MR imaging method
The invention relates to an MR method for generating an MR image of an object situated in an examination volume by using an MR apparatus, wherein MR signal data is first acquired by partial sampling of the spatial frequency space associated with the examination volume and wherein an intermediate MR image is then reconstructed from this MR signal data with account taken of the non-idealities of the MR apparatus.
The invention also relates to an MR apparatus for performing the method and to a computer program for an apparatus of this kind.
In MR imaging, the location of the magnetization of nuclei within the examination volume is usually performed by means of temporally variable, spatially inhomogeneous magnetic fields (magnetic field gradients). The MR signals used for reconstructing the images are recorded, in the time domain, in the form of a voltage that is induced in a high-frequency coil arrangement surrounding the examination volume under the influence of a suitable sequence of magnetic-field gradient pulses and high-frequency pulses. The actual reconstruction of the image is then generally performed by Fourier transformation of the time-related signals. The sampling of the spatial frequency space (what is called the k- space), i.e. the space by which the volumetric region to be imaged, or field of view (FOV), and the image resolution are deterarined, is preset by the number of magnetic-field gradient pulses and high-frequency pulses used, the interval of time between them and their duration and strength. Requirements that have to be met by the image size and the image resolution preset the number of phase encoding steps in the sampling of the spatial frequency space and hence the duration of the imaging sequence. One of the principal problems with MR imaging arises as a direct result of this because it generally takes an undesirably long time to acquire an image of the complete FOV of a resolution that is adequate for diagnostic purposes.
A large number of further technical developments in the field of MR imaging are aimed at shortening the image acquisition times. Developments to the apparatus that make it possible for the magnetic field gradients to be switched as fast as possible have today reached the limits of what is feasible in technical terms and of what is physiologically acceptable for the patient. However, for a large number of applications the acquisition times are still too long.
It is known to be possible in principle for a complete MR image to be reconstructed from MR signal data that is acquired by only partial sampling of the spatial frequency space associated with the examination volume. The MR signal data exists in the form of complex values after demodulation, and because of the Hermitian symmetry of this data it is enough, in principle, for only half the spatial frequency space associated with the examination volume to be sampled. The image acquisition time can therefore be reduced by half in this way. There is, however, a problem in this case in that the Hermitian symmetry of the MR signal data is almost always disrupted in practice due to non-idealities in the MR apparatus used. These non-idealities become apparent as phase errors in the MR signal data that is acquired. Phase errors of this kind come into being as a result of, for example, inhomogeneities in the static magnetic field of the kind that are unavoidable, particularly in medical MR imaging, due to the differing individual magnetic susceptibility characteristics of the patients who are examined. Phase errors also come into being as a result of so-called Maxwell fields in the examination volume, or else as a result of eddy currents that are induced in the metallic structures of the MR apparatus when the magnetic field gradients are switched.
Known from US 4,912,413 is an MR method for reconstructing an MR image from MR signals acquired by partial sampling of the spatial frequency space. There are approaches in this known method to ways of taking account of the problems described above that result from the disruption of the Hermitian symmetry in the MR signals that are acquired. There is, however, a disadvantage in the known method in that account is not taken of all the phase errors that occur in practice. In particular, the known method is unable to compensate for phase errors that are caused by position-dependent inhomogeneities in the static magnetic field within the examination volume. Inhomogeneities of this kind have a particularly adverse effect on the quality of the reconstructed MR image when the sampling of the spatial frequency space takes place in a non-Cartesian fashion, such as spirally, for example.
Against this background, it is an object of the invention to provide an MR method that allows MR images to be reconstructed from MR signal data acquired by partial sampling of the spatial frequency space, wherein the quality of the reconstructed MR image is to be improved in comparison with the MR images that can be reconstructed by conventional methods.
This object is achieved by an MR method as defined in claim 1.
In accordance with the invention, an intermediate MR image is first reconstructed from the MR signal data acquired by partial sampling of the spatial frequency space, this being done with account taken of the non-idealities of the MR apparatus. The possibility exists of, for example, the MR signal data that is acquired being subjected directly to a Fourier transformation for the purposes of reconstruction, with the signal being set to zero, prior to the Fourier transformation, in the unsampled regions of the spatial frequency space. Where required, it may be necessary for an apodization of the MR signal data to take place in some suitable way before the Fourier transformation. Alternatively, the MR signal data may be supplemented in the unsampled regions of the spatial frequency space, from the MR signal data that was actually acquired, by taking advantage of the Hermitian symmetry that, at least in theory, exists. This necessarily produces errors in the intermediate MR image that is reconstructed. These, however, are eliminated in the further course of the method.
In the method according to the invention, account is taken of the non-idealities of the MR apparatus used in a known manner when reconstructing the intermediate MR image. What is termed a conjugate-phase reconstruction may, for example, take place to take account of local inhomogeneities in the static magnetic field. The image errors caused by eddy currents may be compensated for in the reconstruction by convoluting the MR signal data with a suitable correcting function.
The next step is, in accordance with the invention, for the reconstructed intermediate MR image to be used to synthesize therefrom the MR signal data in the unsampled regions of the spatial frequency space. The idea underlying this procedure is for the intermediate MR image to be considered a first approximation of the definitive MR image. The missing MR signal data is synthesized by calculating it from the intermediate MR image by, as it were, reverse reconstruction. Account also has to be taken of the non- idealities of the MR apparatus in this reverse reconstruction. The procedure followed for this purpose is advantageously similar to that followed in reconstructing the intermediate MR image from the MR signal data.
The quality of the intermediate MR image can be improved iteratively by repeating the steps of the method that are described above one or more times, until a presettable cessation criterion is reached. For this purpose, an improved intermediate MR image is reconstructed each time from the data that is supplemented with the synthetic MR signal data. From the intermediate MR image that has been improved in this way, synthetic MR signal data can then be calculated again each time to supplement the MR signal data originally acquired.
The iteration ceases, for example, when the differences between the intermediate MR images that are reconstructed in succession drop below a presettable threshold value.
If, in the method according to the invention, account is taken of the non- idealities of the MR apparatus by performing phase correction both in the reconstruction of the intermediate MR image and in the calculation of the synthetic MR signal data, then it is useful if MR signal data from the central regions of the spatial frequency space is acquired beforehand, so that the phase errors occurring as a result of the non-idealities can be determined from this data. These errors can then be taken as a basis for the phase correction. To determine the phase errors, it is enough for low-resolution MR signal data solely from the central regions of the spatial frequency space to be analyzed. The acquisition of this MR signal data to allow the phase errors to be determined takes only a minimal amount of measuring time.
The method described above is particularly suitable for generating MR images from MR signal data that is acquired by non-Cartesian, and particularly spiral, sampling of the spatial frequency space. When the spatial frequency space is sampled spirally, a possibility that is open is for the sampling to take place at a higher density in the center of the spatial frequency space than at the periphery of this space. The phase errors required for taking account of the non-idealities of the MR apparatus may advantageously be determined directly from the MR signal data that is acquired at higher density in the center of the spatial frequency space. Also, synthetic MR signal data has to be calculated solely at the periphery of the spatial frequency space, which means that, overall, fast convergence can be expected in the iterative improvement of the MR image in accordance with the invention.
An MR apparatus suitable for performing the method according to the invention has a main field coil for generating a homogeneous static magnetic field in an examination volume, a plurality of gradient coils for generating magnetic field gradients in the examination volume, at least one high-frequency coil for generating high-frequency fields in the examination volume and/or for receiving MR signals from the examination volume, and a central control unit for operating the gradient coils and the high-frequency coil, plus a reconstruction and display unit for processing and showing the MR signals. The method described above can be performed on the MR apparatus according to the invention by means of a suitable programmed control means for the reconstruction and display unit.
The method according to the invention may be made available to the users of pieces of MR apparatus in the form of a corresponding computer program. The computer
program may be stored on suitable data carriers, such as CD-ROMs or floppy disks, for example, or it may be downloaded to the reconstruction and display unit of the MR apparatus over the Internet.
These and other aspects of the invention are apparent from and will be elucidated with reference to the embodiments described hereinafter.
In the drawings:
Fig. 1 is a diagrammatic representation of the procedure followed in the method according to the invention.
Fig. 2 shows a first variant of partial spiral sampling according to the invention of the spatial frequency space.
Fig. 3 shows a second variant of spiral sampling of the spatial frequency space. Fig. 4 shows an MR apparatus according to the invention.
The method shown in Fig. 1 begins with the acquisition of an MR signal data set 1 by partial sampling of the spatial frequency space associated with the examination volume. The sub-region of the spatial frequency space that is sampled is denoted by reference numeral 2 in Fig. 1. From the MR signal data set 1, a supplemented MR signal data set 3 is generated by adding the missing MR signal data in the sub-region 4 of the spatial frequency space that was not sampled by taking advantage of the assumed Hermitian symmetry of the MR signal data. An intermediate MR image 5 is then reconstructed from the supplemented MR signal data set 3 with account taken of the non-idealities of the MR apparatus used. Account is taken of the non-idealities by making a phase correction by reference to a phase- correcting data set 6. The phase-correcting data set 6 may, for example, be generated beforehand by determining phase errors by analyzing MR signal data from central regions of the spatial frequency space. In the next step of the method, a synthetic MR signal data set 7 is generated from the intermediate MR image 5 by reverse reconstruction. In this case too, account is taken, on the basis of the phase-correcting data set 6, of the phase errors that can be expected due to the non-idealities of the MR apparatus. The MR signal data in the sub- region 4 of the supplemented MR signal data set 3 is then replaced by the corresponding synthetic MR signal data forming data set 7. An improved intermediate MR image 5 is then reconstructed in turn from the supplemented MR-signal data set 3 in the manner described above. The steps of the method that are described above are then repeated iteratively until
such time as the intermediate MR image is of the required quality. A suitable cessation criterion is laid down for this purpose.
Fig. 2 shows the sampling of the spatial frequency space in a non-Cartesian, namely spiral, fashion. The solid line represents the trajectory along which the spatial frequency space is partially sampled. Along the trajectory shown as a dashed line, the MR signal data is supplemented by the method according to the invention in the manner described above. The co-ordinate axes of the spatial frequency space are identified as kx and ky.
In the spiral sampling pattern shown in Fig. 3, the sampling of the center of the spatial frequency space takes place at higher density than at the periphery of the space. The trajectory shown as a dashed line makes it clear that the MR signal data only needs to be supplemented with synthetic data outside the center of the spatial frequency space, as a result of which particularly fast convergence is obtained for the method according to the invention. The measuring time required for the acquisition of MR signal data with the sampling pattern shown in Fig. 3 is longer by only an immaterial amount than sampling following the pattern shown in Fig. 2.
Fig. 4 shows, as a block diagram, an MR apparatus on which the method according to the invention can be performed. The MR apparatus comprises a main field coil 8 for generating a homogeneous static magnetic field in an examination volume in which a patient 9 is situated. The MR apparatus also has gradient coils 10, 11 and 12 for generating magnetic field gradients in different directions in space within the examination volume. The pattern followed by the magnetic field gradients with time and in space within the examination volume is controlled by means of a central control unit 13, which is connected to the gradient coils 10, 11 and 12 via a gradient amplifier 14. Also forming part of the MR apparatus shown is a high-frequency coil 15 for generating high-frequency fields in the examination volume and for receiving MR signals from the examination volume. The high- frequency coil 15 is connected to the control unit 13 via an emitting unit 17. The MR signals acquired by the high-frequency coil 15 are demodulated and amplified by means of a receiving unit 16 and fed to a reconstruction and display unit 18. The MR signals processed by the reconstruction and display unit 18 can be shown by means of a screen 19. The reconstruction and display unit 18 has a suitable programmed control means to allow the method described above to be performed.