EPI with zigzag trajectory and parallel imaging

Abstract

Methods for echoplanar MR imaging, in which MR signals are acquired in a raw data space with a zigzag trajectory, comprising the following steps: - Acquisition of MR raw data points with a sequence of readout gradients and phase encoding gradients such that a zigzag undersampled trajectory (52, 73) is captured in a captured raw data space (51, 71) such that the signal echoes are shifted by up to a quarter of the captured raw data space in the direction of a readout axis, - Reconstructing an additional zigzag trajectory (53, 72) in the acquired raw data space (51, 71) using parallel imaging techniques based on the acquired zigzag trajectory, - Reconstructing at least one MR image based on an interlaced Fourier transform, iFT, wherein prior to reconstructing the at least one MR image, the raw data points in the raw data space (50, 70) are extended such that an extension segment (55, 75) is created, wherein this extension is carried out by applying a point symmetry of the raw data space to a raw data center to create an extended raw data space that includes the acquired raw data space and the extension segment, wherein the raw data in the extension segment are computed using a point reflection about the raw data center.

Description

[0001] The present invention relates to a method for echo-planar MR imaging in which MR signals are acquired with a zigzag trajectory in the raw data space. The invention further relates to the associated magnetic resonance system for carrying out the method. A computer program and an electrically readable data carrier are also provided.

[0002] In magnetic resonance imaging (MRI), rapid acquisition methods based on the EPI (Echo Planar Imaging) principle are known. One possible EPI acquisition technique uses the so-called zigzag trajectory, employing continuous phase encoding with simultaneous sinusoidal readout gradients. This technique has the advantage of generating less noise during MR signal acquisition and achieving a shorter echo time due to more efficient, continuous data acquisition. However, with such continuous phase encoding, the data points do not lie on a Cartesian grid, which is a fundamental requirement for using Cartesian parallel imaging such as GRAPPA (Generalized Autocalibrating Partially Parallel Acquisition). German patent DE 10 2013 100 349 A1 discloses the use of phase encoding configured such that the data lie on a mathematically correct zigzag trajectory.Other forms can also be chosen, they just need to be exactly a scaled amount of the selection gradient.

[0003] It is possible to reconstruct an MR image from the average non-undersampled MR data using an interlaced Fourier transform (iFT). However, with undersampled, i.e., accelerated, data acquisition, the data can no longer be reconstructed using the iFT. In such a case, GRAPPA reconstruction must be used, taking into account that with an assumed acceleration factor of 2 during acquisition, the GRAPPA reconstruction must be applied with a reconstruction factor of 4. This is because the GRAPPA reconstruction is applied separately to the even or odd k-space rows in the raw data space. These separated datasets are already 2x undersampled for an unaccelerated acquisition. For a 2x accelerated acquisition, they are already 4x undersampled.Due to the higher reconstruction factor in GRAPPA reconstruction, the so-called g-factor increases, which, together with the acquisition acceleration factor, describes the signal-to-noise loss caused by parallel imaging reconstruction. If additional data is reconstructed in undersampled data using parallel imaging, a region with twofold undersampling now exists in the middle of the k-space lines. This means that combining GRAPPA reconstruction with an acceleration factor and the use of iFT is not possible. Further prior art applications include DE 10 2013 100 349 A1 and US 2015 / 0 253 408 A1.

[0004] An objective of the present invention is to use accelerated acquisition with the use of parallel imaging techniques, while simultaneously not increasing the noise in the acquired MR image too much.

[0005] This objective is achieved through the features of the independent claims. Preferred embodiments of the invention are described in the dependent claims.

[0006] A method for echo-planar MR imaging is provided, in which MR signals are acquired in a raw data space with a zigzag trajectory. Here, raw MR data points are acquired with a sequence of readout gradients and phase-encoding gradients such that a zigzag undersampled trajectory is captured in the acquired raw data space.

[0007] The acquisition process involves shifting the signal echoes by up to a quarter of the acquired tube data space in the direction of a readout axis. Furthermore, an additional zigzag trajectory is reconstructed within the acquired raw data space using parallel imaging techniques and the acquired zigzag trajectory. Subsequently, at least one MR image is reconstructed based on an interlaced Fourier transform (iFT).

[0008] By shifting the signal echoes by up to a quarter of the captured raw data space in the direction of the readout axis, the signal echoes are positioned in a region of the raw data space where the spacing of the trajectories in the phase-encoding direction corresponds to Δky. This means that a new raw data center is located where the sampling density is sufficient to allow reconstruction using the iFT. As is known, Δky is proportional to the integral of the gradient along the phase-encoding axis multiplied by the time the gradient is applied. The difference between two trajectories in the direction of the phase-encoding axis is now less than Δky, which means that the Nyquist theorem is satisfied here.If the echo is shifted by less than a quarter, the raw data center lies in a region where the distance between two points in the phase-encoding direction is alternately smaller and larger than the minimum distance required by the Nyquist theorem for a convolution-free reconstruction. This means that the mean distance between adjacent points in the phase-encoding direction meets the Nyquist criterion, and successful data reconstruction is possible.

[0009] Preferably, prior to the reconstruction of the at least one MR image, the raw data in the raw data space are extended in such a way that an extension segment is created, wherein this extension is carried out by applying a point symmetry of the raw data space to a raw data center to create an extended raw data space that includes the acquired raw data space and the extension segment, wherein the raw data in the extension segment are calculated using a point reflection about the raw data center.

[0010] The raw data in the extension segment can be determined using the so-called zero-filling technique or the partial Fourier technique.

[0011] The signal echoes can be shifted along a readout axis by shortening or lengthening the dephasing pre-gradient before signal readout. With a reduced dephasing pre-gradient, signals occur earlier compared to when the signal echo occurs in the middle of a readout window. Here, the signal echo is shifted in a first direction along the readout axis, for example, by a quarter. The extension segment then also lies in the first direction of the readout axis relative to the acquired raw data space.

[0012] For example, an iFT can be applied to the acquired raw data space to create a hybrid space, and an iFT can be applied to the extension segment to create a hybrid extension segment, with the hybrid space and the hybrid extension segment being combined to form an overall hybrid space. This overall hybrid space can be subjected to a Fourier transform along this axis to generate at least one MR image.

[0013] The use of the extension segment offers two possibilities. In the first, the acquisition time is reduced while maintaining the same resolution. In the second, the acquisition time remains the same, but the resolution is increased along the readout axis. In the first case, this is achieved by ensuring that at least one reconstructed MR image has the desired resolution. The acquisition time required for this desired resolution is reduced by not acquiring the entire raw data space necessary for that resolution. The raw data space required for the desired resolution corresponds to the extended raw data space with the extension segment.Since only a sub-area was captured in the extended raw data space, the acquisition time is reduced relative to an acquisition in which the entire extended raw data space with MR signals would be captured for the desired resolution.

[0014] In the second option with the increased resolution, the MR image is acquired with a defined acquisition time. Using the raw data acquired within this defined acquisition time, without using the extension segment, an MR reference image with a defined resolution can be reconstructed. The MR image reconstructed with iFT then has a higher resolution than the defined resolution when the extension segment is used, specifically a higher resolution in the direction of the readout axis.

[0015] Preferably, the additional zigzag trajectory is reconstructed using the GRAPPA method. However, other methods can be used. In general, Cartesian parallel imaging techniques can be employed, e.g., SENSE (SENSitivity Encoding).

[0016] The additional zigzag trajectory can be reconstructed such that the slope of the last segment of the captured zigzag trajectory in the direction of the positive phase encoding direction has the opposite sign compared to the slope of the last segment of the reconstructed zigzag trajectory in the direction of the positive phase encoding direction. In other words, this means that the captured trajectory and the reconstructed trajectory end and begin, respectively, at opposite corners of the raw data space in the phase encoding direction.

[0017] Furthermore, a magnetic resonance imaging system is provided with an image sequence control for the zigzag trajectory and a computing unit for the reconstruction of at least one MR image as described above and below in detail.

[0018] Furthermore, a computer program product is provided, comprising a program that can be directly loaded into the memory of a magnetic resonance imaging (MRI) system, with programmatic means to perform the steps described above and those described below. An electronically readable data carrier contains electronically readable control information stored on it, designed to execute the procedure described above when the data carrier is used in the MRI system.

[0019] The features set out above and those described below can be used not only in the corresponding explicitly set out combinations, but also in further combinations, unless explicitly stated otherwise, or individually, without leaving the scope of protection of the present invention.

[0020] The invention is explained in more detail below with reference to the accompanying drawings. Fig. Figure 1 schematically shows an MR system with which MR images can be calculated according to the invention. Fig. Figure 2 schematically shows a raw data space and the relationship between zigzag trajectories and their distance from each other. Fig. Figure 3 schematically shows a raw data space with a recorded and reconstructed trajectory, where the in Fig. The raw data space shown in Figure 3 cannot be reconstructed without artifacts. Fig. Figure 4 shows a raw data space which, by expansion in the readout direction, is filled with zigzag-like trajectories in such a way that even with undersampling an artifact-free reconstruction of an MR image is possible. Fig. Figure 5 shows an example of a raw data space that was fully captured, but where artifact-free reconstruction is not possible with undersampling. Fig. Figure 6 shows the recording of a raw data space, where, for a given recording time, the resolution can be increased by extending it in the readout direction. Fig. Figure 7 shows the recording of a raw data space, where the recording time can be reduced while maintaining the same resolution. Fig. Figure 8 schematically shows a flowchart with steps that are carried out for artifact-free reconstruction of MR images by extending the raw data space in the readout direction.

[0021] The present invention is explained in more detail below with reference to preferred embodiments and the drawings. In the figures, identical reference numerals denote identical or similar elements. Furthermore, the figures are schematic representations of various embodiments of the invention. Elements depicted in the figures are not necessarily shown to scale. Rather, the elements depicted in the figures are represented in such a way that their function and purpose are understandable to a person skilled in the art. The connections between functional units or other elements shown in the figures can also be implemented as indirect connections, which can be wireless or wired. Functional units can be implemented as hardware, software, or a combination of hardware and software.

[0022] In relation to Fig. Figure 1 describes an MRI system with which, as explained below, MRI images are acquired using echo-planar imaging and undersampling, whereby artifact-free images are generated despite the undersampling. The MRI system 10 has a magnet 11 for generating a polarization field B0, wherein a patient 13, arranged on a table 12, is moved into the magnet 11 to acquire spatially encoded magnetic resonance signals from the patient 13. The coils used for signal acquisition, such as a whole-body coil or local coils, are not shown for clarity. The invention can be applied in so-called parallel imaging, in which the MRI signals are acquired simultaneously with several local coils, a coil array of local coils.By applying high-frequency pulses and switching magnetic field gradients, the magnetization generated by the polarization field B0 can be deflected from its equilibrium position and spatially encoded, and the resulting magnetization is detected by the receiving coils. How MR images can be generated by applying RF pulses and switching magnetic field gradients in various combinations and sequences is generally known to those skilled in the art and will not be explained in detail here.

[0023] The MRI system also includes a control unit 20, which can be used to control the MRI system 10. The control unit 20 includes a gradient control unit 15 for controlling and switching the necessary magnetic field gradients. An RF control unit 14 is provided for controlling and generating the RF pulses for deflecting the magnetization. An image sequence control unit 16 controls the sequence of the magnetic field gradients and RF pulses and thus indirectly the gradient control unit 15 and the RF control unit 14. An operator can control the MRI system via an input unit 17, and MRI images and other information necessary for control can be displayed on a display unit 18. A computing unit 19 with at least one processor unit (not shown) is provided for controlling the various units in the control unit 20. Furthermore, a storage unit 21 is provided, in which, for example, program modules or...Programs can be stored which, when executed by the computing unit 19 or its processor unit, can control the operation of the MRI system. As explained below, the computing unit 19 is configured to calculate the MRI images from the acquired MRI signals according to the invention in such a way that undersampling is possible in the case of a zigzag trajectory, while simultaneously achieving a substantially artifact-free reconstruction with a satisfactory signal-to-noise ratio of the MRI images.

[0024] As explained in detail below, combining the interlaced Fourier transform (iFT) and parallel imaging requires acquiring the accelerated data in such a way that, after parallel image reconstruction, the conditions for the iFT are met. Specifically, the distance of the raw data unit to the center of the raw data space, i.e., k-space, satisfies the Nyquist theorem, or in other words, is less than or equal to Δky. This can be achieved in two ways. The first method involves acquiring only 2 / 3 of the raw data or k-space in the readout direction. This reduces the acquisition time, and after partial Fourier reconstruction in the readout direction of a readout axis, it yields the same resolution as originally desired for k-space. The k-space center then lies in a region where the Nyquist criterion is satisfied.Another way to achieve this is to keep the acquisition time constant and shift the acquired k-space by changing the pre-gradient in the direction of the readout axis. Since the position of the k-space center does not change in this case, the k-space is ultimately sampled asymmetrically. This again ensures that the k-space center lies in a region where the Nyquist criterion is satisfied.

[0025] In connection with the Fig. 2 and Fig. Section 3 explains under which conditions and trajectories artifact-free reconstruction of an MR image using iFT is possible or not. It is important to consider that artifact-free reconstruction depends particularly on the level of noise present in the generated MR image. In the following discussion, it is assumed that for a zigzag trajectory spacing of raw data segments, the Nyquist theorem is satisfied if the spacing in the direction of the phase-encoding axis is 1 Δky (one Δky). 2 Δky (two Δky) represents simple undersampling, etc. Fig. Figure 2 shows a raw data space 30 in the kx and ky directions, where kx is the readout axis and ky is the phase-encoding axis, and a recorded trajectory 31 fills this raw data space. The maximum spacing in the center of the raw data space is 1 Δky, while at the edge it is 2 Δky. This raw data set can be reconstructed using the iFT such that the signal-to-noise ratio is satisfactory and no significant artifacts appear, which is referred to below as artifact-free. Furthermore, a raw data zigzag trajectory 32 is shown as a dashed line, where the recording was made with an acceleration factor of 2, i.e., simple undersampling. As can be seen, the individual segments of the trajectory 32 have a spacing of 2 Δky in the center and a maximum spacing of 4 Δky at the edge. This data set 32 ​​cannot be reconstructed into an image without artifacts using iFT.

[0026] In Fig. Figure 3 shows a raw data space 40 with a recorded zigzag trajectory 41. As is known to those skilled in the art, the zigzag trajectory is created by a periodically switched readout gradient and a phase-encoding gradient that corresponds to the absolute value of the readout gradient with a correspondingly lower amplitude. The trajectory was recorded with an acceleration factor of 2, i.e., only 50% of the raw data was recorded, while the trajectory 42 was reconstructed using parallel imaging techniques, e.g., GRAPPA. This in Fig. The raw data set shown in section 3 cannot be reconstructed into an artifact-free image using iFT, as the maximum distance is 2 Δky in the center and 2 Δky at the edge. Another problem is that the intersection points in the central raw data area lead to artifacts in iFT. A further GRAPPA reconstruction would be necessary for an artifact-free reconstruction.

[0027] In Fig. Figure 4 now shows a raw data space in which the reconstruction is performed such that even with undersampling by a factor of 2, artifact-free reconstruction of MR images is possible. The raw data space has a captured raw data space 51 in which a zigzag trajectory 52 was recorded as shown. By shortening the dephasing pre-gradient relative to the gradient during the readout time, the signal echo occurs earlier and not in the geometric center of the captured raw data space, but in the raw data center 60. Furthermore, in the example shown, similar to in Fig. 3, a calculation of an additional zigzag trajectory 53, for example using GRAPPA reconstruction. In this captured raw data space, the maximum distance is 2 Δky at the center and at the edge, so that an artifact-free reconstruction, as in connection with Fig. 2 and Fig. As explained in section 3, this is not possible. According to the invention, the raw data space is now read out asymmetrically, and an extended raw data space can be created in which two-thirds comprise the acquired raw data space 51 and one-third an extension segment 55. The extended raw data space 50 now includes the acquired raw data space 51 and the extension segment 55. The raw data points in the extension segment 55 can be determined by so-called zero filling or with partial Fourier analysis. The point symmetry of the extended raw data space is used for this purpose. The raw data center 60 is used for this. The raw data points in the extension segment 55 are calculated by point reflection across the raw data center 60. When the extended raw data space 50 is now considered, it can be seen that the distance in the raw data center is only 1 Δky, while at the edge it is a maximum of 2 Δky.Furthermore, the intersection points of the individual zigzag trajectories are not located at the raw data center 60. This allows for artifact-free reconstruction with iFT using the extended raw data space 50. Therefore, it is possible to acquire a 2x accelerated MR image using echoplanar imaging with GRAPPA reconstruction.

[0028] A in Fig. The raw data space 61 shown in Figure 5 has the same extent in the direction of the readout axis kx as the recorded raw data space 51. Fig. 4. However, it can be seen that both in the raw data space or k-space center and at the edge, the maximum distance between the segments of a trajectory is 2 Δky, so that again an artifact-free reconstruction is not possible. Without the displacement of the signal echo, an artifact-free reconstruction is not possible.

[0029] In Fig. 6 is an embodiment similar to Fig. Figure 4 shows that, in the direction of readout, the space is again subdivided between a recorded raw data space 71 and an extension segment 75. The length of the recorded raw data space 71 with the recorded trajectory 73 and the reconstructed trajectory 72 corresponds in turn to the length of the raw data space 61 of Fig. 5. Through asymmetric signal readout and extension with point reflection around the center 76 in the kx direction, with filling in the extension segment 75, the raw data space is modified again such that a maximum distance of 1 Δky exists in the raw data center, while at the edge the maximum distance is 2 Δky. The extended raw data space 70 can thus be reconstructed without artifacts. The acquisition time is the same as for the acquisition of the raw data space from Fig. 5, however, the resolution in the kx direction has improved compared to an MR image that uses the raw data space of Fig. The calculation is increased by 5.

[0030] The present invention makes it possible to undersample raw data points using continuous phase encoding. With this continuous phase encoding, the raw data points are not located on a Cartesian grid, which is a prerequisite for using Cartesian parallel imaging such as GRAPPA. Due to the extension in the kx direction and the asymmetric acquisition, the k-space center is no longer located in the center of the acquired raw data space. This reduces the maximum spacing of the raw data points and also means that the intersection points of the individual trajectory segments are not located in the raw data center. It is now possible, for example, to perform an iFT only in the segment containing the k-space center to obtain an image. Additionally, an image can also be generated using iFT in the extension segment, which primarily contains high-frequency k-space information.This image could be combined with the first image to increase the resolution. Another possibility is not to increase the resolution, but to acquire less data in the kx direction. In this case, the acquisition time would be reduced compared to a fully sampled raw data space. The missing segment, the extension segment, can either be replaced by zero filling or by a partial Fourier reconstruction from the data of the two acquired segments, since the data are point-symmetric about the k-space center. This embodiment is described in [reference missing]. Fig. 7 described. The one in Fig. The entire raw data space 90 shown in Figure 7 has the same extent in the kx direction as the acquired raw data space 71. However, compared to the acquired raw data space 71, only the raw data space 91 is acquired, and the raw data in the extension segment 92 are reconstructed using point symmetry and point reflection at the raw data center 93. Compared to the acquired raw data space 71, the same resolution in the kx direction is now achieved as with an MR image that would be obtained using only the acquired raw data space 71, but the acquisition time is reduced in comparison. In comparison to the embodiment of Fig. 6 will be in Fig. 7 uses a shorter signal readout time for each signal echo, thereby reducing the overall recording time.

[0031] The procedure is in Fig. 8 summarized. The procedure starts in step S80, and in step S81, zigzag segments of a trajectory are recorded. After signal acquisition, the raw data set is divided into even and odd echoes. The signal is acquired such that the maximum signal intensity, the center of the signal echo, is shifted, preferably by a quarter of the acquired raw data space in the direction of the readout axis. In step S82, an additional trajectory is reconstructed for each of the even and odd acquired raw data points using the parallel acquisition technique. In step S83, the raw data space is extended in the readout direction, as described in connection with Fig. 4, Fig. 6 and Fig. As described in section 7. Subsequently, in step S84, an iFT is performed for each segment of the extended raw data space along the phase encoding direction. This means that, for example, an iFT is performed for segment 71 of Fig. 6 and separately for the extension segment 75 or for segment 51 of Fig.4 and separately for segment 55. After this step S84 and the iFT, a partially transformed raw data space and a partially transformed extension segment now exist. They are partially transformed because a Fourier transform was performed only along one direction, the phase-encoding direction. In step S85, the partially transformed raw data space and the partially transformed extension segment are used to apply an inverse Fourier transform back to the raw data space (inverse Fourier transform in the phase-encoding direction). This results in a hybrid space and a hybrid extension segment after step S85. In step S86, the elements generated in step S85—the hybrid space and the hybrid extension segment—are combined to form a complete hybrid space.After the data are combined in step S86, the two-dimensional Fourier transform can be performed in step S87, as is known, to generate at least one MR image. The procedure ends in step S88.

[0032] The method described above allows for the application of a technique known from parallel imaging to EPI data acquired along a zigzag trajectory. Here, iFT can be used, in which the even and odd echoes are phase-corrected individually so that they lie on Cartesian coordinates. This allows for higher acceleration factors and improved image quality. Furthermore, partial Fourier tracing in the kx direction can be performed, which further reduces the acquisition time.

Claims

[1] Method for echo-planar MR imaging in which MR signals are acquired in a raw data space with a zigzag trajectory, comprising the following steps: - recording MR raw data points with a sequence of readout gradients and phase encoding gradients such that a zigzag-like subsampled trajectory (52, 73) is recorded in a recorded raw data space (51, 71) such that the signal echoes are shifted by up to a quarter of the recorded raw data space in the direction of a readout axis, - reconstructing an additional zigzag trajectory (53, 72) in the acquired raw data space (51, 71) by means of parallel imaging methods based on the acquired zigzag trajectory, - Reconstructing at least one MR image based on an interlaced Fourier transform, iFT, wherein, before reconstructing the at least one MR image, the raw data points in the raw data space (50, 70) are extended such that an extension segment (55, 75) is created, wherein this extension is carried out using a point symmetry of the raw data space to a raw data center to create an extended raw data space that includes the acquired raw data space and the extension segment, wherein the raw data in the extension segment are calculated using a point reflection at the raw data center. [2] Method according to claim 1, characterized by that the signal echoes are shifted in a first direction of the readout axis by a quarter of the acquired raw data space, wherein the extension segment lies in the first direction of the readout axis relative to the acquired raw data space. [3] Method according to one of the preceding claims, characterized bythat the signal echoes are only shifted in the direction of the readout axis to such an extent that the distance between a detected trajectory and a reconstructed trajectory in the direction of a phase encoding axis is less than 1 Δky. [4] Method according to one of the preceding claims, characterized by that an iFT is applied to the acquired raw data space to create a hybrid space, wherein an iFT is applied to the extension segment to create a hybrid extension segment, wherein the hybrid space and the hybrid extension segment are combined to create an overall hybrid space, wherein the overall hybrid space is subjected to a Fourier transformation along the readout axis to create the at least one MR image. [5] Method according to one of the preceding claims, wherein the reconstruction of the additional zigzag trajectory (53, 72) is carried out using the GRAPPA method. [6] Method according to one of the preceding claims, characterized by that the additional zigzag trajectory (53, 72) is reconstructed such that a gradient of the last segment of the detected zigzag trajectory (52, 73) in the direction of the positive phase encoding direction has an opposite gradient to the last segment of the reconstructed zigzag trajectory (53, 72) in the direction of the positive phase encoding direction. [7] Method according to one of claims 1 to 6, characterized in that the at least one MR image has a desired resolution, wherein the acquisition time required for the acquisition of the at least one MR image with the desired resolution is shortened in that the raw data space required for the desired resolution is not completely acquired with raw data, but corresponds to the extended raw data space with the extension segment. [8] Method according to one of claims 1 to 6, characterized in that the at least one MR image is acquired in a defined acquisition time, wherein an MR reference image with a defined resolution can be reconstructed with the raw data acquired in the defined acquisition time without using the extension segment, wherein the at least one MR image reconstructed with the iFT using the extension segment has a higher resolution than the defined resolution. [9] Magnetic resonance system (10) which is designed to acquire MR signals in a raw data space with a zigzag trajectory during echo-planar MR imaging, the magnetic resonance system comprising: - an image sequence controller (16) configured to acquire MR raw data points with a sequence of readout gradients and phase encoding gradients such that a zigzag-like subsampled trajectory (52, 73) is acquired in an acquired raw data space (51, 71) such that the signal echoes are shifted by up to a quarter of the acquired raw data space in the direction of a readout axis, - a computing unit (19) which is designed - to reconstruct an additional zigzag trajectory (53, 72) in the acquired raw data space (51, 71) by means of parallel imaging methods based on the acquired zigzag trajectory, - to reconstruct at least one MR image on the basis of an interlaced Fourier transform, iFT, wherein, before reconstructing the at least one MR image, the raw data points in the raw data space (50, 70) are expanded such that an expansion segment (55, 75) is created, wherein this expansion is carried out using a point symmetry of the raw data space to a raw data center to create an expanded raw data space that includes the acquired raw data space and the expansion segment, wherein the raw data in the expansion segment are calculated using a point reflection at the raw data center. [10] Magnetic resonance system according to claim 9, wherein the computing unit is designed to carry out a method according to one of claims 2 to 8. [11] Computer program product comprising a program and being directly loadable into a memory of a magnetic resonance system, with program means for carrying out all steps of the method according to any one of claims 1-8 when the program is executed in the magnetic resonance system. [12] Electronically readable data carrier with electronically readable control information stored thereon, which is designed such that it carries out the method according to one of claims 1-8 when the data carrier (14) is used in a magnetic resonance system.

Images