Diffusion weighted magnetic resonance spectroscopy parallel acquisition and reconstruction method

Abstract

The application discloses a diffusion-weighted magnetic resonance spectrum parallel acquisition and reconstruction method. The method comprises the following steps: adding a pair of diffusion gradients in a single voxel spectrum sequence to obtain a single voxel diffusion-weighted magnetic resonance spectrum sequence; performing point-by-point summation on single-band radio frequency pulses at each voxel position of interest to obtain multi-band radio frequency pulses; obtaining a multi-voxel diffusion-weighted magnetic resonance spectrum signal through the single voxel diffusion-weighted magnetic resonance spectrum sequence and the multi-band radio frequency pulses; obtaining a sensitivity map of each channel of a receiving coil; separating and reconstructing the multi-voxel diffusion-weighted magnetic resonance spectrum signal according to the sensitivity map; and finally obtaining a separated and reconstructed multi-voxel diffusion-weighted magnetic resonance spectrum signal. The application can simultaneously measure diffusion-weighted magnetic resonance spectrum signals of multiple voxels of a human body, and obtain a signal-to-noise ratio and an apparent diffusion coefficient equivalent to those of a single voxel diffusion-weighted magnetic resonance spectrum signal, thereby saving at least half of the scanning time.

Description

Technical Field

[0001] This invention relates to a diffusion-weighted magnetic resonance spectroscopy acquisition and reconstruction method, specifically a diffusion-weighted magnetic resonance spectroscopy parallel acquisition and reconstruction method. Background Technology

[0002] Diffusion-weighted magnetic resonance imaging (DWI) has achieved great success in probing cellular microstructures. However, the widespread presence of water molecules in biological tissues results in numerous diffusion-weighted signal sources, making it difficult to distinguish differences in the diffusion motion of water molecules in different cellular microenvironments. Metabolite molecules, on the other hand, are generally located within confined cellular or tissue spaces. Diffusion-weighted measurements of these molecules can provide richer and more extensive information that DWI cannot detect. DWI has proven its immense potential in exploring various tissue microstructures and physiological information at the cellular and subcellular levels. However, DWI requires the acquisition of multiple diffusion-weighted signals to obtain the apparent diffusion coefficients of metabolite molecules, leading to long scan times, which is currently the main limitation of DWI applications. Several methods for accelerating DWI acquisition have been proposed, such as DWI imaging technology, which can acquire metabolite information from multiple voxels across the entire brain region in a single scan, but its scan time is often very long. Furthermore, due to limited K-space sampling and instability factors such as subject movement, magnetic resonance imaging technology is also subject to negative effects related to the point spread function.

[0003] In certain applications, such as the human brain, the research subject can be limited to several brain regions related to disease or brain activity. This reduces the area scanned and allows limited time to be used to improve scanning accuracy. However, acquiring signals from multiple brain regions in multiple lesions often requires sequential acquisition of single-voxel spectra, which is very time-consuming when the number of voxels is large. Therefore, it is essential to achieve simultaneous acquisition and reconstruction of multi-voxel magnetic resonance spectroscopy signals using non-magnetic resonance spectroscopy imaging techniques. With the use of higher main magnetic field strength and phased array receiving coils with multiple channels, it has been proven that spatial location information can be encoded using the spatial variation of the receiving coil sensitivity profile. This invention utilizes multi-band radio frequency pulses to achieve simultaneous acquisition of multiple voxels in the human body. Subsequently, the diffusion-weighted signals of these multiple voxels are separated and reconstructed using the receiving coil sensitivity information, and the signal-to-noise ratio and apparent diffusion coefficient are quantitatively compared with the corresponding single-voxel diffusion-weighted magnetic resonance spectra.

[0004] Existing diffusion-weighted magnetic resonance spectroscopy techniques are still based on sequential measurement of individual voxels, or on global measurement based on magnetic resonance spectroscopy imaging techniques, rather than simultaneous measurement of multiple specific voxels in the region of interest. Summary of the Invention

[0005] To address the problems existing in the background art, the present invention aims to provide a method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectroscopy. This invention achieves simultaneous acquisition of multiple voxels of the human body using multi-band radio frequency pulses. Subsequently, the diffusion-weighted signals of these multiple voxels are separated and reconstructed using the sensitivity information of the receiving coil. This method can achieve a signal-to-noise ratio and apparent diffusion coefficient comparable to single-voxel diffusion-weighted magnetic resonance spectroscopy while saving at least half of the spectral scanning time.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows, including the following steps:

[0007] Step S1: Add a pair of diffusion gradients to the basic monomer spectral sequence to obtain the monomer diffusion-weighted magnetic resonance spectral sequence.

[0008] The diffusion gradient described herein is a pair of completely symmetrical but oppositely polar diffusion gradients;

[0009] Step S2: Sum the single-band radio frequency pulses at each location of the voxel of interest point by point to obtain multi-band radio frequency pulses. Replace the single-band radio frequency pulses in the single voxel diffusion-weighted magnetic resonance spectrum sequence with multi-band radio frequency pulses to obtain the multi-voxel diffusion-weighted magnetic resonance spectrum sequence, and then obtain the multi-voxel diffusion-weighted magnetic resonance spectrum signal.

[0010] Step S3: Obtain the sensitivity map of each channel of the receiving coil, and perform separation and reconstruction of the multi-voxel diffusion-weighted magnetic resonance spectrum signal based on the sensitivity map, and finally obtain the separated and reconstructed multi-voxel diffusion-weighted magnetic resonance spectrum signal.

[0011] The specific steps of step S2 are as follows:

[0012] S2.1 Determine the number of voxels to be collected as N, and select the initial positions of the N voxels;

[0013] S2.2 Summing the single-band radio frequency pulses corresponding to the N voxel positions point by point to obtain multi-band radio frequency pulses;

[0014] S2.3 Compare the excitation profiles of single-band and multi-band radio frequency pulses:

[0015] If the error between the two is within the preset error range, the waveform of the multi-band radio frequency pulse is considered to be correct, and the process continues to step S2.4.

[0016] Otherwise, the waveform of the multi-band radio frequency pulse is considered incorrect, and the process returns to step S2.2 to acquire a new multi-band radio frequency pulse.

[0017] S2.4 Change the position of each voxel and repeat steps S2.2 to S2.3 to obtain multiple multi-band radio frequency pulses with correct waveforms;

[0018] S2.5 Add all the multi-band radio frequency pulses obtained in step S2.4 to the pulse file of the voxel diffusion-weighted magnetic resonance spectrum sequence, and select one of the multi-band radio frequency pulses from the pulse file to replace the initial single-band radio frequency pulse to obtain the multi-voxel diffusion-weighted magnetic resonance spectrum sequence.

[0019] S2.6. Import the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence into the magnetic resonance scanner, set the number and size of b-values ​​and the number of acquisitions, and then use the magnetic resonance scanner to acquire the multi-voxel diffusion-weighted magnetic resonance spectroscopy signal.

[0020] Specifically, the number of acquisitions for each b value refers to the number of acquisitions of the multi-voxel diffusion-weighted magnetic resonance spectral signal under the same b value; the number of acquisitions is 2 to 16 under the condition of water suppression off; and the number of acquisitions for each b value is 16 to 64 under the condition of water suppression on.

[0021] The specific steps of step S3 are as follows:

[0022] S3.1. Obtain the grayscale image of each channel of the receiving coil using the gradient echo sequence, and then obtain the sensitivity map corresponding to each channel using the grayscale image of each channel.

[0023] S3.2. Perform voxel segmentation and registration on the sensitivity map of each channel to obtain the sensitivity map corresponding to each voxel position. Average these sensitivity maps in the spatial dimension to obtain a sensitivity matrix S with dimension R*N, where R is the total number of channels of the receiving coil.

[0024] S3.3. Take Q pixels at the same edge positions in the grayscale images of each channel, and obtain the noise vector η for each channel image based on the grayscale value of each pixel. r :

[0025] η r =[h1 h2…h q …h Q ]

[0026] Where, η r h is the noise vector corresponding to the r-th channel. q Let be the grayscale value of the q-th pixel in the r-th channel;

[0027] In this invention, the aforementioned pixels are all located at the outermost edge of the channel image, in the corner area outside the human brain image;

[0028] S3.4 Next, the noise vector is processed according to the following formula to obtain the receiving coil noise covariance matrix Ψ of dimension R*R:

[0029]

[0030] Where, ψ i,j Let η be the value of the receiving coil noise covariance matrix Ψ in the i-th row and j-th column. i Let η be the noise vector corresponding to the i-th channel. j This is the noise vector corresponding to the j-th channel;

[0031] S3.5. Obtain the expanded matrix U according to the following formula:

[0032] U=(S H Ψ -1 S) -1 S h Ψ -1

[0033] Wherein, the superscript H indicates the conjugate transpose of the sensitivity matrix S;

[0034] S3.6 Multiply the expanded matrix U with the multi-voxel diffusion-weighted magnetic resonance spectral signal of dimension R*P*A to obtain the multi-voxel diffusion-weighted magnetic resonance spectral signal of dimension N*P*A, where P is the number of sampling points of the multi-voxel diffusion-weighted magnetic resonance spectral signal and A is the number of sampling times of the multi-voxel diffusion-weighted magnetic resonance spectral signal.

[0035] S3.7 Perform frequency and phase correction on all multi-voxel diffusion-weighted magnetic resonance spectral signals corresponding to the same b value;

[0036] S3.8 Arrange all multi-voxel diffusion-weighted magnetic resonance (MDR) spectral signals corresponding to the same b value according to the given peak height of the magnetic resonance spectrum. Select three MDR spectral signals with peak heights of the second, third, and fourth from all the MDR spectral signals. Then, take 60% to 80% of the average peak height of these three MDR spectral signals as a threshold and discard MDR spectral signals below this threshold. Then, average the remaining MDR spectral signals and perform frequency and phase correction again to obtain the reconstructed MDR spectral signal at the given b value.

[0037] The diffusion time of the diffusion gradient is 10–300 ms, and the duration of the diffusion gradient is 2–50 ms.

[0038] In step S2.6, after importing the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence, the number of b-values ​​is set to 2 to 10, and the value of the b-value is set to 0 to 3333 s / mm.2 Under the condition of water suppression off, 2 to 16 multi-voxel diffusion-weighted magnetic resonance spectroscopy signals were collected for each b value; under the condition of water suppression on, 16 to 64 multi-voxel diffusion-weighted magnetic resonance spectroscopy signals were collected for each b value.

[0039] The value of b is obtained according to the following formula:

[0040]

[0041] Where γ is the gyromagnetic ratio of the proton, g is the amplitude of the diffusion gradient, δ is the duration of the diffusion gradient, and Δ is the diffusion time.

[0042] In step S2.2, the multi-band radio frequency pulse is obtained according to the following formula:

[0043]

[0044] Among them, RF N The time-domain waveform of a multi-band radio frequency pulse, RF s f is the time-domain waveform of the initial single-band radio frequency pulse in the sequence. n Let t be the offset frequency at the position of the nth voxel, where n is the ordinal number of the voxel and t is the time.

[0045] Step S2.3 specifically includes:

[0046] S2.3.1. Using the Bloch equation, simulations were performed on single-band and multi-band RF pulses respectively to obtain the simulated excitation profiles of single-band and multi-band RF pulses, and the excitation performance of multi-band RF pulses was preliminarily verified.

[0047] The simulated excitation profiles of single-band and multi-band RF pulses are compared:

[0048] If the error between the simulated excitation profile of a single-band RF pulse and the simulated excitation profile of a multi-band RF pulse is less than the preset simulation profile error value, then the waveform of the multi-band RF pulse is considered to be initially correct, and the process proceeds to step S2.3.2.

[0049] Otherwise, the waveform of the multi-band radio frequency pulse is considered incorrect, and the process returns to step S2.2 to reacquire the multi-band radio frequency pulse.

[0050] S2.3.2. Using a magnetic resonance scanner, actual tests were performed on single-band and multi-band radio frequency pulses to obtain the actual excitation profiles of single-band and multi-band radio frequency pulses, further verifying the excitation performance of multi-band radio frequency pulses:

[0051] Comparison of the actual excitation profiles of single-band and multi-band radio frequency pulses:

[0052] If the error between the actual excitation profile of a single-band RF pulse and the actual excitation profile of a multi-band RF pulse is less than the preset actual profile error value, then the waveform of the multi-band RF pulse is considered correct, and the process proceeds to step S2.4.

[0053] Otherwise, the waveform of the multi-band radio frequency pulse is considered incorrect, and the process returns to step S2.2 to reacquire the multi-band radio frequency pulse.

[0054] In step S3.1, the sensitivity map of each channel of the receiving coil is obtained according to the following formula:

[0055]

[0056] Among them, S r (x,y) represents the sensitivity map corresponding to the r-th channel, I r (x,y) represents the grayscale image received by the r-th channel, where R represents the total number of channels of the receiving coil.

[0057] In step S3.8, if the multi-voxel diffusion-weighted magnetic resonance spectroscopy signal is acquired under water suppression enabled, the given magnetic resonance spectral peak is N-acetylaspartic acid; if the multi-voxel diffusion-weighted magnetic resonance spectroscopy signal is acquired under water suppression disabled, the given magnetic resonance spectral peak is the water peak.

[0058] An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the steps in the above-described method.

[0059] This invention provides a method capable of simultaneously obtaining diffusion-weighted magnetic resonance spectra of multiple voxels in the human body and using the sensitivity information of the receiving coil to separate and reconstruct signals from different voxels. It also obtains the relative concentrations of metabolite molecules of interest within the human body and calculates the apparent diffusion coefficients of the corresponding metabolite molecules based on the attenuation of their relative concentrations. This invention can simultaneously measure the diffusion-weighted magnetic resonance spectra of multiple voxels in the human body and obtain signal-to-noise ratios and apparent diffusion coefficients comparable to those of single-voxel diffusion-weighted magnetic resonance spectra, saving at least half the scanning time.

[0060] This invention is a method for simultaneously acquiring multiple voxels of the human body using multi-band radio frequency pulses, and then using the sensitivity information of the receiving coil to separate and reconstruct the diffusion-weighted signal of these multiple voxels. This method achieves a signal-to-noise ratio and apparent diffusion coefficient comparable to that of single-voxel diffusion-weighted magnetic resonance spectroscopy while saving at least half of the scanning time.

[0061] The innovation of this invention compared with the prior art lies in:

[0062] This invention achieves for the first time parallel acquisition and separation of multi-voxel diffusion-weighted magnetic resonance spectroscopy with relatively small signal leakage. The obtained signal-to-noise ratio and apparent diffusion coefficient are comparable to those of single-voxel diffusion-weighted magnetic resonance spectroscopy, and at least half of the scanning time can be saved. Attached Figure Description

[0063] Figure 1 This is an overall flowchart of the present invention;

[0064] Figure 2 A simplified diagram of a multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence;

[0065] Figure 3 A schematic diagram of the voxel locations of the subjects;

[0066] Figure 4 A schematic diagram of the parameter tabs for the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence added to this invention;

[0067] Figure 5 This is a schematic diagram of the metabolite peak spectrum of the centrum semioval region of the brain of a normal subject reconstructed according to the present invention; wherein, (a) is a schematic diagram of the spectrum obtained by exciting only voxel 1 at voxel 1, (b) is a schematic diagram of the spectrum obtained by exciting only voxel 1 at voxel 2, (c) is a schematic diagram of the spectrum obtained by exciting only voxel 2 at voxel 1, (d) is a schematic diagram of the spectrum obtained by exciting only voxel 2 at voxel 2, (e) is a schematic diagram of the spectrum obtained by exciting both voxels at voxel 1 simultaneously, and (f) is a schematic diagram of the spectrum obtained by exciting both voxels at voxel 2 simultaneously. Detailed Implementation

[0068] The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention. These all fall within the scope of protection of the present invention.

[0069] The embodiments of the present invention take the human brain as an example. Specific embodiments are as follows, such as... Figure 1 As shown:

[0070] Step S1: Recruit 4 normal subjects (aged 23.5 ± 2.6 years).

[0071] By adding a pair of diffusion gradients to the basic monomer spectral sequence, a monomer diffusion-weighted magnetic resonance spectral sequence is obtained; wherein, the diffusion gradients adopt a pair of completely symmetrical but opposite polarity diffusion gradients.

[0072] like Figure 3 As shown, add the corresponding diffusion parameter tab;

[0073] Step S2: Sum the single-band RF pulses at each location of the voxel of interest point by point to obtain multi-band RF pulses, enabling simultaneous acquisition of multiple voxels. Replace the single-band RF pulses in the single voxel diffusion-weighted magnetic resonance spectrum sequence with multi-band RF pulses to obtain a multi-voxel diffusion-weighted magnetic resonance spectrum sequence, and then obtain the multi-voxel diffusion-weighted magnetic resonance spectrum signal; and add a multi-band RF pulse tab.

[0074] The specific steps of step S2 are as follows:

[0075] S2.1 Determine the number of voxels to be collected as N, and select the initial positions of the N voxels;

[0076] In this embodiment, the central region of the semioval fossa, where white matter is abundant, was selected. The two voxels were placed 29.9 mm to the left and 29.9 mm to the right of the center, respectively. The specific positions are as follows: Figure 3 As shown. Figure 3 The black box represents voxel 1, the white box represents voxel 2, and the white dashed box represents the shim position.

[0077] S2.2 Summing up the single-band radio frequency pulses corresponding to the N voxel positions point by point (i.e., Formula 1) to obtain multi-band radio frequency pulses;

[0078] Multi-band radio frequency pulses are obtained according to the following formula:

[0079]

[0080] Among them, RF N The waveform of the multi-band radio frequency pulse is shown in the time domain, where N is the number of voxels to be excited simultaneously. s f is the time-domain waveform of the initial single-band radio frequency pulse in the sequence. n Let be the offset frequency at the position of the nth voxel, where n is the ordinal number of the voxel, t is time, j is the imaginary unit, and e is a natural number;

[0081] S2.3 Compare the excitation profiles of single-band and multi-band radio frequency pulses:

[0082] If the error between the two is within the preset error range, it indicates that the excitation performance of the multi-band radio frequency pulse is good, and the waveform of the multi-band radio frequency pulse is considered to be correct. Proceed to step S2.4.

[0083] Otherwise, it indicates that the excitation performance of the multi-band RF pulse is poor, and the waveform of the multi-band RF pulse is considered incorrect. Return to step S2.2 to obtain a new multi-band RF pulse.

[0084] The specific operation of step S2.3 is as follows:

[0085] S2.3.1. Using the Bloch equation, simulations were performed on single-band and multi-band RF pulses respectively to obtain the simulated excitation profiles of single-band and multi-band RF pulses, and the excitation performance of multi-band RF pulses was preliminarily verified.

[0086] The simulated excitation profiles of single-band and multi-band RF pulses are compared:

[0087] If the error between the simulated excitation profile of a single-band RF pulse and the simulated excitation profile of a multi-band RF pulse is less than the preset simulation profile error value, it indicates that the excitation performance of the multi-band RF pulse is good, and the waveform of the multi-band RF pulse is considered to be initially correct. Proceed to step S2.3.2 for further verification.

[0088] Otherwise, it indicates that the excitation performance of the multi-band RF pulse is poor, and the waveform of the multi-band RF pulse is considered to be incorrect. Return to step S2.2 to reacquire the multi-band RF pulse.

[0089] S2.3.2. Using a magnetic resonance scanner, actual tests were performed on single-band and multi-band radio frequency pulses to obtain the actual excitation profiles of single-band and multi-band radio frequency pulses, further verifying the excitation performance of multi-band radio frequency pulses:

[0090] Comparison of the actual excitation profiles of single-band and multi-band radio frequency pulses:

[0091] If the error between the actual excitation profile of a single-band RF pulse and the actual excitation profile of a multi-band RF pulse is less than the preset actual profile error value, it indicates that the excitation performance of the multi-band RF pulse is good, and the waveform of the multi-band RF pulse is considered to be correct, and proceed to step S2.4.

[0092] Otherwise, it indicates that the excitation performance of the multi-band RF pulse is poor, and the waveform of the multi-band RF pulse is considered incorrect. Return to step S2.2 to reacquire the multi-band RF pulse.

[0093] S2.4 Change the position of each voxel and repeat steps S2.2 to S2.3 to obtain multiple multi-band radio frequency pulses with correct waveforms;

[0094] S2.5 Add all the multi-band radio frequency pulses obtained in step S2.4 to the pulse file of the single-voxel diffusion-weighted magnetic resonance spectrum sequence for selection. Select one of the multi-band radio frequency pulses from the pulse file and replace the single-band radio frequency pulses in the single-voxel diffusion-weighted magnetic resonance spectrum sequence with the selected multi-band radio frequency pulse to obtain the multi-voxel diffusion-weighted magnetic resonance spectrum sequence.

[0095] S2.6. Import the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence into the magnetic resonance scanner, set the number and size of b-values ​​and the number of acquisitions, and then use the magnetic resonance scanner to acquire the multi-voxel diffusion-weighted magnetic resonance spectroscopy signal.

[0096] In step S2.6, multi-voxel diffusion-weighted magnetic resonance (MDR) spectral signals are acquired using a MDR sequence. The steps include:

[0097] S2.6.1. Place the normal subject in the MRI scanner with the head first, in a supine position.

[0098] S2.6.2. Preparation of rapid gradient echo sequence for scanning positioning images and 3D T1-weighted magnetization (imaging field of view: 187*240*247mm) 3 The repeat time is 2300ms, the echo time is 3ms, the flip angle is 9°, and the resolution is 0.96mm*0.96mm*1.06mm) is used to locate the position of the spectral voxels.

[0099] S2.6.3. Set the repetition time of the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence to 1500 ms, the echo time to 64 ms, the mixing time to 45.6 ms, and the voxel size to 20*28*26 mm. 3 The bandwidth is 2000 Hz, the diffusion time is 80 ms, the diffusion gradient duration is 20 ms, and it is applied in the range of 0–3333 s / mm. 2 Five b-values ​​with uniform growth within the range. With water suppression off, 16 spectra were acquired for each b-value condition; with water suppression on, 32 spectra were acquired for each b-value condition. A large shimming frame was selected to cover the locations of two simultaneously excited voxels for shimming.

[0100] Figure 4 The present invention adds a parameter tab for the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence. Among them, "diffusion time" can adjust the diffusion time, "duration time" can adjust the duration of the diffusion gradient, "b_number" can adjust the number of b values ​​acquired in the interleaved sampling, "g" can adjust the amplitude of the diffusion gradient corresponding to different b values, and "choose RF" can select a single-band radio frequency pulse or a multi-band radio frequency pulse with different separation distances.

[0101] By periodically changing the magnitude of the diffusion gradient, the value of b is changed, thereby enabling the cyclic interleaving acquisition of multi-voxel diffusion-weighted magnetic resonance spectral signals corresponding to different b values.

[0102] Specifically, the value of b is obtained according to the following formula:

[0103]

[0104] Taking the stimulated echo acquisition mode sequence as an example, Figure 2 G1 and G2 are the added pair of diffusion gradients, γ is the gyromagnetic ratio of the proton, g is the amplitude of the diffusion gradient, δ is the duration of the diffusion gradient, Δ is the diffusion time, and ε is the rising (or falling) time of the diffusion gradient.

[0105] S2.6.4. Acquire the same monomer diffusion-weighted magnetic resonance spectroscopy signal at the same location using the same experimental parameters as in S2.6.3.

[0106] S3. Separating and reconstructing the human brain spectral signal acquired by multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence using the sensitivity information of the receiving coil, the steps include:

[0107] S3.1. Use the small-angle gradient echo sequence to obtain the low-resolution grayscale image of each channel of the receiving coil, and then use the grayscale image of each channel to obtain the sensitivity map corresponding to each channel.

[0108] The grayscale image acquisition method in step S3.1 is as follows: after all diffusion-weighted magnetic resonance spectroscopy signals have been acquired, the gradient echo sequence is scanned (imaging field of view is 224*224*224mm). 3 The amplitude and phase diagrams of each channel of the receiving coil are obtained with a repetition time of 10ms, an echo time of 2.83ms, a flip angle of 25°, and a resolution of 3.5mm*3.5mm*3.5mm. The amplitude and phase diagrams of each channel are multiplied to obtain the final grayscale image.

[0109] The sensitivity map of each channel of the receiving coil is obtained by processing it according to the following formula:

[0110]

[0111] Among them, S r (x,y) represents the sensitivity map corresponding to the r-th channel, I r (x,y) represents the low-resolution grayscale image received by the r-th channel, where R represents the total number of channels of the receiving coil;

[0112] S3.2. Voxel segmentation and registration are performed on the sensitivity maps of each channel to obtain the sensitivity map corresponding to each voxel position. These sensitivity maps are then averaged spatially to obtain a sensitivity matrix S of dimension R*N. The value in the i-th row and j-th column of the sensitivity matrix S equals the sensitivity value corresponding to the j-th voxel position of the i-th channel. Here, R is the total number of channels in the receiving coil, and N is the number of voxels simultaneously excited. In this embodiment, R = 64, and N = 2.

[0113] S3.3. Take Q pixels at the same edge positions in the grayscale images of each channel, and obtain the noise vector η for each channel image based on the grayscale value of each pixel. r :

[0114] η r =[h1 h2…h q …h Q ]

[0115] Where, η r h is the noise vector corresponding to the r-th channel. q Let be the grayscale value of the q-th pixel in the r-th channel;

[0116] In practice, the aforementioned pixels are all located at the outermost edge of the channel image, in the corner area outside the human brain map;

[0117] S3.4 Next, the noise vector is processed according to the following formula to obtain the receiving coil noise covariance matrix Ψ of dimension R*R:

[0118]

[0119] Where, ψ i,j Let η be the value of the receiving coil noise covariance matrix Ψ in the i-th row and j-th column. i Let η be the noise vector corresponding to the i-th channel. j This is the noise vector corresponding to the j-th channel;

[0120] S3.5. Obtain the expanded matrix U according to the following formula:

[0121] U=(S H Ψ -1 S) -1 S H Ψ -1

[0122] Wherein, the superscript H indicates the conjugate transpose of the sensitivity matrix S;

[0123] S3.6 Multiply the expanded matrix U with the multi-voxel diffusion-weighted magnetic resonance spectral signal of dimension R*P*A to obtain the multi-voxel diffusion-weighted magnetic resonance spectral signal of dimension N*P*A, so as to complete the separation of diffusion-weighted magnetic resonance spectral signals at different voxel positions, where P is the number of sampling points of the multi-voxel diffusion-weighted magnetic resonance spectral signal and A is the number of scans of the multi-voxel diffusion-weighted magnetic resonance spectral signal;

[0124] In this embodiment, P = 1024, A = 80 when scanning the water peak, and A = 160 when scanning the metabolite peak.

[0125] S3.7 Perform frequency and phase correction on all multi-voxel diffusion-weighted magnetic resonance spectral signals corresponding to the same b value;

[0126] S3.8 Arrange all multi-voxel diffusion-weighted magnetic resonance (MDR) spectral signals corresponding to the same b value according to the given peak height of the magnetic resonance spectrum. Select three MDR spectral signals with peak heights of the second, third, and fourth from all the MDR spectral signals. Then, take 60% (under water suppression enabled) or 80% (under water suppression disabled) of the average peak height of these three MDR spectral signals as the threshold. Discard MDR spectral signals with peak heights lower than the threshold. Then, average the remaining MDR spectral signals and perform frequency and phase correction again to obtain the reconstructed MDR spectral signal under the given b value.

[0127] S3.9 Repeat steps S3.7 to S3.8 to obtain the reconstructed multi-voxel diffusion-weighted magnetic resonance spectral signals for all b values;

[0128] Step S4: Quantitatively calculate the relevant parameters of the reconstructed multi-voxel diffusion-weighted magnetic resonance spectroscopy signal, and compare its performance with that of single-voxel diffusion-weighted magnetic resonance spectroscopy to verify the feasibility of this technique.

[0129] S4.1 Calculate the g-factor to assess the difference in the receiving coil sensitivity profile corresponding to the location of each voxel. The formula is as follows:

[0130]

[0131] According to the above formula, we get g = 1.0122 ± 0.0047, which indicates that the difference in the sensitivity profile of the receiving coil corresponding to the two voxel locations is large enough to facilitate signal separation.

[0132] S4.2. The signal leakage ratio was calculated from the area under the water peak. When only voxel 1 was excited, a signal leakage of 3.92% ± 0.56% was observed at voxel 2. When only voxel 2 was excited, a signal leakage of 7.05% ± 1.16% was observed at voxel 1.

[0133] The specific calculation method for the leakage ratio is as follows:

[0134] When using a single-voxel diffusion-weighted magnetic resonance spectroscopy sequence to excite only one voxel to obtain the water peak, the above reconstruction method is used to obtain the signal of each voxel. The signal observed at the unexcited voxel is the leakage signal. The leakage ratio of the unexcited voxel relative to the excited voxel can be obtained by dividing the area under the water peak of each unexcited voxel by the area under the water peak of the excited voxel.

[0135] S4.3 The signal-to-noise ratio (SNR) is calculated by dividing the area under the water peak by the standard deviation of the last 256 points of the corresponding metabolite's spectral time-domain signal. The SNR obtained from multi-voxel excitation is divided by the SNR obtained from the corresponding single-voxel excitation; the results are shown in Table 1 below.

[0136] Table 1 compares the spectral signal-to-noise ratios obtained from multi-voxel excitation and monovoxel excitation.

[0137]

[0138] This verifies that the signal-to-noise ratio of the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence developed in this invention is comparable to that of the single-voxel diffusion-weighted magnetic resonance spectroscopy sequence.

[0139] S4.4. Use LCModel software to quantify the relative concentrations of each metabolite molecule, and calculate the apparent diffusion coefficient of each metabolite molecule according to the following formula:

[0140]

[0141] Where S0 is the signal received without applying a diffusion gradient, S(b) is the signal received with an applied diffusion gradient, D is the apparent diffusion coefficient, and b represents the magnitude of the diffusion weight.

[0142] Apparent diffusion coefficients of water molecules and major metabolite molecules (in μm) 2 The calculation results ( / ms) are shown in Table 2 below:

[0143] Table 2 compares the apparent diffusion coefficients of different molecules obtained by multi-voxel excitation and monovoxel excitation.

[0144]

[0145] After paired-samples t-test, the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence developed in this invention showed no significant difference between the ADCs of water molecules and major metabolite molecules obtained at voxels 1 and 2 and the single-voxel diffusion-weighted magnetic resonance spectroscopy sequence.

[0146] Figure 5 This is a schematic diagram of the metabolite peak spectrum of the centrum semiovale region of a normal subject's brain, reconstructed according to this invention. The spectral signal leakage at different voxels is minimal, and the multi-voxel diffusion-weighted magnetic resonance (MDR) spectral signal is essentially consistent with the corresponding single-voxel diffusion-weighted magnetic resonance (SVR) spectral signal. Using a single-voxel diffusion-weighted magnetic resonance (SVR) spectral sequence, only voxel 1 was excited; the reconstructed spectrum was obtained at voxel 1. Figure 5 (a) Signal leakage is observed at voxel 2. Figure 5 (b) Using a single-voxel diffusion-weighted magnetic resonance spectroscopy sequence, only voxel 2 was excited. After reconstruction, the spectrum was obtained, and signal leakage was observed at voxel 1. Figure 5(c) is obtained at voxel 2. Figure 5 (d); Using the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence developed in this invention, voxels 1 and 2 are simultaneously excited, and the reconstructed spectrum is obtained. The spectrum at voxel 1 is obtained... Figure 5 (e) is obtained at voxel 2. Figure 5 (f). Figure 5 (a)~ Figure 5 The spectra in (f) are at the same scale.

[0147] The parallel acquisition and reconstruction method of multi-voxel diffusion-weighted magnetic resonance spectroscopy proposed in this invention achieves a signal-to-noise ratio and apparent diffusion coefficient comparable to that of single-voxel diffusion-weighted magnetic resonance spectroscopy while saving at least half of the scanning time, demonstrating its clinical application prospects and feasibility.

Claims

1. A method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectra, characterized in that, Includes the following steps: Step S1: Add a pair of diffusion gradients to the basic monomer spectral sequence to obtain the monomer diffusion-weighted magnetic resonance spectral sequence. The diffusion gradient described herein is a pair of completely symmetrical but oppositely polar diffusion gradients; Step S2: Sum the single-band radio frequency pulses at each location of the voxel of interest point by point to obtain multi-band radio frequency pulses. Replace the single-band radio frequency pulses in the single voxel diffusion-weighted magnetic resonance spectrum sequence with multi-band radio frequency pulses to obtain the multi-voxel diffusion-weighted magnetic resonance spectrum sequence, and then obtain the multi-voxel diffusion-weighted magnetic resonance spectrum signal. The specific steps of step S2 are as follows: S2.1 Determine the number of voxels to be collected as N, and select the initial positions of the N voxels; S2.2 Summing the single-band radio frequency pulses corresponding to the N voxel positions point by point to obtain multi-band radio frequency pulses; S2.3 Compare the excitation profiles of single-band and multi-band radio frequency pulses: If the error between the two is within the preset error range, the waveform of the multi-band radio frequency pulse is considered to be correct, and the process continues to step S2.

4. Otherwise, the waveform of the multi-band radio frequency pulse is considered incorrect, and the process returns to step S2.2 to acquire a new multi-band radio frequency pulse. S2.4 Change the position of each voxel and repeat steps S2.2 to S2.3 to obtain multiple multi-band radio frequency pulses with correct waveforms; S2.5 Add all the multi-band radio frequency pulses obtained in step S2.4 to the pulse file of the voxel diffusion-weighted magnetic resonance spectrum sequence, and select one of the multi-band radio frequency pulses from the pulse file to replace the initial single-band radio frequency pulse to obtain the multi-voxel diffusion-weighted magnetic resonance spectrum sequence. S2.

6. Import the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence into the magnetic resonance scanner, set the number and size of b values ​​and the number of acquisitions, and then use the magnetic resonance scanner to acquire the multi-voxel diffusion-weighted magnetic resonance spectroscopy signal. Step S3: Obtain the sensitivity map of each channel of the receiving coil, and perform separation and reconstruction of the multi-voxel diffusion-weighted magnetic resonance spectrum signal based on the sensitivity map, and finally obtain the separated and reconstructed multi-voxel diffusion-weighted magnetic resonance spectrum signal. The specific steps of step S3 are as follows: S3.

1. Obtain the grayscale image of each channel of the receiving coil using the gradient echo sequence, and then obtain the sensitivity map corresponding to each channel using the grayscale image of each channel. S3.

2. Perform voxel registration on the sensitivity maps of each channel to obtain the sensitivity map corresponding to each voxel position. Average these sensitivity maps in the spatial dimension to obtain a sensitivity matrix S with dimension R*N, where R is the total number of channels of the receiving coil. S3.

3. Take Q pixels at the same edge positions in the grayscale images of each channel, and obtain the noise vector for each channel image based on the grayscale value of each pixel. : =[h1 h2 … h q … h Q ], in, h is the noise vector corresponding to the r-th channel. q Let be the grayscale value of the q-th pixel in the r-th channel; S3.4 Next, the noise vector is processed according to the following formula to obtain the receiving coil noise covariance matrix of dimension R*R. : ， in, The noise covariance matrix of the receiving coil The value in the i-th row and j-th column, Let i be the noise vector corresponding to the i-th channel. This is the noise vector corresponding to the j-th channel; S3.

5. Obtain the expanded matrix U according to the following formula: ， Wherein, the superscript H indicates the conjugate transpose of the sensitivity matrix S; S3.6 Multiply the expanded matrix U with the multi-voxel diffusion-weighted magnetic resonance spectral signal of dimension R*P*A to obtain the multi-voxel diffusion-weighted magnetic resonance spectral signal of dimension N*P*A, where P is the number of sampling points of the multi-voxel diffusion-weighted magnetic resonance spectral signal and A is the number of sampling times of the multi-voxel diffusion-weighted magnetic resonance spectral signal. S3.7 Perform frequency and phase correction on all multi-voxel diffusion-weighted magnetic resonance spectral signals corresponding to the same b value; S3.8 Arrange all multi-voxel diffusion-weighted magnetic resonance (MDR) spectral signals corresponding to the same b value according to the given peak height of the magnetic resonance spectrum. Select three MDR spectral signals with peak heights of the second, third, and fourth from all the MDR spectral signals. Then, take 60% to 80% of the average peak height of these three MDR spectral signals as a threshold and discard MDR spectral signals below this threshold. Then, average the remaining MDR spectral signals and perform frequency and phase correction again to obtain the reconstructed MDR spectral signal at the given b value.

2. The method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectroscopy according to claim 1, characterized in that: In step S2.6, after importing the multi-voxel diffusion-weighted magnetic resonance spectroscopy sequence, the number of b-values ​​is set to 2 to 10, and the value of the b-value is set to 0 to 3333 s / mm. 2 Under the condition of water suppression off, 2 to 16 multi-voxel diffusion-weighted magnetic resonance spectroscopy signals were collected for each b value; under the condition of water suppression on, 16 to 64 multi-voxel diffusion-weighted magnetic resonance spectroscopy signals were collected for each b value.

3. The method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectroscopy according to claim 1, characterized in that: The value of b is obtained according to the following formula: ， in, γ The gyromagnetic ratio of the proton. g The magnitude of the diffusion gradient, δ The duration of the diffusion gradient, Δ This refers to the diffusion time.

4. The method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectroscopy according to claim 1, characterized in that: In step S2.2, the multi-band radio frequency pulse is obtained according to the following formula: ， in, The time-domain waveform of a multi-band radio frequency pulse. This is the time-domain waveform of the initial single-band radio frequency pulse in the sequence. Let t be the offset frequency at the position of the nth voxel, where n is the ordinal number of the voxel and t is the time.

5. The method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectroscopy according to claim 1, characterized in that: Step S2.3 specifically includes: S2.3.

1. Using the Bloch equation, simulations were performed on single-band and multi-band RF pulses respectively to obtain the simulated excitation profiles of single-band and multi-band RF pulses, and the excitation performance of multi-band RF pulses was preliminarily verified. The simulated excitation profiles of single-band and multi-band RF pulses are compared: If the error between the simulated excitation profile of a single-band RF pulse and the simulated excitation profile of a multi-band RF pulse is less than the preset simulation profile error value, then the waveform of the multi-band RF pulse is considered to be initially correct, and the process proceeds to step S2.3.

2. Otherwise, the waveform of the multi-band radio frequency pulse is considered incorrect, and the process returns to step S2.2 to reacquire the multi-band radio frequency pulse. S2.3.

2. Using a magnetic resonance scanner, actual tests were performed on single-band and multi-band radio frequency pulses to obtain the actual excitation profiles of single-band and multi-band radio frequency pulses, further verifying the excitation performance of multi-band radio frequency pulses: Comparison of the actual excitation profiles of single-band and multi-band radio frequency pulses: If the error between the actual excitation profile of a single-band RF pulse and the actual excitation profile of a multi-band RF pulse is less than the preset actual profile error value, then the waveform of the multi-band RF pulse is considered correct, and the process proceeds to step S2.

4. Otherwise, the waveform of the multi-band radio frequency pulse is considered incorrect, and the process returns to step S2.2 to reacquire the multi-band radio frequency pulse.

6. The method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectroscopy according to claim 1, characterized in that: In step S3.1, the sensitivity map of each channel of the receiving coil is obtained according to the following formula: ， in, This is the sensitivity plot for the r-th channel. Let R be the grayscale image received by the r-th channel, where R represents the total number of channels of the receiving coil.

7. The method for parallel acquisition and reconstruction of diffusion-weighted magnetic resonance spectroscopy according to claim 1, characterized in that: In step S3.8, if the multi-voxel diffusion-weighted magnetic resonance spectroscopy signal is acquired under water suppression enabled, the given magnetic resonance spectral peak is N-acetylaspartic acid; if the multi-voxel diffusion-weighted magnetic resonance spectroscopy signal is acquired under water suppression disabled, the given magnetic resonance spectral peak is the water peak.

8. An electronic device, characterized in that: It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the steps of the method according to any one of claims 1-7.

Images