MR method utilizing multi-dimensional RF pulses
The invention relates to an MR (MR = magnetic resonance) method utilizing multi-dimensional RF pulses, to an MR apparatus for carrying out the MR method, and to a computer program for the control unit of an MR apparatus of this kind.
As is known, RF pulses are designated to be "multi-dimensional" if they act on an examination zone simultaneously with at least two gradient fields with temporally and spatially different gradients. The nuclear magnetization can be excited in a spatially limited part of the examination zone by means of such multi-dimensional RF pulses. The nuclear magnetization is not excited in the remaining part of the examination zone, even though the RF pulse and the magnetic gradient fields also act on this remaining part. Multi-dimensional RF pulses of this kind are subject to limitations in that on the one hand the RF pulse may have only a given duration and in that on the other hand the capacity of the gradient system is limited (or a given speed of change of the gradient may not be exceeded for medical reasons). Consequently, the trajectory defined by the variation in time of the magnetic gradient fields cannot scan the (excitation) k space with an arbitrary density. This gives rise to a limitation of the resolution as well as to the excitation of the nuclear magnetization outside the desired imaging zone, so that the MR signals subsequently acquired can be corrupted by backfolding effects.
In order to solve this problem, it is known from United States patent application 09/728111 (PHD 99.170) to generate two (or more) MR signals by means of two (or more) multi-dimensional RF pulses while configuring the variation in time of the gradient fields in such a manner that the k space is scanned along trajectories which are mutually offset. When these MR signals (or their Fourier transform) is combined, an excitation profile is obtained with an improved spatial resolution so that the excitation of structures outside the desired region is then prevented to a substantial degree. This approach, however, has the drawback that the acquisition of each time two or more MR signals necessitates a doubling or multiplication of the acquisition time.
Therefore, it is an object of the present invention to conceive an MR method utilizing multi-dimensional RF pulses in such a manner that the acquisition time is not
doubled (or multiplied), while the undesirable excitation of structures outside the desired imaging zone is reduced nevertheless.
This object is achieved by means of an MR method in accordance with the invention which includes the steps of: generating at least one RF pulse which acts on an examination zone, and generating at least two gradient magnetic fields with gradients which vary differently in time and in space, which act on the examination zone simultaneously with the RF pulse, and which vary in time in such a manner that during the RF pulse a trajectory with a spatially varying density is followed in the k space, that is, a trajectory which notably has a density which is higher in a central zone of the k space than in the zones outside this central zone. In conformity with the method of the invention there is formed a trajectory which exhibits a variable density in the k space instead of the customary constant density. The invention is based on the recognition of the fact that not all k values or all spatial frequency components are of equal importance for the excitation of a desired imaging zone. For most objects the central zone of the excitation k space which is associated with the low spatial frequencies is most important, because the energy is concentrated essentially in the central zone, of the k space.
Consequently, an increased density of the k space trajectory is used for this zone, with the result that the undesirable excitation of nuclear magnetization is reduced on the one hand and is pushed to a zone which is situated further from the desired imaging zone on the other hand, so that the backfolding or aliasing artifacts are reduced. In the outer zones of the k space the density of the k space trajectory is still not sufficient, however. Therefore, the RF pulse for this spatial frequency domain can still produce an undesirable transverse magnetization which corrupts the MR signal by way of backfolding effects. However, the RF energy deposited in this spatial frequency domain is comparatively small, so that the effect of the excitation of transverse magnetization in the outer zones is negligibly small in a first approximation.
It is to be noted that it is known, for example, from an article by D. Spielman et al, MRM 34, 388 (1995), to impose such a variation in time on the magnetic gradient fields which are active during the reading out of the MR signals that a trajectory with a spatially variable density is followed in the k space, for example, a spiral whose turns are situated nearer to one another at the center than at the periphery. However, the cited article concerns only the variation in time of the magnetic gradient fields which are active during the
reading out of the MR signals (that is, after the RF pulse), whereas the invention concerns the variation in time of the magnetic gradient fields which are active during the RF pulse.
As has already been stated, the trajectory should preferably have a density which is higher in the central zone of the k space (that is, at low spatial frequencies) than in zones situated outside this central zone. However, the examination zone may also contain (for example, periodic) structures which could be excited better in a spatial frequency domain outside the center. Generally speaking, however, the excitation with a higher density of the trajectory in the central zone of the k space is the optimum approach.
Claim 2 discloses an advantageous further embodiment of the invention. Granted, when the density is higher at the center of the k space than outside the center of such a spiral-shaped trajectory, the number of turns of the spiral-shaped trajectory increases (in comparison with a spiral-shaped trajectory with a constant spacing of the turns), but the duration of the RF pulse is not prolonged to the same extent as a result thereof. This because during the scanning of the central zone of the k space the magnetic gradient fields still have a very small gradient only, so that at a given maximum rate of variation of the gradient said inner turns can be traversed significantly faster than the outer turns.
In principle, however, the trajectory may also follow a different course in the k space; for example, in conformity with claim 3 it may comprise a set of parallel straight lines, as in the cited US patent application 09/728111 and analogous to the EPI sequence during the reading out of MR signals, whose spacing is less at the center of the k space than in the zones outside the center.
The density of the trajectory may vary in steps (claim 4) or continuously (claim 5).
In conformity with claim 6, the invention is also suitable for use in the case of so-called "Transmit Sense" methods in which the multi-dimensional RF pulses generated by a plurality of RF coils act simultaneously on the examination zone. As in the case of the conventional Transmit Sense methods, the RF pulses may then exhibit a different variation in time. However, as opposed to the conventional Transmit Sense method, in conformity with claim 7 they may also exhibit the same variation in time. Claim 8 discloses an MR apparatus which is suitable for carrying out the method in accordance with the invention and claim 9 defines a computer program for the control unit of an MR apparatus of this kind.
The invention will be described in detail hereinafter with reference to the drawings. Therein:
/
Fig. 1 shows the block diagram of an MR apparatus which is suitable for carrying out the invention,
Fig. 2 shows the variation in time of a sequence with a two-dimensional RF pulse, Fig. 3a shows a spiral-shaped trajectory in the k space with a constant density,
Fig. 3b shows the associated intensity profile,
Fig. 4a shows a spiral-shaped trajectory in accordance with the invention, and
Fig. 4b shows the associated intensity profile.
The reference numeral 1 in Fig. 1 denotes a diagrammatically represented main field magnet which generates a steady, essentially uniform magnetic field of a strength of, for example, 1.5 Tesla which extends in the z direction in an examination zone (not shown). The z direction then extends in the longitudinal direction of an examination table (not shown) on which a patient is accommodated during an MR examination.
Also provided is a gradient coil system 2 which includes three coil systems which are capable of generating gradient magnetic fields Gx, Gy and Gz which extend in the z direction and have a gradient in the x direction, the y direction and the z direction, respectively. The currents for the gradient coil system 2 are supplied by a gradient amplifier 3. Their variation in time is controlled by a waveform generator 4, that is, separately for each direction. The waveform generator 4 is controlled by an arithmetic and control unit 5 which calculates the variation in time of the magnetic gradient fields Gx, Gy, Gz required for a given examination method and loads these values into the waveform generator 4. These signals are read out from the waveform generator 4 for the MR examination so as to be applied to the gradient amplifiers 3 which generate therefrom the currents that are required for the gradient coil system 2. The control unit 5 also acts on a workstation 6 which includes a monitor 7 for the display of MR images. Entries can be made via a keyboard 8 or an interactive input unit 9.
The nuclear magnetization in the examination zone can be excited by RF pulses from an RF coil 10 which is connected to an RF amplifier 11 which amplifies the output signals of an RF transmitter 12. In the RF transmitter 12 the (complex) envelopes of the RF pulses are modulated with the carrier oscillations which are supplied by an oscillator and whose frequency corresponds to the Larmor frequency (approximately 63 MHz in the case of a main field of 1.5 Tesla). The arithmetic and control unit loads the complex envelope into a generator 14 which is coupled to the transmitter 12. Instead of one RF coil with one RF
transmit channel there may also be provided a plurality of RF coils which comprise a respective RF transmission channel, each of which is provided with an RF coil.
The MR signals generated in the examination zone are picked up by a receiving coil 20 and amplified by an amplifier 21. The amplified MR signal is demodulated in a quadrature demodulator 22 by way of two 90° mutually offset carrier oscillations of the oscillator 13, so that two signals are generated which may be considered as the real part and the imaginary part of a complex MR signal. These signals are applied to an analog-to-digital converter 23 which forms MR data therefrom. The MR data is subjected to various processing operations in an evaluation unit 24, that is, inter alia a Fourier transformation. It is also possible to provide a plurality of RF receiving channels for a plurality of receiving coils. Fig. 2 shows the variation in time of a sequence which includes a two- dimensional RF pulse. An RF pulse RF0 then acts on the examination zone, the envelope of said RF pulse being loaded into the generator 14 by the control unit 5, and simultaneously with the RF pulse two magnetic gradient fields Gx0 and Gyo, act on the examination zone, their variation in time being imposed by the waveform generator 4 under the control of the control unit 5. The magnetic gradient fields Gxo and Gy0 are formed by oscillations with an amplitude which decreases in time and with a distance between the zero-crossings which also decreases. For the zero-crossing of one oscillation Gx0 or Gy0 the respective other oscillation always exhibits a relative maximum. The variation in time of the envelopes of the RF pulse RFo is tuned to the variation in time of the magnetic gradient fields Gx0 and Gy0 in such a manner that the nuclear magnetization is excited in a zone which is bounded in space in the x direction and the y direction.
Subsequent to this multi-dimensional RF pulse there is generated a refocusing 180° pulse KF\ in conjunction with a magnetic gradient field Gzι. A slice or a layer is thus selected from the zone previously bounded in the x direction and the y direction. The MR signals subsequently read out are determined essentially exclusively by the nuclear magnetization from this zone.
The subsequent reading out of the spatial distribution of the excited nuclear magnetization takes place in the form of a so-called EPI sequence. The magnetic gradient field Gx is then generated with a periodic variation whose polarity continuously reciprocates between a positive value and a negative value. At the same time the magnetic gradient field Gy is active in the form of short pulses ("blips") which occur each time at the zero-crossings of the magnetic gradient field. In each of the intervals in which the magnetic gradient field Gx has a constant positive or negative value, the signal received by the receiving coil 20, and
subsequently demodulated and digitized, is acquired by the evaluation unit 24. An image of the nuclear magnetization distribution in the previously excited slice, bounded in the x direction and the y direction, can be derived from the totality of MR signals received.
The trajectory in the k space which is associated with the variation in time of the magnetic gradient fields Gx0 and Gyo is described by the following equation:
k(t) = -γ |G(t/) dt/ (1) t
In this respect it is assumed that the RF pulse RF0 commences at the instant = 0 and terminates at the instant = T. The letter G represents the magnetic gradient field resulting from the superposition of Gx0 and Gy0. The variation of the magnetic gradient fields Gx0 and Gyo as shown in Fig. 2 results in a spiral-shaped trajectory which is followed inwards from the outside. Fig. 3 a shows such a trajectory with a constant density of the spiral turns. This course of the trajectory yields excitation profile shown in Fig. 3b; this trajectory represents the variation of the transverse magnetization along a straight line extending in the x direction through the center (at x = 0.5) in the case of a homogeneous object. A pronounced maximum of the transverse magnetization can be recognized at the center, the width of this maximum being co-determined by the envelope of the RF pulse RF0- However, secondary maxima of lower amplitude are situated to both sides of this main maximum.
This means that the nuclear magnetization outside the desired region is also excited. Even when during the reading out of the excited nuclear magnetization subsequent to the multi-dimensional RF pulse the so-called field of view is chosen to be such that it is wider than the main maximum and narrower than the distance between the secondary maxima, backfolding (aliasing) effects will corrupt the received MR signal by way of components which originate from such secondary maxima. It is Icnown that such secondary maxima can be pushed further outwards by utilizing a higher, uniform density of the spiral turns (in given circumstances they can even be pushed out of the object). However, this would require either a corresponding prolongation of the duration of RF0, Gx0 and Gy0 (which is undesirable) or stronger gradients; the latter is not possible, or even not permissible, for a given MR system.
Fig. 4a shows the course of the trajectory of a multi-dimensional RF pulse in accordance with the invention. In comparison with the trajectory shown in Fig. 3a, the density of the spiral turns is doubled in a central zone which corresponds to the low k values or the low spatial frequencies, whereas outside this central region it corresponds to the density of the spiral turns in Fig. 3a. Assuming that Fig. 2 shows the variation in time of the
magnetic gradient fields corresponding to Fig. 3a, it is necessary for such a trajectory that as from a given instant the envelope of the magnetic gradient fields Gx0 or Gy0 decreases more slowly than shown in Fig. 2 (the pulse RF0 must then be prolonged accordingly and its variation must be adapted to the changed variation of Gx0 and Gy0). Granted, the larger number of spiral turns of this trajectory imposes a prolongation of the duration of the RF pulse, but this prolongation is not proportional to the (increased) number of spiral turns, because the magnetic gradient fields of the central zone of the k space can be traversed significantly faster than its outer zones because of the predetermined maximum rate of change (slew rate) of the magnetic gradient fields. Fig. 4b shows the excitation profile associated with the trajectory shown in
Fig. 4a. It will be recognized that at the locations in which pronounced secondary maxima are present in the excitation profile shown in Fig. 3b, secondary maxima are still present; however, they have a significantly reduced amplitude. Such secondary maxima are caused by the higher spatial frequency components where the trajectory passes through the k space with the same density as in fig. 3 a. The maxima imposed by the increased (but still finite) density of the trajectory in the range of the lower spatial frequencies have been pushed further outwards, so that they are no longer visible in the representation of Fig. 4b. Undesirable aliasing effects are thus suppressed practically completely.
The invention can be used not only for an imaging sequence as shown in Fig. 2, but also for the generating of navigator pulses. The width of the main maximum in the excitation profile is then even significantly smaller than shown in Fig. 3b or Fig. 4b, and no slice-selective pulse (RFj, Gzl, Fig. 2) is then required. Moreover, only one MR signal is acquired.
The invention can also be used for the so-called Transmit Sense, where a plurality of RF coils, having different spatial sensitivities, simultaneously generate multidimensional pulses, for each RF coil there being provided a separate variation in time of the RF pulse RF (see Katscher et al., Proc. ISMRM 2002, page 189). Such Transmit Sense enables a reduction of the duration of the RF pulses while maintaining the spatial resolution of the excitation profile. The application of the invention to Transmit Sense again yields a reduction of backfolding or aliasing artifacts in conformity with the described principle.
A further advantage of the use of the invention in conjunction with Transmit Sense consists in that the complex calculation of the individual RF time functions can be dispensed with if the same time function is used for each individual transmission coil. This is because in accordance with the invention the sub-sampling occurring in the case of Transmit
Sense is canceled at least for the central zone of the k space. As a result, aliasing artifacts which would otherwise occur within the excitation profile in this case are also minimized in this manner.
The invention can be used not only for multi-dimensional excitation pulses, but also for multi-dimensional focusing pulses.