Convex optimization of MRI exposure for RF-heating mitigation of leaded implants: extended coverage of clinical scenarios at 128 MHz
Earl Zastrow1,2, Juan Córcoles3, and Niels Kuster1,2

1IT'IS Foundation, Zurich, Switzerland, 2Department of Information Technology and Electrical Engineering, ETH-Zurich, Zurich, Switzerland, 3Department of Electronic and Communication Technology, Universidad Autónoma de Madrid, Madrid, Spain

Synopsis

Interactions of long insulated implants with conductive wires (e.g., cardiac pacemaker and deep-brain stimulator) with RF during MRI can lead to excessive local heating of tissue at the vicinity of the implant and is one of the contraindication to MRI. We present the preliminary results of a convex optimization method that can be used to suppress the local deposited power in tissue in a controllable manner. The performance of the proposed method is evaluated, as a function of the trade-off between homogeneity of |B1+| and the mitigated RF-induced power deposition caused by the implant, for multiple clinical scenarios at 128 MHz.

Purpose

Potential health risk from local tissue heating caused by implant-RF interactions during Magnetic Resonance Imaging (MRI) is a well-recognized subject in MRI safety assessments studies. The ASTM-21821 and ISO/TS 109742 are the current guiding documents for the assessment of RF-heating risk; both limit the exposure conditions for safety assessment to a nominal circular polarization (CP) RF fields from 1.5T or 3T cylindrical bore MR systems. Our recent study3 has shown that RF-heating from medical implants may be mitigated in a controllable manner through careful adjustment of RF exposure, if some clinically specific parameters are known a priori and feasibility was demonstrated for a limited set of parameters at 64 MHz. We have extended our evaluation to 128 MHz and the coverage of possible clinical scenarios by increasing the permutations of RF-coil, patient’s anatomy, and imaging positions. The results of the study may be used to determine whether a reduced access to a priori knowledge can be tolerated and to what extent the sacrifice in heating-mitigation performance shall be.

Method

Three generic 2-channel birdcage coils, operating at 128 MHz (Fig 1, top-right); two virtual patients; and four imaging positions marked (a) – (d) (Fig 1, left) are considered. Inclusion of a 400-mm generic leaded-cardiac implant, similar to that described in previous work3, is assumed in both patients. Ten unique implant routings with proximal termination in the left side of the chest and the electrode terminated inside the heart, are considered (Fig 1, bottom-right).

The convex optimization strategies described in previous work3, based on a common formulation for RF shimming4-6 with a constraint based on the piece-wise excitation model of implant-induced RF-heating7, are used. The piece-wise excitation model of the considered implant at 128 MHz is illustrated in Fig 2.

The coefficient of variation of |B1+| (B1+,cov) over a region of interest (RoI) and the estimated local power deposited in tissue by the implant (Pind) are evaluated and used as figures of merit for comparison. The RoI is a rectangular cuboid centered at the isocenter of the RF coil with dimensions of 300 mm (coronal) x 300 mm (sagittal) x L/2 mm (axial), where L is the length of the RF coil.

B1+,cov and Pind are evaluated under the RF-exposure conditions, defined by the driving vectors of the two-port RF coils obtained from the following methods:

(1) Unconstrained optimization: based on common method used in RF-shimming4-6. The solution provides an exposure with the best homogeneity of |B1+|.

(2) |B1+| optimization with implant-induced heating null constraint: a null-constraint on the induced electric field at the electrode is introduced. A straightforward matrix solution provides an exposure with the lowest implant-induced heating.

(3) |B1+| optimization with implant-induced heating constraint: the constraint is adjusted to compromise between implant-induced deposited power and |B1+| homogeneity. The problem is solved with routines from CVXOPT package8.

(4) CP exposure: the implant-induced heating and |B1+| homogeneity under nominal exposure condition considered by safety standards.

(5) Maximum implant-heating exposure: a generalized eigenvalue problem. The solution provides an exposure with the largest implant-induced deposited power.

Results and Conclusions

All results are normalized to the first-level controlled operating mode of the MRI9. Fig 3 shows the averaged Pind over 10 unique routings (<Pind>) and B1+,cov, obtained from the methods summarized in (1) – (5). The results for both patients are illustrated together for each imaging position. The polarizations of the driving vectors obtained from (1) – (5) are illustrated top-right of each figure. As expected, (1) provides the best B1+,cov whereas (2) provides the lowest <Pind> but with the trade-off on reduced |B1+| homogeneity, for all imaging positions; and the upper bound for <Pind> is established by the maximum implant-heating exposure. For all imaging positions, (3) successfully established solutions that provide lower B1+,cov and <Pind> than CP, and <Pind> can be reduced by 2 – 8 dB while maintaining similar B1+,cov as CP-exposure, depending on the imaging scenarios.

It is evident from Fig 4 that different RF coil lengths do not cause large variation of <Pind> and the variation introduced by the different anatomy is much more dramatic in comparison. It is also worth noting that CP-exposure gives a fairly good approximation of maximum implant heating (within 1dB of difference).

The results show that a reduction of a priori information is possible and some common parameters describing the exposure conditions (e.g. RF coil length) may be assumed. The permutation of patient’s body type must be increased to determine the extent of the variation introduced by the different anatomy and other implant locations must be considered.

Acknowledgements

The work of J. Córcoles was partially supported by the Spanish Ministry of Education under grant CAS12/00216 from the José Castillejo Research Mobility Programme, CICYT under contract TEC2013- 47106-C3-2-R and Banco Santander-UAM under contract 2015/ASIA/03.

References

1. ASTM F2182-11, Standard test method for measurement of radio frequency induced heating near passive implants during magnetic resonance imaging. ASTM International 2011.

2. ISO/TS. 10974:2012, Requirements for the safety of magnetic resonance imaging for patients with an active implantable medical device. ISO/TS 10974 2012.

3. Córcoles J, Zastrow E, Kuster N. Convex optimization of MRI exposure for mitigation of RF-heating from active medical implants. Phys Med Biol 2015; 60:7293–7308.

4. Ibrahim TS. Ultrahigh-field MRI whole-slice and localized RF field excitations using the same RF transmit array IEEE Trans Med Imaging 2006; 25:1341–7.

5. Van den Berg CAT, van den Bergen B, Van de Kamer JB, Raaymakers BW, Kroeze H, Bartels LW and Lagendijk JJW. Simultaneous B1+ homogenization and specific absorption rate hotspot suppression using a magnetic resonance phased array transmit coil. Magn Reson Med 2007; 57:577–86

6. Setsompop K, Wald L, Alagappan V, Gagoski B and Adalsteinsson E. Magnitude least squares optimization for parallel radio frequency excitation design demonstrated at 7 Tesla with eight channels. Magn Reson Med 2008; 59:908–15

7. Park SM, Kamondetdacha R, Nyenhuis JA. Calculation of MRI-induced heating of an implanted medical lead wire with an electric field transfer function. J Mag Res Imag 2007; 26:1278–1285.

8. Andersen MS, Dahl J and Vandenberghe L. 2013 CVXOPT: a python package for convex optimization, version 1.16, available at http://cvxopt.org

9. IEC. Medical electrical equipment - Part 2-33: Particular requirements for the basic safety and essential performance of magnetic resonance equipment for medical diagnosis, Edition 3.2. IEC 60601-2-33:2010+AMD1:2013+AMD2:2015 CSV 2015.

Figures

Fig 1. Left: Imaging positions (a) – (d) of both patients inside RF-coils of length L = 50, 60, and 70 cm. Top-right: Dimensions of the RF-coil. Bottom-right: Example of 10 unique implant routings in DUKE.


Fig 2. The piece-wise excitation model at 128MHz of the 400-mm generic implant considered in this study.

Fig 3. <Pind> (W) of considered implant as a function of B1+,cov (%), resulting from excitation vectors obtained from (1) – (5) for imaging positions (a) – (d) of both patients, inside L=50 cm RF-coil. Polarizations of driving vectors obtained from (1) – (5) are illustrated top-right corner of each figure.

Fig 4. <Pind> (W) of considered implant as a function of imaging position for RF-coil length, L=50, 60, and 70 cm. Results obtained from (3) are indicated by the shaded regions. The maximum <Pind> of all coil lengths are illustrated with solid lines for (1), (4), and (5), for each imaging position.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
0913