'Microsoft Word - rad82E10.tmp.DOC' by Administrator

FDTD Simulation of an 11T Re-entrant Cavity Volume Coil

J. Caserta1, B. L. Beck1, J. R. Fitzsimmons1

University of Florida, Gainesville, FL, United States

Abstract: To investigate the cause of image inhomogeneities, and to begin to explore possible solutions, finite-difference time-domain simulations

were performed on a re-cav volume coil. Simulations at 470MHz are compared to 11T images. The simulations show inhomogeneities, in both the

sagittal and axial planes and for all field components (Bx, By and B+), which are comparable to images.

Introduction: As static field strength and operating frequency increase, wavelengths, especially in biological materials with high relative

permittivities, decrease to fractions of the sample size. Wave behavior [1], such as constructive and destructive interference, can occur, reducing B1

homogeneity. Especially for rectangular loads, the Bx and By should be observed, because of the boundary conditions of the load occur at fixed x

and y locations. However, only circularly polarized (B+, B-) components are available from MR images, highlighting the usefulness of field

modeling. Modeling provides magnitudes and phases of all components of the B1 field.

Methods: A re-entrant cavity (re-cav) [2] 11T volume coil was modeled using the XFDTD software package (Remcom Inc. State College, PA). The

coil has an inner radius of 10cm, an outer radius of 12.25cm, an imaging length of 20cm and a shield length of 35cm. Figure 1a-c illustrates the

geometry used. A 2.5mm Yee-cell [3] size was used, and the problem space was 200x200x200 cells. The legs are 3.5 cm wide with 6 breaks for

3.8pF capacitors. The coil is tuned to 470 MHz using Gaussian pulses [4] and observing leg current magnitudes and phases. Capacitor values are

adjusted to their appropriate value. Figure 1d shows |Bx|, at the coil center, versus frequency for Gaussian input pulses and four different capacitor

values. Simulations were then performed with a 12x10x18 cm load approximating beef (εr = 62 and σ = 0.8.). Both linear and quad drives were used.

Results and Discussion: Inhomogeneities appear in sagittal and axial views of both simulations and images. Figure 2 shows simulation results of the

quad drive, for both sagittal (a-c) and axial (d-f) views. Figure 3a shows a sagittal image, while Figures 3b-d show axial slices at different locations.

The simulated field patterns resemble the images, showing both bright and dark regions, however differences are expected because of differences in

phantom dimensions. Additional simulations using different drive points and different phantoms, as well as imaging with the exact phantom used for

the simulations, are needed. Also, current densities along the z-axis of the re-cav are being investigated. Any attempts to reduce inhomogeneities

should account for both axial and sagittal inhomogeneities, and consider both linearly (Bx, By) and circularly (B+, B-) polarized components of the B1

field.

Fig. 1a

Fig 1b.

Fig 1c.

Fig1d.

Figure 1 Geometry (1a - 1c) and Frequency domain results (1d)

Fig. 2a |Bx|

Fig. 2b |By|

Fig. 2c |B+|

Fig. 2d |Bx|

Fig. 2e |By|

Fig. 2f |B+|

Figure 2 Sagittal y=100 (2a 2c) and axial z=80 (2d 2f) simulation results

Fig. 3a

Fig. 3b

Fig. 3c

Fig. 3d

Figure 3 Sagittal 3a and axial (3b 3d) images

References:

Acknowledgements:

[1] QX Yang, et. al., MRM 47:982-989 (2002)

This work was supported by the NIH (P41 RR16105,

[2] Beck, et. al, 10th ISMRM, Honolulu, HI #765 (2002)

R01 NS41094-01).

[3] KS Yee IEEE Trans. on Ant. and Prop. 14(3):302-307 (1966)

[4] TS Ibrahim, et. al. MRI 19:1339-1347 (2001)

Proc. Intl. Soc. Mag. Reson. Med. 11 (2003)

2647