Elena Lucano1,2, Mikhail Kozlov3, Eugenia Cabot4, Sara Louie5, Marc Horner5, Wolfgang Kainz1, Gonzalo G Mendoza1, Aiping Yao4,6, Earl Zastrow4,6, Niels Kuster4,6, and Leonardo M Angelone1
1Center for Devices and Radiological Health, Office of Science and Engineering Laboratories, U.S. Food and Drug Administration, Silver Spring, MD, United States, 2Department of Information Engineering, Electronics and Telecommunications, University of Rome "Sapienza", Rome, Italy, 3MR:comp GmbH, Gelsenkirchen, Germany, 4IT'IS foundation, Zurich, Switzerland, 5ANSYS, Inc., Canonsburg, PA, United States, 6Department of Information Technology and Electrical Engineering, ETH, Zurich, Switzerland
Synopsis
Preliminary
results of an ongoing inter-laboratory study are presented. The
eventual purpose of the effort is to develop a methodology that
harmonizes RF-modeling and RF-testing protocol for use in RF exposure
assessment. In this phase of the study, numerical and experimental data
were collected from four laboratories for unloaded and loaded coil
conditions. Only information about the geometry and resonance frequency
of the physical coil was provided. Qualitatively good agreement across
all teams was found. Subsequent phases of the study shall include a
methodology on uncertainty analysis associated with the numerical and
experimental methods that can be used in practicePurpose
Several
electromagnetic (EM) simulation platforms have been used in literature to
calculate the EM field generated by a magnetic resonance imaging (MRI) RF coil.
We present preliminary results of an ongoing inter-laboratory study that aims to
compare the numerical data obtained by users of different software platforms modeling
a commercially available RF coil, i.e., the MITS system. The computational data
was then compared to measurements obtained using two MITSs systems in two different
laboratories.
Methods
The Medical Implant
Test System (MITS1.5) for 1.5 Tesla RF Safety Evaluation (Zurich Med Tech) (Figure
1a) consists of a 16 rung high-pass birdcage coil (inner diameter = 740 mm, length
= 650 mm, Figure 1c) [1]. The coil is shielded by a metallic enclosure and is
driven at two ports physically shifted by 90o (I and Q in Figure1a). A
simplified geometric model of the MITS1.5 coil (Figure 1b) was distributed in CAD
format to four teams: FDA (Team 1), IT’IS (Team 2), MRCOMP (Team 3), and ANSYS
(Team 4). No further information on the matching and tuning network, and the EM
fields from the physical coils were provided. EM simulations were conducted
independently with three software platforms: xFDTD (Remcom Inc.) (Team 1), SEMCAD-X (Schmid & Partner Engineering) (Team 2), both of which employed the finite
difference time domain method, and HFSS (ANSYS Inc.), which employed finite
element method (Teams 3 and 4). The EM fields were measured with the DASY52 NEO
robotic measurement system (SPEAG) [2]. In the first step, numerical and measured data
were collected for coil unloaded and loaded with ASTM phantom (Figure 1d). Metallic
components were simulated as either copper or perfect electric conductor. The
coil model was tuned at 64MHz. Numerical and measured data were reported for
axial and coronal planes for the unloaded condition (Figure 2a), and for
coronal planes for the loaded condition (Figure 2b). For a quantitative comparison,
all results were rescaled to the norm of the magnetic field vector at the
center of the RF-coil (|H_iso|). Numerical and measured data were compared using
the SMAPE error approach [3].
Results
Figure 3 and 4 show
comparison of the numerical and measured electric field in both loading
conditions obtained by each team (i.e., one for each quadrant). Herein results
are compared only with one of the two measured datasets. Each point in the
figure is drawn with the coordinates based on the simulated value in the x-axis
and the same measured value in the y-axis. The closer the distribution is to
the diagonal, the closer the numerical data are to the measurements (1:1 line).
Conversely, a point closer to the ordinate (i.e., measure axis) is
representative of a higher measured value with respect to the simulated one,
and vice versa. Point distributions were similar for all teams, with an overall
better data match for the coronal planes and the axial planes far from the
isocenter. Results for the central axial plane (i.e. plane C in Figure 3) were affected
by the low electric field values. Figure 5 shows the magnetic and electric
field distributions inside the ASTM phantom. The measured magnetic field
distribution was more asymmetric than the simulated one. Overall, good
agreement was found for the electric field distribution. The mean SMAPE [3] (see equation in Figure 5)
inside the plane was equal to 15.7%, 19.5%, 34.71% for Teams 1, 2 and 3,
respectively. Consolidation for results from Team 4 is still pending. SMAPE variations may be caused by large
sensitivity of |H_iso| on magnetic field symmetry.
Discussion and Conclusions
Further
analysis will be conducted to find reasons for the magnetic field asymmetry
around the iso-center. Figure 4c suggests that the correlation between
numerical and measured data for all teams may be affected by the asymmetry of
the magnetic field. Additional post-processing will be performed to take such
asymmetry into account.
Further
steps of the project will include simulation of MITS guided by additional MITS
information: coil S-parameters and field distribution. The results from all
software platforms will be compared, and uncertainty levels will be defined
between the measurements datasets. This will allow a complete comparison
including different level of accuracy of the field distributions and power
requirements for both numerical and physical coils.
The
mention of commercial products, their sources, or their use in connection with
material reported herein is not to be construed as either an actual or implied
endorsement of such products by the Department of Health and Human Services
Acknowledgements
No acknowledgement found.References
[1] Lucano
et al., ISMRM (2014), 4903
[2]
DASY
5NEO, SPEAG, Zurich, Switzerland
[3] J. S. Armstrong, Long-Range Forecasting: From Crystal Ball to Computer: Wiley 1985