3D navigator-based 3D gradient spin-echo diffusion imaging method, and device
By using a 3D gradient spin echo diffusion imaging sequence based on a 3D navigator, and by acquiring and correcting phase errors in segments, the problems of motion artifacts and phase errors in 3D diffusion imaging were solved, and high signal-to-noise ratio high-resolution diffusion imaging of the whole brain was achieved.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- ZHEJIANG UNIV
- Filing Date
- 2024-12-27
- Publication Date
- 2026-07-02
AI Technical Summary
Existing 3D diffusion magnetic resonance imaging technology suffers from motion artifacts and image distortion during whole-brain acquisition due to the lengthy phase encoding process. Furthermore, traditional 1D or 2D navigators cannot effectively correct phase errors in 3D k-space, affecting image quality.
Using a 3D gradient spin echo diffusion imaging sequence based on a 3D navigator, the whole brain signals are acquired in segments and phase errors are corrected using 3D navigation echo signals. Combined with global saturation, diffusion preparation, fat saturation and gradient spin echo modules, accurate reconstruction of 3D K-space data is achieved.
It improves the signal-to-noise ratio of whole-brain magnetic resonance diffusion imaging, eliminates phase errors between multiple acquisitions, realizes high-resolution diffusion imaging of the whole brain, and enhances the clinical applicability of imaging.
Smart Images
Figure CN2024143371_02072026_PF_FP_ABST
Abstract
Description
3D Gradient Spin Echo Diffusion Imaging Method and Device Based on 3D Navigator Technical Field
[0001] This invention relates to the field of magnetic resonance technology, and more particularly to the field of diffusion magnetic resonance imaging. Background Technology
[0002] Diffusion-weighted MRI (dMRI) can detect tissue microstructures based on the restricted diffusion characteristics of water molecules within the microenvironment of biological tissues. 3D MRI pulse sequences offer high signal-to-noise ratio (SNR) efficiency for high-resolution acquisition, particularly when using 3D k-space coding to define very thin slices, while two-dimensional multilayer acquisition often suffers from suboptimal slice resolution and low SNR. Therefore, 3D acquisition is ideal for high-resolution dMRI, mitigating SNR loss caused by diffusion-weighted attenuation and capturing the tissue's microstructure in continuous 3D space.
[0003] However, due to the lengthy phase encoding process in 3D imaging, whole-brain acquisition often requires segmented acquisition. In 3D-dMRI, additional eddies and motion introduce severe phase errors between segments during segmented acquisition, leading to motion artifacts and distortion in the images, thus posing challenges to the clinical application of this technology in humans. Adding navigation echoes to segmented acquisition is an effective method for correcting phase errors between segments. However, previous research has largely been limited to 1D or 2D navigators, which track the phase of the signal along the readout direction, requiring little or no phase encoding in one direction. Therefore, for diffusion-weighted 3D gradient-spin echo sequences (DW-GRASE), how to track the phase changes of the image signal across the entire 3D k-space and improve the accuracy of inter-segment phase error correction is a pressing technical problem that needs to be solved. Summary of the Invention
[0004] To achieve high-resolution 3D diffusion imaging of the whole brain, this invention proposes a 3D gradient spin echo diffusion imaging sequence (3D DW-GRASE) based on a 3D navigator to address the problems of the aforementioned dMRI sequences and improve the clinical applicability of 3D diffusion imaging.
[0005] To achieve the above objectives, the present invention employs the following technical solution:
[0006] In a first aspect, the present invention provides a 3D gradient spin echo diffusion imaging method based on a 3D navigator, comprising:
[0007] S1: Whole-brain signals are acquired in segments by repeatedly executing a 3D gradient spin echo diffusion imaging sequence based on a 3D navigator. After each execution of the imaging sequence, a 3D GRASE readout signal and a 3D navigation echo signal are obtained. All 3D GRASE readout signal segments constitute whole-brain 3D K-space data.
[0008] The 3D gradient spin echo diffusion imaging sequence based on a 3D navigator consists of a global saturation module, a diffusion preparation module, a fat saturation module, a gradient spin echo module, and a 3D navigation echo module in sequence.
[0009] The global saturation module is used to destroy residual lateral magnetization by applying a gradient across the entire 3D imaging space.
[0010] The diffusion preparation module is used to embed a pair of diffusion gradients into a 90° angle. x -180° y -90° -x In radio frequency pulses, the diffusion coding and signal acquisition are separated; the diffusion gradient is a trapezoidal cosine oscillation diffusion gradient or a pulse diffusion gradient.
[0011] The fat saturation module is used to suppress fat signals;
[0012] The gradient spin echo module is used to acquire signals in 3D K-space to obtain a 3D GRASE readout signal.
[0013] The navigation echo module is used to acquire signals from the 3D navigator to obtain a 3D navigation echo signal.
[0014] S2. Perform a general automatic calibration parallel acquisition and reconstruction operation on the whole brain 3D K-space data. Each 3D GRASE readout signal in the whole brain 3D K-space data is corrected for phase error using the 3D navigation echo signal acquired during the same imaging sequence. The gaps in each 3D GRASE readout signal after phase error correction are filled. Finally, a 3D Fourier transform is performed on all the filled 3D GRASE readout signals to obtain the reconstructed image.
[0015] Based on this scheme, the following preferred implementation methods can be further provided for each step. It should be noted that the technical features in each preferred method can be combined with each other as long as there is no conflict. Of course, these preferred methods can also be implemented in other ways that achieve the same technical effect, and this does not constitute a limitation.
[0016] Preferably, the process executed in the global saturation module of the imaging sequence is as follows:
[0017] At the beginning of the imaging sequence, three B1 radio frequency pulses are applied, and at the same time as each B1 radio frequency pulse is applied, gradients with completely equal amplitudes are applied in the X, Y, and Z directions of the 3D imaging space to eliminate lateral magnetization. After the last B1 radio frequency pulse is executed, the system waits for the longitudinal magnetization vector to recover according to the preset post-saturation delay time (PSD) before executing the next diffusion preparation module.
[0018] Preferably, the process executed in the diffusion preparation module of the imaging sequence is as follows:
[0019] First, an RF excitation pulse with a 90° flip angle is applied along the X-axis; then, a diffusion gradient, which is a trapezoidal cosine oscillation diffusion gradient or a pulse diffusion gradient, is applied along a preset diffusion direction; next, an RF refocusing pulse with a 180° flip angle is applied along the Y-axis, and then the exact same diffusion gradient is applied again; finally, a stabilizer gradient (G) is applied along the Z-axis. stb Finally, a radio frequency excitation pulse with a 90° flip angle along the -X axis is used to convert the transverse magnetization vector into a longitudinal magnetization vector.
[0020] Preferably, the execution flow in the fat saturation module of the imaging sequence is as follows:
[0021] Three frequency-selective Gaussian pulses with 95° flip angles are applied to flip the cos(95°) part of the longitudinal magnetization vector of fat three times and set it to 0 after waiting for T1*ln(2) time, so that most of the longitudinal magnetization vector of fat is converted into the transverse magnetization vector. Gradients are applied synchronously in the X, Y and Z axis directions while applying the Gaussian pulses to completely eliminate the transverse magnetization vector of fat; where T1 is the longitudinal relaxation time of fat.
[0022] Preferably, the execution flow of the gradient spin echo sequence module in the imaging sequence is as follows:
[0023] First, signals are acquired using three navigation echoes. Then, 3D GRASE readout is achieved by performing echo plane imaging (EPI) encoding in the Y direction and fast spin echo (TSE) encoding in the Z direction, resulting in a 3D GRASE readout signal. During the 3D GRASE readout process, linear encoding is performed along the EPI direction, and center encoding is performed along the TSE direction. Parallel GRAPPA accelerated imaging is performed in both the EPI and TSE directions. Simultaneously, a stabilizer gradient needs to be applied along the Z-axis before and after each EPI readout moment, and the magnitude of the stabilizer gradient is the same as the magnitude of the stabilizer gradient applied in the diffusion preparation module.
[0024] Preferably, the processing flow in the navigation echo module of the imaging sequence is as follows:
[0025] N segments of spin echo signals centered in the Z direction of k-space are acquired, and EPI encoding is performed along the X and Y directions, while TSE encoding is performed in the Z direction to obtain a 3D navigation echo signal. A stabilizer gradient needs to be applied along the Z-axis before and after each EPI readout moment, and the gradient magnitude of the stabilizer gradient is the same as the magnitude of the stabilizer gradient applied in the diffusion preparation module. N is an even number not less than 4.
[0026] Preferably, the specific process for performing a general automatic calibration parallel acquisition and reconstruction operation on the whole-brain 3D K-space data in step S2 is as follows:
[0027] S21. Extract a 3D GRASE readout signal and a 3D navigation echo signal obtained after each execution of the imaging sequence from the whole brain 3D K-space data. Use the signals collected at the three navigation echo locations in the 3D GRASE readout signal to estimate the phase change between the odd and even rows of EPI. Then use GRAPPA technology to fill in the missing rows in k-space where parallel accelerated imaging acquisition is performed along the EPI direction.
[0028] S22. For each 3D GRASE readout signal after the missing rows are filled, the phase error between multiple excitations of the 3D GRASE readout signal is corrected using the corresponding 3D navigation echo signal. Then, the GRAPPA technique is used to fill in the missing rows of each 3D GRASE readout signal after the phase error is corrected by parallel accelerated imaging acquisition along the TSE direction, thus obtaining a complete 3D GRASE readout signal.
[0029] S23. Reassemble all complete 3D GRASE readout signal segments into complete whole-brain 3D K-space data, and obtain the reconstructed image through 3D Fourier transform.
[0030] Preferably, in step S22, the specific method for correcting the phase error of the 3D GRASE readout signal using the 3D navigation echo signal is as follows:
[0031] First, both the 3D GRASE readout signal and the 3D navigation echo signal are transformed from k-space to the xyz three-dimensional image space to obtain the first phase signal of the 3D navigation echo signal for each voxel (x,y,z) in the image space. The second phase signal ψ of the 3D GRASE readout signal for each voxel (x,y,z) in the image space. i,s (x,y,z);
[0032] Then, by subtracting the second phase signal from the first phase signal, the corrected phase ψ' of the 3D GRASE readout signal for each voxel (x,y,z) in the image space is obtained. i,s (x,y,z)=ψ i,s (x,y,z)-
[0033] Finally, the three-dimensional image of the phase-corrected 3D GRASE readout signal is transformed back to k-space to obtain the phase-corrected 3D GRASE readout signal.
[0034] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, controls an external device to implement the 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in any of the solutions of the first aspect above.
[0035] Thirdly, the present invention provides a magnetic resonance imaging device, which includes a magnetic resonance scanner and a control unit. The control unit stores a computer program, which, when executed, controls the magnetic resonance scanner to implement the 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in any of the solutions of the first aspect above.
[0036] Compared with the prior art, the present invention has the following beneficial effects:
[0037] This invention proposes a 3D gradient spin echo diffusion imaging sequence (3DDW-GRASE) based on a 3D navigator, which for the first time combines diffusion gradient encoding, 3D gradient spin echo imaging, and 3D navigation echo sequences. Based on this sequence, whole-brain signals can be divided into N segments for sequential signal acquisition, obtaining N segments of 3D GRASE readout signals and N segments of 3D navigation echo signals. The phase of the 3D GRASE readout signals can be corrected based on the 3D navigation echo signals. Compared with traditional 2D-EPI acquisition used on 3T clinical systems, the 3D DW-GRASE sequence not only improves the signal-to-noise ratio of whole-brain magnetic resonance diffusion imaging but also, due to its higher resolution, enables the detection of diffusion time dependence of the gray matter cortex in t-space. The 3D DW-GRASE sequence based on a 3D navigator effectively eliminates phase errors between multiple acquisitions, enabling high-resolution diffusion imaging of the whole brain. Attached Figure Description
[0038] Figure 1 is a timing diagram of the 3D DW-GRASE sequence based on a 3D navigator.
[0039] Figure 2 is a magnified timing diagram of the global saturation module, diffusion preparation module, fat saturation module, and gradient spin echo module within the 3D DW-GRASE sequence in Figure 1.
[0040] Figure 3 is an enlarged timing diagram of the 3D navigation echo module within the 3D DW-GRASE sequence in Figure 1.
[0041] Figure 4 shows the b0 and b1000 images obtained using the 3D DW-GRASE sequence, and illustrates the results with and without phase correction using 1D, 2D, and 3D navigation echoes.
[0042] Figure 5 shows the comparison of the signal-to-noise ratios of b0 and b1000 images obtained from 7 subjects using 3D DW-GRASE and 2D EPI sequences, respectively, in the three regions of interest. Detailed Implementation
[0043] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below. Technical features in various embodiments of the present invention can be combined accordingly without mutual conflict.
[0044] This invention designs a 3D gradient spin echo diffusion imaging sequence (3DDW-GRASE) based on a 3D navigator. This 3D DW-GRASE sequence consists of a global saturation module, a diffusion preparation module, a fat saturation module, a gradient spin echo module, and a 3D navigation echo module, arranged sequentially. The functions of each module are as follows:
[0045] A global saturation module is used to destroy residual lateral magnetization by applying a gradient across the entire 3D imaging space;
[0046] The diffusion preparation module is used to embed a pair of diffusion gradients into a 90° angle. x -180° y -90° -x In radio frequency pulses, the diffusion coding and signal acquisition are separated; the diffusion gradient is a trapezoidal cosine oscillation diffusion gradient or a pulse diffusion gradient.
[0047] The fat saturation module is used to suppress fat signals;
[0048] The gradient spin echo module is used to acquire signals in 3D K-space to obtain a 3DGRASE readout signal.
[0049] The navigation echo module is used to acquire signals from the 3D navigator to obtain a 3D navigation echo signal.
[0050] In an embodiment of the present invention, the timing diagram of the 3D DW-GRASE sequence is shown in Figure 1. The various modules in this imaging sequence will now be described in detail with reference to Figure 1.
[0051] In the imaging sequence, the first module is the global saturation module, and the execution flow in this module is as follows:
[0052] At the beginning of the imaging sequence, three B1 radio frequency pulses are applied, and at the same time as each B1 radio frequency pulse is applied, gradients with completely equal amplitudes are applied in the X, Y, and Z directions of the 3D imaging space to eliminate lateral magnetization. After the last B1 radio frequency pulse is executed, the system waits for the longitudinal magnetization vector to recover according to the preset post-saturation delay time (PSD) before executing the next diffusion preparation module.
[0053] In the imaging sequence, the second module is the diffusion preparation module, and the process executed in this module is as follows:
[0054] First, an RF excitation pulse with a 90° flip angle is applied along the X-axis; then, a diffusion gradient, which is a trapezoidal cosine oscillation diffusion gradient or a pulse diffusion gradient, is applied along a preset diffusion direction; next, an RF refocusing pulse with a 180° flip angle is applied along the Y-axis, and then the exact same diffusion gradient is applied again; finally, a stabilizer gradient (G) is applied along the Z-axis. stb Finally, a radio frequency excitation pulse with a 90° flip angle along the -X axis is used to convert the transverse magnetization vector into a longitudinal magnetization vector.
[0055] In the imaging sequence, the third module is the fat saturation module, and the workflow executed in this module is as follows:
[0056] Three frequency-selective Gaussian pulses with 95° flip angles are applied to flip the cos(95°) part of the longitudinal magnetization vector of fat three times and set it to 0 after waiting for T1*ln(2) time, so that most of the longitudinal magnetization vector of fat is converted into the transverse magnetization vector. Gradients are applied synchronously in the X, Y and Z axis directions while applying the Gaussian pulses to completely eliminate the transverse magnetization vector of fat; where T1 is the longitudinal relaxation time of fat.
[0057] In the imaging sequence, the fourth module is the gradient spin echo sequence module, and the execution flow in this module is as follows:
[0058] First, signals are acquired using three navigation echoes. Then, 3D GRASE readout is achieved by performing echo plane imaging (EPI) encoding in the Y direction and fast spin echo (TSE) encoding in the Z direction, resulting in a 3D GRASE readout signal. During the 3D GRASE readout process, linear encoding is performed along the EPI direction, and center encoding is performed along the TSE direction. Parallel GRAPPA accelerated imaging is performed in both the EPI and TSE directions. Simultaneously, a stabilizer gradient needs to be applied along the Z-axis before and after each EPI readout moment, and the magnitude of the stabilizer gradient is the same as the magnitude of the stabilizer gradient applied in the diffusion preparation module.
[0059] In the imaging sequence, the fifth module is the navigation echo module, and the process executed in this module is as follows:
[0060] N segments of spin echo signals centered in the Z direction of k-space are acquired, and EPI encoding is performed along the X and Y directions, while TSE encoding is performed in the Z direction to obtain a 3D navigation echo signal. A stabilizer gradient needs to be applied along the Z-axis before and after each EPI readout moment, and the gradient magnitude of the stabilizer gradient is the same as the magnitude of the stabilizer gradient applied in the diffusion preparation module. N is an even number not less than 4.
[0061] The aforementioned 3D gradient spin echo diffusion imaging sequence based on a 3D navigator can be embedded in a magnetic resonance imaging device, thereby acquiring and reconstructing K-space data based on the imaging sequence to complete magnetic resonance imaging.
[0062] Therefore, in the embodiments of the present invention, based on the above-mentioned 3D gradient spin echo diffusion imaging sequence based on a 3D navigator, a 3D gradient spin echo diffusion imaging method based on a 3D navigator is provided, which includes the following steps S1 and S2.
[0063] S1: Whole-brain signals are acquired in segments by repeatedly executing a 3D gradient spin echo diffusion imaging sequence based on a 3D navigator. After each execution of the imaging sequence, a 3D GRASE readout signal and a 3D navigation echo signal are obtained. All 3D GRASE readout signal segments constitute whole-brain 3D K-space data.
[0064] S2. Perform a general automatic calibration parallel acquisition and reconstruction operation on the whole brain 3D K-space data. Each 3D GRASE readout signal in the whole brain 3D K-space data is corrected for phase error using the 3D navigation echo signal acquired during the same imaging sequence. The gaps in each 3D GRASE readout signal after phase error correction are filled. Finally, a 3D Fourier transform is performed on all the filled 3D GRASE readout signals to obtain the reconstructed image.
[0065] In an embodiment of the present invention, the specific process of performing a general automatic calibration parallel acquisition and reconstruction operation on the whole-brain 3D K-space data in step S2 above is as follows:
[0066] S21. Extract a 3D GRASE readout signal and a 3D navigation echo signal obtained after each execution of the imaging sequence from the whole brain 3D K-space data. Use the signals collected at the three navigation echo locations in the 3D GRASE readout signal to estimate the phase change between the odd and even rows of EPI. Then use GRAPPA technology to fill in the missing rows in k-space that were acquired by parallel accelerated imaging along the EPI direction.
[0067] S22. For each 3D GRASE readout signal after the missing rows are filled, the phase error between multiple excitations of the 3D GRASE readout signal is corrected using the corresponding 3D navigation echo signal. Then, the GRAPPA technique is used to fill in the missing rows of each 3D GRASE readout signal after phase error correction, where parallel accelerated imaging acquisition is performed along the TSE direction, to obtain a complete 3D GRASE readout signal.
[0068] In an embodiment of the present invention, the specific method for correcting the phase error of the 3D GRASE readout signal using the 3D navigation echo signal is as follows:
[0069] First, both the 3D GRASE readout signal and the 3D navigation echo signal are transformed from k-space to the xyz three-dimensional image space to obtain the first phase signal of the 3D navigation echo signal for each voxel (x,y,z) in the image space. The second phase signal ψ of the 3D GRASE readout signal for each voxel (x,y,z) in the image space. i,s (x,y,z);
[0070] Then, by subtracting the second phase signal from the first phase signal, the corrected phase ψ' of the 3D GRASE readout signal for each voxel (x,y,z) in the image space is obtained. i,s (x,y,z)=ψ i,s (x,y,z)-
[0071] Finally, the three-dimensional image of the phase-corrected 3D GRASE readout signal is transformed back to k-space to obtain the phase-corrected 3D GRASE readout signal.
[0072] S23. Reassemble all complete 3D GRASE readout signal segments into complete whole-brain 3D K-space data, and obtain the reconstructed image through 3D Fourier transform, i.e., whole-brain high-resolution diffusion imaging.
[0073] Therefore, by utilizing the aforementioned 3D DW-GRASE sequence based on a 3D navigator, phase errors between multiple acquisitions can be effectively eliminated, enabling high-resolution diffusion imaging of the whole brain.
[0074] In other embodiments, a computer-readable storage medium may be provided, on which a computer program is stored. When the computer program is executed by a processor, it is used to control an external device to implement the 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in steps S1 and S2 above.
[0075] It should be noted that in steps S1 and S2 above, step S1 requires segmented acquisition of whole-brain signals, thus relying on the cooperation of an MRI scanner. Step S2 mainly processes the whole-brain signals acquired in step S1, which relies on a computer program. Therefore, in the above computer program, one subroutine controls the MRI scanner to acquire whole-brain signals segmentally according to step S1, while another subroutine performs a general automatic calibration parallel acquisition and reconstruction operation on the whole-brain 3D K-space data according to step S2 to obtain a reconstructed image.
[0076] In other embodiments, a magnetic resonance imaging device may be provided, which includes a magnetic resonance scanner and a control unit. The control unit stores a computer program through the aforementioned computer-readable storage medium. When the computer program is executed, it is used to control the magnetic resonance scanner to implement the 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in steps S1 and S2 above.
[0077] The hardware structure of the magnetic resonance scanner and control unit described above can be implemented using a conventional magnetic resonance imaging system. The magnetic resonance scanner should include a magnet part and a magnetic resonance spectrometer part, and the control unit should contain computer programs to implement steps S1 and S2 above, as well as control programs necessary for implementing magnetic resonance imaging.
[0078] It should be noted that the aforementioned computer-readable storage medium may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. The aforementioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. Of course, the device should also have the necessary components for program execution, such as a power supply, communication bus, etc.
[0079] The following specific embodiment will demonstrate the detailed implementation and technical effects of the 3D gradient spin echo diffusion imaging method based on a 3D navigator described in S1 to S2 of the present invention.
[0080] Example
[0081] In this embodiment, the 3D gradient spin echo diffusion imaging method based on a 3D navigator includes the following steps:
[0082] Step 1: Continuously execute the 3D gradient spin echo diffusion imaging sequence (3D DW-GRASE) based on the 3D navigator to collect whole brain signals in segments.
[0083] Each execution of the 3D DW-GRASE sequence requires the sequential execution of the global saturation module, diffusion preparation module, fat saturation module, gradient spin echo module, and 3D navigation echo module. The specific process is as follows:
[0084] Step 1.1: Execute a global saturation module at the beginning of the imaging sequence, as shown in Figure 2. The processing flow within the module is as follows:
[0085] At the start of the timing sequence, three B1 RF pulses are applied. Simultaneously with each RF pulse, gradients are applied in the X, Y, and Z directions, with all three gradient amplitudes being exactly equal. After applying the three B1 RF pulses, the residual transverse magnetization from the previous signal acquisition can be considered completely eliminated. After a period known as the post-saturation delay (PSD), the longitudinal magnetization vector is recovered. Only after the PSD is reached can the next diffusion preparation module be executed.
[0086] Step 1.2: Immediately following the PSD of the global saturation module, the diffusion preparation module is executed. Refer to Figure 2 for the following processing flow within the module:
[0087] First, a non-layer-selective hard pulse excitation with a 90° flip angle is applied along the X-axis as the RF excitation pulse. Then, a diffusion gradient is applied along a pre-set diffusion direction (the diffusion direction can be selected according to actual needs; in this embodiment, gradients are applied along the X, Y, and Z directions). The diffusion gradient can be a trapezoidal cosine oscillation diffusion gradient of a certain frequency or a pulse diffusion gradient with a certain diffusion interval. The magnitude of the gradient depends on the pre-set value b. The diffusion time is denoted as t. d Next, an adiabatic tangent pulse with a 180° flip angle along the Y-axis is used as the RF convergence pulse to converge the divergent transverse magnetization vector; then, a diffusion gradient with the same amplitude, shape, duration, and direction as the previously applied gradient is applied; subsequently, a stabilizer gradient (G) is applied along the Z-axis. stb To mitigate signal modulation related to phase error, the gradient amplitude is set to the same value as the stabilizer gradient in the subsequent gradient spin echo module. Finally, a non-layer-selective hard pulse excitation with a 90° flip angle is emitted along the -X axis as an RF excitation pulse to convert the transverse magnetization vector into a longitudinal magnetization vector. The echo time in the diffusion preparation module is denoted as TE1.
[0088] Step 1.3: After the diffusion preparation module, the fat saturation module is executed to suppress fat signals. Refer to Figure 2 for the following processing flow within the module:
[0089] Three frequency-selective Gaussian pulses with 95° flip angles are applied to flip the cos(95°) part of the longitudinal magnetization vector of fat three times and wait for T1*ln(2) time before setting the fat signal to 0, so that most of the longitudinal magnetization vector of fat is converted into the transverse magnetization vector. Gradients in the X, Y and Z axes are applied at the same time as the Gaussian pulses to completely eliminate the transverse magnetization vector of fat; where T1 is the longitudinal relaxation time of fat.
[0090] Step 1.4: After the fat saturation module, execute the gradient spin echo module to acquire 3D K-space signals and obtain 3D K-space data. Refer to Figure 2 for the processing flow within the module:
[0091] 3D GRASE readout is achieved by performing echo plane imaging (EPI) encoding in the Y direction and fast spin echo (TSE) encoding in the Z direction; linear encoding is performed along the EPI direction and center encoding is performed along the TSE direction; parallel GRAPPA accelerated imaging is performed in both directions; a stabilizer gradient is applied along the Z-axis before and after each EPI readout time.
[0092] First, signals are acquired using three navigator echoes (i.e., 3-navigator echoes). Then, 3D-GRASE readout is achieved through echo plane imaging (EPI) encoding in the Y direction and turbine spin echo (TSE) encoding in the Z direction, resulting in a 3D GRASE readout signal (containing signals from the three navigator echoes). During this 3D GRASE readout process, linear encoding is performed along the EPI direction, and center encoding is performed along the TSE direction. The number of phase encodings in the EPI direction is denoted as N. EPI The number of phase codes in the TSE direction is denoted as N. TSE Furthermore, in this embodiment, during the aforementioned 3D GRASE readout process, 2×2 GRAPPA parallel accelerated imaging is performed in both the EPI and TSE directions, i.e., parallel acquisition of N along the EPI direction. EPI Perform interval acceleration factor value (N) a1 N EPI / N a1 Line number acquisition, along the TSE direction for N TSE Perform interval acceleration factor value (N) a2 N TSE / N a2 The method of acquiring the line count, and reading the EPI signal after the first refocusing pulse, is in k z The direction fills the center of the k-space; simultaneously, a stabilizer gradient is applied along the Z-axis before and after each EPI readout moment, and the magnitude of the stabilizer gradient applied here is the same as the magnitude of the stabilizer gradient applied in the aforementioned diffusion preparation module. The echo time in the gradient spin echo module is denoted as TE2, which is equal to the echo interval (ESP) for subsequent signal acquisition in TSE mode. Therefore, the total echo time of the acquired signal is TE1 + TE2.
[0093] Step 1.5: Embed a navigation echo module at the end of the sequence, which executes after the gradient spin echo module, to acquire 3D spin echo navigation signals. This is used to correct the phase error between multiple excitations in the 3D K-space data acquired in the gradient spin echo module. See Figure 3; the processing flow in the navigation echo module is as follows:
[0094] Four navigation echo segments centered in the Z-direction of k-space are acquired. Gradient coding is used to perform EPI encoding in the X and Y directions and TSE encoding in the Z direction, thus acquiring a 3D navigation echo signal. This navigation echo module uses the same echo interval (ESP) as the gradient spin echo module, and a stabilizer gradient is applied along the Z-axis before and after each EPI readout time. The number of phase codes and the acceleration factor value in the EPI direction are consistent with those in the gradient spin echo module, and are denoted as N. EPI and N a1 .
[0095] Each execution of the imaging sequence involves repeating steps 1.1 to 1.5, acquiring a corresponding 3D GRASE readout signal and a 3D navigation echo signal. Repeating steps 1.1 to 1.5 N times divides the whole-brain signal into N segments for sequential signal acquisition. The N segments of 3D GRASE readout signals constitute whole-brain 3D K-space data. Corresponding to this whole-brain 3D K-space data is the 3D navigator data composed of the N segments of 3D navigation echo signals, which can be used for subsequent phase correction. This acquisition method differs from conventional methods in that N... TSE While reducing the acquisition time by N times, this method is more sensitive to motion and requires correction for motion errors between segments. However, it significantly improves the signal-to-noise ratio of the image.
[0096] Step 2: After completing the acquisition of whole-brain 3D K-space data, execute the general automatic calibration parallel acquisition and reconstruction module to reconstruct the whole-brain 3D K-space data to obtain a complete image.
[0097] It should be noted that, since this invention uses N a1 >1 or N a2>1, therefore, a general automatic calibration section is needed for parallel acquisition and reconstruction to fill the k-space data of the unacquired rows. Furthermore, since segmented signals are used for acquisition, 3D navigation echo reconstruction is required to correct motion errors between each acquired segment. Therefore, in the general automatic calibration parallel acquisition and reconstruction module, the 3-navigator-echo technique is first used to estimate the phase change between odd and even EPI rows, and then phase correction is performed on all acquired image and navigation EPI odd and even row signals. The GRAPPA technique is used to fill the k-space data of the unacquired rows, reconstructing each segment of k-space image data and the corresponding navigator data into a three-dimensional image space.
[0098] In this embodiment, the processing flow in the general automatic calibration parallel acquisition and reconstruction module is as follows:
[0099] Step 2.1: Extract a 3D GRASE readout signal and a 3D navigation echo signal obtained after each execution of the imaging sequence from the whole brain 3D K-space data. Use the 3-navigator-echo technique, i.e., use the signals collected at the three navigator echo points in the 3D GRASE readout signal, to estimate the phase change between the odd and even rows of EPI. Then use the GRAPPA technique to fill in the missing rows in k-space caused by the parallel accelerated imaging of GRAPPA along the EPI direction. That is, fill in the missing rows in the 3D GRASE readout signal acquired by the parallel imaging along the EPI direction to obtain the 3D GRASE readout signal after the missing rows are filled.
[0100] Step 2.2: For each 3D GRASE readout signal after filling in the missing rows, the phase error between multiple excitations of the 3D GRASE readout signal is corrected using the corresponding 3D navigation echo signal. Then, the GRAPPA technique is used to fill in the missing rows in k-space caused by parallel accelerated imaging of GRAPPA along the TSE direction. That is, the missing rows acquired by parallel accelerated imaging along the TSE direction in each 3D GRASE readout signal after phase error correction are filled in completely to obtain a complete 3D GRASE readout signal.
[0101] In this embodiment, it is necessary to use 3D navigation echo signals to correct the phase error of the 3D GRASE readout signals acquired during the execution of the same imaging sequence. The specific process of phase error correction is as follows:
[0102] First, both the 3D GRASE readout signal and the 3D navigation echo signal are transformed from k-space to the xyz three-dimensional image space. Let... It is the phase signal of the voxel (x,y,z) in the image space of the s-th segment of the 3D navigation echo signal, ψi,s (x,y,z) is the phase signal of the voxel (x,y,z) of the s-th segment of the 3D GRASE readout signal in the image space. The corrected phase ψ' of the voxel (x,y,z) of the 3D GRASE readout signal in the image space is calculated according to formula (1). i,s (x,y,z):
[0103] In the image space, after calculating the corrected phase for each voxel, the phase correction of the three-dimensional image corresponding to the s-th segment of the 3DGRASE readout signal can be completed.
[0104] Then, the three-dimensional image of the phase-corrected s-th segment of the 3D GRASE readout signal is transformed back to k space to obtain the phase-corrected s-th segment of the 3D GRASE readout signal.
[0105] The correction process for the s-th segment of the 3D GRASE readout signal described above is similar for the remaining segments, i.e., s = 1, 2, ..., N. The above operation is repeated for each segment, and then blank rows are filled in the TSE direction to obtain N complete 3D GRASE readout signals.
[0106] Step 2.3: Reassemble all N complete 3D GRASE readout signals into complete whole-brain 3D K-space data, and obtain the reconstructed high-resolution diffusion imaging of the whole brain through 3D Fourier transform.
[0107] Compared to conventional phase correction methods, 3D navigation echoes are more sensitive to motion, greatly reducing the problem of image motion artifacts. Compared to conventional acquisition methods, the GRAPPA parallel acquisition method significantly improves acquisition efficiency, not only shortening acquisition time but also reducing TE (transmission distortion), thus enabling the acquisition of images with higher signal-to-noise ratios.
[0108] In this embodiment, since the diffusion preparation module of the above-mentioned 3D gradient spin echo diffusion imaging sequence (3D DW-GRASE) based on a 3D navigator has two types of diffusion gradients: trapezoidal cosine oscillation diffusion gradient and pulse diffusion gradient, the 3D DW-GRASE imaging sequence with a trapezoidal cosine oscillation diffusion gradient in the diffusion preparation module is denoted as 3D OG-GRASE, and the 3D DW-GRASE imaging sequence with a pulse diffusion gradient in the diffusion preparation module is denoted as 3D PG-GRASE.
[0109] This embodiment was tested on seven healthy young male volunteers. The specific parameters are described below: Magnetic resonance scanning was performed using a Siemens Prisma 3T scanner (maximum gradient 80 mT / m, maximum switching rate 200 mT / m), and all scans used a 64-channel head coil. Two sets of experiments were conducted in this embodiment, with the commonly used 2D EPI sequence in the prior art used as a comparison.
[0110] Experiment 1: To compare the signal-to-noise ratio (SNR) of 3D PG-GRASE and 2D EPI sequences based on a 3D navigator within the same acquisition time. Using b = 1000 s / mm... 2 30 directions, FOV = 220×200×144mm 3 mm, resolution 1.5×1.5×1.5mm 3 Oversampling in the plane direction = 20%, partial Fourier factor = 76%, N a1 =2 Perform a scan, using the following scheme:
[0111] (1) GRASE readout after 18 excitations, N EPI =113,N TSE =4, navigation echo Z direction N TSE =4. TE1 / TE2 / TR = 38.68 / 43.92 / 2500ms, only one acquisition is performed, scan time = 24min.
[0112] (2) 2D EPI readout of multi-slice scanning, TR / TE = 11250 / 83ms and N EPI =110, perform 4 repeated acquisitions, scan time =24min.
[0113] Experiment 2: Assessing the time-dependent diffusion of gray matter in the cerebral cortex.
[0114] The diffusion time dependence of the apparent diffusion coefficient (ADC) in cortical gray matter regions was evaluated using 3D OG-GRASE and 3D PG-GRASE sequences based on a 3D navigator. PG encoding (δ / Δ = 15 / 25 ms, 0 Hz, Δ) was employed. eff =20ms; δ / △=15 / 35ms,0Hz,Δ eff =30ms; δ / △=15 / 45ms,0Hz,Δ eff =40ms) and OG encoding (25Hz, 2 cycles, Δ eff ≈10ms; 50Hz, 4 cycles, Δ eff Signals were acquired at a frequency of approximately 5ms. All other parameters remained the same for all diffusion times: TR / TE1 / TE2 = 2600 / 84 / 46ms, N EPI=67, 12 segments were acquired along the TSE encoding direction. Average number of acquisitions = 2, b = 500 s / mm 2 Six directions, b=0, averaged four times, N a1 =2, N a2 =2 Perform the scan. The scan time for each diffusion time is 5 minutes. The remaining scan parameters are the same as the 3D PG-GRASE sequence based on the 3D navigator in Experiment 1.
[0115] Figure 4 shows the b0 image (3Dnavigator_b0) and b1000 image (3D navigator_b1000) obtained using 3D PG-GRASE, as well as the b0 image (non navigator_b0) and b1000 image (non navigator_b1000) obtained without navigation echo correction. In contrast, the motion artifacts in the b1000 image after 3D navigation echo correction are corrected; for example, residual artifacts along the z-direction (white arrows in the figure) are better removed.
[0116] Figure 5 shows the signal-to-noise ratio (SNR) of three regions (white matter (WM), basal ganglia (BG), and thalamus (Thal)) measured using 3D PG-GRASE and 2D EPI sequences at the same acquisition time. The results indicate that the SNR of the 3D PG-GRASE sequence is significantly higher than that of the 2D EPI sequence in WM, BG, and Thal in both b0 and b1000 images.
[0117] It should be noted that the above-described embodiments are merely preferred embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained through equivalent substitution or transformation fall within the protection scope of the present invention.
Claims
1. A 3D gradient spin echo diffusion imaging method based on a 3D navigator, characterized in that, include: S1: Whole-brain signals are acquired in segments by repeatedly executing a 3D gradient spin echo diffusion imaging sequence based on a 3D navigator. After each execution of the imaging sequence, a 3D GRASE readout signal and a 3D navigation echo signal are obtained. All 3D GRASE readout signal segments constitute whole-brain 3D K-space data. The 3D gradient spin echo diffusion imaging sequence based on a 3D navigator consists of a global saturation module, a diffusion preparation module, a fat saturation module, a gradient spin echo module, and a 3D navigation echo module in sequence. The global saturation module is used to destroy residual lateral magnetization by applying a gradient across the entire 3D imaging space. The diffusion preparation module is used to embed a pair of diffusion gradients into a 90° angle. x -180° y -90° -x In radio frequency pulses, the diffusion coding and signal acquisition are separated; the diffusion gradient is a trapezoidal cosine oscillation diffusion gradient or a pulse diffusion gradient. The fat saturation module is used to suppress fat signals; The gradient spin echo module is used to acquire signals in 3D K-space to obtain a 3D GRASE readout signal. The navigation echo module is used to acquire signals from the 3D navigator to obtain a 3D navigation echo signal. S2. Perform a general automatic calibration parallel acquisition and reconstruction operation on the whole brain 3D K-space data. Each 3D GRASE readout signal in the whole brain 3D K-space data is corrected for phase error using the 3D navigation echo signal acquired during the same imaging sequence. The gaps in each 3D GRASE readout signal after phase error correction are filled. Finally, a 3D Fourier transform is performed on all the filled 3D GRASE readout signals to obtain the reconstructed image.
2. The 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in claim 1, characterized in that, The process executed in the global saturation module of the imaging sequence is as follows: At the beginning of the imaging sequence, three B1 radio frequency pulses are applied, and at the same time as each B1 radio frequency pulse is applied, gradients with completely equal amplitudes are applied in the X, Y, and Z directions of the 3D imaging space to eliminate lateral magnetization. After the last B1 radio frequency pulse is executed, the system waits for the longitudinal magnetization vector to recover according to the preset post-saturation delay time (PSD) before executing the next diffusion preparation module.
3. The 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in claim 1, characterized in that, The process executed in the diffusion preparation module of the imaging sequence is as follows: First, an RF excitation pulse with a 90° flip angle is applied along the X-axis; then, a diffusion gradient, which is a trapezoidal cosine oscillation diffusion gradient or a pulse diffusion gradient, is applied along a preset diffusion direction; next, an RF refocusing pulse with a 180° flip angle is applied along the Y-axis, and then the exact same diffusion gradient is applied again; finally, a stabilizer gradient (G) is applied along the Z-axis. stb Finally, a radio frequency excitation pulse with a 90° flip angle along the -X axis is used to convert the transverse magnetization vector into a longitudinal magnetization vector.
4. The 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in claim 1, characterized in that, The execution flow in the fat saturation module of the imaging sequence is as follows: Three frequency-selective Gaussian pulses with 95° flip angles are applied to flip the cos(95°) part of the longitudinal magnetization vector of fat three times and set it to 0 after waiting for T1*ln(2) time, so that most of the longitudinal magnetization vector of fat is converted into the transverse magnetization vector. Gradients are applied synchronously in the X, Y and Z axis directions while applying the Gaussian pulses to completely eliminate the transverse magnetization vector of fat; where T1 is the longitudinal relaxation time of fat.
5. The 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in claim 1, characterized in that, The execution flow of the gradient spin echo sequence module in the imaging sequence is as follows: First, signals are acquired using three navigation echoes. Then, 3D GRASE readout is achieved by performing echo plane imaging (EPI) encoding in the Y direction and fast spin echo (TSE) encoding in the Z direction, resulting in a 3D GRASE readout signal. During the 3D GRASE readout process, linear encoding is performed along the EPI direction, and center encoding is performed along the TSE direction. Parallel GRAPPA accelerated imaging is performed in both the EPI and TSE directions. Simultaneously, a stabilizer gradient needs to be applied along the Z-axis before and after each EPI readout moment, and the magnitude of the stabilizer gradient is the same as the magnitude of the stabilizer gradient applied in the diffusion preparation module.
6. The 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in claim 1, characterized in that, The processing flow in the navigation echo module of the imaging sequence is as follows: N segments of spin echo signals centered in the Z direction of k-space are acquired, and EPI encoding is performed along the X and Y directions, while TSE encoding is performed in the Z direction to obtain a 3D navigation echo signal. A stabilizer gradient needs to be applied along the Z-axis before and after each EPI readout moment, and the gradient magnitude of the stabilizer gradient is the same as the magnitude of the stabilizer gradient applied in the diffusion preparation module. N is an even number not less than 4.
7. The 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in claim 1, characterized in that, In step S2, the specific process for performing a general automatic calibration parallel acquisition and reconstruction operation on the whole-brain 3D K-space data is as follows: S21. Extract a 3D GRASE readout signal and a 3D navigation echo signal obtained after each execution of the imaging sequence from the whole brain 3D K-space data. Use the signals collected at the three navigator echoes in the 3D GRASE readout signal to estimate the phase change between the odd and even rows of EPI. Then use GRAPPA technology to fill in the missing rows of parallel accelerated imaging acquisition along the EPI direction in k-space. S22. For each 3D GRASE readout signal after the missing rows are filled, the phase error between multiple excitations of the 3D GRASE readout signal is corrected using the corresponding 3D navigation echo signal. Then, the GRAPPA technique is used to fill in the missing rows of each 3D GRASE readout signal after the phase error is corrected by parallel accelerated imaging acquisition along the TSE direction, thus obtaining a complete 3D GRASE readout signal. S23. Reassemble all complete 3D GRASE readout signal segments into complete whole-brain 3D K-space data, and obtain the reconstructed image through 3D Fourier transform.
8. The 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in claim 7, characterized in that, In step S22, the specific method for correcting the phase error of the 3D GRASE readout signal using the 3D navigation echo signal is as follows: First, both the 3D GRASE readout signal and the 3D navigation echo signal are transformed from k-space to the xyz three-dimensional image space to obtain the first phase signal of the 3D navigation echo signal for each voxel (x,y,z) in the image space. The second phase signal ψ of the 3D GRASE readout signal for each voxel (x,y,z) in the image space. i,s (x,y,z); Then, by subtracting the second phase signal from the first phase signal, the corrected phase ψ' of the 3D GRASE readout signal for each voxel (x,y,z) in the image space is obtained. i,s (x,y,z)=ψ i,s (x,y,z)- Finally, the three-dimensional image of the phase-corrected 3D GRASE readout signal is transformed back to k-space to obtain the phase-corrected 3D GRASE readout signal.
9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, controls an external device to implement the 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in any one of claims 1 to 8.
10. A magnetic resonance imaging device, characterized in that, It includes a magnetic resonance scanner and a control unit, wherein the control unit stores a computer program, which, when executed, controls the magnetic resonance scanner to implement the 3D gradient spin echo diffusion imaging method based on a 3D navigator as described in any one of claims 1 to 8.