You are on page 1of 14

UMN

Modeling Analysis of Inductive Coupling for Biomedical Applications


Final Report
AB Spring 2012

Table of Contents
Introduction.................................................................................................................................................. 2 Coupling Analysis ....................................................................................................................................... 3 Design for Coil Decoupling ....................................................................................................................... 6 Wedge Implementation for Rod Conductors ...................................................................................... 7 Wedge Implementation for Strip Conductors ..................................................................................... 9 Conclusion ................................................................................................................................................. 11 References ................................................................................................................................................ 12 Appendix A ................................................................................................................................................ 13

Introduction
Magnetic resonance imaging (MRI) techniques exploit the excitation of nuclei into coherent precession using a controlled magnetic field. This requires coupling between the nuclei (specimen) and a source of RF power in the form of external circuitry. These devices are called RF coils and are designed to operate in the general band of the Larmor frequency as specified by equation 1 [1]. =
2

0 172 2

(1)

= 42.58 / : gyromagnetic ratio of hydrogen 0 = 4 : uniform magnetic field produced by the main magnet

Although many different RF coil configurations are in existence, the particular type that is of interest is known as a TEM head coil as shown in figure 1. Head coils are geometrically designed for a human head to fit in the imaging area with minimal excess space.

Figure 1: TEM Head Coil [2] The structure is comprised of N = 4 to 16 uniformly spaced RF coils, implemented with copper rods or strips in a cylindrical configuration. Beneath the coils lies a teflon or air substrate separating the conductors from the cylindrical ground plane located on the outer radius. Figure 1 shows the case for 16 RF coils implemented with copper rods using an air substrate. Because of the close coil-to-coil proximity in the cylindrical structure, RF coils couple to one another with the most substantial case being for N = 16. To facilitate new potential imaging applications, it is desired to isolate and individually control each coil requiring a level of decoupling between conductors. To achieve such a design, an analysis of the coil-to-coil coupling must first be analyzed per the number of coils in the structure followed by the modeling of a proposed modification to decrease the current coupling levels.

Coupling Analysis
The head coil to be analyzed has an outer radius of 15 cm at the surface of the ground plane followed by a 2 cm thick substrate making the inner radius 13 cm. Since the inner radius of the cylindrical structure is fixed, the coil-to-coil spacing is determined by the number of coils N in the structure since any number of coils must be uniformly spaced. The upper part of figure 2 illustrates these dimensions with 8 coils comprised of 1 cm diameter copper rods. Using elementary trigonometry, the center-to-center coil spacing is computed using equation 2. () = 2 sin ( ) = 12.5 : MRI cylinder center to coil center spacing = 2 16 : Number of uniformly spaced coils (2)

Figure 2: RF Coil Model Approximation 3

Since the number of coils N is defined as a discrete set of all even numbers from 2 to 16, 8 different coil-to-coil spacing values arise and thus 8 different models must be simulated. For simplicity, the coupling of 2 RF coils will be simulated since the contribution of coils spaced further in the structure will be negligible. Additionally, the coils are mapped to a planar surface while preserving the coil spacing, substrate height, and conductor dimensions in the cylinder as shown in the second part of figure 2. To compute the coupling, the planar structure is simulated using HFSS from 0 to 350 MHz with the solution frequency specified at 172 MHz. To ensure all ports are matched, the reference impedance of each terminal is set to 92.5 Ohms, the simulated port impedance of the copper rods. A total of 8 simulations are performed with the coil spacing adjusted for each per equation 2. The raw data is shown in figure 3 with the coupling values graphed in figure 4 versus the number of coils and coil spacing. N 2 4 6 8 10 12 14 16 X (cm) X (in) Coupling (dB) Return Loss (dB) Insertion Loss (dB) 25.000 9.8425 -50.790 -17.321 -0.2794 17.677 6.9597 -44.405 -16.780 -0.3124 12.500 4.9213 -38.499 -16.786 -0.3119 9.5671 3.7666 -34.108 -16.600 -0.3166 7.7254 3.0415 -30.493 -16.775 -0.3189 6.4705 2.5474 -28.028 -17.220 -0.3182 5.5630 2.1902 -25.764 -17.295 -0.3090 4.8773 1.9202 -23.342 -18.903 -0.2573 Figure 3: Simulated Data for Different RF Coil Rod Spacing at 172 MHz
Coil Spacing (in)
2 -20 3 4 5 6 7 8 9 10 -20

-25

-25

-30

-30

-35

-35

Coupling (dB)

-40

-40

-45

-45

-50 Number of Coils vs. Coupling Coil Spacing vs. Coupling -55 2

-50

10

12

14

-55 16

Number of Coils

Figure 4: Coupling vs. Number of Coils and Coil Spacing (Rod) at 172 MHz

Coupling (dB)

As expected, the coupling increases significantly as the coil spacing decreases with a maximum value of -23.3 dB for N = 16. In addition, the rods remain relatively well matched for a nonplanar structure as the return loss does not go above -16.6 dB. In addition to analyzing the case for copper rods, RF coils can also be implemented with 1 Oz copper strips. Using the same methodology to map the cylindrical structure to a planar surface, the center-to-center strip spacing is still given by equation 2 since the inner radius of the substrate is unchanged. The outer radius of the substrate however is reduced to 1 cm along with that of the ground plane to allow for the strips to be designed to have a 50 Ohm characteristic impedance with a reasonable width. Using LineCalc, a 50 Ohm strip width is computed to be 1.2476 for a substrate height of 1 cm and relative dielectric constant of 2.1 for teflon. As with the previous case, a total of 8 simulations are performed varying the center-tocenter strip spacing determined by the number of elements. ADS is used to model the scattering parameters since the close proximity of the strip ports becomes too small to accommodate recommended HFSS port sizes. The raw simulated data is listed in figure 5 and the coupling is graphed in figure 6.

N 2 4 6 8 10 12 14 16

X (cm) X (in) Coupling (dB) Return Loss (dB) Insertion Loss (dB) 25.000 9.8425 -68.214 -60.255 -0.003 17.677 6.9597 -59.743 -63.325 -0.003 12.500 4.9213 -51.945 -58.259 -0.003 9.5671 3.7666 -46.451 -46.509 -0.004 7.7254 3.0415 -42.413 -45.221 -0.005 6.4705 2.5474 -39.288 -50.636 -0.007 5.5630 2.1902 -36.944 -61.392 -0.011 4.8773 1.9202 -35.188 -49.844 -0.018 Figure 5: Simulated Data for Different RF Coil Strip Spacing at 172 MHz
1 -35 2 3 4 Coil Spacing (in) 5 6 7 8 9 10 -35

-40

-40

-45

-45

Coupling (dB)

-50

-50

-55

-55

-60 Coil Spacing vs. Coupling -65 Number of Coils vs.Coupling -70 2

-60

-65

8 10 Number of Coils (in)

12

14

-70 16

Figure 6: Coupling vs. Number of Coils and Coil Spacing (Strip) at 172 MHz

Coupling (dB)

As expected, the coupling increases significantly as the strip-to-strip spacing decreases with the worst case being -35.188 dB for N = 16. The return loss data also indicates the ports are well matched confirming the strips are properly designed to 50 Ohms. Figures 4 and 6 properly characterize the coupling behavior per the RF coil spacing and thus establish a reference for measuring improvement for the design that will be discussed next.

Design for Coil Decoupling


In order to isolate the individual RF coils, a design must be implemented that prevents the individual electric and magnetic field components produced by one coil from interacting with adjacent coils. To achieve such isolation, a metal wedge connected to the ground plane is inserted between each coil to terminate any fields that may interact with other conductors. Figure 7 illustrates the wedge design and implementation on a planar substrate using both circular rods and strips as RF coils.

Figure 7: Wedge Design and Implementation To simplify modeling, only two coils are simulated since the coupling between adjacent coil pairs is much greater in comparison to others that are further spaced. The coil spacing of particular interest is with N = 8 elements producing a center-to-center conductor spacing of 3.76 inches. Additionally, three wedges are added to the two transmission line circuit such that the overall structure is symmetrical. The presence of the wedge should significantly decouple the RF coils since any fringing fields should terminate on the wedge itself rather than on the adjacent coil. Also, angled edges on the left and right hand side of the wedge create uniformity in the spacing between the coil and the ground plane and allow for the radiated magnetic field to propagate outward. To characterize the impact the wedges has on the coupling, several HFSS 6

simulations are performed with different wedge dimensions varying the width W and the height H in several cases. The solution frequency is set to 172 MHz and the scattering parameters are simulated from 0 to 350 MHz. Both types of RF coils (rod and strip) are studied separately as the wedge presence will have different impacts on the coupling values for each conductor type.

Wedge Implementation for Rod Conductors


Using the methodology illustrated in figure 7, 3 wedges are implemented in the two transmission line circuit. The rod spacing and substrate height are fixed for all simulations while the dimensions of the wedge are varied. A total of 12 different wedge dimensions are simulated each with a separate combination of width W and height H. Figure 8 shows the simulated coupling for each wedge dimension and figure 9 plots the results providing graphical insight into the behavior. Coupling W = 3.5 in W = 4.0 in W = 4.5 in H = 0.2 in -53.25 dB -53.97 dB -54.55 dB H = 0.45 in -56.62 dB -58.02 dB -59.43 dB H = 0.7 in -60.46 dB -62.80 dB -65.28 dB H = 1.4 in -74.92 dB -80.88 dB -87.79 dB

Figure 8: Simulated Coupling for Different Wedge Dimensions at 172 MHz

-50

-55

3.5'' Width 4.0'' Width 4.5'' Width

-60

Coupling (dB)

-65

-70

-75

-80

-85

-90 0.2

0.4

0.6

0.8

1.2

1.4

1.6

Wedge Height (in)


Figure 9: Coupling for Different Wedge Dimensions at 172 MHz

Per inspection of the graph, a dramatic reduction in the coupling is found for larger wedge dimensions, particularly those with a height of 0.7 or 1.4. As shown in figure 3, the coupling for 2 rods with N = 8 spacing without a wedge is -34.10 dB, while the presence of a wedge with (W, H) = (4.5, 1.4) decreases this value by almost 50 to -87.79 dB. Providing further insight into the results, figure 10 provides HFSS screenshots of the simulated model for a wedge with (W, H) = (4.5, 0.7) indicating the magnitude of the E- and H-Fields on the geometry. The illustrations of the field magnitudes indicate that the electric field properly terminates on the copper wedge while the magnetic field is still able to propagate outward. Such behavior meets the design objectives outlined earlier as a much greater level of isolation is achieved with the presence of the wedge.

Figure 10: HFSS Screenshot for (W, H) = (4.5, 0.7) with E- and H-Field Magnitudes

Wedge Implementation for Strip Conductors


As before, the two copper strips are implemented with 3 wedges as illustrated in figure 7. The rod spacing and substrate height are fixed for all simulations while the dimensions for the wedge are varied. A total of 9 different wedge dimensions are simulated each with a different combination of width W and height H parameters. Figure 11 shows the simulated coupling values for each wedge dimension and figure 12 plots the results providing insight into the behavior of the data. Coupling W = 2.5'' W = 3.0'' W = 3.5'' H = 0.45'' -51.34 dB -52.29 dB -54.10 dB H = 0.7'' -55.14 dB -56.98 dB -59.61 dB H = 1.4'' -68.03 dB -71.99 dB -78.35 dB

Figure 11: Simulated Coupling for Different Wedge Dimensions at 172 MHz

-50

-55

2.5'' Width 3.0'' Width 3.5'' Width

-60

Coupling (dB)

-65

-70

-75

-80 0.4

0.5

0.6

0.7

0.8

0.9

1.1

1.2

1.3

1.4

Wedge Height (in)


Figure 12: Coupling for Different Wedge Dimensions at 172 MHz As with the previous trend, a dramatic reduction in the coupling is found for larger wedge dimensions, particularly those with a height of 0.7 or 1.4. As shown in figure 5, the coupling for 2 strips with N = 8 spacing without a wedge is -46.45 dB, while the presence of a wedge with (W, H) = (3.5, 1.4) decreases this value by almost 30 to -78.35 dB. Providing further insight 9

into the results, figure 13 provides HFSS screenshots of the simulated model for a wedge with (W, H) = (3.0, 0.7) indicating the magnitude of the E- and H-Fields on the geometry. The illustrations of the field magnitudes indicate that the electric field properly terminates on the copper wedge while the magnetic field is still able to propagate outward. Such behavior is consistent with the results found for the simulations with the copper rods meeting the design objectives outlined earlier.

Figure 13: HFSS Screenshot for (W, H) = (3.0, 0.7) with E- and H-Field Magnitudes

10

Conclusion
The coupling analysis for 2 RF coils implemented with both copper rods and strips on planar substrates shows a consistent trend with respect to spacing. As the number of uniformly spaced coils is increased in a fixed cylindrical MRI structure, the coils must be placed closer together. As this coil-to-coil space decreases, the coupling increases at a very high rate as theoretically expected. The coil configuration of most interest is that of N = 8 uniformly spaced elements producing a center-to-center coil spacing of 3.76 inches. Per HFSS and ADS simulations, the coupling for the RF coil copper rods is -34.108 dB while the coupling for the RF coil copper strip is -46.451 dB. Upon implementation of the wedge, both cases produced a vast reduction in the coil-to-coil coupling with values as low as -87.79 dB for the rod case, and -78.35 dB for the strip case. In each data set, the decoupling is highest when the wedge width W is wider and the wedge height H is taller with the largest reduction simulated when both parameters were maximized. The design appears to show great promise in isolating each individual RF coil while still allowing the magnetic field to propagate outward towards the specimen.

11

References
1. Jin, Jian-Ming. Electromagnetic Analysis and Design in Magnetic Resonance Imaging. Boca Raton: CRC, 1998. Print. 2. Vaughan, J.T. "7T vs. 4T: RF Power, Homogeneity, and Signal-to-noise Comparison in Head Images." Magnetic Resonance in Medicine 46.1 (2001): 24-30. Print.

12

Appendix A
Simulation files for coupling analysis of 2 coils: Copper Rod (HFSS): 2Coil_Teflon_2cmH_N 2Coil_Teflon_2cmH_N.hfssresults Copper Strip (ADS): Botz_Spring12_MSProject_prj o 2Mstrip_1cm_Teflon Simulation files for coupling analysis of 2 coils with wedge implemented: Wedge Implementation with 2 Copper Rods (HFSS): 2Coil_Teflon_3500WedgeX3_NoBend 2Coil_Teflon_4000WedgeX3_NoBend 2Coil_Teflon_4500WedgeX3_NoBend Wedge Implementation with 2 Copper Strips (HFSS): 2MStrip_Teflon_2500WedgeX3_NoBend 2MStrip_Teflon_3000WedgeX3_NoBend 2MStrip_Teflon_3500WedgeX3_NoBend

13

You might also like