U.S. patent application number 13/491674 was filed with the patent office on 2012-09-27 for cylindrical bi-planar gradient coil for mri.
Invention is credited to Jianyu Lian, Grum Teklemariam.
Application Number | 20120240385 13/491674 |
Document ID | / |
Family ID | 34375480 |
Filed Date | 2012-09-27 |
United States Patent
Application |
20120240385 |
Kind Code |
A1 |
Teklemariam; Grum ; et
al. |
September 27, 2012 |
Cylindrical Bi-Planar Gradient Coil for MRI
Abstract
Cylindrical bi-planar gradient coil assemblies for use in open
magnetic resonance imaging, wherein each of the coil assemblies
contains in sequential order (i) a circular primary coil set placed
flat above a cylindrical planar substrate, (ii) cooling means,
(iii) 0th and 2nd order shims, (iv) shield layers, and (v) 1st
order shims. In use the gradient coil assemblies are disposed
symmetrically to each other about a plane of symmetry parallel to
each.
Inventors: |
Teklemariam; Grum;
(Washington, DC) ; Lian; Jianyu; (Westford,
MA) |
Family ID: |
34375480 |
Appl. No.: |
13/491674 |
Filed: |
June 8, 2012 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10944080 |
Sep 17, 2004 |
8203341 |
|
|
13491674 |
|
|
|
|
60504371 |
Sep 19, 2003 |
|
|
|
Current U.S.
Class: |
29/592.1 |
Current CPC
Class: |
G01R 33/3856 20130101;
G01R 33/385 20130101; G01R 33/3875 20130101; G01R 33/3802 20130101;
Y10T 29/49002 20150115 |
Class at
Publication: |
29/592.1 |
International
Class: |
H05K 13/00 20060101
H05K013/00 |
Claims
1. A method for designing a plurality of coil sets comprising
planar X and Y-primary coil sets and planar X and Y-shield coil
sets of a bi-planar gradient coil assembly said method comprising
the steps of: specifying a surface primary and shield current
distributions as well as a plurality of magnetic fields in polar
coordinates and have reciprocal space expressions related by a
Fourier-Hankel transform; and obtaining a surface transverse
primary and shield spatial current distributions that generate a
plurality of desired transverse gradient magnetic fields by: i.
specifying polar surface current distributions on four coplanar
planes that share a common z-axis through their respective polar
centers so the z-axis is in an orthogonal relation to the planes,
setting the z-axis center equidistant between two inner planes so
that these planes are positioned symmetrically at z=.+-.a, then
defining the current distributions on these planes as the primary
coils where a plurality of primary gradient magnetic fields are
generated therebetween; ii. specifying an outer plane to be
positioned symmetrically at z=.+-.b, defining the current
distributions on these planes as the shield coils so that a
plurality of shield gradient magnetic fields are generated in
opposition to the primary gradient magnetic fields, together
producing a total gradient magnetic field in the space between
z<a and z>-a, and zero magnetic field in the space above
z>b and below z<-b; iii. using a shielding condition of
producing zero magnetic field in the space above z>b and below
z<-b to obtain an algebraic relation between the primary and
shield current distributions in reciprocal polar space; iv. using
this primary and shield current distribution relation to express a
total shielded magnetic field in terms of just the primary current
distribution; v. expressing the shielded gradient magnetic field in
its Fourier-Hankel expansion so that a current density is in its
reciprocal space representation; vi. evaluating the magnetic field
expression at specified discrete values, these points all having a
same axial position z=c, where c<a, and a same polar angle
.phi., but different radial positions .rho.j, forming a term where
for each j the magnetic field value so expressed is subtracted from
the desired magnetic field values at that particular jth location,
then summing all these N terms where each term is multiplied by a
different constant, the Lagrange multipliers, and this sum added to
an energy formula in its reciprocal space representation through a
Fourier-Hankel expansion to finally yield an expression which is an
energy functional; vii. the energy functional differentiated with
respect to the current distribution and set to zero, subsequently
solving the resultant expression to obtain an extremized current
distribution in reciprocal space which becomes a sum over the
Lagrange multipliers multiplied by terms containing a plurality of
corresponding radial constraint points; viii. the extremized
current density substituted back into the expression for the
shielded gradient magnetic field that will contain the sum over the
N discrete radial positions with the associated multiplicative
Lagrange multipliers; ix. the field calculated at each discrete
constraint value and set equal to the corresponding constraint
discrete magnetic field value to form a matrix equation of
calculated field values at the discrete constraint radial positions
times the corresponding Lagrange multipliers equaling the
constraint magnetic field values; x. inverting the matrix equation
to finally obtain the Lagrange multipliers; xi. substituting the
Lagrange multiplier values in the expression for the extremized
current density; xii. the primary and shield magnetic fields
calculated from the extremized current distribution by an inverse
Fourier-Hankel transform to obtain the spatial field distributions,
an efficiency and a linearity of the gradient magnetic fields so
produced calculated, further an inductance also calculated from
said current distribution; xiii. a parameter constructed as a
figure of merit equal to the efficiency squared and divided by the
product of the inductance times the square-root of the linearity
value; and xiv. optimizing the figure of merit by a Nelder-Mead
simplex optimization algorithm that has an associated cost function
to limit the planar or radial extent of the current distribution
whereby in each iteration of the algorithm the extremized current
distribution is obtained first, subsequently the fields, linearity
and inductances calculated to get the figure of merit, then the
radial constraint points with the associated magnetic field values
varied by the simplex algorithm to obtain a new extremized current
distribution from which to calculate a new figure of merit, the
algorithm terminating when no change is obtained in the figure of
merit finally yielding the desired transverse primary and shield
current distributions.
2. A method for designing the Z-primary coil and the Z-shield coil
of the gradient coil assembly of claim 1, said method further
comprising the steps of: expanding an axial magnetic field in terms
of a Fourier-Hankel expansion, inverting this expression in
reciprocal polar space so that a current density can be specified
as a function of the magnetic field; specifying two regions of
constant axial magnetic fields as a function of radial position on
an interior planar surface coaxial and parallel to a plurality of
planar axial primary current distributions; further specifying an
outer axial magnetic field in opposite relation to the previous two
axial magnetic fields to limit a plurality of radial extents of the
planar axial primary and shield current distributions;
Fourier-Hankel transforming the field distributions and inputting
them in said primary current distribution in reciprocal polar
space; obtaining an associated planar axial shield current
distribution from said primary current distribution and calculating
an interior axial shielded magnetic field from said Fourier-Hankel
magnetic field expression so an efficiency and a linearity of the
magnetic fields is found; further calculating a corresponding
inductance from said planar axial primary and shield current
distributions; forming a figure of merit by squaring the efficiency
and dividing it by a product of the inductance and a square root of
the linearity; optimizing the figure of merit by a Nelder-Mead
simplex optimization algorithm where the simplex algorithm varies a
range and a plurality of amplitudes of the three target field
distributions to obtain a new current distribution from which a new
figure of merit is obtained by calculating the corresponding axial
shielded magnetic fields, and thus the efficiency and linearity,
along with the inductance over each iteration, the algorithm
terminating when no change in the figure of merit is reached to
yield an optimized planar axial primary and shield current
distribution.
3. A method to define a planar shim current distribution for a
gradient coil assembly to generate spatial magnetic field
distributions corresponding to terms in a spherical harmonic
expansion of a plurality of main magnetic fields, said method
comprising the steps of: specifying polar, surface shim current
distributions placed on two parallel planes sharing a common z-axis
to produce a shim magnetic field there between; expressing the shim
magnetic field in a Fourier-Hankel expansion, transforming the shim
magnetic fields to its reciprocal space representation, inverting
the expression so that the planar shim current distribution can be
specified as a function of the shim magnetic field in reciprocal
space; each term in the spherical harmonic expansion of the main
magnetic field converted to a plurality of cylindrical coordinates
and multiplied by a radial envelope function to constrain radial
extent, then Fourier-Hankel transformed to reciprocal space, the
function entered as the shim magnetic field in reciprocal field for
the shim current distribution in reciprocal space; and the shim
current distribution inverse Fourier-Hankel transformed to obtain a
continuous radial and angular distribution in cylindrical
coordinates.
4. The planar current distributions solved for in claim 1 for the
transverse, axial and shim currents being continuous surface
distributions are discretized to form conductor placement positions
wherein: if conductors of a same cross-sectional area are to be
used, a continuous central conductor path can be obtained by
constructing a plurality of corresponding stream functions and
generating a plurality of contours of constant intervals of an
integrated current; if the continuous conductor paths are to be
etched from a plurality of solid conductors the etch paths are
formed in between said equal, constant intervals of integrated
currents from the corresponding stream functions.
5. A method of manufacturing the coils of claim 1, comprising the
steps of: cutting a metal plate in the shape of a circular coil
creating a coil and a curf portion; removing the curf position from
the metal plate creating a curf void around the coil; placing a
first insulation layer into the former; placing the coil into the
former over the first insulation layer; placing a second insulation
layer over the coil in the former; and compressing the insulating
layers and the coil with a control weight.
6. A method of manufacturing the coils of claim 1, comprising the
steps of: cutting a metal plate creating a noncontiguous curf
portion and a coil with a plurality of tabs interconnecting
sections of the coil; removing the curf portion from the metal
plate creating a curf void around the coil; placing an epoxy into
the curf void around coil; and removing the plurality of tabs from
the coil.
7. The method of claim 6, further comprising the steps of:
fabricating a former having an open top and a closed planar bottom;
placing a first insulating layer inside the former and onto the
former planar bottom; adhering the coil and epoxy onto the first
insulating layer; adhering a second insulating layer on top of the
coil and epoxy; and compressing the first and second insulating
layers, the coil and the epoxy with a control weight.
8. A method of manufacturing the coils of claim 1, comprising the
steps of: cutting a metal plate in the shape of a circular coil
creating a coil portion and a curf portion; removing the curf
portion from the metal place creating a curf void round the coil;
forming a formed insulation layer having curf tracks; and placing
the coil on the formed insulation between the curf tracks.
9. The method of claim 8, further comprising the steps of:
fabricating a former having an open top and a closed planar bottom;
placing the coil and the formed insulation layer into the former;
adhering a second insulating layer on top of the coil and the
formed insulation layer; and compressing the second insulating
layer, the coil and the formed insulation layer with a control
weight.
10. A method of manufacturing a gradient coil assembly, comprising
the steps of: fabricating a former having an open top and a closed
planar bottom; placing an insulation layer inside the former and
onto the former planar bottom; fabricating and X-primary coil with
a coil portion and a curf void portion; placing the X-primary coil
into the former; forming an insulation layer in the X-primary coil
curf void portion and on top of the X-primary coil portion;
applying control weight while the insulation layer cures;
fabricating a Y-primary coil with a coil portion and a curf void
portion; placing the Y-primary coil into the former; forming an
insulation layer in the Y-primary coil curf void portion and on top
of the Y-primary coil portion; applying control weight while the
insulation layer cures; fabricating a Z-primary coil with a coil
portion and a curf void portion; placing the Z-primary coil into
the former; forming an insulation layer in the Z-primary coil curf
void portion and on top of the Z-primary coil portion; applying
control weight while the insulation layer cures; placing one or
more cooling tubes on top of Z-primary coil portion and the
insulation layer; pouring epoxy on top of and between cooling tubes
forming an insulation layer; applying control weight while the
insulation layer cures; placing a 0.sup.th and a 2.sup.nd order
shim on top of cooling tubes and the insulation layer; pouring
epoxy onto shims forming an insulation layer; applying control
weight while the insulation layer cures; fabricating a Z-shield
coil with a coil portion and a curf void portion; placing the
Z-shield coil into the former; forming an insulation layer in the
Z-shield coil curf void portion and on top of the Z-shield coil
portion; applying control weight while the insulation layer cures;
fabricating an X-shield coil with a coil portion and a curf void
portion; placing the X-shield coil into the former; forming an
insulation layer in the X-shield coil curf void portion and on top
of the X-shield coil portion; applying control weight while the
insulation layer cures; fabricating a Y-shield coil with a coil
portion and a curf void portion; placing the Y-shield coil into the
former; forming an insulation layer in the Y-shield coil curf void
portion and on top of the Y-shield coil portion; applying control
weight while the insulation layer cures; placing 1.sup.st order
shims on top of Y-shield coil portion and insulation layer; pouring
epoxy onto shim forming an insulation layer; applying control
weight while the insulation layer cures; and sealing the top of
former.
11. The method of claim 10, wherein the shims comprise
self-insulated coils.
12. The method of claim 10, wherein the former, the coils, the
shims and the insulation layers have alignment holes to align the
former, the coils, the shims and the insulation layers relative to
each other.
13. The method of claim 10, wherein the insulation layers consist
of a fiber cloth substrate and epoxy.
14. A gradient coil manufacturing method comprising in sequential
order the steps of: i. fabricating a pair of hollow cylindrical
formers with a plurality of mounting holes placed around an edge of
the formers and a plurality of additional mounting holes
strategically placed in a central planar surface of the formers;
ii. etching a X-primary coil from a solid copper plate by a
water-jet cutting machine and placing the X-primary coil first on
the former using a plurality of pins through the central mounting
holes that extend vertically through all the layers to properly
align the placement of the plate on the former; iii. placing a
single layer of thin fiber cloth on the X-primary coil and pouring
an epoxy resin and a hardener mixture over it, then using a heavy,
heated flat plate placed on top to help the epoxy resin settle
forming an insulated layer and a flat controlled surface; iv.
etching a Y-primary coil from a solid copper plate as in step ii
and placing it in an orthogonal relation to the X-primary coil on
top of the insulating layer using the aligning pins as in step ii;
v. applying step iii to the Y-primary coil; vi. etching a Z-primary
coil from a solid copper plate and placing it on top of the
insulating layer above the Y-primary coil using the aligning pins
as in step ii; vii. applying step iii to the Z-primary coil; viii.
placing a plurality of non-conducting cooling conduits on the
insulating layer over the Z-primary coil then applying step iii;
ix. subsequently, placing a preassembled set of shim layers
comprising the zeroth and second order shims on top of the
insulating layer above the cooling layer then applying step iii to
the shim layers; x. a layer made of thick fiber cloth and a filler
material mixed with the epoxy resin and the hardener mixture over
the cooling and insulating layer forms an insulative support layer
when step iii is applied to it to fill the gap between the shim and
the shield layers; xi. an X-shield coil etched from a solid copper
plate is placed on the support layer as in step ii, co-aligned with
the X-primary coil, then step iii is applied to it afterwards; xii.
a Y-shield coil etched from a solid copper plate and placed in an
orthogonal relation to the X-shield coil using step ii and
subsequently applying step iii to it forming an insulation layer;
xiii. the Z-shield coil etched from a solid copper plate is placed
on the insulation layer above the Y-shield coil using step ii and
subsequently step iii is applied to it to form an insulating layer;
xiv. a preassembled set of shim layers comprising first, third and
higher order shim coils is placed on the insulating layer using
step ii; and xv. applying step iii to the shim layer with
additional fiber cloth layers forms a final insulative layer that
caps the entire layer.
15. The gradient coil manufacturing method of claim 14, further
comprising the step of placing the central mounting holes
strategically away from the dense parts of a plurality of current
distributions throughout the entire shielded gradient coils to
minimize disturbing the performance of the shielded coils and limit
a gradient field leakage adjacent the pole faces.
16. The gradient coil manufacturing method of claim 14, further
comprising the step of placing a plurality of center fed busbars
for the X and Y-primary and shield coils in the channels between
the two halves of each respective coil to save vertical stack up
space.
17. The gradient coil manufacturing method of claim 14, further
comprising the step cutting a plurality of slits into a plurality
of conductor tracks of the etched solid copper plates for the
gradient coils.
Description
CROSS REFERENCE TO RELATED APPLICATION
[0001] This application takes priority from and claims the benefit
of U.S. Utility patent application Ser. No. 10/944,080 filed on
Sep. 17, 2004, which in turn takes priority from and claims the
benefit of U.S. Provisional Patent Application 60/504,371 filed on
Sep. 19, 2003.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention is directed to a gradient coil design
and manufacturing process for making self-shielded gradient coil
sets for use in cylindrical bi-planar open magnetic resonance
imaging (MRI).
[0004] 2. Description of the Related Art
[0005] Progress in MRI scanner design has taken two directions. The
first has been towards higher field traditional magnet systems
while the second has been towards low field open magnet systems. By
"high field" magnet systems is meant those wherein the field
strength is greater than 10,000 Gauss and which require
superconducting wire technology to generate the magnetic fields.
"Low field" magnet systems are those in which the field strength is
about 5,000 Gauss and below and use permanent magnets or electrical
coils. In general, the cost of the magnetic field producing
material is a very strong function of the imaging volume. This is
more so the case for open magnet structures than for cylindrical
magnets. Therefore, image space management is crucial.
[0006] In open MRI systems, the gradient and RF coils, which have
to be placed within the imaging volume, take up significant space.
In particular, an efficient and compact structure for stacking the
gradient coil layers, which typically can take up to 20-30% of the
image space in the vertical direction, is a very important goal.
Moreover, the planar extent has to be limited as much as possible
because the gradient coil must be contained within the magnet
poles. The volume, size and weight of the magnets scale
quadratically with the radius of the magnet. Thus this variation is
the dominant factor that controls the cost of the magnet.
Therefore, limiting the planar extent of the magnet and magnet
poles should be a primary goal of any MRI system design.
[0007] In designing open, MRI systems the traditional approach has
entailed using unshielded gradient coils and then trying to design
a magnet system that would minimize the natural interactions
between the gradient fields and magnets. Although the approach has
had some success, it restricts the scanner from performing many MRI
applications, particularly those requiring the use of bipolar
current waveforms to drive the gradient coils due to eddy currents
and residual gradient fields from conductive surfaces and magnet
materials. In contrast, the present invention in which shielded
gradient coils are used is based upon a systems approach that seeks
to avoid all interactions between the magnet and the gradient and
RF coils, thereby enabling all MRI applications to be possible.
[0008] It has also been discovered that the performance of open MRI
systems can be further improved (and the cost lowered) by including
active shims in the shielded gradient coils. The system can achieve
up to two or more orders of magnitude improvement in magnetic field
homogeneity by including 0th, 1st, 2nd and higher order active
shims.
[0009] Furthermore, permanent magnets are heated differentially
from unshielded gradient coils causing inhomogeneities and thermal
drifts in magnet materials. It has been determined that adding
cooling means in the gradient space can alleviate this
significantly. Heretofore the cooling mechanisms had taken up too
much space, so that the tradeoff had not been considered
beneficial.
[0010] The present invention utilizes a gradient coil design that
includes shims and cooling mechanisms located inside the gradient
coil to achieve these improvements and allow a more compact open
magnet design without sacrificing field homogeneity and
performance.
[0011] Aside from physical constraints there are also imaging
requirements on a gradient coil's performance that are met with the
shielded gradient coils which allow the full complement of MRI
applications. The use of shielded gradient coils that include
active shims in an open MRI system has translated into a
significant performance enhancement that combines stable
homogeneity derived from self-shielding and additional active shim
sets.
[0012] Unexpectedly, the present system has also been found to
suppress the acoustic noise generated by the gradient coils. This
is due to the inherent greater physical size of shielded gradient
coils that combined with the opposing gradient fields of a shielded
configuration helps to stiffen the gradient assembly and generate
reduced torque effects.
[0013] The present invention is based upon a design methodology and
a manufacturing process to make an open self-shielded gradient coil
with the following features and benefits: [0014] (1) ultra-fast
switching; [0015] (2) full 0th, 1st, and 2nd order active shims and
the capacity to provide even 3rd, 4th, 5th or higher order active
shims if needed; [0016] (3) compact physical thickness and planar
extent to maximize subject access space and minimize magnet pole
face space; [0017] (4) air or water cooling; and [0018] (5)
construction and mounting method to suppress acoustic noise
generation.
[0019] The method and manufacturing process described enhance open
magnet MRI performance substantially while providing significant
cost benefits due to the substantial reduction in the open magnet
size.
SUMMARY OF THE INVENTION
[0020] The present invention is directed to a gradient coil
assembly for use in magnetic resonance imaging, said coil assembly
including (i) a circular primary coil set placed flat atop a first
and second cylindrical planar substrate disposed symmetrically to
each other about a plane of symmetry parallel to each. The primary
coil set is driven by a pulsed current signal for generating a
spatially varying magnetic field in the central plane.
[0021] The assembly further includes a circular shield coil set for
the primary coil set placed coplanar and atop the outer surfaces of
the primary coil set about the first and second cylindrical planar
substrates. The shield coil set is driven by a pulsed current
signal that is substantially 180.degree. out-of-phase with respect
to the pulsed current signal in the primary coil set for generating
a spatially varying magnetic field in a direction substantially
opposite to the first spatially varying magnetic field and
substantially canceling the magnetic field in a region
cylindrically outside of the outer surfaces of the shield
layers.
[0022] Preferably the assembly further includes a cooling mechanism
placed coplanar and atop the primary coil set between the primary
coil and shield coil sets.
[0023] Also preferably the assembly includes a set of circular
correction coils comprising the 0th, 1st, 2nd, and higher order
Legendre polynomial harmonics totaling about nine or more mutually
orthogonal coil sets for actively shimming the main magnetic field,
placed coplanar to the primary and shield coil sets in the
cylindrical planar substrates symmetrically disposed about the
planar surface. The 1st order correction coil sets are placed atop
the shield coil set and the 0th, 2nd and higher order correction
coils are placed between the primary and shield coil sets and atop
the cooling mechanism. Generally the primary patterns are denser
generating the most heat so placing the cooling mechanism directly
above the primaries is desirable.
[0024] The primary coil sets, the shield coil sets, the cooling
mechanism and the 0th, 1st and 2nd correction coils are supported
within a pair of cylindrical planar formers.
[0025] The primary coil set generally contain three coils each
generating one of three mutually orthogonal linearly-varying
magnetic fields and the shield coil set generally contains three
coils generating a linearly-varying magnetic field to substantially
oppose the complimentary primary coil field above the shield layer
surface. The 0th, 1st and 2nd order correction coils are designed
to have a planar circular footprint.
[0026] The cylindrical bi-planar gradient coil assembly contains in
sequential order (i) a circular primary coil set placed flat above
a first and second cylindrical planar substrate disposed
symmetrically to each other about a plane of symmetry parallel to
each, (ii) cooling means, (iii) oth and 2nd order shims, (iv)
shield layers, and (v) 1st order shims.
[0027] The primary coil set includes an X-primary, followed by a
first insulating layer, then a Y-primary, a second insulating
layer, and a Z-primary.
BRIEF DESCRIPTION OF THE DRAWINGS
[0028] FIG. 1a is a cross-sectional view of a magnetic resonance
imaging (MRI) machine in accordance with the present invention.
[0029] FIG. 1b is a top view of the MRI machine of FIG. 1a.
[0030] FIG. 2 is a cross-sectional view of a gradient coil set
showing the stacking of the elements of the gradient coil set.
[0031] FIG. 3a shows a Z primary axial coil pattern with a circular
footprint.
[0032] FIG. 3b shows one half of an X or Y primary transverse coil
pattern with a circular footprint.
[0033] FIG. 3c shows a Z shield axial coil pattern with a circular
footprint.
[0034] FIG. 3d shows one half of an X or Y shield transverse coil
pattern with a circular footprint.
[0035] FIG. 4 is a block diagram showing the key steps in the
manufacturing process of this invention.
[0036] FIG. 5 is a top view of a cylindrical former which holds the
multiple layers of a gradient coil, serves as a means to attach the
gradient coil to the magnet pole, and has recesses in transverse
patterns for running busbars to connect the gradient coil patterns
while avoiding build-up of vertical size.
[0037] FIGS. 6a and 6b show two alternative cooling tube designs
for use with the cylindrical gradient coil of this invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
[0038] The magnetic resonance imaging apparatus 10 is constructed
as shown in FIGS. 1a and 1b. It has a magnetic yoke 12 magnetically
connected to a magnetic circuit arrangement comprising a pair of
permanent magnet blocks 14 each attached to the surface of a pole
16, which has annular protrusions 17 for field shaping. The magnet
blocks 14 also have space for magnetic field gradient coil
assemblies 18 and RF coil assemblies 20, disposed opposite each
other so as to define a gap which is used as the imaging region.
The permanent magnet blocks 14 are magnetically connected by the
yoke 12 to generate a magnetic field in the gap. The gap has a
sufficient opening to insert a subject therein for the purposes of
obtaining tomographic images. The apparatus is specifically
designed for magnetic resonance imaging (MRI) purposes although
other applications that require the use of a static uniform
magnetic field may also be performed with the system.
[0039] The insertion gap used as an imaging region, is required to
have a magnetic strength greater than 0.1 T (1,000 Gauss) and a
uniformity less than 100 ppm (parts per million) throughout to
obtain images of sufficient clarity. The imaging region is large
enough to image a human being and has a diameter spherical volume
(DSV) of about 10 to about 50 cm, preferably about cm. The DSV is
critical to determining the dimensions of the gap and the diameter
of the pole pieces of a suitable magnet.
[0040] The choice of design methodology and the design innovations
for the planar open gradient coil is based on imaging requirements
and physical constraints. The imaging requirements demand that
there be minimal eddy current and residual magnetization
production, ultra-fast switching capability and high gradient
strengths to yield very high slew rates, and highly linear gradient
fields. Physically, what is required is limiting the gradient coil
stacking thickness and planar extent. These demands are met by
using the target field method specifically solved in polar
coordinates to allow the use of a radial envelope function, which
is the key to limiting the radial extent of the gradient coil.
Moreover, this approach allows a theoretically correct solution for
the shield coil layer similarly limited in radial extent and easily
extended to shim coil designs of any order.
[0041] In the open bi-planar cylindrical systems of the present
invention, axial gradients are made from a pair of identical
coaxial coils disposed symmetrically to each other about a central
plane and driven with opposing currents. A highly linear, very high
efficiency and low inductance axial gradient is produced by using a
Maxwell pair configuration. This condition is met when the coils
are separated by an axial length equal to 2a and their radii are
each given by 2a/ {square root over (3)}, where "a" is one-half the
distance between the upper and lower coils along the axial
direction. Using this configuration for the primary coil the target
field method can be used to design a shield. The resulting axial
gradient is so drastically changed that the performance is very
poor. To produce a similar performance with the added shields
requires finding new radial distributions for the primaries, which
can be done by allowing the current densities to have more than one
radial density distribution. Adding these degrees of freedom in the
target field specification and using a Nelder-Mead simplex
optimization algorithm, one can find an optimal design for the
radial current density distributions that yields very high
efficiency, linearity and low inductance axial shielded
gradients.
[0042] To save vertical space the shields are made as close as
possible to the primaries, but doing so diminishes the performance
of the overall gradient. This is overcome by making the current
densities, especially in the primaries, very dense at two or more
regions. This can be seen in FIG. 3a, a coil pattern for axial
primary coils. The dense regions are obtained from the
aforementioned optimization. The code generates the optimal
position, i.e. the position that yields the lowest inductance,
highest strength gradient and linearity. Constructing such coils
has been found to be a manufacturing challenge. To avoid high
resistances and generating heat, high-pressure water-jet cutting of
a copper plate (or other suitable machining technique) has been
found to be particularly suitable. Alternatively but less
desirably, one could use wires. The construction method is
described further below.
[0043] Furthermore, to limit the radial extent in such a tight
design with very high current densities the last few turns of the
axial gradients in both the primaries and the shields are reversed
so they produce an opposite field to cancel the field spill-out and
provide gradient confinement within the magnet pole space as can be
seen in FIG. 3a. The result is lower energy and lower overall
inductance while achieving very high switching speeds, i.e. about
50 to 150% gain using the same amplifier setting. This is a novel
approach in axial gradient designs and construction.
[0044] The shielded transverse gradients are designed by starting
with the field, inductance and energy expressed in polar
coordinates to maintain a circular geometry and then incorporating
a constrained minimization to keep the energy, and hence the
inductance, as small as possible but where the constraint points
and values where optimized by the Nelder-Mead optimization
algorithm in which the cost function includes the gradient
efficiency, linearity and inductance. This yields an optimal design
for a chosen radial and vertical configuration. Using this
approach, the vertical separations between primary and shield are
fixed and the constraint values are chosen so the radial extents
are also fixed. Thus, for a fixed set of physical values, the
designs yield the optimal performance for the shielded transverse
gradients.
[0045] A further enhancement of the design comes from the fact that
a cylindrical bi-planar configuration is a more natural geometry
for axial-shielded gradients, but an inefficient one for
transverse-shielded gradients. This dichotomy is resolved by
placing both the axial primary and shield coils inside the
transverse primary and shield coils. This entails putting both the
transverse primaries and shields on the outside of the axial
primaries and shields to provide an overall better three-axis
shielded gradient performance. Shields in general deteriorate the
efficiency of the primary gradients; thus the further away they are
the better.
[0046] Using these approaches, a typical gradient coil prepared in
accordance with the present invention has yielded axial and
transverse shielded performances with peak gradients approaching
0.15 mT/m/A, inductances well below 800 uH, linearity of less than
+/-5% on a 30 cm DSV, and shielded performances that suppress both
odd and even eddy field production to less than 1-0.1% of peak
gradients in a physical space that contains the gradients within a
45 mm vertical height, 90 cm diameter and gradient upper to lower
axial separation of 50 cm. Those skilled in the art of gradient
coil design and production will recognize the optimal performance
achieved in a very compact physical space, yet very wide upper to
lower gradient gap.
[0047] The use of the target field method of the present invention
allows the design of shims of any order since the technique is an
analytical one. Therefore, designs for 0th, 1st, 2nd order as well
as 3rd, 4th, 5th, and even higher Legendre polynomial shim coils
are easily generated and incorporated in the gradient coils. To
avoid coupling the 1st order shims to the gradients because they
are also linear or 1st order, these shims are placed above the
axial and transverse shield coil layers and all other order shims
are placed in the space between the primary and shield axial and
transverse coils. This saves space and optimizes the geometric
compactness of the entire assembly.
[0048] The present design contrasts with previous approaches that
have been based on rectangular coordinates and use a Gaussian or
similar tapering function to limit the planar extent of the primary
coil. That approach suffers for two fundamental reasons. First, the
geometric footprint is not circular and consequently not optimal.
Second, it does not allow for solving the shield layer
self-consistently to obtain a shield coil similarly limited in
planar extent.
[0049] A further advantage of the present invention is that it
allows specifying the most linear gradient field production
possible while achieving the ultra-fast switching (by including
energy minimizing constraints) and minimal eddy current production
due to shielding the primary gradient coil layer. The shielding,
minimum energy constraints and radial limits combine to limit the
field spill-out, which consequently result in ultra-fast switching.
While the rate of switching normally depends upon the amplifier
used with the MRI system, the lower the inductance and resistance,
the faster gradients can be switched. Thus the present invention is
capable of producing inductances of 800 uH or less with resistances
well under 250 m.OMEGA. depending on the type and size of the coil
material used.
[0050] Added to the benefits derived from the invention is that it
permits a manufacturing process which provides for the most
efficient physical stacking of gradient, shim and cooling
layers.
[0051] The manufacturing process begins by the specific choice of
former to contain and enclose the various layers. FIG. 4 provides a
summary of the manufacturing process and FIG. 5 shows a former as
produced below. The former is part of the gradient coil assembly.
It is what holds everything together and mounts to the magnet pole
faces. There are mounting holes on the outer edges of the former in
addition to the alignment pinholes, which also serve as mounting
holes when the entire gradient assembly has been constructed. A
fiber cloth braid/weave mesh pattern or other similar insulating
material forms the starting base for the former. An iron form of
the appropriate dimension is used to mold the former including the
mounting and screw holes. Each fiber cloth is laid inside the mold
and is potted by epoxy. When the last layer is reached a pressure
mold is used to obtain a flat control surface. Furthermore, the
iron mold is tapered at an angle of about 3 degrees to release the
former when the epoxy has set and the pressure mold is completed.
The former wall thickness generally ranges from about 1 mm to about
5 mm, preferably about 1-2 mm. The former height generally ranges
from 30 mm-45 mm. The choice of mesh and epoxy forms a strong and
rigid base to hold the weight of the various gradient coil layers
and also dampen acoustic noise generation. Using a fiber cloth
weave design and potting with an epoxy combines the benefits of
being strong yet thin and hence takes up the least amount of
space.
[0052] As best seen in FIG. 5, alignment holes are preferably
strategically placed throughout the former during the molding
process so that the gradient coils that have matching holes
throughout can be aligned with the former and to each other during
the construction process. The holes are placed throughout the
gradient stack such that they can be used as mounting supports to
the magnet pole face to help suppress acoustic noise generation.
The alignment holes alter the original path of the currents and in
a shielded design, where the balance is very delicate, the
placement of the holes is designed to have minimal impact yet
suppress acoustic noise generation as best as possible. Thus the
holes are located away from the densest regions in the coil
patterns. For example, in the axial gradient pattern of FIG. 3a
there is a lot of space where the reverse currents are and placing
the holes in these locations is ideal because it requires minimal
rearrangement of the wires/tracks.
[0053] An important space management technique used to obtain the
most compact stacking possible is by etching the gradient coils
from a thin copper plate typically between 3-5 mm thickness with a
circular footprint. A high-pressure water-jet cutting machine is
used to etch the wire tracks in the plate. Etching the wire
patterns gives a lot of freedom in achieving the many optimal
performance targets because the wire tracks can have variable
widths that are often needed due to tight design requirement. The
etching process is so precise that it allows etching of the
alignment and support holes in the gradient coil plates as
well.
[0054] Once the plates are etched by the above process, they become
loose and wire track positions can change because the whole plate
structure is no longer rigid. Two methods have been developed to
overcome this problem. In the first method, tabs are left
strategically throughout the etching process so that etched curfs
are not completely cutout but rather help keep the plate structure
rigid. Once the plate is removed from the water-jet machine, the
empty curfs are filled with epoxy and allowed to cure so that the
rigidity of the plate is now maintained by the epoxy fill.
Subsequently, the tabs are removed so adjacent wire tracks are not
shorted to each other. In this way, the wire track positions are
kept intact.
[0055] Alternatively, curf paths are molded into place using an
insulating material that has a thin flat base on top of which lies
the curf tracks. This track is then laid inside the former and the
flat portion serves as an insulating layer between two coil layers
while the tracks serve to align the etched plates into their
designed physical positions. In this scheme, the plate structure is
etched continuously without leaving tabs in the curfs. Although,
the plate is not rigid when the etching is initially finished, the
positions are restored to their designed value when placed on the
alignment insulating mold as described above.
[0056] For the axial gradients, care should be taken to avoid
leaving big areas of copper especially in the middle. Either these
pieces are cut out or slits are etched in wide areas of copper
patches to prevent the induction of eddy currents from other
gradients. To prevent a high resistance in the axial gradients, a
thicker copper plate is preferably used for the primary gradient
while a thinner copper plate is preferably used for the shield. For
example, a 5 mm plate can be used for the primary and a 3 mm plate
for the shields. By this approach a reduction of resistance is
gained compared to using thin copper plates in the primary axial
gradients as well.
[0057] This reduction in resistance is very significant because it
means less heat is generated and less power is needed to drive the
coils. Thus the gradients produced by the present invention are
stronger and switch faster.
[0058] After the first gradient coil layer is put in place, a thin
(typically 0.5 mm) insulator layer is formed by a fiber cloth
substrate and potted with epoxy after allowing it to settle. Then
the next layer of the gradient coil can be placed atop the
insulator layer. Subsequently, after another insulator layer, the
third gradient coil is placed atop the previous two. The gradient
coil layers are potted by pouring epoxy between the wire track
etchings that are about 0.8 mm wide. To maintain a tight tolerance,
generally about 0.1 mm variation across any surface, the plates are
stacked in the former placed on a flat control surface with a
tolerance of less than 0.1 mm, and a flat, heavy weight of the same
tolerance and control surface is applied during the epoxy curing
process.
[0059] Because the gradient return loops need to run to the center
of each gradient coil these return tracks are created in between
each semicircle for the X and Y-gradient coils by exploiting their
orthogonality. For the Z-gradient coil, the return track is either
etched in the former if it is the first layer or placed facing away
from the gradients if it is on top. The only separations between
gradient coil layers are the insulation layers. This is a key step
in the stacking efficiency. For such a thin separation, the limit
before exceeding the onset of corona effects is 50 kV/in voltage
gradients. Thus, the insulation thickness has to be consistent with
the amplifier drive voltages used. The voltages can vary anywhere
from 100 to 500 or more volts so long as the voltage divided by the
separation between gradient layers is below 50 kV/in.
[0060] The layers are placed in the former and relative to each
other using alignment pins during the placement of each layer. The
alignment pins differ in height and continue to grow as the layers
are built up. The flat weights have the corresponding holes to
accommodate the alignment holes during the curing process.
[0061] It is a fact that if the shield layers are placed directly
on top of the primary layers, then the fields produced in the
imaging volume will be nulled out. Consequently, the shield layers
must be placed at some distance away from the primary layers,
preferably as far apart as possible. Distancing the shield layers
from the primaries is made easier by using etched plate technology
with its high density stacking of wire tracks because of the ease
in providing variable wire track widths and thus a better
approximation of the theoretical design.
[0062] To obtain the most optimal stacking as called for by the
performance specification, all of the shim and cooling layers are
placed in between the primary layers and the shield layers with the
exception of the 1st order shims because they will couple to the
gradients. The 1st order shims are placed on top of the shield
layers. To make the shims as compact as possible, they are designed
by the same method using polar coordinates and a radial envelope
function to limit their planar extent. In addition, improved
stacking efficiency can be obtained by etching each shim from a 0.1
to 0.5 mm thick insulated copper plate. Preferably the shims have
the same alignment holes as the other layers within the gradient
coil. The shims are pre-processed to be self-insulated. The shims
can then be stacked atop each other and potted by an epoxy pour
running through the etched tracks and a controlled surface weight
applied during the curing process. The surface weight is used for
compression and to maintain the flatness of the system, preferably
to within about 0.06 mm across the entire surface. If too much
weight is used, the epoxy will be pushed out. If too little weight
is used, then the layer buildup will be greater and the imaging
region reduced beyond what was intended. Therefore, the weight of
the compression plates is used to control the surface build up.
[0063] Cooling can be added to the gradient coils by any of several
means. One approach is to use channels in the layer between the
primaries and the shields and then use air to cool the gradients.
This method relies on convective heat current elimination. A more
efficient method is to use copper tube with chilled water running
through and using conductive cooling. The chilled water approach,
however, generates eddy currents in the copper tubes and requires
the physical size of the tubes to be quite large. To avoid
generating eddy currents, a zig-zag pattern or similar patterns
such as those shown in FIGS. 6a and 6b can be used. FIG. 6b
represents a preferred pattern. To pack as much copper tubing
between the primary and shield layers as possible, the bending
radius of the copper tubes, r>=5.times.dia, can be exploited. A
3/16'' diameter copper tube is sufficient for this purpose.
[0064] The cooling scheme of FIG. 6b is preferred for a more even
distribution of cold flow to hot flow between the incoming to
outgoing flow. Furthermore, these tubes can be made of plastic as
well and any pattern can be wound because they are not conductive
and eddy currents will not be induced. Moreover, instead of fluid,
air-cooling can also be used to remove heat although it won't be as
effective as using fluid. Yet even further, instead of tubes just
channels can be formed in the cooling space and used for either
fluid or air-cooling purposes without worrying about inducing eddy
currents.
[0065] The entire gradient coil layer, which contains the shims and
cooling, is potted together to form a very rigid unit. It has
enough weight to withstand severe deformations or vibrations.
Consequently, the potting procedure has the added advantage of
suppressing acoustic noise generation resulting when the gradients
are pulsed against the high static background field.
[0066] Wire interconnects are strategically place between the
various quadrants, top and bottom layers to give an extra degree of
freedom in shunting current away from any quadrant for the delicate
shielding balance required. This balance can be thrown off by
manufacturing imperfections due to translational, rotational or
vertical misplacement or offsets of the many gradient coil layers
throughout. If this passive correction doesn't work, a final active
correction can be applied because of the symmetry of the gradients
and shims. Pulses can be applied to the shims when the gradients
are turned on to provide a final, almost perfect correction which
will cancel any eddies and make the shielded gradient coils work
almost perfectly.
* * * * *