This invention relates to magnetic resonance imaging (MRI). In particular, it relates to control of amplitude and phase of radio frequency coil arrays to achieve enhanced imaging.
MRI is moving inexorably toward higher field strength in search of improved signal-to-noise ratio and spectral resolution. At high fields (>4 Tesla) the precession frequency is higher and therefore the wavelength shorter. The tissue/field interactions become quite pronounced at high field and make obtaining high quality images very difficult. Beck has shown [Beck et al, MRM 51 Pg 1103-1107 2004] that at 11 T image intensity distributions are subject to significant distortions attributable to tissue/field interactions. Several methods such as optimizing the current distribution on the rungs of a multi-element volume coil [Ledden, Proc ISMRM 11 pg 2390 2003], transmitting field of time varying spatial characteristics using two separate pulses [Ledden & Cheng, Proc ISMRM 12 pg 38 2004] and the use of parallel imaging at high field strength [Weisinger et al, Proc ISMRM 12 Pg 323 2004] have been suggested to reduce the distortions.
A wide variety of parallel imaging techniques are available, such as the well known simultaneous acquisition of spatial harmonics (SMASH) [Sodikson & Manning, Magn Reson Med 38 pg 591-603, 1997] and sensitivity encoding (SENSE) [Pruessman et al, Magn Reson Med 42 pg 952-962, 1999]. Rapid parallel imaging is enabled by phased array coils as they allow scan time reduction compared to single coil acquisitions. The known techniques have typically used receive-only phased arrays but transceiver phased arrays are becoming more common.
Array technology was introduced by Roemer, as described in U.S. Pat. No. 4,825,162 assigned to General Electric Company. Roemer describes how images from an array of coils can be combined on a point-by-point basis to produce a single composite image. Roemer also describes how to minimise interactions between adjacent coils.
A significant amount of research directed to improving the usefulness on MRI has focussed on improving the signal to noise ratio (SNR). Larson described a focussing process in U.S. Pat. No. 5,252,922 that used a phased array of antennas to focus in a specific region of a body being imaged. Larson claims to have achieved focussing in volumes as small as 500-3000 cm3.
The SNR can be improved by applying higher fields but this can result in high Specific Absorption Rate (SAR), which may exceed regulatory guidelines. Peterson et al [Invest Rad 38 (7) pg 428-435 2003] has described a transceiver phased array for imaging the spinal cord at 3 Tesla. The Peterson system produced high resolution images. In a related paper [Invest Rad 38 (7) pg 443-451 2003], Peterson describes another transceiver phased array coil for imaging the pelvic region. Once again the phased array coils produced high resolution images.
Various array designs have been conceived for imaging various regions of the body. Persons skilled in the field will be familiar with common circular and rectangular array elements. Planar strip arrays are also known, such as described by Lee in United States patent application number 2003/0214299. Lee includes a useful background to MRI which is incorporated herein by reference. Lee also usefully notes that the current amplitude and phase of each transmit array element can be individually regulated such that the homogeneity of the RF excitation field is optimised in the presence of the patient.
Finally, reference may be had to International patent application number WO 2004/021025 filed by the present applicant. This patent application describes a coil array in which each coil element has its maximum sensitivity close to the centre of the object under study. The described coil array is useful for deep imaging of a body, for example cardiac imaging.
It is an object of the present invention to provide an improved method and apparatus for magnetic resonance imaging.
Further objects will be evident from the following description.
In one form, although it need not be the only or indeed the broadest form, the invention resides in a coil system for use with magnetic resonance imaging to form images of a region of interest, the coil system comprising:
a phased array having at least two coils;
means for driving each coil of said phased array at an amplitude and phase determined to focus at a target region; and
an interface subsystem linking said phased array to a magnetic resonance imaging system,
said magnetic resonance imaging system transmitting a substantially uniform magnetic field through said phased array to said region of interest and receiving a response to said substantially uniform magnetic field from said region of interest.
Preferably the phased array is a transceiver array which both transmits the field and receives the response.
In a further form the invention resides in a method of recording a magnetic resonance image of a target region including the steps of:
transmitting a substantially uniform magnetic field into said target region by driving a phased array having at least two coils at selected phase and amplitude values;
recording signals received from said target region in response to the substantially uniform magnetic field; and
processing said signals to construct an image.
In one alternate form the method may further include the steps of:
driving the phased array with a first set of phase and amplitude values and recording a first image;
driving the phased array with a second set of phase and amplitude values and recording a second image; and
constructing an improved image from said first image and said second image.
Suitably the step of constructing an improved image employs a process of averaging said first image and said second image. A suitable averaging process is point by point sum-of-squares averaging.
In a second alternate form the method may further include the steps of:
driving the phased array with a first set of phase and amplitude values and recording first time-domain signals;
driving the phased array with a second set of phase and amplitude values and recording second time-domain signals;
combining the first time-domain signals and the second time-domain signals; and
performing a multi-dimensional Fourier transform to construct the image.
To assist in understanding the invention preferred embodiments will now be described with reference to the following figures in which:
In describing different embodiments of the present invention common reference numerals are used to describe like features.
Referring to
A schematic block diagram of an MRI system for use with the invention is shown in
The pulse control module 24 also controls a RF Transceiver assembly 30 that provides signals of determined phase and amplitude to a transmit/receive switch 27 which directs the signals to the coils of the coil system 10.
The shimming coil assembly 23, gradient coil assembly 26 and coil system 10 operate in the B0 field of magnet 28, as will be well understood to persons skilled in the field.
The RF transceiver assembly 30 consists of amplitude control 31 and phase control 32 for each signal provided to the coil system 10. One embodiment for amplitude and phase control is shown in
The signals received by the coil system 10 are switched by the T/R switch 27 to a receiver 34 which has quadrature phase detector 35 and A/D converter 36. The T/R switch 27 may be radiofrequency circulators or circulator hybrids, or a group of individual T/R switches. The signals are passed to computer 21 to reconstruct an image of the region of interest.
Separate adjustment of phase and amplitude for each coil facilitates correction of the transmission field to account for tissue/field interactions. To exemplify the effectiveness of the technique, the B1 fields inside a simulated brain are calculated and used to construct signal intensity (SI) maps.
In the simulation of the invention there are four focusing transceive phased array coils and two separate scans of the same slice (Scan_A and Scan_B) with each scan using different excitations during transmission. That is, parallel, transmission phase cycling. Thereafter, the received B1 fields from the two separate scans are used for calculating the signal intensity (SI) maps, which are combined together in the image domain. By optimizing the phases of the transmission systems, signal distortion is significantly reduced.
A hybrid Finite Difference Time Domain (FDTD)/Method of Moments (MOM) approach is adopted in this work to accurately calculate B1 fields inside the brain. Firstly, four square surface coils resonating at 470 MHz, loaded with a four layered spherical phantom with each layer having different dielectric constants and conductivities that approximately resemble a human head is simulated using FEKO, a MOM based RF simulation program to calculate the current distribution on each coil (available from EM Software & Systems-SA (Pty) Ltd of Technopark, Stellenbosch, South Africa; www.feko.co.za). With the coils loaded with the spherical phantom, coil-sample coupling is considered. The dimension of each coil is 120×120 mm, positioned orthogonally and separated from its opposite coil neighbor with a distance of 246 mm as depicted in
a-b) shows the SI maps calculated for Scan_A and Scan_B while 5(c) shows the SI map when Scan_A and Scan_B are combined together using sum-of-squares.
Although a sum-of-squares averaging approach is described an improved image may also be constructed by first combining the time domain signals acquired in multiple scans and then performing multi-dimensional Fourier transforms to construct the image.
Two observations are evident from the simulation. Firstly, control of the phase and amplitudes of each coil in the coil system 10 is used to control the homogeneity of the transmitted field. This means that the position of a substantially homogeneous region is not limited to the geometric centre of the coil system. Secondly, phase amplitudes of successive scans may be controlled to provide complimentary fields that can be averaged to ameliorate unwanted tissue/field effects.
It will be appreciated that the embodiment described above utilizes four transceive phased array coils but the invention is not limited to such an arrangement. All that is required is that there are two or more coils. However, the inventors envisage that the invention is most advantageous when the coil system surrounds the region of interest.
It will further be appreciated that the invention is not limited to transceive coils. The description is applied to transceive coils for convenience. Separate systems can be used for transmit and receive functions.
As discussed above in the background section, previous uses of phased arrays have not controlled the phase and amplitude applied to each coil to achieve local focusing in a sample. It is evident from
A further embodiment of the invention is shown in
It will be noted from Table 2 that both the amplitude and phase values are changed between Scan_C and Scan_D. It is evident that the regions of homogeneity are different between Scan_C and Scan_D, and that the combined image is essentially homogeneous over the entire brain region. This is evident in the plots of
As the invention requires two scans of the same slice, it is advantageous to employ a rapid parallel imaging technique to minimize motion artifacts. A suitable reconstruction algorithm is GRAPPA [Griswold M A, Jakob P M, Hiedemann R M, Nittka M, Jellus V, Wang J M, et al. 2002. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med 47:1202-1210]. In GRAPPA, a small number of extra lines are acquired during the acquisition of the undersampled data. These extra lines are known as autocalibration signal (ACS) lines and are used to reconstruct missing k-space lines of each coil. To determine the weights used for reconstruction GRAPPA uses multiple blocks from all coils to fit to the ACS lines from that coil.
To implement GRAPPA reconstruction, a reduction factor of 2 and a total number of 21 ACS lines are used to implement the reconstruction procedure. Every even line of the synthetically derived k-space data is unsampled, and the 21 ACS lines are acquired at the center of the k-space. The missing lines in each k-space set are reconstructed and the fully reconstructed k-space data is Fourier transformed to produce two individual brain scans which are processed in a similar method to that described above. A brain image using the described method and incorporating GRAPPA is shown in
Another factor which can impact the quality of image obtained with the invention is the placement of the coils. When the coils are positioned at different locations, the B1 field pattern will also change. Therefore, if the coils are arranged in optimum location further improvement of the B1 field can be achieved. This is demonstrated by the plot of
The previous three embodiments have related to imaging of the brain. The invention also has application to other parts of the body. Shown in
The eight element phased array torso coil is first modeled using the FEKO software. The overall dimensions are 392 mm×312 mm×160 mm and the coil shape is designed to fit onto a voxel-based human torso model with frequency-dependent dielectric properties obtained from the United States Air Force (http://www.brooks.af.mil/AFRL/HED/hedr/).
The coil elements are typically positioned about 15 mm from the model. Distributed capacitors are placed at equal intervals in each single coil element to reduce frequency shifts associated with dielectric loading. Using conventional circuitry, the torso phased array coil is tuned to 85 MHz and matched to 50 Ohms when loaded with the body model. The individual coil dimensions are 160 mm×120 mm for coils 81-84 and 160 mm×168 mm for coils 85-88. FEKO automatically includes coil-coil and coil-sample coupling in its calculations, a combination of coil overlapping and the insertion of decoupling capacitors between coils, where they are not overlapped, are used to minimize the extrinsic coupling, as is common in reception-only systems.
An FDTD/MOM technique is used as described above. The outputs of the FDTD calculations are the magnitudes and phases of the sinusoidal steady-state electric and magnetic fields, which are registered within a period of the operating frequency. After the FDTD calculation transient response has passed, the peak EMF data are recorded. The elapsed (transient) time is also recorded to calculate the phases of the steady EMF. In this way, the complex electric and magnetic field vectors at all vertices on the grid of FDTD cells can be generated. The SI are calculated and displayed.
To implement the FDTD calculation, the cell size is 4 mm and the entire computational domain is divided into Nx×Ny×Nz=114×90×80 820.8×103 Yee cells. Ten PML layers with a parabolic conductivity profile are located on the six open sides, which are positioned at the anterior, posterior, the two lateral sides, the top and bottom planes with respect to the torso model. The overall structure is shown in
The benefit of controlling phase and amplitude on each coil are exemplified by applying the three drive schemes tabulated in Table 3.
From the simulations, the magnetic flux density with focused |B1| field is calculated for each coil element and used as the sensitivity profiles for the SENSE reconstruction. Using point-wise multiplication of the sensitivity profiles with the numerical chest images yields individual chest images. A reduction factor of 4 is employed for the simulation of SENSE reconstruction. Signal is separated from the aliased images using an unfolding matrix and SENSE reconstruction of the non-aliased chest image is shown in
The simulation clearly shows that the focusing scheme delivers improved homogeneity across the targeted regions, as demonstrated early with the four-coil system. Furthermore, when used with a multi-scan process the focusing scheme produces improved field homogeneity across a target region and therefore improved SNR with reduced SAR.
Throughout the specification the aim has been to describe the invention without limiting the invention to any particular combination of alternate features.
Number | Date | Country | Kind |
---|---|---|---|
2005901158 | Mar 2005 | AU | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/AU2006/000311 | 3/9/2006 | WO | 00 | 3/31/2008 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO2006/094354 | 9/14/2006 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
3937955 | Comisarow et al. | Feb 1976 | A |
4682112 | Beer | Jul 1987 | A |
5086275 | Roemer | Feb 1992 | A |
5179332 | Kang | Jan 1993 | A |
5510711 | Molyneaux et al. | Apr 1996 | A |
5719499 | Chandrakumar | Feb 1998 | A |
5861749 | Van Heelsbergen | Jan 1999 | A |
5903150 | Roznitsky | May 1999 | A |
6236205 | Ludeke et al. | May 2001 | B1 |
6348791 | Shattil | Feb 2002 | B2 |
6411090 | Boskamp | Jun 2002 | B1 |
6486667 | Wu et al. | Nov 2002 | B1 |
6608480 | Weyers | Aug 2003 | B1 |
6727703 | Lee | Apr 2004 | B2 |
6771067 | Kellman et al. | Aug 2004 | B2 |
6825660 | Boskamp | Nov 2004 | B2 |
6906518 | Leussler | Jun 2005 | B2 |
6952100 | McKinnon et al. | Oct 2005 | B1 |
6995561 | Boskamp et al. | Feb 2006 | B2 |
7012430 | Misic | Mar 2006 | B2 |
7064546 | Feiweier | Jun 2006 | B2 |
7075299 | Peters | Jul 2006 | B1 |
7091721 | Jevtic | Aug 2006 | B2 |
7123010 | Krockel | Oct 2006 | B2 |
7176688 | Sato | Feb 2007 | B2 |
7245752 | Oesingmann | Jul 2007 | B2 |
7253622 | Saylor et al. | Aug 2007 | B2 |
7336074 | Yang et al. | Feb 2008 | B2 |
7446528 | Doddrell et al. | Nov 2008 | B2 |
20030214299 | Lee et al. | Nov 2003 | A1 |
20060261811 | Ham et al. | Nov 2006 | A1 |
Number | Date | Country |
---|---|---|
2227095 | Jul 1990 | GB |
2337819 | Dec 1999 | GB |
WO 9709632 | Mar 1997 | WO |
WO 2005111645 | Nov 2005 | WO |
Number | Date | Country | |
---|---|---|---|
20080272785 A1 | Nov 2008 | US |