Method, apparatus and magnetic resonance imaging device for reconstruction of multi-echo images
By adding constraints of low-rank terms and residual terms to the reconstruction function, the stability and realism issues of multi-echo image reconstruction methods are resolved, achieving higher reconstruction accuracy and stability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI UNITED IMAGING HEALTHCARE
- Filing Date
- 2023-05-18
- Publication Date
- 2026-07-24
AI Technical Summary
Existing multi-echo image reconstruction methods rely on the accuracy of real-world data acquisition, resulting in low stability and a tendency to overfit.
The low-rank and residual terms of the multi-echo image are constrained in the reconstruction function, and the multi-echo image is reconstructed by minimizing the low-rank and residual terms.
It improves the realism and stability of multi-echo image reconstruction, reduces the dependence on the accuracy of real-world data acquisition, and avoids overfitting.
Smart Images
Figure CN116630456B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of magnetic resonance image reconstruction technology, and in particular to a method, apparatus, magnetic resonance imaging device, storage medium and computer program product for reconstructing multi-echo images. Background Technology
[0002] Multi-shot echo planar imaging (EPI) technology can be used to reconstruct multi-echo images. The reconstruction process requires finding a solution that minimizes the reconstruction function, which is then used as the reconstruction result, thus obtaining the multi-echo image.
[0003] Currently, the reconstruction function provided by multi-echo image reconstruction schemes only includes the difference term. The reconstruction process based on this reconstruction function can be described in words as follows: First, assume a possible multi-echo image, transform the possible multi-echo image to K space to obtain the corresponding multi-echo data, and the difference between the multi-echo data and the actual acquired multi-echo data corresponds to the above-mentioned difference term; find a multi-echo image that minimizes the difference term, and complete the reconstruction of the multi-echo image.
[0004] However, the realism of the multi-echo images obtained using this reconstruction function, which only includes the difference term, depends heavily on the accuracy of the actual acquired multi-echo data. The reconstruction method using this reconstruction function has low stability and is prone to overfitting. Summary of the Invention
[0005] Therefore, it is necessary to provide a method, apparatus, magnetic resonance imaging device, storage medium, and computer program product for reconstructing multi-echo images to address the aforementioned technical problems.
[0006] This application provides a method for reconstructing multi-echo images, the method comprising:
[0007] Based on the multi-excitation echo plane imaging sequence, the first multi-echo data corresponding to the echo signal in the preset space is obtained;
[0008] Based on the difference between the second multi-echo data and the first multi-echo data corresponding to the multi-echo image to be reconstructed in the preset space, the basic part for constructing the reconstruction function is obtained;
[0009] A first constraint and a second constraint are added to the basic part to obtain the reconstruction function; the first constraint is a constraint on the low-rank term in the multi-echo image to be reconstructed; the second constraint is a constraint on the residual term in the multi-echo image to be reconstructed.
[0010] The multi-echo image is reconstructed based on the low-rank term and residual term that minimize the reconstruction function.
[0011] This application provides a reconstruction apparatus for multi-echo images, the apparatus comprising:
[0012] The signal observation module is used to obtain the first multi-echo data of the echo signal in a preset space based on the multi-excitation echo plane imaging sequence.
[0013] The reconstruction function processing module is used to obtain the basic part for constructing the reconstruction function based on the difference between the second multi-echo data and the first multi-echo data corresponding to the multi-echo image to be reconstructed in the preset space.
[0014] The reconstruction function processing module is used to add a first constraint and a second constraint to the basic part to obtain the reconstruction function; the first constraint is a constraint on the low-rank term in the multi-echo image to be reconstructed; the second constraint is a constraint on the residual term in the multi-echo image to be reconstructed.
[0015] The reconstruction result acquisition module is used to reconstruct the multi-echo image based on the low-rank term and residual term that minimize the reconstruction function.
[0016] This application provides a magnetic resonance imaging device, including a memory and a processor, wherein the memory stores a computer program and the processor executes the above-described method.
[0017] This application provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor using the methods described above.
[0018] This application provides a computer program product having a computer program stored thereon, the computer program being executed by a processor using the above-described method.
[0019] The aforementioned multi-echo image reconstruction method, apparatus, MRI equipment, storage medium, and computer program product, based on a multi-excitation echo plane imaging sequence, obtains first multi-echo data corresponding to the echo signal in a preset space; based on the difference between the second multi-echo data corresponding to the multi-echo image to be reconstructed in the preset space and the first multi-echo data, a basic part for constructing a reconstruction function is obtained; a first constraint and a second constraint are added to the basic part to obtain the reconstruction function; the first constraint is a constraint on the low-rank term in the multi-echo image to be reconstructed; the second constraint is a constraint on the residual term in the multi-echo image to be reconstructed; the multi-echo image is reconstructed based on the low-rank term and residual term that minimize the reconstruction function. In this application, the reconstruction function includes not only the aforementioned difference term, but also constraints on the low-rank term in the multi-echo image to be reconstructed and constraints on the residual term in the multi-echo image to be reconstructed. The realism of the multi-echo image obtained using this reconstruction function has a lower dependence on the accuracy of the actually acquired first multi-echo data, and the reconstruction method using this reconstruction function has higher stability and is less prone to overfitting. Attached Figure Description
[0020] Figure 1 This is a flowchart illustrating a method for reconstructing multi-echo images in one embodiment;
[0021] Figure 2 This is a schematic diagram of a multi-excitation EPI sequence in one embodiment;
[0022] Figure 3(a) is a schematic diagram of multi-echo data in one embodiment;
[0023] Figure 3(b) is a schematic diagram of filling multi-echo data in one embodiment;
[0024] Figure 3(c) is a schematic diagram of the correspondence between multi-echo data and multi-echo images in one embodiment;
[0025] Figure 4 This is a schematic diagram illustrating how a multi-echo image is constructed in one embodiment;
[0026] Figure 5 This is a schematic diagram illustrating the determination of the basis image weight matrix in one embodiment;
[0027] Figure 6 This is a schematic diagram illustrating the change of signal strength with echo time in gradient echo mode in one embodiment;
[0028] Figure 7 This is a schematic diagram illustrating the change of signal intensity with echo time in gradient echo mode + spin echo mode in one embodiment.
[0029] Figure 8This is a structural block diagram of a multi-echo image reconstruction device in one embodiment;
[0030] Figure 9 This is an internal structural diagram of a magnetic resonance imaging device in one embodiment. Detailed Implementation
[0031] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0032] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.
[0033] In one embodiment, such as Figure 1 As shown, a method for reconstructing multi-echo images is provided, which includes the following steps:
[0034] Step S101: Based on the multi-excitation echo plane imaging sequence, obtain the first multi-echo data corresponding to the echo signal in the preset space.
[0035] Multi-excitation echo-plane imaging sequences, also known as multi-excitation EPI sequences, are EPI sequences that acquire data using multiple excitation pulses. Each acquisition utilizes one excitation pulse, and the signals from multiple excitations are integrated to reconstruct a complete image. A multi-excitation EPI sequence can include multiple excitation pulses. Figure 2 The multi-excitation EPI sequence shown includes two excitation pulses: the first excitation pulse (shot 1) and the second excitation pulse (shot 2).
[0036] Among them, the magnetic field gradient in the Gy direction is designable, specifically reflected in... Figure 2 The height of each small peak in Gy can be set as needed, and is not mandatory.
[0037] Figure 2 Acq is short for Acquire, which represents the signal obtained after applying a corresponding magnetic field. In echo plane imaging technology, the obtained signal is the echo signal, and the shape of a single echo signal is shown in Figure 3(a).
[0038] After applying an excitation pulse, Gy and Gx can be used to cause the scanned object to generate multiple echo signals sequentially. By acquiring these echo signals, discrete echo data can be obtained. Specifically, the first echo signal generated after applying the excitation pulse is the first echo signal, and the time from the excitation pulse to the first echo signal is one echo time, denoted as TE. The second echo signal generated after applying the excitation pulse is the second echo signal, and the time from the excitation pulse to the second echo signal is two echo times, denoted as 2TE.
[0039] Referring to Figure 3(a), after applying the first excitation pulse shot 1, six discrete echo data points are obtained, corresponding to TE, 2TE, 3TE, 4TE, 5TE, and 6TE respectively. After applying the second excitation pulse shot 2, six discrete echo data points are also obtained, corresponding to TE, 2TE, 3TE, 4TE, 5TE, and 6TE respectively.
[0040] The discrete echo data is filled into a preset space (i.e., K-space) by applying Gy and Gx when the corresponding discrete echo data is acquired. The K-space includes k x and k y Two coordinate axes, k x Not shown in Figure 3(b), k y As shown in Figure 3(b), since the echo data has a corresponding echo time, an additional coordinate axis t can be introduced. TE As shown in Figure 3(b), in this case, when filling in the echo data, it is necessary to fill it according to the echo time corresponding to the echo data; referring to Figure 3(b), the coordinate axis t TE The column with the value TE is used to fill in the echo data corresponding to TE under each excitation pulse, such as the echo data corresponding to TE under the first excitation pulse shot 1 and the second excitation pulse shot 2.
[0041] After applying all excitation pulses of the multi-excitation EPI sequence and acquiring the corresponding echo signals, multi-echo data can be obtained. This multi-echo data can be filled into the k-space as described above. Reconstruction is then performed using this multi-echo data to obtain the multi-echo image {I}. TE I 2TE I 3TE I 4TE I 5TE I 6TE}, where I TE Corresponding to TE, I 2TE Corresponding to 2TE, I 3TE Corresponding to 3TE, I 4TE Corresponding to 4TE, I 5TE Corresponding to 5TE, I6TE For 6TE, see Figure 3(c).
[0042] To distinguish it from the other echo data below, the acquired multi-echo data will be referred to as the first multi-echo data.
[0043] Step S102: Based on the difference between the second multi-echo data and the first multi-echo data corresponding to the multi-echo image to be reconstructed in the preset space, the basic part for constructing the reconstruction function is obtained.
[0044] Step S103: Add a first constraint and a second constraint to the basic part to obtain the reconstruction function; the first constraint is a constraint on the low-rank term in the multi-echo image to be reconstructed; the second constraint is a constraint on the residual term in the multi-echo image to be reconstructed.
[0045] The basic process of multi-echo image reconstruction includes: first, assuming a multi-echo image, mapping this image back to k-space to obtain the corresponding multi-echo data, which can be called the second multi-echo data; calculating the difference between the second multi-echo data and the first multi-echo data; the smaller the difference, the closer the previously assumed multi-echo image is to the real multi-echo image. Continuously searching for various possible multi-echo images, the process finds the one that minimizes the reconstruction function to complete the multi-echo image reconstruction.
[0046] Reference Figure 4 The echo images in a multi-echo image can be represented by the following formula: I = LR + S, where L represents the base image, describing the common content among the multi-echo images; R represents the weight; LR refers to the low-rank term of the multi-echo image as it changes with echo time, which represents the part of the multi-echo image that changes slowly with echo time and has obvious low-rank properties; S refers to the residual between the multi-echo image and the low-rank term. Since it generally has good low-rank properties, the residual is generally sparse.
[0047] In constructing I TE At that time, through the weight {R} corresponding to TE 11 R 12 R 13 R 14 The basis images {L1, L2, L3, L4} are weighted and summed. The weighted sum is then added to the corresponding residual S1 to construct I. TE R is pre-set and will not change during the reconstruction process; it is not the solution required for reconstruction. L and S are the solutions required for reconstruction. Therefore, the above search for the multi-echo image that minimizes the reconstruction function is equivalent to searching for L and S.
[0048] In this case, the basic part used to construct the reconstruction function can be expressed as: M(LR+S)-y; where y is the first multi-echo data, and the result of performing a Fourier transform on LR+S using F included in M is the second multi-echo data.
[0049] In constructing the reconstruction function, this application includes not only the aforementioned basic components but also constraints on the low-rank term LR and the residual term S. These two constraints are referred to as the first constraint and the second constraint, respectively. The low-rank term LR can be represented as a matrix, and the first constraint is used to constrain the low-rank property of this matrix, hence the matrix is called a low-rank term. The second constraint is used to constrain the sparsity of the residual term S.
[0050] Furthermore, the first constraint is a norm constraint on the low-rank terms in the multi-echo image to be reconstructed; the second constraint is a norm constraint on the residual terms in the multi-echo image to be reconstructed.
[0051] By applying norm constraints, low-rank constraints are applied to the low-rank term LR, and sparsity constraints are applied to the residual term, thereby reducing processing complexity.
[0052] Furthermore, the first constraint can be a nuclear norm constraint, and the second constraint can be a 1-norm constraint. The resulting reconstruction function can be expressed as:
[0053]
[0054] Where y represents the acquired echo data, which can be obtained through height downsampling, and λ LR and λ S It is a real number that is not less than 0.
[0055] Step S104: Reconstruct the multi-echo image based on the low-rank term and residual term that minimize the reconstruction function.
[0056] After obtaining the basic components, first constraint, and second constraint used to construct the reconstruction function, the reconstruction function is constructed. The L and S that minimize the reconstruction function are then searched. Since R is constant during the reconstruction process, searching for the L that minimizes the reconstruction function is equivalent to searching for the LR that minimizes the reconstruction function. Based on the found L and S that minimize the reconstruction function, LR+S is obtained, thereby reconstructing the multi-echo image.
[0057] In this application, the reconstruction function includes not only the aforementioned difference term (corresponding to the aforementioned basic part), but also constraints on the low-rank term in the multi-echo image to be reconstructed, and constraints on the residual term in the multi-echo image to be reconstructed. The realism of the multi-echo image obtained using this reconstruction function has a lower dependence on the accuracy of the actual first multi-echo data. The reconstruction method using this reconstruction function has higher stability and is less prone to overfitting.
[0058] In one embodiment, prior to step S102, the method provided by this application further includes the following steps: obtaining a multi-echo image to be reconstructed based on the low-rank term in the multi-echo image to be reconstructed and the residual term in the multi-echo image to be reconstructed; mapping the multi-echo image to be reconstructed to the preset space to obtain second multi-echo data.
[0059] We can first assume a set of basis images {L1, L2, L3, L4}, and then use R to perform a weighted summation of these basis images to obtain the low-rank term. Combined with the residual term, we can construct the hypothetical multi-echo image {I}. TE I 2TE I 3TE I 4TE I 5TE I 6TE}, specifically, is as follows:
[0060] L1R 11 +L2R 12 +L3R 13 +L4R 14 +S1=I TE L1R 21 +L2R 22 +L3R 23 +L4R 24 +S2=I 2TE ;
[0061] L1R 31 +L2R 32 +L3R 33 +L4R 34 +S3=I 3TE L1R 41 +L2R 42 +L3R 43 +L4R 44 +S4=I 4TE ;
[0062] L1R 51 +L2R 52 +L3R 53 +L4R 54 +S5=I 5TE L1R 61 +L2R 62 +L3R 63 +L4R 64 +S6=I 6TE
[0063] Next, a hypothetical multi-echo image {I} will be constructed. TE I 2TE I3TE I 4TE I 5TE I 6TE Mapped to K space, I TE The echo data obtained by mapping to K-space corresponds to TE, and this echo data is compared with the echo data corresponding to TE in the first multi-echo data; similarly, I 2TE The echo data obtained by mapping to K space corresponds to 2TE, and this echo data is compared with the echo data corresponding to 2TE in the first multi-echo data.
[0064] Further, the multi-echo image to be reconstructed is mapped to the preset space to obtain the second multi-echo data, specifically including the following steps: the multi-echo image to be reconstructed is mapped to the preset space through the coil sensitivity matrix, Fourier transform and downsampling matrix to obtain the second multi-echo data.
[0065] In the above basic part, M = UFC, where C is the coil sensitivity matrix, F is the Fourier transform, and U is the downsampling matrix. The assumed multi-echo image is mapped to the K space using M to obtain the corresponding second multi-echo data.
[0066] In one embodiment, the method provided by this application further includes: acquiring multiple base images and a base image weight matrix corresponding to multiple echo times; and obtaining the low-rank term in the multi-echo image to be reconstructed based on the multiple base images and the base image weight matrix corresponding to multiple echo times.
[0067] Multiple basis images can be obtained, which can be denoted as {L1, L2, L3, L4}. The number of basis images depends on the number of basis curves, and the number of basis images is the same as the number of basis curves. The process of obtaining the basis curves will be introduced later.
[0068] If there are 6 TEs, the resulting basis image weight matrix is:
[0069]
[0070] The weights in the first row correspond to TE, the weights in the second row correspond to 2TE, the weights in the third row correspond to 3TE, the weights in the fourth row correspond to 4TE, the weights in the fifth row correspond to 5TE, and the weights in the sixth row correspond to 6TE.
[0071] Next, the weights of each row are assigned to the aforementioned base image, resulting in:
[0072] L1R 11 +L2R 12 +L3R 13 +L4R 14 This item is for building ITE The low-rank terms used;
[0073] L1R 21 +L2R 22 +L3R 23 +L4R 24 This item is for building I 2TE The low-rank terms used;
[0074] L1R 31 +L2R 32 +L3R 33 +L4R 34 This term is the low-rank term used to construct the I3TE;
[0075] L1R 41 +L2R 42 +L3R 43 +L4R 44 This item is for building I 4TE The low-rank terms used;
[0076] L1R 51 +L2R 52 +L3R 53 +L4R 54 This item is for building I 5TE The low-rank terms used;
[0077] L1R 61 +L2R 62 +L3R 63 +L4R 64 This item is for building I 6TE The low-rank term used.
[0078] The following describes the process of obtaining the basis image weight matrix corresponding to multiple echo times: Based on the simulation curve cluster of the echo signal midpoint intensity value changing with echo time, multiple basis curves are obtained; based on the echo signal midpoint intensity value corresponding to multiple echo times on multiple basis curves, the basis image weight matrix corresponding to multiple echo times is obtained.
[0079] Referring to the echo signal shown in Figure 3(a), the intensity value of the echo signal changes from large to small over time as it is generated and disappears. The midpoint corresponds to the midpoint of the echo signal, and the intensity value corresponding to the midpoint is the maximum intensity value of the echo signal. The intensity value corresponding to the midpoint is called the midpoint intensity value of the echo signal.
[0080] Once the multi-excitation EPI sequence is determined, the TE can be determined accordingly; based on the midpoint intensity value of the echo signal. The values of M0 and T2* are obtained within a certain range. During simulation, after selecting a certain M0 value and T2* value within the set range, TE is used as the independent variable. Substituting TE, 2TE, ..., nTE respectively, yields the corresponding I, which is the midpoint intensity value of the echo signal and is the dependent variable. This results in a simulation curve in which the midpoint intensity value of the echo signal changes with the echo time. Then, another M0 value and T2* value can be selected within the set range, thus obtaining another simulation curve. By iterating through all the M0 and T2* values within the set range, a cluster of simulation curves can be formed.
[0081] Next, singular value decomposition is performed on the simulated curve cluster to obtain a certain number of basis curves. This means that the simulated curve cluster can be reconstructed using these basis curves. The number of basis curves is equal to the number of basis images mentioned above. If the obtained are as follows... Figure 5 The four base curves {①, ②, ③, ④} are shown. The values corresponding to 2TE are extracted from each base curve, and these values are used as weights corresponding to 2TE to obtain {R}. 21 R 22 R 23 R 24 Similarly, the weights corresponding to TE, 3TE, 4TE, etc., can also be determined, thus obtaining the basis image weight matrix corresponding to the 6 TEs.
[0082] In the above embodiments, multiple base curves determined by a cluster of simulated curves showing the change of midpoint intensity value of echo signal with echo time are used to obtain the base image weight matrix corresponding to multiple echo times, thereby improving the accuracy of the reconstruction results.
[0083] Quantitative calculation of magnetic resonance parameters is performed based on the reconstructed multi-echo images. For example, the B0 field distribution is obtained based on the phase changes between different echo images; QSM (quantitative susceptibility measure) is obtained using the amplitude and phase information of the multi-echo images; T2* and PD quantitative distributions are obtained by exponential fitting of the multi-echo images based on the scanning and reconstruction results of gradient echo EPI; T2*, T2, and PD quantitative distributions are obtained by segmented exponential fitting of the multi-echo images based on the scanning and reconstruction results of EPI acquired together with spin echo and gradient echo.
[0084] In one embodiment, the multi-excitation EPI sequence may have a gradient echo mode. In this case, after reconstructing the multi-echo image, the method provided in this application further includes: performing exponential fitting based on the multi-echo image and the multi-echo time; and determining the transverse relaxation time and proton density based on the exponential fitting result.
[0085] After reconstructing the multi-echo image, the midpoint intensity value of the echo signal corresponding to each echo image can be determined. Since the echo images in the multi-echo image correspond to TE, 2TE, 3TE, etc., the midpoint intensity value of the echo signal corresponding to TE, 2TE, 3TE, etc., can be determined, such as... Figure 6 As shown. Perform an exponential fit according to the following formula.
[0086]
[0087] Where M0 is the proton density, M0 is the transverse relaxation time that needs to be solved, and TE is the echo time. An exponential fit is performed on all echo images to obtain M0 and M1.
[0088] In some scenarios, VFA (variable flip angle) technology can be used on the sequence to change the flip angle size and then collect and reconstruct the data. By performing exponential fitting on the multi-echo data, an additional T1 distribution can be obtained. Similarly, by using IR (inverse recovery) module technology on the sequence and changing the TI duration, exponential fitting on the multi-echo data can also obtain a T1 distribution.
[0089] After adding a diffusion module to this sequence in the spin echo EPI mode, it is possible to measure and reconstruct a multi-echo image under a certain diffusion condition.
[0090] In one embodiment, the method provided in this application further includes: acquiring multi-echo images reconstructed from excitation pulses at different flip angles; acquiring the transverse relaxation time corresponding to the multi-echo images at each flip angle; and performing exponential fitting based on the multi-echo images at each flip angle and the corresponding transverse relaxation time to determine the longitudinal relaxation time.
[0091] In this embodiment, the flip angle α of the excitation pulse is continuously changed and multi-echo images under different flip angles are reconstructed, and finally the quantitative distribution of the T1 relaxation parameter is obtained by fitting.
[0092]
[0093] Where I0 is the signal amplitude when TE=0, calculated according to the example, i.e., the image proton density; M0 is the actual proton density of the material; α is the flip angle; and T1 is the longitudinal relaxation time to be calculated. In this formula, I0 and α are known values, while M0 and T1 are unknowns. Therefore, by continuously adjusting the flip angle α to obtain I0 under different conditions (at least two conditions), it is possible to solve for M0 and T1 while obtaining multi-echo images.
[0094] Furthermore, after reconstructing the multi-echo image under gradient echo and spin echo modes, the method provided in this application further includes: performing piecewise exponential fitting based on the multi-echo image and the multi-echo time to obtain the first transverse relaxation time, the second transverse relaxation time, and the proton density.
[0095] The first lateral relaxation time is The second lateral relaxation time is T2.
[0096] The reconstructed echo images are segmented as follows: Figure 7 As shown, the signal in each segment satisfies:
[0097]
[0098]
[0099]
[0100] Where ρ0 is the signal amplitude when TE = 0, which is the proton density; TE is the echo time; T SE The time at the center of the spin echo; ρ SE It is the signal amplitude at the center of the spin echo. From the above formula, it can be seen that there are four unknowns: ρ SE ,ρ0, T2. By performing exponential fitting on the reconstructed multi-echo image from the example into the three segments mentioned above, the four unknowns can be solved.
[0101] Furthermore, in the gradient echo + spin echo EPI mode, the gradient echo component is discarded, and a diffusion module is added, allowing simultaneous acquisition of diffusion information and multi-echo information. The diffusion information is acquired in the same way as in general multi-shot EPI, yielding ADC quantitative parameters.
[0102] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0103] In one embodiment, such as Figure 8As shown, a reconstruction apparatus for multi-echo images is provided, comprising:
[0104] The signal observation module 801 is used to obtain the first multi-echo data corresponding to the echo signal in a preset space based on the multi-excitation echo plane imaging sequence.
[0105] The reconstruction function processing module 802 is used to obtain the basic part for constructing the reconstruction function based on the difference between the second multi-echo data and the first multi-echo data corresponding to the multi-echo image to be reconstructed in the preset space.
[0106] The reconstruction function processing module 802 is further configured to add a first constraint and a second constraint to the basic part to obtain a reconstruction function; the first constraint is a constraint on the low-rank term in the multi-echo image to be reconstructed; the second constraint is a constraint on the residual term in the multi-echo image to be reconstructed.
[0107] The reconstruction result acquisition module 803 is used to reconstruct the multi-echo image based on the low-rank term and residual term that minimize the reconstruction function.
[0108] In one embodiment, the apparatus further includes a mapping module, configured to obtain a multi-echo image to be reconstructed based on a low-rank term in the multi-echo image to be reconstructed and a residual term in the multi-echo image to be reconstructed; and to map the multi-echo image to be reconstructed onto the preset space to obtain second multi-echo data.
[0109] In one embodiment, the mapping module is further configured to map the multi-echo image to be reconstructed onto the preset space using a coil sensitivity matrix, a Fourier transform, and a downsampling matrix, to obtain second multi-echo data.
[0110] In one embodiment, the first constraint is a norm constraint on the low-rank terms in the multi-echo image to be reconstructed; the second constraint is a norm constraint on the residual terms in the multi-echo image to be reconstructed.
[0111] In one embodiment, the apparatus further includes a low-rank term construction module for acquiring multiple base images and base image weight matrices corresponding to multiple echo times; and obtaining low-rank terms in the multi-echo images to be reconstructed based on the multiple base images and the base image weight matrices corresponding to multiple echo times.
[0112] In one embodiment, the low-rank term construction module is further configured to obtain multiple base curves based on a cluster of simulated curves showing the change of echo signal midpoint intensity value with echo time; and to obtain a base image weight matrix corresponding to multiple echo times based on the echo signal midpoint intensity values corresponding to multiple echo times on the multiple base curves.
[0113] In one embodiment, in gradient echo mode, the device further includes a first processing module for performing exponential fitting based on the multi-echo image and the multi-echo time; and determining the transverse relaxation time and proton density based on the exponential fitting result.
[0114] In one embodiment, the first processing module is further configured to acquire multi-echo images reconstructed from excitation pulses at different flip angles; acquire the lateral relaxation time corresponding to the multi-echo images at each flip angle; and perform exponential fitting based on the multi-echo images at each flip angle and the corresponding lateral relaxation time to determine the longitudinal relaxation time.
[0115] In one embodiment, in gradient echo and spin echo modes, the device further includes a second processing module for performing piecewise exponential fitting based on the multi-echo image and multi-echo time to obtain a first transverse relaxation time, a second transverse relaxation time, and a proton density.
[0116] Specific limitations regarding the reconstruction apparatus for multi-echo images can be found in the limitations of the reconstruction method for multi-echo images described above, and will not be repeated here. Each module in the aforementioned multi-echo image reconstruction apparatus can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in the MRI equipment, or stored in software within the memory of the MRI equipment, so that the processor can call and execute the corresponding operations of each module.
[0117] In one embodiment, a magnetic resonance imaging device is provided, the internal structure of which can be shown as follows: Figure 9 As shown, this magnetic resonance imaging (MRI) device includes a processor, memory, and network interface connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores reconstructed data from multi-echo images. The network interface allows communication with external terminals via a network connection. The device also includes input / output interfaces (I / O interfaces), which are circuits connecting the processor and external devices for information exchange. These I / O interfaces are connected to the processor via a bus. When the computer program is executed by the processor, it implements a multi-echo image reconstruction method.
[0118] Those skilled in the art will understand that Figure 9The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the magnetic resonance imaging device to which the present application is applied. A specific magnetic resonance imaging device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0119] In one embodiment, a magnetic resonance imaging device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the various method embodiments described above.
[0120] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the various method embodiments described above.
[0121] In one embodiment, a computer program product is provided having a computer program stored thereon, the computer program being executed by a processor of the steps described in the various method embodiments above.
[0122] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.
[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0124] The above embodiments are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A method for reconstructing multi-echo images, characterized in that, The method includes: Based on the multi-excitation echo plane imaging sequence, the first multi-echo data corresponding to the echo signal in the preset space is obtained; Based on the low-rank term and the residual term in the multi-echo image to be reconstructed, the multi-echo image to be reconstructed is obtained. The multi-echo image to be reconstructed is then mapped to a preset space to obtain the second multi-echo data. Based on the difference between the second multi-echo data and the first multi-echo data, the basic part used to construct the reconstruction function is obtained; A first constraint and a second constraint are added to the basic part to obtain the reconstruction function; the first constraint is a constraint on the low-rank term in the multi-echo image to be reconstructed; the second constraint is a constraint on the residual term in the multi-echo image to be reconstructed. The multi-echo image is reconstructed based on the low-rank term and residual term that minimize the reconstruction function; The low-rank term is obtained based on multiple base images and the base image weight matrix corresponding to multiple echo times. The method for determining the base image weight matrix includes: obtaining multiple base curves based on a cluster of simulated curves showing the change of echo signal midpoint intensity value with echo time; and obtaining the base image weight matrix corresponding to multiple echo times based on the echo signal midpoint intensity value corresponding to multiple echo times on the multiple base curves.
2. The method according to claim 1, characterized in that, The first constraint is a norm constraint on the low-rank terms in the multi-echo image to be reconstructed; the second constraint is a norm constraint on the residual terms in the multi-echo image to be reconstructed.
3. The method according to claim 1, characterized in that, In gradient echo mode, after reconstructing the multi-echo image based on the low-rank term and residual term that minimize the reconstruction function, the method further includes: Based on the multi-echo images and multi-echo times, perform exponential fitting; Based on the exponential fitting results, the transverse relaxation time and proton density were determined.
4. The method according to claim 3, characterized in that, The method further includes: Acquire multi-echo images reconstructed from excitation pulses at different flip angles; Obtain the lateral relaxation time corresponding to the multi-echo image at each flip angle; Based on the multi-echo images corresponding to each flip angle and the corresponding lateral relaxation time, exponential fitting is performed to determine the longitudinal relaxation time.
5. The method according to claim 1, characterized in that, In gradient echo and spin echo modes, after reconstructing the multi-echo image based on the low-rank term and residual term that minimize the reconstruction function, the method further includes: Based on the multi-echo images and multi-echo times, piecewise exponential fitting is performed to obtain the first transverse relaxation time, the second transverse relaxation time, and the proton density.
6. The method according to claim 1, characterized in that, The multi-echo image to be reconstructed is mapped to a preset space to obtain the second multi-echo data, including: By using the coil sensitivity matrix, Fourier transform, and downsampling matrix, the multi-echo image to be reconstructed is mapped to a preset space to obtain the second multi-echo data.
7. A reconstruction apparatus for multi-echo images, characterized in that, The device includes: The signal observation module is used to obtain the first multi-echo data of the echo signal in a preset space based on the multi-excitation echo plane imaging sequence. The mapping module is used to obtain the multi-echo image to be reconstructed based on the low-rank term and the residual term in the multi-echo image to be reconstructed, and to map the multi-echo image to be reconstructed to a preset space to obtain the second multi-echo data. The reconstruction function processing module is used to obtain the basic part for constructing the reconstruction function based on the difference between the second multi-echo data and the first multi-echo data; The reconstruction function processing module is further configured to add a first constraint and a second constraint to the basic part to obtain a reconstruction function; the first constraint is a constraint on the low-rank term in the multi-echo image to be reconstructed; the second constraint is a constraint on the residual term in the multi-echo image to be reconstructed. The reconstruction result acquisition module is used to reconstruct the multi-echo image based on the low-rank term and residual term that minimize the reconstruction function; The low-rank term is obtained based on multiple base images and the base image weight matrix corresponding to multiple echo times. The method for determining the base image weight matrix includes: obtaining multiple base curves based on a cluster of simulated curves showing the change of echo signal midpoint intensity value with echo time; and obtaining the base image weight matrix corresponding to multiple echo times based on the echo signal midpoint intensity value corresponding to multiple echo times on the multiple base curves.
8. The apparatus according to claim 7, characterized in that, The mapping module is used for: By using the coil sensitivity matrix, Fourier transform, and downsampling matrix, the multi-echo image to be reconstructed is mapped to a preset space to obtain the second multi-echo data.
9. The apparatus according to claim 7, characterized in that, The first constraint is a norm constraint on the low-rank terms in the multi-echo image to be reconstructed; the second constraint is a norm constraint on the residual terms in the multi-echo image to be reconstructed.
10. A magnetic resonance imaging device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.