: rotate the current trajectory spoke by the golden angle in the clockwise or counterclockwise direction to obtain the next trajectory spoke. Repeat the modules M3.1 and M3.2 until the trajectory meets the sampling rate requirement, and record the polar coordinate angle corresponding to the kth ideal trajectory line in the ideal square kyz plane as .

[0171] Module M4: mapping the ideal trajectories to the real rectangular kyz plane by affine transformation to generate the base trajectories. The module M4 includes the following sub-modules: Module M4.1: for each ideal trajectory spoke in the ideal rectangular kyz plane, calculate the spoke angle of the base trajectory in the real rectangular kyz plane using the affine transformation obtained by module M1 according to the polar coordinate angle, for the kth ideal trajectory spoke, the corresponding polar coordinate angle of the ideal trajectory spoke in the ideal rectangular kyz plane is , and wherein, is the corresponding polar coordinate angle of the kth base trajectory spoke in the real rectangular kyz plane. Module M4.2: obtain the kth base trajectory spoke in the real rectangular kyz plane according to the polar coordinate angle . Repeat the module M4.1 and the module M4.2 until all the ideal trajectories are mapped to the base trajectories in the real rectangular kyz plane.

[0172] Module M5: establish a sector search window around each spoke of the base trajectory within the angle to form n sector-ring intersection areas with the n variable-density concentric ring areas, and optimize the position of the sampling points in each sector ring based on the historical sampling frequency and the principle of giving priority to the low-frequency region near the k-space center to obtain the optimized sampling trajectory. The module M5 includes: Module M5.1: for each spoke of the base trajectory, establish a sector search window around it within the angle in the polar coordinate representation to form n sector-ring intersection areas with the n variable-density concentric ring areas, and the intersection area of the kth search window and the i th variable-density concentric ring is denoted as the sector-ring search window area : , wherein, is the polar radius, is the polar angle, is the specified angle tolerance window half-width, i.e., the maximum absolute value of the acceptable deviation angle, , wherein, denotes the i th variable-density concentric ring area. Module M5.2: grid the base trajectory to the Cartesian coordinate system, and obtain the i th candidate point set of the kth spoke for the sector-ring search window area , i.e., wherein, denotes taking the set of discrete points in the Cartesian coordinate system within the area. Module M5.3: confirm and optimize the sampling point position of the target trajectory, dynamically select the optimal sampling point from the candidate point set The search rule is determined according to a trajectory application scene, and includes preferentially selecting a point with the least number of previous accumulative samplings, preferentially selecting a point close to a central region, and preferentially selecting a point close to a high-frequency region in a k-space periphery , preferentially selecting a point close to a central region, and preferentially selecting a point close to a high-frequency region in a k-space periphery , and randomly selecting a single sampling point from a point set with the least number of samplings.Module M5.4: For a sequence that needs to continuously collect data, is easy to produce eddy current, and includes three-dimensional heart movie imaging, an alternate sampling sequence of periphery-center-periphery is used to recombine adjacent trajectory lines, and sampling points of the kth trajectory line after recombination have: wherein n is a total number of concentric rings, indicates a sampling point of the kth optimized trajectory on the ith variable-density concentric ring.

[0173] Module M6: scanning according to the adjusted optimized sampling trajectory, performing binning on the acquired k-space data by a self-gated binning algorithm, estimating the respiratory motion field based on the binned data, and reconstructing a three-dimensional image by a motion compensated image reconstruction algorithm. The module M6 comprises the following sub-modules: Module M6.1: scanning using the optimized sampling trajectory, and acquiring magnetic resonance k-space data. Module M6.2: analyzing the principal components of the k-space center signal to obtain each principal component containing motion information. The analysis of the principal components of the k-space center signal comprises: performing inverse Fourier transform on the full-sampling k-space center line acquired by each trajectory to obtain a one-dimensional projection image of the current layer block along the frequency encoding direction, which reflects the motion state of the current layer block image along the frequency encoding direction. Arranging the projection images obtained at each time along the time dimension to obtain a dynamic motion image of the projection image of the current layer block along the frequency encoding direction over time. Performing principal component analysis on the obtained one-dimensional frequency encoding direction projection-time image to obtain each principal component containing motion information. Module M6.3: performing frequency analysis on each extracted principal component to obtain a self-navigated respiratory curve and a cardiac curve. Module M6.4: after binning processing of the respiratory curve and the cardiac curve, obtaining undersampled k-space data of each respiratory / heart dimension bin. Module M6.4 comprises: according to the self-navigated respiratory curve, removing abnormal data whose amplitude exceeds a preset threshold of 5%, equally dividing the respiratory curve according to the height, and m is the number of respiratory bins. According to the height of each point on the respiratory curve, the k-space data corresponding to the points on the respiratory curve is divided into the corresponding respiratory bin Bin, and the k-space data of the respiratory dimension is completed. According to the self-navigated cardiac curve, recording the interval between every two maximum values, and recording the interval distance between each point and the previous maximum value point, and dividing the k-space data corresponding to each cardiac curve point into the corresponding cardiac phase bin according to the distance, and completing the k-space data binning of the cardiac dimension. After extracting the self-navigated respiratory curve and cardiac curve using the k-space center signal, the acquired k-space data is simultaneously subjected to respiratory and cardiac binning, and undersampled k-space data of each respiratory / heart dimension bin is obtained. Module M6.5: using the respiratory dimension binned k-space data, using a parallel imaging reconstruction method to estimate coil sensitivity and reconstruct images of the respiratory bin, and obtaining images of each respiratory state. Using an image registration method, specifying the end-expiratory period as a target respiratory state, and registering the images of the remaining respiratory states to the target respiratory state to obtain the respiratory displacement field of the remaining respiratory state images registered to the target respiratory state. Module M6.6: using the estimated respiratory displacement field, using a motion compensated image reconstruction method to perform motion compensated image reconstruction on the undersampled k-space data of each respiratory / heart dimension bin to obtain a three-dimensional cardiac magnetic resonance image without respiratory and cardiac artifacts.

[0174] Those skilled in the art know that, in addition to implementing the system provided by the present application and each device, module and unit thereof in the form of pure computer readable program code, the system provided by the present application and each device, module and unit thereof can also be implemented in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded microcontrollers, etc. by logically programming the method steps to achieve the same functions. Therefore, the system provided by the present application and each device, module and unit thereof can be considered as a hardware component, and the devices, modules and units included therein for achieving various functions can also be considered as structures within the hardware component; the devices, modules and units for achieving various functions can also be considered as both software modules implementing methods and structures within hardware components.

[0175] The specific embodiments of the present application are described above. It needs to be understood that the present application is not limited to the specific embodiments described above, and various changes or modifications can be made by those skilled in the art within the scope of the claims, which does not affect the essential content of the present application. The embodiments of the present application and the features in the embodiments can be combined with each other in any manner without conflict.

Claims

1. A self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling, characterized in that, include: Step S1: Establish the ideal square kyz plane, the real rectangular kyz plane, and the affine transformation relationship between the two; Step S2: Divide the real rectangular kyz plane into n concentric annular regions based on a preset area ratio, where n is the number of kyz plane points collected for each heartbeat of the trajectory; Step S3: Form an ideal radial spoke trajectory rotating along the golden angle in the ideal square kyz plane; Step S4: Through affine transformation, the ideal trajectory is mapped onto the real rectangular kyz plane to generate the basic trajectory; Step S5: Near each spoke of the base trajectory A fan-shaped search window is established within the angle, forming n fan-ring intersection areas with the concentric annular area. Within each fan-ring, the position of the sampling point is optimized based on the historical sampling number and the principle of prioritizing low-frequency areas near the center of k space, thus obtaining the optimized sampling trajectory. Step S6: Scan according to the adjusted optimized sampling trajectory, divide the collected k-space data into bins using a self-gated binning algorithm, estimate the respiratory motion field based on the binned data, and reconstruct the three-dimensional image using a motion-compensated image reconstruction algorithm.

2. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 1, characterized in that, Step S1 includes the following sub-steps: Step S1.1: Establish an ideal square kyz plane with an aspect ratio of 1:1; Step S1.2: Establish a real rectangular kyz plane, wherein the number of points on the long side of the real rectangular kyz plane is equal to the actual number of sampling points in the ky direction, and the number of points on the wide side is equal to the actual number of sampling points in the kz direction; Step S1.3: Establish the affine transformation relationship between the real rectangular kyz plane and the ideal square kyz plane, specifically for the polar coordinate angles in the ideal square kyz plane. After affine transformation, it has: in, represents the polar coordinate angle corresponding to the real rectangular kyz plane, and ny and nz represent the actual number of sampling points in the ky and kz directions of the real rectangular kyz plane, respectively.

3. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 1, characterized in that, Step S3 includes the following sub-steps: Step S3.1: In an ideal square kyz plane, for each heartbeat, a spoke is collected from the center of the k-space to the surrounding area along a fixed direction; Step S3.2: Rotate the current trajectory spokes at a golden angle in a clockwise or counterclockwise direction to obtain the next trajectory spokes; Repeat steps S3.1 and S3.2 until the trajectory meets the sampling rate requirement. Record the polar coordinate angle of the k-th ideal trajectory line in the ideal square kyz plane as . .

4. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 1, characterized in that, Step S4 includes the following sub-steps: Step S4.1: For each ideal trajectory spoke in the ideal square kyz plane, calculate the spoke angle of the base trajectory in the real rectangular kyz plane using the affine transformation obtained in step S1, based on the polar coordinate angle. For the k-th ideal trajectory spoke, the polar coordinate angle corresponding to the ideal trajectory spoke in the ideal square kyz plane is... ,have: in, Let be the polar coordinate angle of the k-th basic trajectory spoke in the real rectangular kyz plane; Step S4.2: Based on polar coordinate angles This yields the k-th basic trajectory spoke within the real rectangular kyz plane; Repeat steps S4.1 and S4.2 until all ideal trajectories are mapped to the base trajectories in the real rectangular kyz plane.

5. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 1, characterized in that, Step S5 includes: Step S5.1: For each spoke of the basic trajectory, in its vicinity in polar coordinates... A fan-shaped search window is established within the angle, forming n fan-ring intersection regions with n concentric ring regions of varying density. The intersection region of the k-th search window and the ith concentric ring of varying density is denoted as the fan-ring search window region. : in, This represents the coordinates of a point in polar coordinates. Polar radius, Polar angle, The tolerance window half-width for a specified angle, i.e., the maximum absolute value of the acceptable deviation angle. This represents the polar coordinate angle of the k-th basic trajectory spoke within the real rectangular kyz plane. This represents the i-th concentric ring region with varying density; Step S5.2: Grid the basic trajectory to a Cartesian coordinate system, targeting the fan-ring search window region. This yields the i-th candidate point set for the k-th spoke. ,Right now: in, Indicates taking The set of discrete points within a region in Cartesian coordinates; Step S5.3: Confirm and optimize the sampling point locations of the target trajectory, and combine historical access counts with search rules to select candidate points from the set. Dynamically select the optimal sampling point ; Step S5.4: For sequences that require continuous data acquisition and are prone to eddy currents, including those used in 3D cardiac cine imaging, an alternating sampling sequence of peripheral-central-periphery is used to reconstruct adjacent trajectory lines. The reconstructed k-th trajectory line... The sampling points are: Where n is the total number of concentric rings, This represents the sampling point of the k-th optimized trajectory on the i-th variable-density concentric ring.

6. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 5, characterized in that, The search rules are determined based on the trajectory application scenario, including prioritizing the point with the fewest cumulative sampling times and selecting low-frequency regions near the k-space center. Prioritize selecting points close to the central region and high-frequency regions on the outer periphery of the k-space. A single sampling point is randomly selected from the set of points with the fewest sampling times.

7. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 1, characterized in that, Step S6 includes the following sub-steps: Step S6.1: Use the optimized sampling trajectory to scan and acquire magnetic resonance k-space data; Step S6.2: Analyze the principal components of the k-space center signal to obtain each principal component containing motion information; Step S6.3: Perform frequency analysis on each extracted principal component to obtain the self-navigation respiratory curve and cardiac curve; Step S6.4: After binning the respiratory curve and cardiac curve, undersampled k-space data of each respiratory / cardiac dimension bin are obtained; Step S6.5: Using the k-space data of the respiratory dimension bins, a parallel imaging reconstruction method is used to estimate the coil sensitivity and reconstruct the images of the respiratory bins to obtain images of each respiratory state; using the image registration method, the end of expiration is designated as the target respiratory state, and the images of the other respiratory states are registered to the target respiratory state to obtain the respiratory displacement field of the images of the other respiratory states registered to the target respiratory state. Step S6.6: Using the estimated respiratory displacement field, the undersampled k-space data of each respiratory / cardiac dimension bin is reconstructed using a motion-compensated image reconstruction method to obtain a three-dimensional cardiac magnetic resonance image with respiratory and cardiac artifacts removed.

8. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 7, characterized in that, The analysis of the principal components of the k-space center signal includes: Perform an inverse Fourier transform on the center line of the fully sampled k-space of each trajectory to obtain a one-dimensional projection map of the current block along the frequency coding direction. The projection map reflects the motion state of the current block along the frequency coding direction. Arrange the projection images obtained continuously at each time point along the time dimension to obtain a dynamic motion image of the projection image of the current block along the frequency coding direction as time changes. Principal component analysis was performed on the obtained one-dimensional frequency-encoded directional projection-time image to obtain the principal components containing motion information.

9. The self-navigation cardiac magnetic resonance imaging method based on adaptive sector sampling according to claim 7, characterized in that, Step S6.4 includes removing abnormal data whose respiratory curve amplitude exceeds a preset threshold of 5% based on the self-navigation respiratory curve, dividing the respiratory curve into m equal parts according to the curve height, where m is the number of respiratory boxes; and dividing the k-space data corresponding to the respiratory curve point into the corresponding respiratory box Bin according to the height of each point on the respiratory curve, thus completing the k-space data binning of the respiratory dimension. Based on the self-navigation heartbeat curve, the distance between every two maxima is recorded. Starting from the first maxima point, the distance between each point and the previous maxima point is recorded. The k-space data corresponding to each heartbeat curve point is divided into the corresponding heartbeat phase bins according to the distance, thus completing the k-space data binning of the heartbeat dimension. Simultaneously, after extracting the self-navigation respiratory and cardiac curves using the k-space center signal, respiratory and cardiac binning are applied to the collected k-space data to obtain undersampled k-space data for each respiratory / cardiac dimension bin.

10. A self-navigation cardiac magnetic resonance imaging system based on adaptive sector sampling, characterized in that, include: Module M1: Establishes the ideal square kyz plane, the real rectangular kyz plane, and the affine transformation relationship between the two; Module M2: Divide the real rectangular kyz plane into n concentric annular regions based on a preset area ratio, where n is the number of kyz plane points collected for each heartbeat of the trajectory; Module M3: Forms an ideal radial spoke trajectory rotating along the golden angle in the ideal square kyz plane; Module M4: Through affine transformation, the ideal trajectory is mapped onto the real rectangular kyz plane to generate the basic trajectory; Module M5: Near each spoke of the basic trajectory A fan-shaped search window is established within the angle, forming n fan-ring intersection areas with the concentric annular area. Within each fan-ring, the position of the sampling point is optimized based on the historical sampling number and the principle of prioritizing low-frequency areas near the center of k space, thus obtaining the optimized sampling trajectory. Module M6: Scans according to the adjusted and optimized sampling trajectory, divides the collected k-space data into bins using a self-gated binning algorithm, estimates the respiratory motion field based on the binned data, and reconstructs a 3D image using a motion-compensated image reconstruction algorithm.

Images