WO1996004566A1

WO1996004566A1 - A method of magnetic resonance imaging employing velocity refocussing

Info

Publication number: WO1996004566A1
Application number: PCT/GB1995/001829
Authority: WO
Inventors: Richard Alan Fox; Steven Frederick Tanner; Nicola Jane Taylor; Martin Osmund Leach
Original assignee: British Technology Group Limited
Priority date: 1994-08-02
Filing date: 1995-08-02
Publication date: 1996-02-15
Also published as: IL114797A0; GB9415574D0

Abstract

The invention relates to an improved method of magnetic resonance imaging in which gradient motion refocussing of a slice selection imaging technique is employed. A problem with existing slice selection magnetic field imaging techniques has been that voids or image artifacts occur if motion is present within the object being imaged. The problem is to some extent avoided by using gradient motion refocussing techniques at the expense of the sequence having a fixed and relatively longer echo time thereby increasing the time required to acquire an image. The invention reduces the relative imaging time and is capable of performing gradient motion refocussing using a shorter and variable echo time.

Description

A METHOD OF MAGNETIC RESONANCE IMAGING

EMPLOYING VELOCITY REFOCUSSING This invention concerns improvements in or relating to magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS), and more particularly to an improved method of gradient motion refocussing (GMR) of a slice selection pulse sequence, which is particularly suitable for use in spin-echo (SE) imaging.

The spin-lattice relaxation time (T,) is the time characterising the return to equilibrium conditions of the longitudinal magnetisation for a particular substance after the perturbation resulting from the application of a radio frequency (rf) pulse. The transverse relaxation time (T₂) is the time characterising the decay of the magnetic resonance (MR) signal in a particular substance.

Magnetic resonance images contain areas of different contrast that are indicative of different properties or of different material types or substances. The contrast results from differences in the relaxation times (T„ j) of the nuclei being imaged; from differences in the density of hydrogen nuclei in different parts of the object; and from other differences in the properties of the nuclei being imaged. These differences enable one material type to be distinguished from another after the relevant MR signals have been processed to form an image.

The bulk magnetisation resulting from placing an object in a magnetic field can be sampled in an MR experiment by the application of one or more rf pulses. These pulses are applied at known instances and have a predetemiined phase relationship with respect to one another. When carrying out imaging, both rf pulses and magnetic field gradients are used to generate a signal which is spatially encoded and which on processing forms the image.

One example of a type of imaging sequence is the spin-echo (SE) imaging sequence. This consists of a 90° rf pulse applied prior to at least one 180° rf pulse. Different types of magnetic field gradients may then be applied at predetermined times, and for predetermined durations during the SE sequence. When the SE imaging sequence is used, a signal is usually acquired after a 180° pulse in a form known as an echo. If more than one

180° pulse is applied then it is possible to acquire more than one echo. Each echo is acquired in a period that is characterised by a time known as the echo time (TE). When using SE imaging sequences it is sometimes desirable to set the value of TE to be as short as possible. There are however limitations as to how short the TE period can be. One

-1- SUBSTITUTE SHEET (fiULE 26) limitation is the finite rise and fall times associated with turning the slice selective magnetic field gradient (G_$) on and off.

The slice selective magnetic field gradient (G_s) used in an SE imaging sequence conventionally consists of positive magnetic field gradients applied during the 90° and 180° pulses. An example of such a switching sequence is shown in Figure 1 a and is described in detail below. These gradient lobes, applied in isolation, result in a dephasing of the received signal and a consequent loss in signal intensity. The necessary rephasing, at TE, can be achieved for spins that are not moving by judicious setting of the time that the lobe is high during the 180° pulse and by applying a negative gradient lobe just after the 90° pulse with an area equal to half that applied during the 90° pulse. Although this rephasing works well for static spins, any movement within the object being imaged, causes moving spins to experience a changing field with a consequent loss in phase coherence and degradation of signal at the echo.

Images acquired using conventional SE techniques therefore can contain signal voids and image artefacts consisting of blurring and streaks propagating through the image, if motion is present within the object being imaged. This problem is to some extent mitigated by using gradient motion refocussing methods. An example of prior art for imaging blood entitled: "A rapid echo flow rephased spin-echo imaging technique", is described in a paper, by Jung g_ al., published in Magnetic Resonance Imaging Vol. 1 1 (1993) at pages 301-309.

The present invention arose from consideration of the aforementioned problems and provides an improved gradient switching sequence which is more flexible than the existing sequences; is capable of giving images with a shorter minimum echo time than existing sequences; and which may be used to image a moving object such that static spins and spins having a constant velocity are refocussed at the time of the echo.

According to the present invention there is provided a method of magnetic resonance imaging comprising the steps of: applying a first magnetic field gradient (G_sl) lobe, having a first polarity; applying a first rf pulse; subsequently applying a second magnetic field gradient lobe (G^), with opposite polarity to that of the first lobe; applying a third magnetic field gradient lobe (G_s3), with opposite polarity to the first lobe; and applying a second r.f. pulse. In an alternative embodiment GMR can be effected using only two gradient lobes with opposite polarity; the second lobe applied during the 180° pulse. This alternative embodiment effectively coalesces second and third lobes, into a single lobe, albeit of different shape. Preferably the first rf pulse is a 90° pulse and the second rf pulse is a 180° pulse.

Subsequent rf pulses may be applied after application of the second rf pulse. These subsequent pulses may be 180° pulses.

Preferably the method is employed in a spin echo imaging sequence. However, the sequence may be employed in other types of magnetic resonance imaging such as echo planar imaging, fast spin echo or in sequences for magnetic resonance angiography. The sequence may have applications in magnetic resonance spectroscopy where spin echo techniques are employed.

According to another aspect of the present invention there is provided a method of Magnetic Resonance Imaging comprising applying a gradient motion refocusing sequence, characterised in that a variable echo time is provided in said sequence.

According to the present invention there is provided apparatus for magnetic resonance imaging comprising means for establishing a magnetic field such that an object to be imaged may be positioned within said field; means for producing orthogonal magnetic field gradients; means for varying the polarity of said gradients; and means for producing radio frequency (rf) pulses, capable of achieving, in use, an image of or signal from a selected slice of the object characterised in that the means for varying magnetic field gradients is arranged to apply an rf pulse with a first magnetic field gradient at a first polarity and a second rf pulse with a second magnetic field gradient at a polarity, opposite to the first polarity. Because gradient lobes are applied at known instances with a predetermined relative polarity, the sequence is capable of obtaining images or signals that are gradient motion refocussed with shorter minimum TE values than obtained by the method described in the prior art. Further the present invention is capable of obtaining an image or signal with variable TE, without recourse to changing the magnitude of any slice select gradient lobe (G_j) or changing the position of the first G_s gradient lobe (G_sl ) relative to that of the second G₅ gradient lobe (G_s2). In an alternative embodiment two G_s gradient lobes with opposite polarity are applied. In this implementation an even shorter minimum TE is obtainable whilst still retaining the property of GMR, however this alternative implementation has to be used at a particular value of TE unless the magnitudes or durations of the two gradient lobes are changed.

Embodiments of the present invention will now be described; by way of example only and with general reference to the Figures, and particular reference to Figures 2, and Figures 4 to 7 in which:

Figure la shows an example of a PRIOR ART switching sequence for G, in a SE sequence, wherein no gradient motion refocussing (GMR) is implemented;

Figure 1 b shows the phase shift of spins moving with a component of velocity along G_s using the arrangement of gradient depicted in Figure la;

Figure 2a illustrates diagrammatically an embodiment of the G_s switching sequence according to the present invention; Figure 2b shows the phase shift of spins moving with a component of velocity along G_s, using the sequence of gradient lobes depicted in Figure 2a;

Figure 3 shows an example of a PRIOR ART switching sequence for G_s in a SE sequence containing GMR;

Figure 4 is a photograph of an image affected by motion artefacts acquired using a standard SE sequence;

Figure 5 is a photograph of the same subject as in Figure 4, but obtained using die present invention;

Figure 6 is an alternative embodiment to Figure 2 and illustrates a switching sequence termed HYBRID with gradient motion refocussed projections; and Figure 7 is a graph of blood and brain Gd-DTPA concentrations obtained using the sequence as shown in Figure 6.

In each case (Figures 1 and 2) the gradient lobes are outlined in bold lines. The hatched area outlined with a dashed line following the centre of the 180° rf pulse shows the effective reversal in phase effect of the illustrated gradient lobe as a result of the inversion of phase caused by the 180° rf pulse.

Motion of spins during a spin echo imaging sequence results in a loss of signal due to a phase cancellation and misregistration caused by variation in a magnetic field experienced by the spins as the spins move through applied field gradients as explained above. In cases where it is desirable to restore a signal from moving tissue, for example blood, it is necessary to correct magnetic field gradients to overcome effects of velocity, by using velocity refocussing techniques also called gradient motion refocussing (GMR). The use of GMR also has the effect of reducing artefacts along a phase encoding axis that arise as a result of motion. The present invention is concerned with applying GMR on slice selection gradient directions, affecting the dephasing of components of motions in the gradient directions. A slice selection gradient for a spin echo sequence normally consists of positive magnetic field gradients applied during the 90° and 180° pulses. Rephasing for stationary spins may be achieved by applying a negative gradient, with an area equal to half of that of the positive gradient during the 90° pulse, immediately after application of the positive gradient. An example of this is shown in Figure la, in which other gradients required to form an image have been omitted for clarity. Thus the second half of the 90° gradient is completely balanced by the negative lobe. Due to the inversion of magnetization, caused by the 180° rf pulse, the first half of the 180° gradient is balanced by the second half of the 180° gradient as illustrated by shaded area 10 in Figure la. Displacing matter, i.e. spins, from the slice at a constant velocity destroys this balance, since the magnetic field increases (or decreases) as spins progress through the field gradient.

Figure lb shows the effect of a constant velocity on the field experienced by moving nuclei. The relevant areas 11 have again been shaded. However, now it is clear that phases will no longer be balanced since respective areas do not cancel. Balance may be achieved by adding additional negative gradient both before and after the 180° pulse. If these have equal areas, phase balance for stationary spins is caused by inversion of magnetization at the 180° pulse. The gradient lobe after the 180° pulse increase or decrease the effect on the phase for spins moving with constant velocity. By appropriate selection of the separation of these two negative gradient lobes it is possible to rebalance the dephasing effect of the normal sequence shown in Figures la and lb. The aforementioned method, described with reference to Figures la and lb forms the basis of the REFRESH sequence proposed by Jung si al- as described in MAGNETIC RESONANCE IMAGING 1 1, 1993 at pages 301 to 309. The method however has two distinct disadvantages. Firstly there is a need for additional negative gradient lobes both before and after the 180° pulse appearing as references 19 and 20 in Figure 3. This extends the minimum echo time (TE) achievable with this sequence. Thus a minimum echo time of 9.7 msec was required. The additional negative lobe required after the 180° pulse is particularly important in this respect due to the length of time that needs to be allowed at the end of the slice selection gradient for the application of a read gradient. Secondly the sequence generally only operates at a fixed value of echo time (TE), since any change in separation of the two additional negative lobes destroys the phase balance. A more compact and flexible velocity refocussed sequence can be achieved by applying a negative, rather than a positive gradient simultaneously with the 180° pulse. Although it is common to use gradients of the same polarity during the 90° and 180° pulses, there is no reason why this should have to be the case provided the frequency of the selective rf pulse is appropriately adjusted. The present invention will now be described with specific reference to Figure 2 which illustrates a simplified sequence that balances for both stationary spins and those moving at constant velocity. It is apparent from Figure 2a that stationary spins have completely balanced phases. It is also the case that the positive and negative shaded areas in Figure 2b completely balance. This is because the additional phase change caused by flow during the application of the first two gradients is exactly balanced by an equal and opposite phase change resulting from flow during the third gradient. Consequently, this sequence is capable of balancing both stationary spins and spins moving at constant velocity without the need for additional lobes in the gradient. It is now also possible to separate 90° and 180° pulses by an arbitrary amount without destroying this balance. This renders the sequence more flexible than may be expected and it is felt the sequence of the present invention is particularly well suited for use in other imaging techniques, examples of which are described below.

In practice gradients cannot be switched instantaneously, so that invariably time is lost in ramping the gradients up and down. Eddy currents may result in a further loss of time in practical situations, as this results in the gradients taking even longer to achieve their desired value. Theoretical calculations have been performed to consider the advantages of applying a negative gradient on the 180° pulse. To minimise the total value of TE it has been assumed that the 90° and 180° pulses have a duration of only 0.8 msec and the gradient can be switched using a linear ramp wi a total switching time of 0.5 msec. These values are similar to those used by Jung ≤l al. as referred to above.

Using a negative gradient on the 180° pulse it is possible to balance both stationary protons and those moving with constant velocity using a TE of 6.3 msec. This allows the read gradient to be applied 2.05 msec before the echo position, therefore allowing 4.1 msec in which to read out the echo. If a larger TE is required it is only necessary to separate the gradients on the 90° and 180° pulses by the appropriate time. Thus to increase the value of TE to 10 msec the two gradients need to be separated by 1.85 msec. The invention permits this and is thus considerably more flexible than existing sequences.

If the effects of eddy currents are added it is necessary to increase the minimum value of TE. Thus for eddy currents which introduce a time constant of 0.2 msec, the minimum TE must be increased to 7.46 msec.

If the situation where a positive gradient is used on both 90° and 180° pulses is considered, the minimum TE in which both stationary protons and those moving at constant velocity can be balanced is determined by the time available between the 180° pulse and the echo. If the first half of the read gradient requires 1.92 msec then the minimum time between the 180° pulse and the echo must be 3.82 msec. This allows a minimum of 1 msec for the negative lobe after the 180° pulse. Hence the minimum possible echo time is 7.64 msec.

It is believed that an alternative embodiment of GMR can be achieved using only two gradient lobes with opposite polarity; the second lobe applied during the 180° pulse. In this alternative embodiment the second and third lobes, mentioned above, (G_s2 and G_s3 in Figure 2a) are effectively coalesced.

Figures 4 and 5 show images obtained using a conventional switching sequence and a sequence according to the present invention, respectively. It is apparent that the ghosting in Figure 4 is not present in Figure 5. In an alternative embodiment, described with reference to Figure 6, the use of

Gd-DTPA enhanced MRI has led to the publication of several new methods for acquiring data on contrast uptake as described for example by Taylor el al- in Magnetic Resonance Medicine Vol. 30 (1993) at pages 744 to 750. Such data may be quantified to acquire physiological parameters such as the permeability of impaired blood brain barriers or of tumours. Since changes may occur in permeability as a result of treatment, monitoring changes with MRI gives a non-invasive, quantitative assessment of the effects of treatment. Figure 7 shows a graph of results obtained.

The present invention improves the quantification of results obtained using the above sequence, described by Taylor, where gradient echo projections were interleaved with each Fourier line of a gradient echo image. This sequence was prone to susceptibility-induced and flow induced signal dephasing. The gradient echo projection section has been replaced by a flow refocussed spin echo (SE) section comprising a sequence according to the present invention. The new sequence (HYBRID) monitors blood and tissue concentrations simultaneously and accurately after a bolus injection of Gd-DTPA, thereby allowing measurement of the blood input function and the tissue uptake curve. The HYBRID sequence shown in Figure 6 has very good spatial resolution, good temporal resolution and can be applied to a variety of regions within a patient's body.

The spin echo section 30 collects 1-D projections and is gradient motion refocussed in order to obtain signals from flowing blood. Two slice selective 90° saturation pulses 32 are placed over the projection slice extending 30mm to the proximal side of the selected slice to ensure complete saturation of spins moving into the projection slice, thus optimising the effects of the contrast agent. The spins are left to recover after saturation for

50ms prior to the start of the SE section. Saturation appears to take place after the initial

SE but repetition of the sequence ensures that it also precedes the relevant projection acquisition. The acquired SE signal is thus proportional to the amount of relaxation that has taken place, and hence to the amount of Gd-DTPA present. During this blood relaxation interval, a line of a gradient echo image of a tumour slice 34 is obtained. The projection time resolution is 120ms, and image time resolution is 34 seconds for a full 256² data set, with 17 seconds for a Half Fourier acquisition. A complete sequence yields 256 projections and one image per acquisition, with 15-30 acquisitions being obtained in succession in one investigation. This allows monitoring of a first pass of the Gd-DTPA bolus and the start of the washout process. Figure 7 shows results obtained using the above sequence and brief reference will now be made to Figure 7. It will be appreciated that the information is auxiliary to the present invention and has been included as an illustrative example of the type of information which may be derived by the novel switching sequence. The time dependent raw projection data are Fourier transformed and stacked to give intensity/time "images" from which the lines of data corresponding to a blood vessel are identified and extracted. Reference axial images are also used to identify the blood vessels.

The data must then be corrected to remove the contribution to the signal intensity which arises from the superimposed tissue. For brain studies, the internal carotid arteries are the most suitable vessels, as they are straight and do not bifurcate significantly. The tissue correction is made by subtracting the mean signal from the voxels immediately to each side of the blood vessel. The remaining blood signal intensities are then corrected for geometrical factors, such as those arising from the projection of a cylinder onto a line, by normalising the vessel's signal to that from blood in an image of the sagittal sinus which is acquired using an identical refocussed and saturated sequence.

All data, from both images and projections, are adjusted to allow for the effects of inhomogeneity within the main field and the coil (not shown), and for tissue dependent proton densities. An identical process is carried out on two reference samples consisting of different concentrations of Gd-DTPA in water, placed on either side of the head and neck. The ratio of the reference-corrected signal intensities (C.S) to their expected intensities (E), when plotted against their respective T,'s, gives a line from which the blood or tissue T, values at all time points can be found from their C.S/E values by iteration. The T,'s are then converted into Gd-DTPA concentrations, given die relaxivity of Gd-DTPA and the initial T, values. Software written for this procedure then deconvolves the blood curve from the tissue curve and fits the resultant curve to an adaptation of the model described by Tofts and Kermode in Magnetic Resonance in Medicine Vol. 17 (1991) at pages 357 to 367. This gives values for the rate constants Kl and K2 for the blood brain barrier permeability.

The above technique was tested on both flow phantoms and on normal volunteers. The processing of the results yielded T, values for phantoms (not shown) and brain tissue within 6% of those measured during the same investigation by a highly accurate SE/IR T, sequence. Preliminary clinical studies were then performed, which gave excellent data as depicted in Figure 7. The high resolution of MRI allows the exploration of important heterogeneities within tumours, whereas position emission tomography (PET) may only permit mean permeabilities to be found. The mean uptake permeability (Kl) of the melanoma metastasis, as seen from Figure 7, was found to be 0.0093 ml/min/g, whilst for a rapidly enhancing region, K 1=0.0152 ml/min/g. Normal white matter within the same slice gave K 1=0.0009 ml/min g. This agrees with PET tumour permeability data and shows the potential of this method for routine clinical use.

It will be appreciated that the present invention has been described by way of example only and variation to the embodiments described above may be possible without departing from the scope of the invention. In particular the sequences described may be used with a myriad of other imaging sequences. Some of which may use reversed polarity gradients for 90° and 180° pulses in spin-echo and multi-echo sequences.

The use with other spin echo, fast spin-echo or spin-echo- EPI sequences has been mentioned. The invention may also be employed in spectroscopy sequences where slice selection employs a spin-echo as a part of me localisation sequence, with a first 90° and a second 180° rf pulse applied in the same plane.

Claims

CLAIMS 1. A method of magnetic resonance imaging comprising the steps of : applying a first magnetic field gradient (G_sl) lobe, having a first polarity, applying a first rf pulse; subsequently applying a second magnetic field gradient lobe (G_s2), with opposite polarity to that of die first lobe; applying a third magnetic field gradient lobe (G_s3) with opposite polarity to the first lobe; and applying a second rf pulse.

2. A me od of magnetic resonance imaging according to Claim 1 wherein a first magnetic field gradient lobe is applied and a second magnetic field gradient lobe is applied subsequent to the first said lobe, the polarity of the second said lobe being opposite to the first said lobe and applying a 180° radio frequency (rf) slice selection pulse during application of the second said lobe.

3. A method according to Claim 1 or Claim 2 wherein the first rf pulse is a 90° if pulse and the second rf pulse is 180° rf pulse.

4. A method of magnetic resonance imaging according to any preceding claim wherein subsequent rf pulses are applied after application of the second rf pulse.

5. A method of magnetic resonance imaging according to Claim 4 wherein the subsequent rf pulses are 180° pulses.

6. A method of magnetic resonance imaging according to any preceding claim wherein the sequence of pulses is employed in a spin echo imaging sequence.

7. A method of magnetic resonance imaging according to any of Claims 1 to 5 wherein the imaging sequence is employed in echo planar imaging.

8. A method of magnetic resonance imaging according to any preceding claim wherein the method is employed in magnetic resonance angiography.

9. A method of magnetic resonance imaging comprising applying a gradient motion refocusing sequence, characterised in that a variable echo time is provided in said sequence.

10. Apparatus for magnetic resonance imaging comprising: means for establishing a magnetic field such that an object to be imaged may be positioned within said magnetic field; means for producing at least one orthogonal magnetic field gradient; means for varying said at least one magnetic gradient; and means for producing radio frequency (rf) pulses, the said pulses being capable of achieving, in use, an image of, or signal from, a selected slice of the object to be imaged, characterised in that the means for varying said at least one magnetic field gradient is arranged to apply an rf pulse with a first magnetic field gradient at a first polarity and a second rf pulse with a second magnetic field gradient at a polarity opposite to the first polarity.