Method for acquiring a magnetic resonance image with an extended field of view

Abstract

The invention relates to a method for acquiring a magnetic resonance image with a field of view (S) expanded in at least one direction using a magnetic resonance tomograph (1), the method comprising the following steps: (a) providing a directory correction function for the expanded field of view, (b) providing (24) at least one optimized position along the slice selection direction, wherein the static magnetic field has a high degree of homogeneity within the slice at the optimized position in at least one predetermined region (40a, 40b) in the periphery of the field of view (S), (c) providing (25) an optimized readout gradient, wherein the strength and polarity of the optimized readout gradient are selected such that distortions in the magnetic resonance image caused by inhomogeneities of the static magnetic field and distortions caused by non-linearities of the readout gradient field,are compensatingly superimposed within the at least one predetermined region (40a, 40b), (d) acquiring the magnetic resonance image at the optimized position along the slice selection direction with the optimized readout gradient, and (e) performing a directory correction of the acquired magnetic resonance image based on the directory correction function. The invention is also directed to a computer program and a magnetic resonance tomograph.

Description

[0001] Regardless of the grammatical gender of a particular term, persons with male, female or other gender identities are included.

[0002] Positron emission tomography (PET) is an important modality for functional imaging. Due to its low spatial resolution, it is often combined with another, higher-resolution imaging modality, such as CT (computed tomography) or magnetic resonance imaging (MRI), with both imaging modalities being combined in a single clinical whole-body system for simultaneous acquisition of CT or MRI and PET images. Integrated magnetic resonance (MR) / PET systems provide excellent tissue contrast and significantly reduce radiation dose. Therefore, MR / PET is increasingly being used for diagnostic purposes, as well as for the planning and monitoring of radiotherapy. Therefore, the accuracy and reproducibility of PET images acquired with an MR / PET system are of utmost importance for the planning and delivery of radiotherapy.

[0003] In quantitative positron emission tomography (PET) or in MRI-based radiotherapy planning, it is extremely important to accurately measure and compensate for the attenuation of photons absorbed in the tissue. If PET is combined with another imaging technique, it is common practice to first acquire an image of the body part using the other technique and then use this image to generate an attenuation correction map (AC map), which can then be used to correct the PET data. In MRI / PET systems or MRI-based methods for radiotherapy planning (such as synthetic CT), the calculation of the attenuation coefficient map is therefore based on a magnetic resonance image. However, magnetic resonance systems have an inherent limitation of the field of view, as the static magnetic field (B0 field) is sufficiently homogeneous only within a certain sensitive area in the magnetic resonance imaging (MRI) scanner.The same applies to magnetic gradient fields, which are also only truly linear within a central sensitive area. Outside and in the periphery of the sensitive area, magnetic field inhomogeneities and gradient nonlinearities lead to image distortions in MR imaging. Therefore, an MR-based AC map can be distorted at the edges of the field of view, and consequently, a PET image or synthetic CT image corrected with an MR-based AC map can be inaccurate. This effect is particularly pronounced at the edges of the field of view in the right-left direction, where the patient's arms are typically located.

[0004] A fully MR-based method for extending the conventional MR field-of-view (FOV), the so-called B0 homogenization using gradient enhancement (HUGE), has been proposed and is described in the following publications: Blumhagen JO, Ladebeck R, Fenchel M, Scheffler K. “MRbased field-of-view extension in MR / PET: B0 homogenization using gradient enhancement (HUGE)”. Magn Reson Med. 2013 Oct;70(4):1047-57. doi: 10.1002 / mrm.24555. Epub 2012 Nov 30. PMID: 23203976 and Blumhagen JO, Braun H, Ladebeck R, Fenchel M, Faul D, Scheffler K, Quick HH. “Field of view extension and truncation correction for MR-based human attenuation correction in simultaneous MR / PET imaging”. Med Phys. 2014 Feb;41(2):022303. doi: 10.1118 / 1.4861097. PMID: 24506641.

[0005] In this method, the amplitude of the readout gradient is optimized for both sides of the body during image acquisition to locally compensate for the B0 field inhomogeneity, thus achieving the lowest possible distortion and thus expanding the FOV. The method is based on the fact that B0 field inhomogeneities and gradient nonlinearities can be compensated for by choosing an appropriate readout gradient strength and polarity. This is possible because the gradient error field scales linearly with the gradient amplitude, while the B0 field inhomogeneity is independent of the gradient amplitude.

[0006] However, depending on the properties of the underlying B0 and gradient fields, the method proposed by Blumhagen et al. may still retain significant bias.

[0007] With regard to the state of the art, reference is made to the following two patent applications, which describe methods for imaging partial areas of an examination object in a magnetic resonance system: DE 10 2010 006 431 A1 DE 10 2010 044 520 A1

[0008] It is an object of the invention to provide a method for planning and performing an MRI acquisition with an extended field of view, which achieves a further improvement in image quality. A further goal is to provide a method that is less sensitive to manufacturing variations.

[0009] The invention solves this problem by a method for acquiring a magnetic resonance image according to claim 1, a computer program according to claim 12 and a magnetic resonance tomograph according to claim 15.

[0010] According to a first aspect, the invention is directed to a method for acquiring a magnetic resonance image with a field of view expanded in at least one direction, wherein the magnetic resonance image is acquired with a magnetic resonance tomograph capable of generating a static magnetic field and magnetic gradient fields in the expanded field of view, which fields extend in a slice selection direction, a readout direction, and at least one phase encoding direction, wherein an MR signal from signal-generating spins is acquired in the expanded field of view during the acquisition, and wherein the method comprises the following steps: (a) Providing a directory correction function for the extended field of view, (b) providing at least one optimized position along the slice selection direction, wherein the static magnetic field has a high degree of homogeneity within the slice at the optimized position in at least one predetermined region in the periphery of the field of view, wherein optionally the optimized position is simultaneously optimized to minimize artifacts due to signal-generating spins outside the slice, (c) providing an optimized readout gradient, wherein the strength and polarity of the optimized readout gradient are selected such that distortions in the magnetic resonance image caused by inhomogeneities of the static magnetic field and distortions caused by nonlinearities of the readout gradient field are superimposed in a compensating manner within the at least one predetermined range, (d) acquiring the magnetic resonance image at the optimized position along the slice selection direction with the optimized readout gradient, and (e) performing a directory correction of the acquired magnetic resonance image based on the directory correction function.

[0011] The invention is therefore based on a two-stage optimization method: in a first stage, an optimal position along the slice selection direction is used at which the static magnetic field within the selected slice has a particularly high homogeneity, in particular in at least one region in the periphery of the field of view, in other words in an edge region. In a second stage, an optimized readout gradient is provided for the selected slice at the optimal position such that distortions are compensated, especially in the edge region of the field of view. For this purpose, the aforementioned HUGE technique can be used, for example. The two articles by Blumhagen et al. cited in this regard are hereby incorporated into this disclosure. In both stages, the optimization is preferably carried out using a directory correction function extended to an expanded field of view.

[0012] The invention is based on the finding that the static magnetic field, particularly along the z-direction, but also in other directions, in most magnetic resonance imaging scanners does not exhibit the best homogeneity in the very peripheral regions at the central position z=0 (hereinafter also referred to as the isocenter). This is due to the design of the main magnets, whose static magnetic field (also referred to as the main magnetic field) usually exhibits many small fluctuations with relatively strong local maxima and minima at the edges of the field of view. On a coronal MR image, these B0 inhomogeneities become visible through local compressions and expansions at the edges of the field of view. At the position z=0 (isocenter), a strong expansion effect is typically found, while somewhat further away along the z-direction, i.e., parallel to the static magnetic field, the distortion takes the form of a compression.According to the invention, a position along the slice selection direction, typically along the z-direction, is selected at which expansion and compression at the edges of the field of view are balanced, so that a high homogeneity of the static magnetic field is present locally within this slice.

[0013] Optionally, a further criterion for selecting the optimal position in the slice selection direction is the minimization of signal contributions from spins outside the selected slice position. These can arise if the assignment of frequency to location in the slice selection direction is not clear, e.g., due to the non-optimal properties of the B0 field and the gradient field in the slice selection direction (particularly the z-direction). As a result, MR signals from spins located outside the selected slice can contribute to the signal and lead to artifacts. According to one embodiment, the position in the slice selection direction is therefore selected such that these unwanted signal contributions are minimized or avoided. In particular, the slice position is optimized such that homogeneity within the slice is maximized and unwanted signal contributions from outside the slice are minimized.

[0014] Then, in a second step, an optimized readout gradient is determined for this slice at the optimized slice position. Both the strength and polarity of the readout gradient are selected such that the distortions caused by inhomogeneities in the static magnetic field and the distortions caused by nonlinearity of the readout gradient field are superimposed within a predetermined range in a compensating manner. The predetermined range lies in the periphery of the field of view, in other words, at the edge of the extended field of view. According to one embodiment, the properties of the slice selection gradient, in particular its polarity and strength or amplitude, are also selected such that signal contributions from spins outside the selected slice position are minimized.

[0015] An “extended field of view” is understood here to be a field of view that is enlarged compared to the normal field of view of a magnetic resonance imaging scanner, also known as an MR scanner. In particular, a “normal” field of view is understood to be the maximum, conventional field of view of the MR scanner. This has, for example, the approximate shape of a sphere with its center at the isocenter. The sphere typically has a diameter of approximately 50-55 cm. The extended field of view is then enlarged at least along one axis by, for example, 2-7 cm, preferably 4-6 cm, and preferably by half at both peripheral ends. The field of view is preferably enlarged at least in the x-direction and / or in the y-direction. In one embodiment, the extended field of view has the approximate shape of a sphere with a diameter of 55 to 62, preferably 58-60 cm.In some embodiments, the predetermined area and / or periphery of the field of view lies outside the "normal field of view" but within the extended field of view. The periphery is the area by which the extended field of view is expanded compared to the normal field of view.

[0016] According to the invention, a directory correction of the acquired magnetic resonance image is performed. This occurs, in particular, on the basis of a directory correction function for the expanded field of view. This means that the MR image in the periphery is also directory-corrected. Such a directory correction function can mathematically correct previously reconstructed MR images so that they better reflect the scanned object geometry. Such directory correction functions often utilize the spatial distribution of the gradient fields, also referred to here as gradient field maps, and in particular the nonlinear field components, which in turn can be measured or calculated from the geometry of the gradient coils of the MR scanner. A suitable method for directory correction is disclosed, for example, in US Pat. No. 8,054,079 B2.Such directory correction functions are routinely present in clinical MR devices and are automatically used for directory correction of MR images in the (normal) field of view of the MR device.

[0017] According to the invention, a directory correction function for the extended field of view is provided. This can be obtained in various ways. According to a first method, extended gradient field maps are used. The extended gradient field maps extend to the extended field of view. The extended gradient field maps can be obtained in various ways. For example, they can be simulated or calculated from the geometry of the gradient coils of the MR device. According to a further embodiment, the extended gradient field maps are extrapolated from maps of the gradient fields for the normal field of view. For example, the gradient fields can be extrapolated from a sphere with a diameter of 55 cm to a sphere with a diameter of 60 cm. The image data of the magnetic resonance image can then be directory-corrected using the coefficients that model these fields.According to a second method, the directory correction function routinely provided in the MR scanner is extended or extrapolated to the expanded field of view. This can be done by extrapolating the coefficients of the directory correction function to the expanded field of view.

[0018] The HUGE method does not perform directory correction because it assumes that B0 inhomogeneities and gradient nonlinearities can compensate one-to-one. However, this assumption is only valid if the two are of the same magnitude and have a similar spatial distribution, which is often not the case. Therefore, the invention also applies directory correction, specifically in the expanded, enlarged field of view. The optimization of the readout gradient according to step (c) is performed additionally. Experiments have shown that although step (c) alone does not produce optimal results, excellent distortion correction can be achieved in combination with directory correction.

[0019] The optimized position and the optimized readout gradient are then used to acquire a magnetic resonance image. This image delivers improved quality compared to the HUGE technique, and in particular, improved geometric accuracy in the periphery of the field of view. Due to the dual optimization of both the shift in the slice selection direction and the amplitude of the readout gradient, the invention makes the HUGE method more stable and less susceptible to image artifacts. The invention makes it possible to expand the field of view of MR images by several cm, in particular by 2-7 cm, preferably 4-6 cm, compared to previously used fields of view. This allows, for example, a field of view up to 60 cm wide to be achieved in a clinical magnetic resonance imaging system with a tube diameter of 60 cm, instead of the previous 55 cm.

[0020] According to the invention, several optimization steps are applied, which optimize the slice position on the one hand, and the amplitude and polarity of the slice readout gradient on the other, as well as optionally the amplitude and polarity of the slice selection gradient. This makes it possible to reduce image artifacts, minimize distortions, and improve the stability of the method.

[0021] A magnetic resonance image acquired in this way is preferably used to determine AC maps for the correction of PET images, and in particular MR / PET images. According to a further embodiment, it can also be used to generate MR-based synthetic CT images. To generate AC maps or synthetic CT images, the magnetic resonance images according to the invention can advantageously be combined with other images, for example, with a conventionally acquired VIBE-DIXON image, which can be used to differentiate between the various tissues (fat, water, bone).

[0022] To acquire a magnetic resonance image with a larger field of view, including in the slice selection direction, one slice at a time can be acquired at the determined optimal position (acquisition and acquisition are used synonymously here). Between or during acquisitions, the patient can be moved on the patient couch through the sensitive area along the slice selection direction, with a stack of images being acquired along the length of the body. Preferably, all images are therefore acquired at the optimized position in the slice selection direction. To cover a specific body region, the position of the patient couch changes between individual acquisitions, or it moves continuously during the acquisition of multiple slices. The stack of images can cover part of the patient's anatomy, e.g., the head or torso, or even a whole-body scan.These images can then be combined to form a three-dimensional (3D) image from which an AC map can be determined.

[0023] The required optimizations are preferably performed only once for a specific type of magnetic resonance imaging scanner in order to determine the hardware-specific position along the slice selection direction that is optimized for this type. The optimized readout gradient also depends on the gradient coils and the main magnet used and is therefore preferably determined only once for a specific type of magnetic resonance imaging scanner. For the acquisition of MR images on a specific MR scanner, the optimal position in the slice selection direction is then known and can be stored in the MR scanner or retrieved from another data storage device, e.g., remote storage, a cloud, another PC, USB stick, etc. The optimal strength and amplitude of the readout gradient and, optionally, the slice readout gradient can also be stored in the MR scanner or retrieved from a data storage device.Alternatively, the optimizations can be repeated at intervals to take into account possible drifts in the magnetic fields.

[0024] A magnetic resonance image with a field of view extended in at least one direction is understood to be a two-dimensional (2D) or three-dimensional (3D) MR image which has a particularly large field of view, also called the field of view (FOV). The field of view is preferably particularly large in the readout direction; in particular, it has at least one predetermined region at the edges in this direction. The predetermined region preferably lies outside the "normal" field of view, which could be acquired by conventional MR imaging without major distortions. The periphery is understood here to be an outer region or edge region of a field of view. The predetermined region lies, in particular, at the edge of the field of view in one direction, this direction preferably being the readout direction. In other embodiments, however, the direction can also be the phase encoding direction.The slice selection direction is preferably the z-direction, but the invention is not limited to this. Rather, depending on the desired field of view, it may also be advantageous to provide the slice selection direction in the x-direction or the y-direction. Preferably, two predetermined regions are present, one on each side of the field of view. This has the advantage that the edge regions on both sides can be captured with improved quality. If the readout direction lies along a left-right direction, the two predetermined regions are arranged, for example, in the region of the patient's outer shoulders or arms. Preferably, the predetermined regions each form a strip at the edge of the field of view.

[0025] The determination of the optimized position along the slice selection direction and / or the optimized readout gradient can be carried out using an at least partially computer-implemented method. In particular, it is possible to calculate the homogeneity of the static magnetic field in the at least one predetermined region within a selected slice from the map of the static magnetic field. For example, a homogeneity index can be determined by summing the squares of the deviation from the mean magnetic field at a series of specified positions within the predetermined region or within the entire selected slice. This index can be calculated for slices at different positions along the slice selection direction. The slice with the smallest index then has the greatest homogeneity, and the corresponding position along the slice selection direction is the optimized position.

[0026] Alternatively, the optimized position can also be determined experimentally. In this method, for example, MR images of a geometrically uniform phantom are acquired at various positions along the slice selection direction. By quantifying the distortions, the degree of inhomogeneity of the static magnetic field can be determined, and the best image or its slice position can be identified. In particular, the slice position is selected at which, on the one hand, distortions are minimized and, on the other hand, artifacts from signal-generating spins outside the slice are minimized. Since the optimized position is preferably determined only once for a type of magnetic resonance scanner and is then simply retrieved when the method is carried out, the optimized position can also be determined experimentally. MR images are preferably evaluated for this purpose.

[0027] In a further step, an optimized readout gradient is determined, or a previously determined optimized readout gradient is provided for compensation, which is used when acquiring the magnetic resonance image at the optimized position determined in step (b). The strength and polarity of the optimized readout gradient were selected such that distortions in the magnetic resonance image caused by inhomogeneities in the static magnetic field and distortions caused by nonlinearities in the readout gradient field are superimposed in a compensating manner within the at least one predetermined range.

[0028] According to one possible embodiment, the optimized readout gradient is selected according to the method of Blumhagen et al. in Magn. Reson. Med. 70:1047-1057 (2013).

[0029] According to a further embodiment, the optimized readout gradient is determined once experimentally for a specific device type and is then made available for acquisitions according to the invention. In particular, the optimization can be performed by acquiring and evaluating MR images with different gradient strengths and / or polarities.

[0030] According to a preferred embodiment, the at least one optimized position in step (b) and / or the optimized strength and polarity of the slice selection gradient was / were determined by evaluating MR images that were corrected using the directory correction function for the extended field of view. In other words, to determine the optimized position and / or the optimized slice selection gradient, those MR images that have already been directory-corrected are evaluated. Since the periphery of the field of view is important here, the directory correction function for the extended field of view is preferably used for this purpose.

[0031] This first corrects the gradient error mathematically through the directory correction. Any remaining distortions / artifacts are then minimized by optimizing the slice position and / or the readout gradient.

[0032] According to one embodiment, the readout direction is aligned along a left-to-right direction of a subject lying in the magnetic resonance imaging scanner, wherein the slice selection direction is preferably aligned along a z-direction of the magnetic resonance imaging scanner. This orientation has the advantage that the field of view is extended in the left-to-right direction, in which many patients are wider. In particular, when imaging the torso, it is necessary to take the patient's arms lying next to the body into account when calculating the AC map. Furthermore, it is advantageous if the slice selection direction is in the z-direction, since a whole-body scan of a patient can then be acquired by advancing the patient bed in the z-direction and repeatedly acquiring an MR image at the optimized slice position. This is particularly important when planning radiation therapy.

[0033] According to a preferred embodiment, an optimized position along the slice selection direction was determined in step (b) such that the static magnetic field has a high degree of homogeneity within the slice at the optimized position in two predetermined regions, wherein the two regions are positioned on either side of the field of view in the readout direction. Thus, in this embodiment, an optimized position is determined that is the same for the predetermined regions on either side of the field of view. In other words, a single optimized position is determined for two predetermined regions. This simplifies implementation and generally yields very good results because the main magnet is typically symmetrical.

[0034] According to an alternative embodiment, two optimized positions along the slice selection direction were determined such that the static magnetic field has a high degree of homogeneity within a first predetermined region within a first slice at a first optimized position and a high degree of homogeneity and minimized artifacts within a second predetermined region within a second slice at a second optimized position, wherein the first and second predetermined regions are arranged on both sides of the field of view in the readout direction. In this embodiment, optimal slice positions are determined separately for each of the two sides. This allows for particularly high image quality to be achieved.

[0035] According to one embodiment, the step of determining the at least one optimized position in step (b) is performed by analyzing another magnetic resonance image acquired in a plane aligned along the slice selection and readout gradient directions, and determining the position along the slice selection direction at which distortion of the magnetic resonance image at the periphery of the field of view in the readout direction is minimal. This allows the optimal slice position to be determined, at which neither expansion nor compression of the image occurs. In particular, this can be done by analyzing a magnetic resonance image of a phantom with a known geometry.

[0036] According to one embodiment, step (c) of determining an optimized readout gradient is performed using B0 homogenization with gradient enhancement (HUGE), as described in the two cited articles by Blumhagen et al. This allows for a significant reduction in distortions at the edge of the field of view. Typically, the optimal polarity of the readout gradient is opposite for both sides of the magnet.

[0037] According to one embodiment, the optimized readout gradient in step (c) has been determined separately for both sides. This means that when acquiring the magnetic resonance image, the two sides must be acquired separately. For example, one side can be scanned first, then the other, and the images located at the same slice position are then combined in a post-processing step.

[0038] According to one embodiment, the method comprises a further step of providing an optimized slice selection gradient given a predetermined slice thickness to be used in acquiring the magnetic resonance image, wherein the strength and polarity of the slice selection gradient are optimized such that no interfering signal from spins outside of this slice is introduced into the slice. This also optimizes a third parameter, namely the amplitude and polarity of the slice selection gradient, typically the gradient in the z-direction. The amplitude of this gradient depends on both the slice thickness and the duration of the radiofrequency (RF) pulse. For the desired slice thickness, the RF pulse duration should therefore be adjusted to achieve the desired gradient amplitude.Optimizing the gradient amplitude and polarity in the z-direction is a further improvement of the proposed method, as this parameter influences the position in z where the compression and expansion effects described above occur. This means that both the B0 inhomogeneities and the gradient amplitude and polarity in the z-direction determine the optimal position in the z-direction. Furthermore, the gradient nonlinearities can cause unwanted signal distortion in the outer parts of the sensitive region, which can be reduced by increasing the amplitude of the slice selection gradient. Before determining the optimal position in the z-direction, the distortion of a phantom of known geometry can be measured with both polarities of the slice selection gradient, as described above.The polarity with the least distortion can then be taken, and based on this, the optimal position in the z-direction and the readout gradient are determined.

[0039] The acquisition of the magnetic resonance image of a slice at the at least one optimized position along the slice selection direction and using the optimized readout gradient is preferably carried out using a spin-echo sequence. With this method, the optimized slice position determined at an earlier time as well as the optimized readout gradient can be used to acquire a magnetic resonance image. This can be a 2D image or a 3D image. Preferably, a series of images, in particular 2D images, is acquired simultaneously while the subject is moved in the slice selection direction. This allows a stack of magnetic resonance images with an expanded field of view to be acquired. The subject is preferably a human or animal, for example a patient. The field of view can depict any anatomy in the subject.The sequence is preferably of the spin-echo type, i.e., a sequence with an excitation pulse followed by one or more refocusing pulses. It is preferably a turbo spin-echo sequence.

[0040] According to one embodiment, a magnetic resonance image is acquired separately for each of the two predetermined regions at the edge of the field of view on both sides in the readout direction. This makes it possible to use an optimized readout gradient, in particular with optimized polarity, for each side.

[0041] Preferably, the acquired magnetic resonance image is further processed using directory correction methods to further improve the results. These are typically applied during image reconstruction, toward the end of the reconstruction chain. These directory correction methods are typically based on gradient field maps. Compared to conventional maps, these maps are expanded to cover the expanded field of view. The distortions resulting from these maps can be predicted and corrected accordingly.

[0042] According to one embodiment, a series of magnetic resonance images is acquired at the at least one optimized position along the slice selection direction, with the subject moving forward along the slice selection direction between the acquisition of individual images of the series. This allows a larger area of ​​the body or the entire body of the subject to be scanned. The series of magnetic resonance images can be combined into an overall image, from which an AC map can be derived. Such an AC map is preferably used to correct PET images or to generate MR-based synthetic CT images for radiation treatment planning. Accordingly, the invention is also directed to the use of a magnetic resonance image generated in this way with an expanded field of view to generate an AC map of the subject.

[0043] According to a further aspect, the invention is directed to a computer program with program code that causes a computer to execute the optimization steps described here for planning the acquisition of a magnetic resonance image. In particular, this method can be computer-implemented to determine the optimized position and the optimized readout gradient from the maps of the static magnetic field and the readout gradient field. The determination can be carried out, for example, using numerical optimization methods. According to another embodiment, the determination of the optimized position and / or the determination of an optimized readout gradient can also be carried out by a trained algorithm, in particular a trained deep learning model such as a neural network. The computer can be any data processing device, for example a CPU, GPU, PC, laptop, cloud computer, or a mobile device.

[0044] According to a further aspect, the invention is directed to a computer program with program code that causes a magnetic resonance imaging scanner to perform the method described herein. The magnetic resonance imaging scanner can, in particular, be a combined MR / PET scanner. This allows the acquired magnetic resonance image to be used to determine AC maps and to correct the PET images.

[0045] The invention is also directed to a non-transitory computer-readable medium containing a computer program described herein. The computer-readable medium is, in particular, a digital storage medium, for example, an optical, magnetic, or semiconductor storage medium. In particular, it may be a hard disk, cloud storage, a floppy disk, CD-ROM, SD card, SSD card, or other memory stick.

[0046] Finally, the invention is also directed to a magnetic resonance imaging scanner configured to carry out the method described herein, wherein the magnetic resonance imaging scanner comprises: a main magnet generating a static magnetic field; gradient coils capable of generating magnetic gradient fields extending in a slice selection direction, a readout direction, and at least one phase encoding direction; a radio-frequency controller configured to drive a radio-frequency coil, a gradient controller configured to control the gradient coils; and a control unit configured to control the radio-frequency controller and the gradient controller. The magnetic resonance imaging scanner may be any commercial MR scanner, for example, a 0.5 T, 1.5 T, 3 T, or 7 T device.The magnetic resonance imaging device preferably also has a ring of PET sensors, so it is preferably a combined MR / PET device.

[0047] All advantages and features described with respect to the methods according to the invention also apply to the computer program, the computer-readable medium and the magnetic resonance imaging device and vice versa.

[0048] The invention will now be explained in more detail using exemplary embodiments with reference to the accompanying drawings. In the drawings: Fig. 1 a magnetic resonance imaging apparatus according to an embodiment of the invention; Fig. 2 an exemplary map of the static magnetic field in a magnetic resonance imaging scanner in the xz plane; Fig. 3 an axial magnetic resonance image of a phantom at position z = 0: Fig. 4 a coronal magnetic resonance image of the phantom of the Fig. 3; Fig. 5 is a schematic flow diagram of a method according to an embodiment of the invention; Fig. 6 shows an axial magnetic resonance image of a phantom acquired using a method according to an embodiment of the invention; and Fig. 7 an axial magnetic resonance image of the same phantom as in Fig. 6, recorded using a conventional method.

[0049] Fig. 1 schematically shows a magnetic resonance imaging system, also called an MR device 1, according to one embodiment of the invention 1. The MR device 1 has a main magnet 3 that generates the constant magnetic field within the sensitive area 2, a gradient coil arrangement 5 that generates gradient fields, one or more radio-frequency coils 7 for transmitting and receiving radio-frequency signals (RF pulses), and a control computer 9 that is configured to carry out the method according to the invention. The radio-frequency coils 7 can be configured as a multi-channel coil arrangement with at least two coils. The MR device is preferably an MR / PET device. In this case, PET detectors (not shown) are also arranged in a ring around the sensitive area 2.

[0050] To acquire MR data from an examination subject U, for example, a patient or a phantom, the examination subject U is placed on a couch B in the sensitive area 2 of the MR device. The slice S shown is an example of a 2D slice from which MR data can be acquired using a method according to an embodiment of the invention. The subject U can be moved along the z-direction on the couch B so that the slice S lies at different heights of the body and thus a stack of magnetic resonance images is acquired. The control computer 9 controls the MR device 1 and can control the gradient coil arrangement 5 with a gradient controller 5' and the RF coil 7 with an RF transmit / receive controller 7'. The control computer 9 also has a processor 15 that can perform the optimization steps described here.A control unit 13 of the control computer 9 is configured to perform all controls and calculations required for the acquisitions. Intermediate and final results required for this purpose or determined during the process can be stored in a memory 11 of the control computer 9. In particular, the optimized position in the slice selection direction and the optimized readout gradient for this MR device can be stored there and retrieved for the acquisition of the MR image. A user can enter control commands and / or view displayed results, e.g., image data, via an input / output interface (I / O). A non-transitory data carrier 4 can be loaded into the control computer 9. This data carrier can be encoded with programming instructions (program code) that cause the control computer 9 and its various functional units described above to execute one or all embodiments of the method according to the invention.

[0051] Fig. Figure 2 schematically shows the course of the main magnetic field of a magnetic resonance imaging scanner 1 in the sensitive area 2 in the xz plane. The strength of the main magnetic field is indicated by contour lines. It is clearly evident that strong fluctuations in the main magnetic field are present at the edges of the sensitive area 2. At z = 0, there is a local maximum of the static magnetic field, which can cause significant distortion, as shown in the Fig. 3 and Fig. 4. Furthermore, Fig. 2 also the field of view of the layer S and the predetermined regions 40a, 40b in the periphery of the field of view of the layer S. According to the invention, the position of this layer S is optimized in the z-direction.

[0052] Fig. Figure 3 shows a magnetic resonance image of an axial slice through a special phantom arranged in the sensitive area 2. The phantom comprises, on the one hand, a section 30 with a grid-like arrangement of fluid-filled upright tubes. To the right of this section is an arm phantom 32. This arm phantom 32 comprises a series of parallel tubes within a larger fluid-filled tube. Fig. Figure 4 shows the same phantom in a coronal section. Fig. Figure 4 clearly shows that strong distortions are present at the edge of the sensitive area 2. These distortions vary greatly in the z-direction, leading to a strong expansion of the image at z = 0 and a strong compression just a few centimeters away in the z-direction. According to one embodiment of the invention, a position in the z-direction is therefore selected at which optimal homogeneity of the static magnetic field is present precisely in this peripheral area.

[0053] An example of the method according to the invention is shown in Fig. 5. First, in step 22, a directory correction function for the expanded field of view is received. This can be simulated or generated from a map of the readout gradient field 21 generated by the magnetic resonance tomograph (optional step). In step 24, an optimized position along the slice selection direction with a high degree of homogeneity in the predetermined region(s) is determined. In step 25, an optimized readout gradient is determined for the predetermined region(s) within a slice at this optimized position. These optimized values ​​are stored in a memory 4 or 11. At a later time, in step 26, a magnetic resonance image is acquired on a patient using the optimized values ​​from the memory 4, 11. The magnetic resonance image is directory-corrected in step 28. An AC map, for example, can then be created from such an image.This AC card can be used for attenuation correction of PET images acquired on this device.

[0054] The Fig. 6 and Fig. 7 show a comparison between the proposed technique ( Fig. 6) and the conventional HUGE method ( Fig. 7). An axial magnetic resonance image of the arm phantom 32 is shown. It can be seen that the proposed method reduces both image artifacts and distortions compared to the conventional HUGE method.

Claims

[1] Method for acquiring a magnetic resonance image with a field of view (S) extended in at least one direction, using a magnetic resonance tomograph (1) which is capable of generating a static magnetic field and magnetic gradient fields in the extended field of view (S) which extend in a slice selection direction, a readout direction and at least one phase encoding direction, wherein an MR signal from signal-generating spins in the extended field of view is acquired during the acquisition, and wherein the method comprises the following steps: (a) Providing a directory correction function for the extended field of view, (b) providing (24) at least one optimized position along the slice selection direction, wherein the static magnetic field has a high degree of homogeneity within the slice at the optimized position in at least one predetermined region (40a, 40b) in the periphery of the field of view (S), (c) providing (25) an optimized readout gradient, wherein the strength and polarity of the optimized readout gradient are selected such that distortions in the magnetic resonance image caused by inhomogeneities of the static magnetic field and distortions caused by non-linearities of the readout gradient field are superimposed in a compensating manner within the at least one predetermined region (40a, 40b), (d) acquiring the magnetic resonance image at the optimized position along the slice selection direction with the optimized readout gradient, and (e) performing a directory correction of the acquired magnetic resonance image based on the directory correction function. [2] The method according to claim 1, wherein the readout direction is aligned along a left-right direction of a subject lying in the magnetic resonance imaging apparatus, and wherein the slice selection direction is preferably aligned along a Z-direction of the magnetic resonance imaging apparatus. [3] Method according to one of claims 1 to 2, wherein the optimized position along the slice selection direction in step (b) has been selected such that the static magnetic field has a high degree of homogeneity within the slice at the optimized position in two predetermined regions (40a, 40b), the two regions being positioned on either side of the field of view (S) in the readout direction. [4] Method according to one of claims 1 to 2, wherein two optimized positions along the layer selection direction have been selected such that the static magnetic field has a high degree of homogeneity within a first predetermined area (40a, 40b) within a first layer at a first optimized position and a high degree of homogeneity within a second predetermined area (40a, 40b) within a second layer at a second optimized position, wherein the first and second predetermined areas (40a, 40b) are arranged on both sides of the field of view (S) in the readout direction. [5] Method according to one of the preceding claims, wherein the optimized position along the slice selection direction is further optimized to minimize artifacts due to signal-generating spins outside the slice. [6] Method according to one of the preceding claims, wherein the at least one optimized position in step (b) was determined by analyzing another magnetic resonance image acquired in a plane aligned along the slice selection and readout gradient directions and by determining the position along the slice selection direction at which distortion of the magnetic resonance image at the periphery of the field of view in the readout direction is minimal. [7] A method according to any one of the preceding claims, wherein the optimized readout gradient was determined using B0 homogenization with gradient enhancement (HUGE). [8] A method according to any one of the preceding claims, comprising a further step of providing an optimized slice selection gradient in view of a predetermined slice thickness used in acquiring the magnetic resonance image, the strength and polarity of the slice selection gradient being optimized to minimize interfering signals from spins outside that slice. [9] Method according to one of the preceding claims, wherein the at least one optimized position in step (b) and / or the optimized strength and polarity of the slice selection gradient was / were determined by evaluating magnetic resonance images which were corrected by means of the directory correction function for the extended field of view. [10] Method according to one of claims 3 to 9, wherein a magnetic resonance image is acquired separately for each of the two predetermined regions (40a, 40b) at the edge of the field of view (S) on both sides in the readout direction. [11] The method of any one of claims 2 to 10, wherein a series of magnetic resonance images is acquired at the at least one optimized position along the slice selection direction, wherein the subject is moved forward along the slice selection direction between the acquisition of individual images of the series. [12] Use of a magnetic resonance image with an extended field of view (S) obtained by the method of any one of claims 1 to 11 to generate an AC map of the subject. [13] Computer program with program code which causes a magnetic resonance imaging device (1) to carry out the method according to one of claims 1 to 11. [14] A non-transitory computer-readable medium containing a computer program according to claim 13. [15] A magnetic resonance imaging apparatus configured to carry out the method according to any one of claims 1 to 11, wherein the magnetic resonance imaging apparatus (1) comprises: - a main magnet (2) which generates a static magnetic field, - gradient coils (5) capable of generating magnetic gradient fields extending in a slice selection direction, a readout direction and at least one phase encoding direction, - a gradient control (5') configured to control the gradient coils (5), - a high-frequency control (7') configured to control a high-frequency coil (7), and - a control unit (13) configured to control the radio frequency control (7') and the gradient control (5').

Images