CN102540115A - Magnetic resonance imaging method and system - Google Patents

Abstract

The invention discloses a magnetic resonance imaging method, which comprises the steps of sparsely sampling dynamic imaging objects to obtain dynamic image data and navigation data, wherein the navigation data are obtained by radial collection; reconstructing the dynamic image data and the navigation data based on a partially separable function model so as to obtain signal data; and processing the signal data by Fourier transform so as to generate magnetic resonance images. According to the magnetic resonance imaging method and the magnetic resonance imaging system, the navigation data are obtained by radial collection in the dynamic imaging objects, so that each collected navigation datum can cover from a low frequency to a high frequency, therefore, space frequency component in a space-time frequency joint domain which can be covered by the navigation data obtained by collection is increased, the capability to capture the time-space signal change of movement imaging objects is greatly improved, and the imaging resolution ratio is increased.

Description

Magnetic resonance imaging method and system

【Technical field】

The invention relates to imaging technology, in particular to a magnetic resonance imaging method and system.

【Background technique】

Magnetic resonance imaging technology plays an important role in clinical diagnosis because of its good soft tissue resolution, multi-directional and multi-parameter imaging and no radiation radiation. However, due to the limitation of the magnetic resonance imaging technique, a long scanning time is usually required in the actual imaging process, for example, it takes about 15-30 seconds to acquire a spin echo image. The slow imaging speed limits the time resolution of images in the dynamic imaging of magnetic resonance imaging, and will also cause serious motion artifacts in the images, seriously reducing the image quality, which in turn limits the application of magnetic resonance imaging in the heart, coronary Applications in areas such as arteries and other moving organs and neurological imaging.

The magnetic resonance imaging method based on Partially Separable Functions (PSF for short) is to perform sparse sampling imaging on moving objects, so it can be used to realize dynamic imaging in magnetic resonance imaging technology, and is applied to cardiac dynamic imaging and perfusion imaging. .

However, in the magnetic resonance imaging method based on partially separable function theory, only a few low-frequency lines in K space can be collected, so that the spatial frequency components in the space-time frequency joint domain that the acquired navigation data can cover Less, severely restricts the ability to capture changes in spatio-temporal signals, resulting in images that cannot accurately reflect the real situation of the imaging object's motion.

【Content of invention】

Based on this, it is necessary to provide a magnetic resonance imaging method that can accurately reflect the real situation of the motion of the imaging object.

In addition, it is also necessary to provide a magnetic resonance imaging system that can accurately reflect the real situation of the motion of the imaging object.

A magnetic resonance imaging method, comprising the steps of:

Sparsely sampling the dynamic imaging object to obtain dynamic image data and navigation data, the navigation data is obtained through radial acquisition;

performing interpolation recovery through the dynamic image data and navigation data to obtain signal data;

performing Fourier transform on the signal data to generate a magnetic resonance image.

Preferably, the step of performing sparse sampling on the dynamic imaging object to obtain dynamic image data and navigation data is:

The dynamic image data is collected according to the Cartesian sampling trajectory, and the navigation data is collected according to the radial sampling trajectory.

Preferably, the acquisition of the dynamic image data and the acquisition of the navigation data are interleaved.

Preferably, the step of performing sparse sampling on the dynamic imaging object to obtain dynamic image data and navigation data is:

The collection of dynamic image data and navigation data is interleaved according to preset time intervals.

Preferably, the step of performing sparse sampling on dynamic imaging objects to obtain dynamic image data and navigation data includes:

Collect dynamic image echo signals at odd times to obtain dynamic image data;

The navigation data is obtained by collecting the echo signals of the navigation data according to the radial sampling trajectory at even times.

A magnetic resonance imaging system, characterized in that it comprises:

The data collection module is used to perform sparse sampling on the dynamic imaging object to obtain dynamic image data and navigation data, and the navigation data is obtained through radial collection;

A reconstruction module, used to obtain signal data by reconstructing the dynamic image data and navigation data based on a partially separable function model;

An image generating module, configured to perform Fourier transform on the signal data to generate a magnetic resonance image.

Preferably, the data collection module is further configured to collect dynamic image data according to Cartesian sampling trajectories, and collect navigation data according to radial sampling trajectories.

Preferably, the collection of dynamic image data and the collection of navigation data in the data collection module are interleaved.

Preferably, the data collection module is also used to interleave the collection of dynamic image data and navigation data at preset time intervals.

Preferably, the data collection module includes:

The image data acquisition unit is used to collect dynamic image echo signals at odd times to obtain dynamic image data;

The navigation data collection unit is used to collect navigation data echo signals according to radial sampling trajectories at even times to obtain navigation data.

The above-mentioned magnetic resonance imaging method and system obtain navigation data through radial acquisition in the dynamic imaging object, so that each navigation data collected can cover from low frequency to high frequency, thereby increasing the space covered by the collected navigation data. - Spatial frequency components in the time-frequency joint domain greatly improve the ability to capture changes in spatio-temporal signals and accurately reflect the real movement of the imaging object.

【Description of drawings】

Fig. 1 is the flowchart of the magnetic resonance imaging method in an embodiment;

Fig. 2 is a schematic diagram of sparse sampling in an embodiment;

Fig. 3 is the flow chart of the method for obtaining dynamic image data and navigation data by sparsely sampling the dynamic imaging object in Fig. 1;

Fig. 4 is a sequence diagram of sparse sampling in an embodiment;

Fig. 5 is a schematic structural diagram of a magnetic resonance imaging system in an embodiment;

Fig. 6 is a schematic structural diagram of the data acquisition module in Fig. 5;

Fig. 7 is the ideal image of water mold in an embodiment;

Fig. 8 is the ideal image of water mold in another embodiment;

Fig. 9 is the ideal image of water mold in another embodiment;

Fig. 10 is the ideal image of water mold in another embodiment;

Figure 11 is an ideal image to be reconstructed in one embodiment;

Fig. 12 is the magnetic resonance image reconstructed by the magnetic resonance imaging method of the present invention;

Fig. 13 is the magnetic resonance image reconstructed by the traditional magnetic resonance imaging method;

Fig. 14 is a magnetic resonance image reconstructed by the magnetic resonance imaging method of the present invention in an in vivo experiment in an embodiment;

Fig. 15 is a magnetic resonance image reconstructed by the magnetic resonance imaging method of the present invention in an in vivo experiment in another embodiment;

Fig. 16 is the magnetic resonance image reconstructed by the traditional magnetic resonance imaging method;

FIG. 17 is a magnetic resonance image reconstructed by a traditional magnetic resonance imaging method;

Fig. 18 is a schematic diagram of the SNR comparison of the reconstructed image.

【Detailed ways】

In one embodiment, as shown in Figure 1, a kind of magnetic resonance imaging method, comprises the steps:

In step S110, the dynamic imaging object is sparsely sampled to obtain dynamic image data and navigation data, and the navigation data is obtained through radial acquisition.

In this embodiment, in order to realize dynamic imaging in magnetic resonance, a small amount of image data is obtained by sparsely sampling the dynamic imaging object, and then reconstruction is performed based on the small amount of collected image data, which will greatly reduce the acquisition time. The image data obtained by sparse sampling includes dynamic image data and navigation data. Specifically, dynamic image data with high spatial resolution and low temporal resolution and navigation data with low spatial resolution and high temporal resolution are two complementary data. In the sampling process, the amount of data to be collected increases exponentially with the increase of the physical dimension, which will cause a contradiction between the spatial resolution and the temporal resolution. In the traditional sparse sampling process, if the To obtain an MRI image with high spatial resolution, the time resolution of the MRI image needs to be sacrificed; if the time resolution is increased, the spatial resolution of the MRI image will be reduced, and it is impossible to achieve a balance between the time resolution and the spatial resolution. At the same time, high resolution is achieved. In order to solve this contradiction, the navigation data is obtained by radially collecting dynamic imaging objects.

Radial acquisition is a sampling method that traverses the K space radially. Through this sampling method, the sampled navigation data lines can be covered from low frequency to high frequency, so as to improve the spatial-time frequency joint domain that navigation data can cover. Spatial frequency components, thereby improving the resolution of magnetic resonance images.

In one embodiment, the specific process of the above step S110 is: collecting dynamic image data according to a Cartesian sampling trajectory, and collecting navigation data according to a radial sampling trajectory.

In this embodiment, in the Cartesian sampling trajectory, the K space is regarded as the Fourier transform domain space of the image, and a plurality of horizontal lines are equally spaced along the column direction, that is, the vertical direction, wherein each horizontal line is the phase of the K space coded line. Collecting dynamic image data according to the Cartesian sampling trajectory and sampling navigation data according to the radial sampling trajectory can ensure more accurate spatio-temporal information collected from moving sampling objects, thereby ensuring the accuracy of reconstructed images.

In one embodiment, the acquisition of dynamic image data and the acquisition of navigation data are interleaved. Specifically, the collection of dynamic image data and navigation data is interleaved according to preset time intervals.

In this embodiment, it is relatively difficult to simultaneously collect dynamic image data and navigation data during the actual data collection process. data collection. In the time dimension, dynamic image data and navigation data can be collected in turn according to preset time intervals. In one embodiment, as shown in Figure 2, the dynamic image data can be collected through the Cartesian sampling trajectory at one radio frequency repetition time, and the navigation data can be collected through the radial sampling trajectory at the next radio frequency repetition time, so as to realize the interaction between the dynamic image data and the navigation data collection.

Through the interleaving acquisition between dynamic image data and navigation data, the temporal and spatial information of moving sampling objects can be obtained at the same time, ensuring that the dynamic image data and navigation data have the same and smaller RF repetition time and echo time, reducing the rapid gradient Toggle the effect of accompanying eddy currents on image reconstruction.

In one embodiment, as shown in FIG. 3, in order to realize the interlaced collection of dynamic image data and navigation data, the specific process of the above step S110 includes:

Step S111 , collecting dynamic image echo signals at odd-numbered moments to obtain dynamic image data.

In this embodiment, as shown in FIG. 4 , after transmitting the sequence to the sampling object, the reflected dynamic image echo signals are collected at odd times to obtain dynamic image data, and the reflected navigation echo signals are collected at even times.

Step S113, collecting navigation data echo signals according to the radial sampling trajectory at even times to obtain navigation data.

In step S130, signal data is obtained by reconstructing the dynamic image data and the navigation data based on a partially separable function model.

和时间基函数

这两个独立变量函数之和，即如以下公式所示：In this embodiment, in the partially separable function model, it is considered that the spatial variation and temporal variation of the image function are L-order separable. Using the properties of the partially separable function and the linear characteristics of Fourier transform, the signal data S (k, t) is expressed as a spatial basis function

and time basis functions

The sum of the functions of these two independent variables, that is, as shown in the following formula:

和时间基函数

即可实现高时间高空间覆盖密度的基于部分可分离函数模型重建，得到完全的信号数据。Through the above formula, the complex motion of the signal in the joint dimension of time and space can be transformed into a relatively simple mathematical problem in which the signal at each point in the space changes with time. It only needs to accurately predict the frequency component parameter L and the spatial basis function

and time basis functions

The partial separable function model reconstruction based on high temporal and high spatial coverage density can be realized, and complete signal data can be obtained.

Specifically, the dynamic image data and navigation data meet the following three conditions during the acquisition process: (1) The pulse repetition time T _R must meet the time Nyquist rate of the navigation data; (2) The sampling interval _Δky in the phase encoding direction The spatial Nyquist rate of the dynamic image data must be met; (3) The sampling frame number N obtained from the dynamic image data must be greater than or equal to the order L. The signal data can be described in terms of solving the optimization problem as:

In order to solve the above problems, the collected navigation data is extracted into a navigation data matrix:

其中，{λ_l}为C的按降序排列的奇异值，{μ_l}和{v_l}分别是C的左特征向量和右特征向量，L为模型的频率成分函数。取前L个左特征向量为部分可分离函数模型的时间基函数，即：Then perform singular value decomposition on the navigation matrix C to get

Among them, {λ _l } is the singular value of C in descending order, {μ _l } and {v _l } are the left eigenvector and right eigenvector of C, respectively, and L is the frequency component function of the model. Take the first L left eigenvectors as the time basis function of the partially separable function model, namely:

The frequency component function L can be determined according to the following formula and the noise level of the measured image data:

${<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; rank</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; min</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}$

or

${<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; rank</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; min</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\sqrt{\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; Q</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; \}</span>}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}$

After obtaining the time basis function and frequency component parameters, use the time basis function and dynamic image data to predict the space basis function. The detailed solution process is shown in the following formula:

为已经获得的时间基函数(l＝1，2，...，L，n＝1，2，...，N)，

是待预测的空间基函数

为已采集的动态图像数据(p＝1，2，...，P，n＝1，2，...，N)。为了叙述方便，上式可简写为：in,

For the obtained time basis functions (l=1, 2,..., L, n=1, 2,..., N),

is the spatial basis function to be predicted

is the collected dynamic image data (p=1, 2, . . . , P, n=1, 2, . . . , N). For the convenience of description, the above formula can be abbreviated as:

${\mathrm{<span\; class="notranslate">\; \&Phi;</span>}}_{\mathrm{<span\; class="notranslate">\; p</span>}}{\stackrel{<span\; class="notranslate">\; \&Right\; Arrow;</span>}{\mathrm{<span\; class="notranslate">\; c</span>}}}_{\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; p</span>}}$

Step S150, performing Fourier transform on the signal data to generate a magnetic resonance image.

In this embodiment, the inverse Fourier transform is performed on the signal data to obtain a magnetic resonance image with high temporal resolution and high spatial resolution.

In one embodiment, as shown in FIG. 5 , a magnetic resonance imaging system includes a data acquisition module 10 , a reconstruction module 30 and an image generation module 50 .

The data collection module 10 is used to perform sparse sampling on the dynamic imaging object to obtain dynamic image data and navigation data, and the navigation data is obtained through radial collection.

In this embodiment, in order to realize the dynamic imaging in magnetic resonance, the data acquisition module 10 performs sparse sampling on the dynamic imaging object to obtain a small amount of image data, and then reconstructs according to the small amount of image data collected, which will greatly reduce the acquisition time. reduce. The image data obtained by the data acquisition module 10 through sparse sampling includes dynamic image data and navigation data. Specifically, dynamic image data with high spatial resolution and low temporal resolution and navigation data with low spatial resolution and high temporal resolution are two In the traditional sampling process, the amount of data to be collected increases exponentially with the increase of the physical dimension, which will cause a contradiction between the spatial resolution and the temporal resolution. In the traditional If a magnetic resonance image with high spatial resolution is obtained during the sampling process, the temporal resolution of the magnetic resonance image needs to be sacrificed; if the temporal resolution is increased, the spatial resolution of the magnetic resonance image will be reduced. It is impossible to achieve high resolution at the same time, in order to solve this contradiction, the data acquisition module 10 obtains navigation data by radial acquisition on the dynamic imaging object.

Radial acquisition is a sampling method that traverses the K space radially. Through this sampling method, the sampled navigation data lines can be covered from low frequency to high frequency, so as to improve the spatial-time frequency joint domain that navigation data can cover. Spatial frequency components, thereby improving the resolution of magnetic resonance images.

In one embodiment, the data collection module 10 is further configured to collect dynamic image data according to Cartesian sampling trajectories, and collect navigation data according to radial sampling trajectories.

In this embodiment, in the Cartesian sampling trajectory, the K space is regarded as the Fourier transform domain space of the image, and a plurality of horizontal lines are equally spaced along the column direction, that is, the vertical direction, wherein each horizontal line is the phase of the K space coded line. The data acquisition module 10 collects dynamic image data according to the Cartesian sampling trajectory, and samples navigation data according to the radial sampling trajectory, which can ensure more accurate spatio-temporal information collected from moving sampling objects, thereby ensuring the accuracy of reconstructed images.

In one embodiment, the collection of dynamic image data and the collection of navigation data in the data collection module 10 are interleaved. Specifically, the data collection module is also used to interleave the collection of dynamic image data and navigation data according to preset time intervals.

In this embodiment, it is relatively difficult to simultaneously collect dynamic image data and navigation data during the actual data collection process. In order to make the above-mentioned data collection process easy to implement, the data collection module 10 can perform dynamic Acquisition of image data and navigation data. In the time dimension, dynamic image data and navigation data can be collected in turn according to preset time intervals. In one embodiment, the data collection module 10 can collect dynamic image data through Cartesian sampling trajectory at one radio frequency repetition time, and collect navigation data through radial sampling trajectory at the next radio frequency repetition time, so as to realize interactive collection of dynamic image data and navigation data.

The data acquisition module 10 can acquire the time-space information of the moving sampling object at the same time through the interleaving acquisition between the dynamic image data and the navigation data, so as to ensure that the dynamic image data and the navigation data have the same and smaller radio frequency repetition time and echo time , to reduce the influence of eddy currents associated with rapid gradient switching on image reconstruction.

In one embodiment, as shown in FIG. 6 , the data collection module 10 includes an image data collection unit 110 and a navigation data collection unit 130 .

The image data collection unit 110 is configured to collect dynamic image echo signals at odd times to obtain dynamic image data.

In this embodiment, the image data collection unit 110 collects reflected dynamic image echo signals at odd-numbered moments to obtain dynamic image data after transmitting a sequence to the sampling object, and the navigation data collection unit 130 collects reflected navigation echo signals at even-numbered moments.

The navigation data collection unit 130 is configured to collect navigation data echo signals according to radial sampling trajectories at even times to obtain navigation data.

The reconstruction module 30 is used for reconstructing the dynamic image data and the navigation data based on a partially separable function model to obtain signal data.

和时间基函数

这两个独立变量函数之和，即如以下公式所示：In this embodiment, in the partially separable function model, it is considered that the spatial variation and temporal variation of the image function are L-order separable. Using the properties of the partially separable function and the linear characteristics of Fourier transform, the signal data S (k, t) is expressed as a spatial basis function

and time basis functions

The sum of the functions of these two independent variables, that is, as shown in the following formula:

和时间基函数

即可实现高时间高空间覆盖密度的基于部分可分离函数模型重建，得到完全的信号数据。Through the above formula, the complex movement of the signal in the joint dimension of time and space can be transformed into a relatively simple mathematical problem that the signal at each point in the space changes with time. The reconstruction module 30 only needs to accurately predict the frequency component parameter L and the spatial basis function

and time basis functions

The partial separable function model reconstruction based on high temporal and high spatial coverage density can be realized, and complete signal data can be obtained.

Specifically, the dynamic image data and navigation data meet the following three conditions during the acquisition process: (1) The pulse repetition time T _R must meet the time Nyquist rate of the navigation data; (2) The sampling interval _Δky in the phase encoding direction The spatial Nyquist rate of the dynamic image data must be met; (3) The sampling frame number N obtained from the dynamic image data must be greater than or equal to the order L. The signal data can be described in terms of solving the optimization problem as:

In order to solve the above problems, the collected navigation data is extracted into a navigation data matrix:

其中，{λ_l}为C的按降序排列的奇异值，{μ_l}和{v_l}分别是C的左特征向量和右特征向量，L为模型的频率成分函数。取前L个左特征向量为部分可分离函数模型的时间基函数，即：Then perform singular value decomposition on the navigation matrix C to get

Among them, {λ _l } is the singular value of C in descending order, {μ _l } and {v _l } are the left eigenvector and right eigenvector of C, respectively, and L is the frequency component function of the model. Take the first L left eigenvectors as the time basis function of the partially separable function model, namely:

The frequency component function L can be determined according to the following formula and the noise level of the measured image data:

${<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; rank</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; min</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}$

or

${<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; rank</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; min</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\sqrt{\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; Q</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; \}</span>}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}}$

After obtaining the time basis function and frequency component parameters, use the time basis function and dynamic image data to predict the space basis function. The detailed solution process is shown in the following formula:

为已经获得的时间基函数(l＝1，2，...，L，n＝1，2，...，N)，

是待预测的空间基函数

为已采集的动态图像数据(p＝1，2，...，P，n＝1，2，...，N)。为了叙述方便，上式可简写为：in,

For the obtained time basis functions (l=1, 2,..., L, n=1, 2,..., N),

is the spatial basis function to be predicted

is the collected dynamic image data (p=1, 2, . . . , P, n=1, 2, . . . , N). For the convenience of description, the above formula can be abbreviated as:

${\mathrm{<span\; class="notranslate">\; \&Phi;</span>}}_{\mathrm{<span\; class="notranslate">\; p</span>}}{\stackrel{<span\; class="notranslate">\; \&Right\; Arrow;</span>}{\mathrm{<span\; class="notranslate">\; c</span>}}}_{\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; p</span>}}$

The image generation module 50 is configured to perform Fourier transform on the signal data to generate a magnetic resonance image.

In this embodiment, the image generation module 50 performs an inverse Fourier transform on the signal data to obtain a magnetic resonance image with high temporal resolution and high spatial resolution.

In the following, numerical simulation and in vivo experiments are used to verify the above magnetic resonance imaging method. Figures 7 to 10 are the ideal images of the water model constructed by numerical simulation under different motion states. Among them, the water model is composed of three different moving objects, that is, the square in the upper left corner changes from large to small and then becomes larger, and the motion cycle is 13, the larger elliptical motion cycle in the middle is 15, and the innermost small circle motion cycle is 11.

By comparing Figures 12 and 13, it can be seen that the magnetic resonance image reconstructed by the above magnetic resonance imaging method has fewer artifacts, and the image quality is significantly improved.

In the verification of the in vivo experiment, the signal-to-noise ratio of the magnetic resonance images obtained by the above-mentioned magnetic resonance imaging method (Fig. 14 and Fig. 15) and the magnetic resonance image obtained by the traditional magnetic resonance imaging method (Fig. It can be concluded that the signal-to-noise ratio of the magnetic resonance image obtained by the above magnetic resonance imaging method is higher than that of the magnetic resonance image obtained by the traditional magnetic resonance imaging method, as shown in FIG. 18 .

The above-mentioned magnetic resonance imaging method and system obtain navigation data through radial acquisition in the dynamic imaging object, so that each navigation data collected can cover from low frequency to high frequency, thereby increasing the space covered by the collected navigation data. - Spatial frequency components in the time-frequency joint domain greatly improve the ability to capture changes in spatio-temporal signals and accurately reflect the real movement of the imaging object.

The above-mentioned embodiments only express several implementation modes of the present invention, and the description thereof is relatively specific and detailed, but should not be construed as limiting the patent scope of the present invention. It should be pointed out that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent for the present invention should be based on the appended claims.

Claims (10)

1. A magnetic resonance imaging method comprising the steps of:

sparse sampling is carried out on a dynamic imaging object to obtain dynamic image data and navigation data, and the navigation data are obtained through radial acquisition;

reconstructing the dynamic image data and the navigation data based on a partial separable function model to obtain signal data;

fourier transforming the signal data to generate a magnetic resonance image.

2. The magnetic resonance imaging method as claimed in claim 1, wherein the step of sparsely sampling the dynamic imaging subject to obtain dynamic image data and navigation data is:

dynamic image data are collected according to a Cartesian sampling track, and navigation data are collected according to a radial sampling track.

3. A magnetic resonance imaging method according to claim 1 or 2, characterized in that the acquisition of the dynamic image data and the acquisition of the navigation data are interleaved.

4. The magnetic resonance imaging method as claimed in claim 3, wherein the step of sparsely sampling the dynamic imaging subject to obtain dynamic image data and navigation data is:

and interleaving the dynamic image data and the navigation data according to a preset time interval to acquire the dynamic image data and the navigation data.

5. The magnetic resonance imaging method as set forth in claim 4, wherein the step of sparsely sampling the dynamic imaging subject to obtain dynamic image data and navigation data includes:

acquiring dynamic image echo signals at odd moments to obtain dynamic image data;

and acquiring the navigation data echo signal at even number moment according to the radial sampling track to obtain navigation data.

6. A magnetic resonance imaging system, comprising:

the data acquisition module is used for carrying out sparse sampling on the dynamic imaging object to obtain dynamic image data and navigation data, and the navigation data is obtained through radial acquisition;

the reconstruction module is used for reconstructing the dynamic image data and the navigation data based on a partial separable function model to obtain signal data;

and the image generation module is used for carrying out Fourier transform on the signal data to generate a magnetic resonance image.

7. The magnetic resonance imaging system of claim 6, wherein the data acquisition module is further configured to acquire dynamic image data according to a Cartesian sampling trajectory and navigation data according to a radial sampling trajectory, respectively.

8. A magnetic resonance imaging system according to claim 6 or 7, wherein the acquisition of dynamic image data and the acquisition of navigation data in the data acquisition module are interleaved.

9. The system of claim 8, wherein the data acquisition module is further configured to interleave the acquisition of the dynamic image data and the navigation data at predetermined time intervals.

10. The magnetic resonance imaging system of claim 9, wherein the data acquisition module comprises:

the image data acquisition unit is used for acquiring dynamic image echo signals at odd moments to obtain dynamic image data;

and the navigation data acquisition unit is used for acquiring the navigation data echo signal at even number moment according to the radial sampling track to obtain navigation data.

Images