One-stop cardiac magnetic resonance examination technique

Abstract

The application relates to the field of medical image examination and discloses a one-stop cardiac magnetic resonance examination technology, which comprises the following steps: S1, high-integration degree nuclear magnetic resonance scanning technology: a high-integration degree imaging sequence is adopted to collect original data of multiple cardiac nuclear magnetic resonance examinations through one-time scanning; the original data comprises different contrast images for measuring myocardial tissue T1, T2 and T2 * , and movie images for evaluating the function of a cardiac ventricle; S2, expansion type deep learning reconstruction: all different T1, T2, T2 * weighted images and movie images in quantitative imaging and movie imaging data are reconstructed by adopting a deep learning-based reconstruction technology; S3, quantitative image reconstruction: T1, T2 and T2 * quantitative images are reconstructed by adopting a deep learning-based quantitative image reconstruction technology. The method realizes movie imaging for cardiac function examination, T1 quantitative imaging for myocardial tissue definition, T2 quantitative imaging and T2 * quantitative imaging in one-time scanning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical imaging, in particular to a one-stop cardiac magnetic resonance examination technology. BACKGROUND

[0002] Magnetic resonance imaging (MRI) technology uses two relaxation times generated in the interaction process between hydrogen protons in the human body and an external magnetic field, and obtains different contrast images of tissues by modulating multiple scanning parameters, which can perform multi-dimensional examination on structure, function, composition, blood flow and metabolites, and is an important medical imaging examination technology. Compared with CT, ultrasound and PET imaging technologies, MRI has the advantages of high sensitivity, high resolution, high spatial coverage, multi-angle, no radiation and good soft tissue contrast, and has been widely used in the examination of different diseases.

[0003] In heart disease, magnetic resonance is the gold standard method for evaluating structure, function and myocardial infarction. In recent years, quantitative imaging technology (English: mapping) measures the tissue magnetic resonance relaxation time of myocardial tissue under a fixed magnetic field, mainly the longitudinal relaxation time T1, the transverse relaxation time T2 and T2 * time. Compared with conventional contrast examination technology, this quantitative technology has the advantages of objectivity, independence from lesion severity and no need for reference tissue, and solves the limitation of magnetic resonance that cannot examine diffuse lesions, and has great potential clinical value in the early examination of heart disease. At the same time, this quantitative property helps to achieve accurate diagnosis and prognosis, and to carry out multi-center and long-term research. Due to the physical properties of T1, T2 and T2 * The sensitivity of these parameters to the microenvironment of myocardial tissue is different, so the changes in these tissue parameters caused by fibrosis, edema, iron deposition and fat infiltration in the tissue are inconsistent, so in the diagnosis of diseases, the clinician often needs to integrate the change information of these tissue parameters to make an accurate assessment.

[0004] T1, T2 or T2 * Quantitative imaging technology needs to acquire multiple images of the heart in the same motion state with different contrasts. In image post-processing, the signals of each pixel are fitted to the model of signal evolution in the image acquisition process to obtain the corresponding T1, T2 and T2 *The above process is fitted for all pixels to obtain a set of heart images, and the value of each pixel is the relaxation time of the corresponding tissue, which is called Map, and the whole process is called quantitative image reconstruction. At present, cardiac magnetic resonance needs to complete the examination of these different quantitative results through multiple independent scanning processes, which has the following problems: 1) there is no registration between different contrast images, which limits the further development of tissue feature analysis based on pixel level; 2) the execution of different scanning sequences, the switching between sequences and the preparation time of each sequence cause the problem of frequent patient breath holding and long total scanning time, which leads to low scanning efficiency, heavy economic burden on patients, etc.; 3) there is data redundancy between different imaging methods, and in the process of image reconstruction, the temporal and spatial redundancy information is not fully utilized to shorten the amount of data required for each examination; 4) the commonly used quantitative image reconstruction method (mainly least squares method) is sensitive to the initial value and is prone to fitting errors.

[0005] Therefore, it is urgent to develop a fast and high-integration cardiac magnetic resonance examination technology, which can realize one-stop cardiac function and multiple quantitative time examination in one scan, and a robust quantitative image reconstruction algorithm method, to solve the problems existing in the above-mentioned cardiac magnetic resonance quantitative technology and improve the scanning efficiency and disease diagnosis ability. SUMMARY

[0006] In view of the deficiencies of the prior art, the one-stop cardiac magnetic resonance examination technology provided by the present application mainly solves the problems of low scanning efficiency, multiple patient breath holding, long total breath holding time and incomplete registration between images when completing the examination in different scanning processes through different scanning methods.

[0007] In order to achieve the above-mentioned application purposes, the technical scheme adopted by the present application is as follows: a one-stop cardiac magnetic resonance examination technology, comprising the following steps:

[0008] S1, high-integration magnetic resonance imaging sequence: using a high-integration imaging sequence to collect original data of different cardiac magnetic resonance examinations through one scan, which includes quantitative imaging data for measuring T1, T2, T2 * of myocardial tissue and movie imaging data for evaluating ventricular function;

[0009] S2, extended deep learning reconstruction: using a deep learning-based reconstruction technology to reconstruct all different T1, T2, T2 * weight images and movie images in the quantitative imaging data and movie imaging data;

[0010] S3, deep learning-based quantitative image reconstruction: using a deep learning-based quantitative image reconstruction technology to reconstruct T1, T2 and T2 * quantitative images and improve image quality.

[0011] Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 * Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 * Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 * Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2

[0012] Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 * Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2

[0013] Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 * Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2

[0014] Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 * Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 * Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2

[0015] Further, the one-stop heart magnetic resonance examination technique comprises a T1 and T2 quantitative imaging module, a T2 *The quantitative imaging module has a golden angle of 124° between the two readout radio frequencies, a rotation angle of 0° between the multiple echoes, and a golden angle between the data of the same echo in the adjacent two cardiac cycles, so that the data of the same echo from multiple cardiac cycles can cover the entire K-space.

[0016] Further, in the one-stop cardiac magnetic resonance examination technology, when the respiratory motion is suppressed by the breath-hold method, the number of breath-hold is 1, and the acquisition time of the T1 and T2 quantitative imaging module is 4-6 cardiac cycles, i.e. 4-6 seconds. * The acquisition time of the image acquisition module is 4 cardiac cycles, i.e. 4 seconds, and the acquisition time of the cine imaging module is 4 cardiac cycles, i.e. 4 seconds.

[0017] When the respiratory motion is suppressed by the respiratory navigation, the T1 and T2 quantitative imaging module and the T2 * The diaphragmatic respiratory navigation is placed before the image acquisition or before the execution of the T2 preparation pulse, and the cine imaging is performed by the self-navigation method, the self-navigation data acquisition direction is the head-foot direction, and the execution cycle is 100-300 ms.

[0018] Further, in the one-stop cardiac magnetic resonance examination technology, the gradient echo acquisition method includes a fast gradient echo sequence and a balanced free induction gradient echo sequence, when the fast gradient echo sequence is used, the readout angle is 6-10°, and when the balanced free induction gradient echo sequence is used, the readout angle is 35°-40°.

[0019] Further, in the one-stop cardiac magnetic resonance examination technology, when the imaging sequence is applied to three-dimensional imaging, the image readout is performed by the multiple excitation method, the readout angle is 15°-20°, the respiratory motion is suppressed by the respiratory navigation, the heavy cardiac motion in free breathing is suppressed by reducing the acquisition window, the combination of the saturation pulse and the T2 preparation pulse is used as the signal preparation method for the simultaneous T1 and T2 quantitative imaging to reduce the signal waiting time, and the number of mixed T1 / T2 weighted images is reduced to accelerate the three-dimensional imaging.

[0020] Further, in the one-stop cardiac magnetic resonance examination technology, in S2, the unfolded deep learning reconstruction is reconstructed by an unfolded convolutional neural network deep learning model, the unfolded convolutional neural network deep learning model includes a convolutional neural network and a data fidelity term for removing image noise, and the convolutional neural network and the data fidelity term simulate the iterative optimization process of the traditional algorithm by cascading.

[0021] Further, in the one-stop heart magnetic resonance examination technology, in S3, the quantitative image reconstruction is reconstructed by a deep neural network, the deep neural network comprises a recurrent neural network and a multilayer fully connected neural network, the recurrent neural network comprises a plurality of nodes, and each node comprises a gray signal, a flip recovery time, an echo time of a T2 preparation pulse and an echo time of multiple echoes.

[0022] Further, in the one-stop heart magnetic resonance examination technology, the quantitative image reconstruction method comprises a least square method, and the least square method is specifically as follows: when a combination of a flip pulse and a T2 preparation pulse is used, a signal formula of T1 and T2 at the same time is formula (1):

[0023]

[0024] Wherein, A is an image signal corresponding to a longitudinal magnetization vector entering a steady state after multiple image acquisitions; T INV i is a flip recovery time corresponding to the i-th image; TEprepi is an echo time of a T2 preparation pulse;

[0025] T2 * A calculation formula of a quantitative image is formula (2):

[0026]

[0027] Wherein, Mss is a signal value of a longitudinal magnetization vector entering a steady state after an excitation pulse is flipped to a horizontal direction magnetization vector;

[0028] When a combination of a saturation pulse and a T2 preparation pulse is used, a signal formula of T1 and T2 at the same time is formula (3):

[0029]

[0030] Wherein, A is an image signal corresponding to a longitudinal magnetization vector entering a steady state after multiple image acquisitions; T SAT i is a flip recovery time corresponding to the i-th image; TEprepi is an echo time of a T2 preparation pulse.

[0031] Advantages of the present application are as follows: the method realizes that movie imaging for heart function examination, T1 quantitative imaging for myocardial tissue specification, T2 quantitative imaging and T2 * Imaging in a clinical magnetic resonance conventional examination scanning protocol are completed in one fast scanning, solves the problems of low scanning efficiency, many times of patient breath holding, long total breath holding time and patient economic burden in the current clinical scanning technology; a technology capable of obtaining a plurality of completely registered heart examination images is provided, and the problem of unregistration of results obtained by using different sequences in different scanning to complete the examination in the current clinical is solved, and the identification of a small abnormality from different contrast images is limited. BRIEF DESCRIPTION OF DRAWINGS

[0032] Fig. 1 is a flow chart of the present method;

[0033] Fig. 2 is a deployment of the fast high-integration cardiac scan method in the breath-hold manner in the present invention;

[0034] Fig. 3 is a deployment of the fast high-integration cardiac scan method in the free-breathing manner using diaphragmatic respiratory navigation and self-navigation in the present invention;

[0035] Fig. 4 is a deployment of the fast high-integration cardiac scan method in the free-breathing manner using diaphragmatic respiratory navigation and self-navigation and using saturated pulses as T1-weighting preparation pulses in the present invention;

[0036] Fig. 5 is an unrolled deep learning reconstruction network model based on convolutional neural network in the present method;

[0037] Fig. 6 is a T1, T2 and T2 * quantitative image reconstruction network model based on recurrent neural network and fully connected neural network in the present method;

[0038] Fig. 7 is a plurality of examination result images obtained in a single breath-hold cardiac magnetic resonance scan in the present method. DETAILED DESCRIPTION

[0039] The specific embodiments of the present invention are described below to facilitate the understanding of the present invention for those skilled in the art, but it should be clear that the present invention is not limited to the scope of the specific embodiments, and for those skilled in the art, all the changes that are obvious within the scope of the present invention defined and determined by the appended claims are within the scope of the present invention.

[0040] As shown in Figs. 1-7 , the present embodiment provides a one-stop cardiac magnetic resonance examination technology, including the following steps:

[0041] S1, high-integration magnetic resonance imaging sequence: using a high-integration imaging sequence to acquire original data of different cardiac magnetic resonance examinations by a single scan, which includes T1, T2, T2 * quantitative imaging data for measuring myocardial tissue and cine imaging data for evaluating ventricular function;

[0042] S2, unrolled deep learning reconstruction: using a deep learning-based reconstruction technique to reconstruct all different T1, T2, T2* Weighted images and movie images;

[0043] S3, deep learning based quantitative image reconstruction: using deep learning based quantitative image reconstruction technique to reconstruct T1, T2 and T2 * quantitative images, and improve image quality.

[0044] The method contains simultaneous T1, T2 and T2 * quantitative imaging and movie imaging parts; from the aspect of examination content, the method can evaluate both myocardial tissue composition and ventricular function; from the aspect of function, the method can check quantitative parameters (such as ejection fraction, output) of fibrosis, fat infiltration, edema, iron deposition and inflammation existing in myocardium, and the contraction and diastolic function of ventricle. In addition, the movie imaging part of the method can be used to reconstruct real-time movie images for checking arrhythmia and other problems.

[0045] The method contains a fast and high-integration imaging sequence, which contains a simultaneous T1-T2 quantitative imaging module, a T2 * quantitative imaging module and a movie imaging module, which are sequentially executed.

[0046] The imaging method, in terms of image acquisition method, is able to be compatible with simultaneous T1-T2 quantitative imaging, T2 * quantitative measurement and movie scanning, gradient echo with dispersion effect is adopted. Gradient echo is less affected by B0 field and B1 field inhomogeneity, ensuring the accuracy of T1 and T2. The excitation angle range of gradient echo acquisition sequence should be 6°-10°, and the specific angle can be optimized according to whether the system enters steady state before acquisition. *

[0047] In the method, the images of the simultaneous T1-T2 quantitative imaging module and the movie imaging module adopt single-shot single-echo signal readout mode with a single excitation, the acceleration factor is 2-4, and the data acquisition window size is 190-220 milliseconds; the T2 * quantitative imaging adopts multiple-echo mode, the acceleration factor is 3-5, the number of echoes is 8-10, the repeated acquisition time is 27-30 milliseconds, the growth interval of echo time is 2 milliseconds, and the data acquisition window size is 190-220 milliseconds. T2 * The signal formula of the weighted image is: Here Mss is the component of the longitudinal magnetization vector after multiple imaging, which is flipped to the horizontal plane after entering the steady state, and TEi is the echo center time of the ith echo. Each T2 * The data of the quantitative image is acquired on multiple cardiac cycles by segmentation. ​

[0048] In the magnetic resonance data acquisition of the method, a radiation trajectory with a golden angle rotation angle is designed to fill the K-space data, and the radiation trajectory has the ability to realize high multiple accelerated acquisition. For a single echo sequence, the rotation angle between every two excitations is designed to be 124°; in T1, T2 and movie imaging, the repetition acquisition time (TR) is 3 ms, and the echo time is 2-4 ms; for T2 * In the multiple echo acquisition mode of quantitative imaging, the number of echoes is 8-10, the repetition acquisition time (TR) is 27-30 ms, and the growth interval of the echo time is 2 ms; the rotation angle between echoes is 0. However, for the same echo image, the rotation angle between the adjacent two readouts is 124°. In the present application, the field of view (FOV) of the image is designed to be 228 mm x 228 mm, and the resolution is 2 x 2 mm 2 Therefore, the full sampling data of each image is 114 radiation lines.

[0049] In order to reduce the influence of heart motion, electrocardiogram gating is used in the present application. Specifically, during image acquisition, the electrocardiogram gating checks the R wave of each cardiac cycle and triggers the start time of subsequent image acquisition, so that the single excitation image comes from the middle or end of the diastolic period of the cardiac cycle, and the single excitation image acquisition window width is 190-220 ms, which ensures that the heart motion is small. During the data acquisition process of the movie imaging module, the electrocardiogram gate is not triggered, but the signal of the electrocardiogram gate will be recorded and used in the retrospective data processing.

[0050] Specifically, the simultaneous T1-T2 quantitative imaging module proposed in the present method acquires 4-6 mixed T1 / T2 weight images on 4-6 cardiac cycles, and the repetition time of the single excitation image is designed to be 3 ms. Therefore, on one cardiac cycle, the number of radiation lines of K-space of each image is 72, and the acceleration is 2 times.

[0051] The mixed weight signal of simultaneous T1 and T2 quantitative imaging is prepared by using the combination of T1 preparation pulse and T2 preparation pulse, and the flip pulse is used as the T1 preparation pulse; the T2 weight is prepared by the T2 preparation pulse with the Carr-Purcell-Meiboom-Gill signal characteristics. In sequence, the T1 preparation pulse is first executed to make the longitudinal magnetization vector start to relax recovery, and then the T1 weight image is acquired; from the second image, the T2 preparation pulse is applied before acquisition to introduce the T2 weight, and the longitudinal magnetization vector before image acquisition is: Here M ZSS is the value of the longitudinal magnetization vector entering the steady state after multiple image acquisitions, T INV is the time between the flip pulse and the image acquisition or the T2 preparation pulse, and TE prepis the echo time of the T2 preparation pulse. Both the flip pulse and the T2 preparation pulse are designed as adiabatic pulses, and the echo time of the T2 preparation pulse is optimized in the range of 25 ms to 60 ms.

[0052] T2 * The quantitative imaging part acquires the signal values at equal intervals on the free induction decay (FID) curve in a multi-echo manner to obtain different T2 * weight images. In this method, T2 * weight images are acquired in 4 consecutive cardiac cycles, 8 shots are performed in each cardiac cycle, and the distance between two shots is 25-30 ms. Therefore, 9 echoes are continuously acquired at equal intervals after each shot, and the step length of the echoes is 2.5-3 ms, ensuring that the image acquisition window in each cardiac cycle does not exceed 210 ms. The data from the same echo in different cardiac cycles are complementary in K-space, thereby ensuring that the image of each echo has 32 radial lines, with a 4-fold acceleration, to ensure that the image can be reconstructed.

[0053] The cine imaging module of this method acquires data in 4 consecutive cardiac cycles, and the time resolution of each cardiac phase is designed to be 42 ms. The K-space radial lines acquired in each cardiac cycle are 12, and the acquisition time of each line is 3 ms. During the acquisition process, the patient's electrocardiogram signal is recorded, and in the image post-processing, the data of the same phase in 4 cardiac cycles are retrospectively combined into one K-space according to the recorded electrocardiogram signal and the phase window length, with a 2-fold acceleration.

[0054] The execution sequence of the image acquisition process of this method is simultaneous T1 and T2 quantitative imaging, T2 * quantitative imaging, and cine imaging. This sequence ensures that after the simultaneous T1 and T2 quantitative imaging, the longitudinal magnetization vector enters a steady state, which does not affect the subsequent T2 * quantitative imaging. The number of times the radio frequency pulses are executed in T2 * quantitative imaging is sparse, about 8-9 times, so the disturbance to the longitudinal magnetization vector is small, which allows the longitudinal magnetization vector to be preserved before cine imaging, ensuring that the cine imaging has a high signal-to-noise ratio.

[0055] When using breath-holding to suppress respiratory motion: the length of the acquisition process is 14 cardiac cycles, ensuring that the single breath-holding time does not exceed the average breath-holding capacity of the patient, and the breath-holding lengths of the various modules (simultaneous T1-T2 quantitative imaging, T2 * quantitative imaging, and cine imaging) are optimized to be 40%, 30%, and 30%.

[0056] This method employs an unfolded deep learning reconstruction network model based on convolutional neural networks (CNNs) for image reconstruction. This reconstruction network model consists of a U-shaped CNN and a data consistency module, which are cascaded multiple times to simulate the iterative optimization process in parallel imaging. Here, the CNN is used to remove aliasing artifacts caused by high acceleration, while the data consistency module ensures that the original data remains unchanged before and after image denoising. This invention trains the neural network using volumetric data. Specifically, the images of the Cartesian acquisition trajectory are resynthesized into a dataset of radiation trajectories; the fully acquired radiation trajectory data is used to reconstruct the reference image; simulated down-acquisition is then performed on the fully acquired data to obtain training data.

[0057] This method constructs a deep learning model based on recurrent neural networks and fully connected networks to reconstruct quantitative images. Each input endpoint in the recurrent neural network contains a grayscale signal S and corresponding time information t (flip recovery time, T2 echo time of the preparation pulse, or T2...). * The recurrent neural network (RNN) contains multiple hidden layers and is designed to be bidirectional to mine the relationship between the acquired signal and its corresponding time (center time of multiple echoes). The input of the fully connected network is the output of the RNN, and the output of the fully connected network is T1, T2, or T2... * Fully connected neural networks contain 8-10 hidden layers, each with multiple nodes. The activation function between each hidden layer in both recurrent neural networks and fully connected neural networks is based on a modified linear unit function. The recurrent neural networks and fully connected neural networks in this method are trained using simulated, phantom, and in vivo data. During the generation of phantom and simulated data, different B1 values, partial resonance, noise levels, and heart rates are incorporated to improve the robustness of the network model and the quality of the images.

[0058] When acquiring data using free breathing, motion artifacts are introduced, necessitating optimization of the length and number of repetitions for each module. Due to the need for repetitive acquisition, appropriate idle times are set based on the preparation method of the T1 / T2 weighted signals to ensure consistency of the longitudinal magnetization vector between different repetitions. When using diaphragmatic breathing navigation gating, data rejection and re-acquisition processes occur, requiring optimization of the T1 / T2 weighted signal preparation to ensure that the data re-acquired after rejection contributes significantly to the signal and T1 / T2 calculations compared to the rejected data. When using respiratory navigation, different respiratory navigation methods can be employed depending on the task of each module; for T1 / T2 / T2... *The module employs diaphragmatic breathing navigation; for cine data acquisition, cardiac self-navigation is used. When using diaphragmatic breathing navigation, it is positioned at the junction of the liver and lungs. Timing-wise, diaphragmatic breathing gating is placed before image acquisition or before the T2 pulse. Breathing navigation can be used to determine whether subsequent image data is accepted or reacquired; it can also be used to track the real-time position of the heart and adjust the imaging plane. When using breathing navigation for cine imaging, cardiac self-navigation is used, with an execution cycle of 100-500ms.

[0059] The imaging method proposed in this paper can employ Cartesian trajectories for K-space filling. When the trajectory is a Cartesian trajectory, the accelerated image acquisition method is designed as compressed sensing or parallel imaging; simultaneously, the acceleration factor for T1 and T2 quantitative imaging is designed to be 2-3; T2 * The acceleration factor for quantitative imaging is 4.5x; the acceleration factor for real-time cinematic imaging is 14x; and the acceleration factor for cinematic imaging through multi-segment reconstruction is 3.5x.

[0060] In this method, simultaneous T1 and T2 quantitative imaging, when preparing a mixed-weighted signal using a combination of saturation pulses and T2 preparation pulses, involves the following timing sequence: neither the saturation pulse nor the T2 preparation pulse is used before the first image acquisition; both preparation pulses are executed simultaneously before subsequent image acquisitions. More specifically, the T2 preparation pulse is executed before image acquisition begins, and the saturation pulse is placed before each T2 preparation pulse. The time interval between the saturation pulse and the T2 preparation pulse is set to T. SAT The signal for the image is: Here, Mz0 is the value after the longitudinal magnetization vector is fully recovered.

[0061] This method can also employ a least squares fitting algorithm for calculating T1 and T2. Specifically, the acquired signal, T1 relaxation recovery time T... INV The echo time TE of the T2 preparation pulse prep Fit to The parameter model. If a saturation pulse and T2prep are used to prepare the pulse, the signal formula is: For T2 * Quantitative calculation, the fitting formula is:

[0062] When the scanning technology in this method is deployed in three dimensions, a multi-excitation approach is used to complete the scanning of the entire heart. In addition to using acceleration of 2x or higher, similar to two-dimensional data acquisition, three-dimensional data requires multiple excitations to complete the acquisition of a single image; respiratory movements are synchronized using diaphragmatic navigation; furthermore, the acquisition duration of each excitation is controlled to be around 100ms; and the number of images with different T1 / T2 weights can be reduced to 3-4.

[0063] The above merely illustrates the embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical scope disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A one-stop cardiac magnetic resonance examination technique, characterized in that, The method comprises the following steps: S1, high-integration magnetic resonance imaging sequence: acquiring raw data for different cardiac magnetic resonance examinations by one scan using a high-integration imaging sequence, which includes T1, T2, T2 * quantitative imaging data and cine imaging data for evaluating ventricular function; The imaging sequence comprises three parts of a simultaneous T1 and T2 quantitative imaging module, a T2* quantitative imaging module and a cine imaging module; the sequence of the three parts is designed as the order of the T1 and T2 quantitative imaging module, the T2* quantitative imaging module and the cine imaging module to be completed in one sequence. The simultaneous T1 and T2 quantitative imaging module prepares T1 and T2 simultaneous quantitative imaging signals by the combination of a flip pulse and a T2 preparation pulse; the data of the simultaneous T1 and T2 quantitative imaging module and the cine imaging module are collected by a single-echo gradient echo signal readout mode; the data of the T2* quantitative imaging module are collected by a multi-echo gradient echo signal readout mode; the T1 and T2 simultaneous quantitative imaging signals can also be prepared by the combination of a saturation pulse and a T2 preparation pulse. The original data adopt a golden rotation angle radiation track, for the simultaneous T1 and T2 quantitative imaging module and the cine imaging module by two-dimensional single-echo, the rotation angle between every two readout radio frequencies is a golden angle; for the T2* quantitative imaging module by multi-echo, the rotation angle between two readout radio frequencies is a golden angle 124°, and the rotation angle between multi-echoes is 0°; the rotation angle between the data of the same echo in adjacent two cardiac cycles is a golden angle, so that the data of the same echo from multiple cardiac cycles can cover the entire K-space. S2, unfolded deep learning reconstruction: reconstructing all different T1, T2, T2 * weight images and movie images; S3, deep learning based quantitative image reconstruction: T1, T2 and T2 * quantitative images, and improve image quality.

2. One-stop cardiac magnetic resonance examination technique according to claim 1, characterized in that, T1 and T2 quantitative imaging modules and T2 * The T1 and T2 quantitative imaging modules acquire data in the mid-diastole or end-diastole phase of the cardiac cycle by electrocardiogram gating, while the cine imaging module acquires electrocardiogram data continuously over multiple cardiac cycles. Based on the electrocardiogram signal, the data is retrospectively sorted into different cardiac phases, avoiding image artifacts caused by motion in adjacent phases.

3. The one-stop cardiac magnetic resonance examination technique of claim 1, wherein, The simultaneous T1 and T2 quantitative imaging module, T2 * The quantitative imaging module and the cine imaging module realize fast scanning by using different image acceleration multiples; the acceleration multiple of the simultaneous T1 and T2 quantitative imaging module is 2-4 times; the acceleration multiple of the T2 * The acceleration multiple of the quantitative imaging module is 3-5 times; and the acceleration multiple of the cine imaging module is 11-16 times.

4. The one-stop cardiac magnetic resonance examination technique of claim 1, wherein, When the respiratory motion is suppressed in the breath-hold manner: the breath-hold times is 1, at this time the acquisition time length of the T1 and T2 quantitative imaging modules is 4-6 cardiac cycles, i.e. 4-6 seconds; the T2 * The acquisition time length of the image acquisition module is 4 cardiac cycles, i.e. 4 seconds; the acquisition time length of the cine imaging module is 4 cardiac cycles, i.e. 4 seconds; With respiratory navigation to suppress respiratory motion: simultaneous T1 and T2 quantitative imaging module and T2 * The diaphragmatic respiratory navigation is placed before the image acquisition or before the T2 preparation pulse execution by the quantitative imaging module. The cine imaging adopts a self-navigation mode, the self-navigation data collection direction is the head-foot direction, and the execution period is 100-300 ms.

5. The one-stop cardiac magnetic resonance examination technique of claim 1, wherein, The gradient echo collection mode comprises two modes of a fast gradient echo sequence and a balanced free precession gradient echo sequence; when the fast gradient echo sequence is adopted, the readout angle is 6-10°; when the balanced free precession gradient echo sequence is adopted, the readout angle is 35°-40°.

6. The one-stop cardiac magnetic resonance examination technique of claim 1, wherein, When the imaging sequence is applied to three-dimensional imaging, the image readout adopts a multi-stimulated mode, the readout angle is 15°-20°, the respiratory motion is inhibited by using respiratory navigation, and the cardiac motion with aggravated free breathing is inhibited by reducing the collection window. The combination of a saturation pulse and a T2 preparation pulse is used as the signal preparation mode of the simultaneous T1 and T2 quantitative imaging, the signal waiting time is reduced, the number of mixed T1 / T2 weight images is reduced, and the three-dimensional imaging is accelerated.

7. The one-stop cardiac magnetic resonance examination technique of claim 1, wherein, In S2, the unfolded deep learning reconstruction is reconstructed by an unfolded convolutional neural network deep learning model, the unfolded convolutional neural network deep learning model comprises a convolutional neural network and a data fidelity term for removing image noise, and the convolutional neural network and the data fidelity term simulate the iterative optimization process of a traditional algorithm in a cascaded manner.

8. The one-stop cardiac magnetic resonance examination technique of claim 1, wherein, In S3, the quantitative image reconstruction is reconstructed by a deep neural network, the deep neural network comprises a recurrent neural network and a multi-layer fully connected neural network, the recurrent neural network comprises a plurality of nodes, and each node comprises a gray signal, a flip recovery time, an echo time of a T2 preparation pulse and an echo time of multi-echoes.

9. The one-stop cardiac magnetic resonance examination technique of claim 1, wherein, The quantitative image reconstruction method includes a least square method, and the least square method is specifically: when a combination of a flip pulse and a T2 preparation pulse is used, a signal formula of T1 and T2 at the same time is formula (1): ； wherein A is an image signal corresponding to the entry of the longitudinal magnetization vector into the steady state after a plurality of image acquisitions; T INV i is the inversion recovery time corresponding to the i-th image; TEprepi is the echo time of the T2 preparation pulse; T2 * The calculation formula of the quantitative image is formula (2): ； Wherein Mss is a signal value of a longitudinal magnetization vector after entering a steady state and being flipped to a horizontal direction by an excitation pulse; When a combination of a saturation pulse and a T2 preparation pulse is used, a signal formula of T1 and T2 at the same time is formula (3): ； wherein A is the image signal corresponding to the full recovery of the longitudinal magnetization vector; T SAT i is the saturation recovery time corresponding to the i-th image; TEprepi is the echo time of the T2 preparation pulse.

Images