EM field distribution and SAR in a Human Head with MRI Coil

In  Ultra-High Field MRI  systems, e.g. for B0 beyond 7 T, the performance of the RF coils is essential to provide a great image quality with a reasonable B1-field homogeneity in the imaging area, low SARs in the  biological  tissues  and  a  good  signal  to  noise  ratio (SNR). Despite the many advantages in increasing the B0-field strength (higher resolution and SNR, reduced scan time), operating at higher frequencies adds significant technical complexity to NMR experiments and consequently to the RF coils design. As frequencies increase, the wavelength shortens (60 cm in the vacuum and 8.5 cm inside the head at 500 MHz), and becomes comparable to the electrical dimensions of the head/body and the RF coil. RF fields interact more strongly with human tissues and wave behaviour of the B1-field should affect strongly its homogeneity inside the highly permittive and lossy tissues. While B1 inhomogeneities cause non uniform intensity distribution in MR images, the RF electric field counterpart, potentially inhomogeneous too, should expose biological tissues to an excessive RF power deposition with induced local heating. Coupled to detailed anatomical human models, computational methods offer an indispensible tool for the investigatation of the distribution of the magnetic field or the maximum local SAR values inside the tissues and for the design and evaluation of the RF coils' performance.

This article provides a summary of the work carried out by CEA Saclay, France using CST MICROWAVE STUDIO® (CST MWS) and the HUGO Dataset with the permission and courtesy of Xavier Hanus and his colleagues[1].

CST MWS was used in conjunction with the 3D anatomical data set HUGO to simulate the time varying fields of an RF coil loaded by a human head section of the HUGO model with a mass of 7.81kg. A Dell 650 station was used with CPU P4 Xeon 2.8 GHz, 2 Go of RAM and SCSI HD. The model setup is shown in Figure 1 with the lines defined for the fields' extraction. 

For the example presented here, the total number of mesh cells over the entire volume of calculation is 1,982,178 as shown in Figure 2, with about 1,100,000 cells for the tissues corresponding to a 3 by 3 by 5 mm average resolution (roughly 18 voxels per 1 cm3 of tissue).

All the outer surfaces of the calculation volume are set to open boundaries with PML layers. As shown in Figure 3, four discrete ports have been defined at the bottom of the coil at 90° in the axial plane, with adjustable impedance and lumped capacitance in parallel for better matching when loading with the head.

For the high meshing intensive calculations, the excitations are performed over 0-800 MHz through the ports 1 and 3 in quadrature. As a result of its polarization, this mode, called B1, is not transmitted between the ports in quadrature as indicated by the transmission S 3,1 for the loaded coil at 480 MHz as shown in Figure 4.

The  fields  and  power  loss  density  are  obtained  at  a  specific frequency  for  each port excitation.  As shown Figure 5, the B1-mode presents a magnetic vector oriented linearly along -Oy when excited (Lin1P1) by the port 1 along x (cf. Figure 3.a and 5.a). The  results  are  then combined  as  a  post-processing  step  with  specific  amplitudes  compensating  for reflections (|S11| and |S33|), and phase-shifting (90°) in order to create the circularly polarized excitation. When animated with wt, the magnetic vector field appears rigorously circular in the axial plane for the empty coil (Fig. 5.a3) but quite disturbed in the head (Fig. 5.b3). In MRI, only the strictly circular component of the rotating magnetic field is effective on the spins excitation and participates to the formation of the images. 

Figure 6 shows the spatial variation of the magnitude (left) and phase (right) of the H+1 vector along the transverse axes for both empty and loaded coil.  |H+1| is enhanced with a sharp focusing effect at the centre of the brain : up to 1.2 A/m for the loaded coil compared to the 0.95 A/m uniform profile of the unloaded coil (Fig. 6.a). When loading with the head, phases are not symmetrically distributed over -45° and a strong phase-shifting is acting at the high/low permittivity interfaces of the tissues white and grey matter/bones and fat (positions 150/250 mm curve blue for the x-direction Fig. 6.b).

Numerical investigations of SAR and B1 profiles provide crucial information about head loaded coils designed for UHF-MRI. |B1| and SAR values normalized to an effective power injected of 1 Watt CW give the necessary amplification for given tip angles  and  allow  the  calculations  of  expected  SARs  for  MRI  typical  pulse  sequences  in  accordance  to  guidelines.   More studies have to be carried out in order to improve the performances of the coil. Such studies include the increasing of the Q (Qloaded=40), the influence of the ports disposition, the investigation of multi-excitation scheme, composing a better uniform and circular rotating B1+ becoming a major challenge in UHF-MRI with the inherent reduction of the wavelength.

[1] X. Hanus, M. Luong, F. Lethimonnier, "Electromagnetics Fields and SAR Computations in a Human Head with a Multi-port Driven RF Coil at 11.7 Tesla", Proc. Intl. Soc. Mag. Reson. Med. 13 (2005).