Zero echo time MR imaging with water/fat separation

Abstract

The invention relates to a method of MR imaging of an object positioned in an examination volume of a MR device (1), the method comprises the steps of:subjecting the object (10) to an imaging sequence of RF pulses (20) and switched magnetic field gradients(G), which imaging sequence is a zero echo time sequence comprising: i) setting a readout magnetic field gradient (G) having a readout direction and a readout strength; ii) radiating a RF pulse (20) in the presence of the readout magnetic field gradient (G); iii) acquiring a FID signal in the presence of the readout magnetic field gradient (G), wherein the FID signal represents a radial k-space sample; iv) gradually varying the readout direction; v) sampling a spherical volume in k-space by repeating steps i) through iv) a number of times, with the readout strength being varied between repetitions;reconstructing a MR image from the acquired FID signals, wherein signal contributions of two or more chemical species to the acquired FID signals are separated. It is an object of the invention to enable silent ZTE imaging in combination with water/fat separation. This is achieved by varying the readout strength such that each position in k-space is sampled at least two times, each time with a different value of the readout strength. Moreover, the invention relates to a MR device and to a computer program for a MR device.

Claims

1. A method of magnetic resonance (MR) imaging of an object positioned in an examination volume of a MR device, the method comprising the steps of: subjecting the object to an imaging sequence of radio frequency (RF) pulses and switched magnetic field gradients (G), which imaging sequence is a zero echo time sequence comprising: i) setting a readout magnetic field gradient (G) having a readout direction and a readout strength; ii) radiating a RF pulse in the presence of the readout magnetic field gradient (G); iii) acquiring a free induction decay (FID) signal in the presence of the readout magnetic field gradient (G), wherein the FID signal represents a radial k-space sample; iv) gradually varying the readout direction; v) sampling a spherical volume in k-space by repeating steps i) through iv) a number of times, with the readout strength being varied between repetitions; wherein the readout strength is varied such that individual positions in k-space are sampled at least two times, each time with a different value of the readout strength, such that said k-space position is sampled at two or more different sampling times and reconstructing a MR image from the acquired FID signals, wherein signal contributions of two or more chemical species to the acquired FID signals are separated, wherein the signal contributions of the two or more chemical species to the FID signals are derived from phase differences of the acquired FID signals induced by the variation of the readout strength and the separation of the signal contributions is performed on the basis of a signal model including at least the MR spectrum of each of the chemical species.

2. The method of claim 1, wherein the signal model further includes the inhomogeneity of the main magnetic field in the examination volume.

3. The method of claim 2, wherein a phase map is derived from the acquired FID signals, wherein the inhomogeneity of the main magnetic field is derived from the phase map by exploiting that the phase shift induced by the inhomogeneity of the main magnetic field varies smoothly over space.

4. The method of claim 1, wherein the readout strength is varied by switching it between two or more pre-selected values.

5. The method of claim 4, wherein k-space is sampled in a segmented fashion, each segment having the shape of a hollow sphere of a given wall thickness, wherein a different combination of the two or more pre-selected values is applied in sampling of each segment.

6. The method of claim 1, wherein the spherical k-space volume is sampled by randomly varying the readout direction and the readout strength.

7. The method of claim 1, wherein compressed sensing is employed for reconstructing the MR image and/or for separating the signal contributions of the two chemical species.

8. A magnetic resonance (MR) device comprising at least one main magnet coil for generating a uniform, steady magnetic field within an examination volume, a number of gradient coils for generating switched magnetic field gradients in different spatial directions within the examination volume, at least one radio frequency (RF) coil for generating RF pulses within the examination volume and/or for receiving MR signals from an object positioned in the examination volume, a control unit for controlling the temporal succession of RF pulses and switched magnetic field gradients, and a reconstruction unit, wherein the MR device is arranged to perform the following steps: subjecting the object to an imaging sequence of RF pulses and switched magnetic field gradients, which imaging sequence is a zero echo time sequence comprising: i) setting a readout magnetic field gradient having a readout direction and a readout strength; ii) radiating a RF pulse in the presence of the readout magnetic field gradient (G); iii) acquiring a free induction decay (FID) signal in the presence of the readout magnetic field gradient (G), wherein the FID signal represents a radial k-space sample; iv) incrementally varying the readout direction; v) sampling a spherical volume in k-space by repeating steps i) through iv) a number of times, with the readout strength being varied between repetitions wherein the readout strength is varied such that individual positions in k-space are sampled at least two times, each time with a different value of the readout strength, such that said k-space position is sampled at two or more different sampling times and; reconstructing a MR image from the acquired FID signals, wherein signal contributions of two or more chemical species to the acquired FID signals are separated.

9. A non-transitory computer readable medium to be run on a magnetic resonance (MR) device, which comprises instructions for: generating an imaging sequence of radio frequency (RF) pulses and switched magnetic field gradients (G), which imaging sequence is a zero echo time sequence comprising: i) setting a readout magnetic field gradient (G) having a readout direction and a readout strength; ii) radiating a RF pulse in the presence of the readout magnetic field gradient; iii) acquiring a free induction decay (FID) signal in the presence of the readout magnetic field gradient, wherein the FID signal represents a radial k-space sample; iv) incrementally varying the readout direction; v) sampling a spherical volume in k-space by repeating steps i) through iv) a number of times, with the readout strength being varied between repetitions wherein the readout strength is varied such that individual positions in k-space is sampled at least two times, each time with a different value of the readout strength, such that said k-space position is sampled at two or more different sampling times and; wherein the signal contributions of the two or more chemical species to the FID signals are derived from phase differences of the acquired FID signals induced by the variation of the readout strength and the separation of the signal contributions is performed on the basis of a signal model including at least the MR spectrum of each of the chemical species and reconstructing a MR image from the acquired FID signals, wherein signal contributions of two or more chemical species to the acquired FID signals are separated.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The enclosed drawings disclose preferred embodiments of the present invention. It should be understood, however, that the drawings are designed for the purpose of illustration only and not as a definition of the limits of the invention. In the drawings:

(2) FIG. 1 schematically shows a MR device for carrying out the method of the invention;

(3) FIG. 2 shows a diagram illustrating the ZTE sequence applied according to the invention;

(4) FIG. 3 illustrates the radial sampling of k-space according to an embodiment of the invention using two different readout strengths;

(5) FIG. 4 illustrates the segmented k-space sampling approach of the invention;

(6) FIGS. 5 and 6 illustrate an iterative scheme for separating chemical shift from spatial inhomogeneity of the main magnetic field in the image reconstruction step of the method of the invention;

(7) FIG. 7 illustrates random k-space sampling according to a further embodiment of the invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

(8) With reference to FIG. 1, a MR device 1 which can be used for carrying out the method of the invention is shown. The device comprises superconducting or resistive main magnet coils 2 such that a substantially uniform, temporally constant main magnetic field B.sub.0 is created along a z-axis through an examination volume. The device further comprises a set of (1.sup.st, 2.sup.nd, andwhere applicable3.sup.rd order) shimming coils 2, wherein the current flow through the individual shimming coils of the set 2 is controllable for the purpose of minimizing B.sub.0 deviations within the examination volume.

(9) A magnetic resonance generation and manipulation system applies a series of RF pulses and switched magnetic field gradients to invert or excite nuclear magnetic spins, induce magnetic resonance, refocus magnetic resonance, manipulate magnetic resonance, spatially and otherwise encode the magnetic resonance, saturate spins, and the like to perform MR imaging.

(10) More specifically, a gradient pulse amplifier 3 applies current pulses to selected ones of whole-body gradient coils 4, 5 and 6 along x, y and z-axes of the examination volume. A digital RF frequency transmitter 7 transmits RF pulses or pulse packets, via a send-/receive switch 8, to a -body RF coil 9 to transmit RF pulses into the examination volume. A typical MR imaging sequence is composed of a packet of RF pulse segments of short duration which taken together with each other and any applied magnetic field gradients achieve a selected manipulation of nuclear magnetic resonance. The RF pulses are used to saturate, excite resonance, invert magnetization, refocus resonance, or manipulate resonance and select a portion of a body 10 positioned in the examination volume. The MR signals are also picked up by the body RF coil 9.

(11) For generation of MR images of limited regions of the body 10 by means of parallel imaging, a set of local array RF coils 11, 12, 13 are placed contiguous to the region selected for imaging. The array coils 11, 12, 13 can be used to receive MR signals induced by body-coil RF transmissions.

(12) The resultant MR signals are picked up by the body RF coil 9 and/or by the array RF coils 11, 12, 13 and demodulated by a receiver 14 preferably including a preamplifier (not shown). The receiver 14 is connected to the RF coils 9, 11, 12 and 13 via send-/receive switch 8.

(13) A host computer 15 controls the current flow through the shimming coils 2 as well as the gradient pulse amplifier 3 and the transmitter 7 to generate a ZTE imaging sequence according to the invention. The receiver 14 receives a plurality of MR data lines in rapid succession following each RF excitation pulse. A data acquisition system 16 performs analog-to-digital conversion of the received signals and converts each MR data line to a digital format suitable for further processing. In modern MR devices the data acquisition system 16 is a separate computer which is specialized in acquisition of raw image data.

(14) Ultimately, the digital raw image data is reconstructed into an image representation by a reconstruction processor 17 which applies an appropriate reconstruction algorithm. The MR image represents a a three-dimensional volume. The image is then stored in an image memory where it may be accessed for converting projections or other portions of the image representation into appropriate format for visualization, for example via a video monitor 18 which provides a human-readable display of the resultant MR image.

(15) FIG. 2 shows a diagram illustrating the ZTE sequence applied according to the invention. The essence of the silent ZTE technique is that excitation RF pulses 20 are transmitted simultaneously with frequency-encoding readout magnetic field gradients G being switched on. The readout magnetic field gradient G is not intended as a slice-selection gradient which implies that the RF pulses 20 have to be extremely short (typically 1 s to 8 s) in order to achieve sufficient excitation bandwidth. The readout of FID signals takes place during intervals 21 in the presence of the readout magnetic field gradients G immediately after the RF pulses 20. Each interval 21 has a duration between 100 s and 3 ms. The readout magnetic field gradient G has a readout strength and a readout direction both staying substantially constant over each excitation/readout cycle. After each cycle the readout direction is varied only very gradually. The readout direction changes only slightly, e.g. by a few degrees (e.g. 2). In a practical example, the magnetic field gradient in one spatial direction ramps up from zero to full in about 45 ms. For a full sampling of k-space the readout direction is varied until a spherical volume is covered with sufficient density.

(16) A known constraint of the ZTE technique is that there is a finite time between the center of each RF pulse 20 and the start of the sampling interval 21. Depending on the equipment used, this dead time may be anything between 2 s and 20 s. This means that the center of k-space cannot be scanned. However, it has to be taken into account that the size of the central k-space volume that cannot be sampled depends on the readout strength. The lower the strength of the magnetic field gradient, the smaller is the central k-space region that will not be sampled during the dead time. On the other hand, it is not feasible to apply as weak as possible readout gradients.

(17) According to the invention, the strength of the readout magnetic field gradient G is varied between repetitions of the ZTE sequence. This is illustrated in the diagram of FIG. 3 showing the interdependence of the k-space position k and the sampling time t (k actually represents three dimensions from which only one is drawn for the purpose of illustration). The application of different readout strengths low G and high G implies that each k-space position is sampled at two or more different sampling times (i.e. the time interval between the RF pulse and the sampling of a given k-space position). As can be seen in FIG. 3, the k-space position k.sub.example is visited two times during the scan, namely at t.sub.s1 (using readout strength high G) and at t.sub.s2 (using readout strength low G). The sampling of each k-space position with two or more different sampling times results in a specific phasing of the acquired FID signals which is induced by the precessional frequency difference of, e.g., hydrogen in fat and water. This is exploited in accordance with the invention to separate the signal contributions from fat and water as in the per se known phase unwrapping techniques applied in Dixon-type MR imaging.

(18) FIG. 4 illustrates an embodiment of the invention employing a segmented k-space sampling approach, wherein each segment has the shape of a hollow sphere of a given wall thickness. k-space is to be sampled up to k.sub.max. The required gradient strength would be (approximately):

(19) $\frac{k_{\max }}{T_R},$
wherein is the gyro-magnetic ratio and T.sub.R is the repetition time of the ZTE sequence. This value will be referred to as:
G.sub.ref
A variable is introduced:

(20) $a=\sqrt{\frac{T_R}{T_{\mathrm{deadtime}}}},$
wherein T.sub.deadtime is the dead time during which no signal acquisition is possible. A typical value of is 5. In this embodiment of the invention, FID signals are acquired with the following set of gradient strengths:
.sup.1G.sub.ref, .sup.0G.sub.ref, .sup.1G.sub.ref, .sup.2G.sub.ref, .sup.3G.sub.ref,

(21) Mathematically, this is an infinite series. However, in practice acquisition may be stopped beyond .sup.3G.sub.ref or .sup.4G.sub.ref. One additional acquisition should be performed with G=0.

(22) It has to be noted that this proceeding does not result in a large number of extra acquisitions in comparison to a conventional ZTE scan (employing only acquisitions with .sup.0G.sub.ref). Considering the required sampling density of the inner k-space spheres of the proposed segmentation, only a limited number of additional radial k-space samples need to be acquired. Hence, the total number of required cycles of the ZTE sequence may be only about twice the number of cycles in a conventional ZTE scan with comparable imaging parameters.

(23) As can be seen in FIG. 4, each segment 1-4 is sampled with a different combination of two different readout strengths. Simultaneously, an optimal coverage of central k-space (segment 4) is achieved. One might easily increase the number of readout strengths per segment, for example by choosing

(24) $a=\sqrt[3]{\frac{T_R}{T_{\mathrm{deadtime}}}}$
and starting at .sup.2G.sub.ref.

(25) With reference to FIGS. 5 and 6 an iterative scheme for separating chemical shift from main magnetic field inhomogeneity in the image reconstruction step of the method of the invention is explained in the following.

(26) In this embodiment, the reconstruction and the water/fat separation consists of two steps: (a) estimating a phase map, i.e. a map reflecting both main magnetic field inhomogeneity and chemical shift effects (and maybe further phase shift-inducing effects), and (b) separating chemical shift from main magnetic field inhomogeneity by the assumption that the latter varies smoothly over space. Step (b) constitutes the well-known phase unwrapping problem of Dixon water/fat imaging. Since suitable algorithms are well-known and available in existing MR environments this does not need to be further elaborated here.

(27) Step (a) is performed iteratively. The reconstruction step comprises calculating two sets of information over space: (i) the magnetization density (i.e. the water and fat MR image), and (ii) an estimate of the phase map. At each iteration step, these sets are calculated up to a given resolution (i.e. within a full sphere in k-space).

(28) It is assumed that initially estimates of the above two sets of information are available for a small central region of k-space region. In the embodiment shown in FIG. 5, both magnetization density and the phase map are assumed to be known for regions 4 and 3 (indicated by the bold horizontal line 50 at t=0). As a next step of the iteration, estimates are to be computed including region 2.

(29) The dashed line in FIG. 5 represents the average sampling time for the set of k-space samples of region 2. The phase map is known for the sphere enclosed by region 2 (i.e., regions 4 and 3 in this embodiment). This knowledge is applied in reconstructing region 2. Both datasets of region 2 are reconstructed as if they were acquired using the average (dashed) timing of k-space sampling, for example by using a segmented homogeneity correction method (see Douglas C. Noll et al., IEEE Transactions on Medical Imaging, 10, 629-637, 1991). For this purpose, it is useful to sub-segment region 2 into regions 2a, . . . 2d, as depicted in FIG. 5. In this way, some distortion caused by magnetic field inhomogeneity is intentionally left in the data. The data reconstructed up to this point behaves as if it were acquired with the sampling timing shown in FIG. 6.

(30) For the central k-space region, both the magnetization density and the phase map are known, as mentioned before. Hence, the signal data can be simulated at any sampling time. In this way, simulated data 60 is added in the central k-space region as indicated by the bold dotted lines in FIG. 6. From these two data sets, the average and the difference, are calculated. Transforming the average and the difference to the spatial domain enables calculation of high-resolution (i.e. including region 2) estimates of the magnetization density and the phase map. In the next step, this process is performed including region 1, and the reconstruction step (a) is accomplished. The iteration may start by estimating a 0-th order estimate of the phase map from the G=0 k-space sample, which can be considered as the most central region in k-space.

(31) On this basis, the separation of chemical shift from main magnetic field inhomogeneity can be performed in step (b), as mentioned above, by the assumption that the latter varies smoothly over space. Algorithms known in the art for Dixon water/fat imaging may be employed for reconstructing separate water and fat images from the magnetization density and the (inhomogeneity-corrected) phase map.

(32) Another embodiment of the invention is in the following discussed with reference to FIG. 7. In this embodiment, the gradient coils along the x, y and z-axes are controlled such that the readout strengths in the respective directions assume mutually independent random values between repetitions of the ZTE sequence, with the noise being frequency-restricted to about 15 Hz or less, in order not to be audible. FID signals are acquired, with a typical duration of each cycle of one millisecond. After, e.g., 200 seconds of scan time, 200.000 FID signals are available, acquired with a distribution of readout directions and readout strengths.

(33) The diagram of FIG. 7 shows the sampling time t in relation to the tangential component in k-space for a given radius k.sub.r. The central dashed line represents the average sampling time or reference sample time t.sub.ref. Each of the dots in the diagram represents a FID signal having its characteristic sampling time t at the moment of reaching k.sub.r. The resulting data can be considered as comprising a few million points, each with its characteristic values of k.sub.x, k.sub.y, k.sub.z and t. Their (complex) values are
s.sub.i=s(k.sub.i,t.sub.i)
with i being the index of the point.

(34) For the step of reconstruction and water/fat separation, again the component (a) of estimating the phase map is focused on in the following.

(35) A region size in k-space is defined such that it can be made sure that it includes, in most cases, at least two points s.sub.i with substantially different values of t. For each point,

(36) $s_{A,i}=\frac{1}{\underset{k_{j}N\left(k_i\right)}{.Math.}{}_j}{s}_i$
is calculated. Herein, N(k.sub.i) should be read as neighborhood of k.sub.i, and .sub.i should be read as point present in neighborhood. In essence, s.sub.A,i can be interpreted as density-compensated data-point.

(37) Further, s.sub.B,i is calculated as

(38) $\frac{\left(t_i-t_{\mathrm{ref}}\right).Math.\underset{k_{j}N\left(k_i\right)}{.Math.}{}_j-\underset{k_{j}N\left(k_i\right)}{.Math.}\left(t_j-t_{\mathrm{ref}}\right)}{\left(\underset{k_{j}N\left(k_i\right)}{.Math.}{}_j\right).Math.\underset{k_{j}N\left(k_i\right)}{.Math.}{\left(t_j-t_{\mathrm{ref}}\right)}^2-{\left(\underset{k_{j}N\left(k_i\right)}{.Math.}\left(t_j-t_{\mathrm{ref}}\right)\right)}^2}{s}_i,$
which can be interpreted as signal weighted by the difference of the actual sampling time of the data point and the average sampling time in the neighborhood, normalized over the local variance of the sampling time. In essence, it represents the slope of the signal with respect to sampling time.

(39) As a next step s.sub.C,i is calculated as

(40) $\frac{\left(t_i-t_{\mathrm{ref}}\right).Math.\underset{k_{j}N\left(k_i\right)}{.Math.}{}_j-\underset{k_{j}N\left(k_i\right)}{.Math.}\left(t_j-t_{\mathrm{ref}}\right)}{\left(\underset{k_{j}N\left(k_i\right)}{.Math.}{}_j\right).Math.\underset{k_{j}N\left(k_i\right)}{.Math.}{\left(t_j-t_{\mathrm{ref}}\right)}^2-{\left(\underset{k_{j}N\left(k_i\right)}{.Math.}\left(t_j-t_{\mathrm{ref}}\right)\right)}^2}{s}_i.Math.\left(t_i-t_{\mathrm{ref}}\right),$
which provides an estimate of how much the signal deviates from what it would have been if it had been measured at t.sub.ref.

(41) A three-dimensional image I.sub.ref is reconstructed from the difference s.sub.A,i-s.sub.B,i. This reconstruction may be performed using a compressed sensing approach. A further three-dimensional image I.sub.slope is reconstructed from the data points s.sub.B,i. As a next step

(42) $-i\frac{I_{\mathrm{slope}}}{I_{\mathrm{ref}}}$
is calculated, wherein i is the imaginary unit and y is the gyro-magnetic ratio. The result is a direct estimate of the phase map (in units of Tesla). On this basis, again, the separation of chemical shift from main magnetic field inhomogeneity can be performed by assuming that the latter varies smoothly over space, and per se known algorithms can then be employed for reconstructing separate water and fat images.