Pulsed Field MRI System With Spatial Selection

1. Field of the Invention

This invention relates to magnetic resonance imaging systems. In a primary application this invention involves a pulsed magnetic field used to polarize the magnetic moments of an object. As the pulsed field collapses, a selection gradient is used to cause the magnetic moments to vary spatially, providing slice selection.

2. Description of Prior Art

The basic concepts in magnetic resonance are described in a series of papers in the June 1980 series of the IEEE Transactions on Nuclear Science, Vol. NS-27, pp 1220-1255.

Exisiting MRI systems have a large number of theoretical and practical problems. The main magnet is a very costly item, primarily because of the very high degree of homogeniety required. Also, the radio frequency transmitter requires relatively high power, and must provide this at high uniformity. In addition, regions of inhomogeniety due to susceptibility variations can produce significant distortions. These problems were addressed in U.S. Pat. No. 5,057,776 by the same inventor. Here a pulsed magnet is used to polarize the moments, providing immunity to inhomogeniety. However, except for systems which used radio frequency excitation, this approach did not describe methods for slice selectivity. Of course 3D readouts of the entire volume can be used, but these often require some type of slab selection to avoid aliasing artifacts.

An object of this invention is to provide a method of slice selection in pulsed magnetic resonance imaging systems without r.f. excitation.

A further object of this invention is to provide NMR images of selective sections without requiring r.f. excitation.

A further object of this invention is to provide 3-dimensional images of selected slabs without causing aliasing.

Briefly, in accordance with the invention, a magnetic gradient is applied prior to the collapse of the main pulsed field. During the collapse, the magnetic moments become spatially varying in the vicinity of the null plane of the gradient field. When the gradient field is normal to the pulsed field a slab surrounding the null plane has its moments parallel to the pulsed field and therefore precesses. When the gradient field is parallel to the pulsed field the magnetic moments on either side of the null plane precess in opposite directions. Subtracting excitations with different null planes provides the desired slice.

For a more complete description of the invention reference can be made to the following detailed description of several embodiments thereof which is given in conjunction with the accompanying drawings of which:

FIG. 1 is a schematic drawing illustrating an embodiment of the invention.

FIG. 2 is a set of graphs of signals used in an embodiment of the invention.

FIG. 3 is a block diagram of a signal processing embodiment of the invention.

FIG. 4 is a plot of the selection gradient field.

FIG. 5 is a plot of the precessing moment distribution in an embodiment of the invention.

FIG. 6 is a set of graphs of signals used in an alternate embodiment of the invention; and

FIG. 7 is a plot of the precessing moment distribution in the alternate embodiment of the invention.

An understanding of the broad aspects of the invention can best be had by reference to FIG. 1. Here we basically see the pulsed MRI system first described in U.S. Pat. No. 5,057,776. The pulsed field created by coils 11, connected by terminals A, is used to polarize the magnetic moments in object 10, usually a portion of the human anatomy. A bias field generated by coils 14, connected by terminals B, can be used to enable the moments to precess about its axis. A set of gradients 12 provide fields pointed normal to the precessing moments. These are made spatially varying to provide imaging information while the moments are precessing. Using the eight coils, variations can be made in the x, y, or z axes representing the three gradient fields, all pointing in the y direction. For example, to create a y gradient coils ABCD and EFGH are driven out of phase. Similarly for a z gradient coils ABEF and CDGH are driven out of phase, and coils ADEG and BCFH are driven out of phase for an x gradient. These provide gradient fields as given by:

These gradient fields are used following the polarizing pulse, while the moments are precessing, to create linear space-varying fields to provide imaging. If a two-dimensional slice is selected, only the two gradient fields in the plane of the slice are required to make an image of the slice. Thus if a slice is created in an xy plane, at some value of z, only G.sub.x and G.sub.y are required. In many cases a 3D set is required of a volume in the body. In that case a slab is selected representing the volume of interest. This is followed by a gradient set in all three axes which cover the desired range of k-space. Failure to select the slab can result in inadequate sampling or aliasing where portions of the image will overlap. The resultant signals, representing the spatial frequencies, are received by coil 16 and appear on terminal 17. They are then processed to create an image.

In U.S. Pat. No. 5,057,776 the gradients can be used to provide 3D imaging of the volume. For 2D imaging of a slice, the only slice selection methods shown require r.f. excitation. In this invention we show general methods of exciting one or more slices without requiring r.f. excitation.

Basically we make use of the fact that, as fields are changed adiabatically, the magnetic moments do not precess but continues to point in the direction of the net field. When they are changed non-adiabatically the field continues to point in the same direction as prior to the change and the moments will precess if there is a field component in quadrature with the moments. These properties are used in this invention to enable slice selection.

One embodiment is described using FIG. 2. The pulse 19, is applied to terminal 13 to provide the pulsed polarizing field. The ramp-up 20 is limited in slope by dB/dt considerations to avoid inducing neural currents in the body. The "on" period, 21, is determined by T.sub.1 considerations for the materials involved. The ramp-down or collapse is the most significant consideration. The first part, 22, represents an adiabatic decrease where dB/dt<<.gamma.B.sup.2. During this period the magnetic moments will point in the direction of the field and not precess. Prior to the ramp-down, as shown, the selection gradient field G.sub.s, represented by waveform 33, is turned on. The turn-on is adiabatic so as not to initiate precession.

The selection gradient field is shown graphically in FIG. 4 and uses gradient coils 12; the same ones used in the receiving mode. Basically the gradient coils 12 provide a By field which varies with x, y, or z, depending on the direction of slice desired. For illustrative purposes the gradient direction shown in FIG. 4 is z, although it could equally well be pointed in any other direction. As shown the gradient field 30 is zero, or has a null plane, at z=z.sub.0. Thus moments at z.sub.0 will be unaffected by this added field and those at other z planes will have a transverse magnetic field added to the longitudinal pulsed field. At the point where the gradient field is equal to the pulsed field B.sub.z ', the amplitude of the magnetic moments in the z direction is down by 3 db. This may arbitrarily be defined as the slice width.

Returning to FIG. 2, we see that the pulsed field is adiabatically reduced to a value B.sub.z ', in the presence of the transverse field from gradient G.sub.s. Those moments at z=z.sub.0 will be unaffected and continue to point in the z direction. However, those moments at planes well removed from z.sub.0, where the transverse field B.sub.y is significantly greater than B.sub.z ', will have their moments pointing in the y direction. We will thus have a distribution of pointing angles being primarily in the z direction around z.sub.0 and primarily in the y direction well removed from z.sub.0.

At time T the pulsed field is collapsed at greater than the adiabatic rate where, in segment 23, dB/dt>>.gamma.B.sup.2. At the same time the selection gradient field 26 is turned off and the bias field 24 and gradient field 25 are turned on. Thus we have solely transverse fields in the y direction. These will cause any moment in the z direction to precess, since it is in quadrature. Thus the moments around z.sub.0 will precess and provide signals while those well removed from z.sub.0 will not precess since they are in the same direction as the transverse field. This provides the desired slice selection in the vicinity of z.sub.0. The distribution of precessing spins is shown in FIG. 5 as graph 32. The amplitude of the Mz component, corresponding to the amplitude of precessing spins, is given by: ##EQU1## As shown the 3 db or 71% point occurs at .DELTA.z=B.sub.z 'G.sub.s. For z values far removed from the slice, the spin amplitude decreases as 1 .DELTA.z.

Following slice selection the spin distribution of the selected slice is read out to provide an image. This can be done using bias field B.sub.b, signal 24, to place the image information on a carrier, as described in U.S. Pat. No. 5,057,776. To scan k-space an array of readout gradients are used which vary on each excitation. To fill k-space we use the basic relationship:

where k is the spatial frequency, r is the spatial dimension, G is the time-varying readout gradient signal and .gamma. is the gyromagnetic ratio.

A variety of imaging or readout gradient waveforms can be used as indicated in U.S. Pat. No. 5,057,776. The signals representing each k-space value are received using coil 16 and appear on terminal 17. The output signal 17 is processed as shown in FIG. 3. Here the signal is processed by quadrature demodulation at the bias frequency .gamma.B.sub.b in demodulator 26. The in-phase and quadrature signals are sampled and stored in 27, and transformed in 28. The magnitude of the resultant transformed signal is found in 29 and displayed in 30.

As shown this is an effective mechanism for slice selection. It may be particularly effective for "slab" selection for 3D studies where the volume of interest is selected to avoid aliasing. If an inadequate number of samples are taken in k-space, the imaged volume can overlap with other parts of the volume. To avoid this, without requiring excess sampling, we excite only the portion of the volume of interest by the method described. For those applications where a very sharp slice is desired, as in multi-slice imaging, the embodiment in FIG. 6 may be preferable. Here a new gradient system is required, as compared to the previous embodiment which used gradient coils 12, the same set used for the receiving operation. Gradient system 18 is added which is identical to that of 12 but has the field pointing in the same direction as the pulsed field. The same sets of coils can be used to provide gradients in any direction. Alternatively, if slice selection is only desired in one or two directions, a limited set of coils can be used.

The polarizing pulse 35 is essentially the same as that of 19 except it continues its adiabatic decline to zero. The gradient selection field 37 in coils 18 is turned on prior to the collapse of the pulsed field and turned off adiabatically after the field 35 has collapsed. The gradient field G.sub.s is identical to that shown in FIG. 4, providing a null field at z=z.sub.0. Note that this leaves the magnetic moments on either side of z.sub.0 facing in opposite directions since, as previously indicated, during adiabatic variations the moments follow the field direction. We are thus left with a distribution shown in the solid line 40 in FIG. 7. When the image of the volume is then recorded, it corresponds to a projection in the z direction, with the image in the xy plane weighted as shown in 40. The entire imaging is then repeated with gradient G.sub.s changed with a null field at z.sub.0 '. The difference between these two acquisitions forms the final image, representing a sharp slice in the cross-hatch labeled 42.

For a multislice acquisition n acquisitions can be taken to form n-1 images. Different slice widths can be obtained by subtracting different acquisitions. This can be used to increase the signal-to-noise-ratio at the expense of resolution in the slice-select direction.

Both slice-selecting embodiments are quite robust and retain their properties with variations in the various parameters. For the embodiment of FIG. 2 the slice center is determined completely by the G.sub.s gradient null plane. The width of the slab will vary somewhat with the B.sub.z ' of the pulsed field which is a negligible effect. For the embodiment of FIG. 6, the slices depend solely on the gradient G.sub.s and are independent of all other parameters. If there is an extraneous transverse field, such as the earth's field, it can slightly effect the edge of the slice by causing magnetic moments immediately adjacent to the null plane to rotate, and therefore not precess when the B.sub.b bias and/or the readout gradients are turned on.

Although the descriptive material included a bias field, as explained in U.S. Pat. No. 5,057,776, a system needn't have a bias field and can operate with the readout gradient as the sole transverse field.