Database Construction for Local SAR Prediction: Preliminary Assessment of the Intra and Inter Subject SAR Variability in Pelvic Region
Ettore Flavio Meliadò1, Alexander J.E Raaijmakers1, Matthew C. Restivo1, Matteo Maspero1, Peter R. Luijten1, and Cornelis A.T. van den Berg1

1Center for Image Sciences, University Medical Center Utrecht, Utrecht, Netherlands

Synopsis

A reliable technique to assess the 10g averaged Specific Absorption Rate is necessary for parallel transmit ultra-high field body MRI. We believe that the best solution is to build a database with many different models. We have created nine dedicated body models and performed FDTD simulations on them to evaluate the inter-subject variability for prostate imaging at 7T using fractionated dipole antennas. Maximum SAR10g ranges from 1.64 to 2.48W/kg with 8x1W input power. No relationship was found between BMI and maximum SAR10g. Intra-subject variability (caused by slight antenna positioning variability, +/-2cm) was also investigated showing up to 67.8% SAR variability.

Purpose

A reliable technique to assess the 10g averaged Specific Absorption Rate (SAR10g) is necessary in order to benefit from the advantages offered by ultra-high field MRI. To increase the signal penetration and to reduce the signal voids, 7T body imaging is done using on-body parallel transmit arrays. The on-body operation in conjunction with the parallel transmit drive requires careful assessment and monitoring of local SAR levels. The SAR10g is usually determined by simulations using generic dielectric patient models. However, often these models do not well represent the features of the body under consideration, and thus the assessment could be inaccurate. Currently it is still not possible to make a real time SAR10g assessment using subject-specific online-generated models of subjects undergoing MRI. The aim of this study is to build an extensive database of representative body models for prostate imaging, that should make local SAR estimates more accurate by using the simulation results of the body model that fits best to the imaging subject. In this first phase we will use the database that we have obtained so far to evaluate the inter-subject variability. In addition, we have investigated the intra-subject variability, meaning the variation in SAR10g level if the array elements are placed in a slightly different position (+/- 2cm). In a second stage, once the database has sufficient body models, we will test the predictive value of the database. Preferably, we would find measurable features (e.g. BMI, fat-layer thickness, coupling matrix) capable to characterize a priori, pre-examination, the maximum SAR10g.

Methods

A group of 9 volunteers with Body Mass Index (BMI) from 22.5 to 28 and age between 40 and 61 was included in this study (Table 1). Similar to [1], we have created dedicated body models by segmentation in four tissue types (muscle, fat, cortical bone and skin). The volunteers were scanned with at 1.5T MR scanner (Ingenia, Philips Healthcare, The Netherlands) with mock ups of our array of eight antennas in place [2] in order to guarantee a perfect matching between the antennas and the body surface. (Figure 1). 3D multi echo FFE images (TR/TE1/TE2 = 5.56/1.64/3.76ms) using a resolution of 1.7x1.7x2.5mm3 were acquired. A Dixon reconstruction was performed resulting in water/fat/IP/OP images. The Dixon fat image and the IP/OP images were used respectively to differentiate soft and adipose tissue, and the bone structures [3]. In order to have a correct antenna positioning on the models, 4 MR visible markers have been placed in the corners of each element. An automatic procedure to localize the antenna and position it in the simulation geometry was implemented. The Medical Image Segmentation Tool Set (ISEG, ZurichMedTech, Switzerland) was used in order to obtain the tissue type contours and to import them into the simulation software. Electromagnetic simulations (Sim4Life, ZurichMedTech, Switzerland), was performed on these realistic subject-specific models in order to evaluate the E-Field distributions used to calculate the 10g averaged Q-Matrices [4]. Afterwards, the SAR10g distribution for prostate shimmed phase settings was calculated for each model. To assess the impact of antenna positioning errors (intra-subject variability), we have evaluated the SAR10g distribution for volunteer M04 repeatedly while moving two antennas +/- 2cm around the correct positions.

Results&Discussion

In Figure 2 are reported the body models and the SAR10g distributions with 8x1W accepted power and phase settings to give the maximum average $$$B_1^+$$$ in the prostate region. All images show the SAR10g distribution in the plane where the prostate is located and the maximum value in the whole body. As highlighted in Figure 3, in this limited group no relation was found between BMI and maximum SAR10g; Nevertheless, the regions with a high SAR10g level seem to appear in almost all models at similar anatomical locations (gluteus, pubic region and iliac regions). It is necessary to include many more subjects in this study to obtain more reliable statistics on maximum SAR10g variability and correlate features (e.g. thickness of subcutaneous fat layer) to SAR characteristics. In Figures 4 the SAR10g distribution and the maximum SAR10g value are reported where the top or bottom antenna position is shifted +/- 2cm around the correct position. The maximum SAR10g variability for the indicated range of positions is about 68% on the belly and 45% on the back.

Conclusions

Maximum SAR10g ranges from 1.64 to 2.48 W/kg with 8x1W input power. No relationship was found between BMI and maximum SAR10g. The influence of element placement is also investigated showing up to 67.8% SAR variability for slightly different positioning (+/- 2cm), which should be diminished by using a placeholder frame.

Acknowledgements

The research leading to these results has received funding from the ARTEMIS Joint Undertaking under grant agreement no 332933.

References

[1] O. Ipek, A.J.E. Raaijmakers, J.J. Lagendijk,P.R. Luijten, C.A.T. van den Berg. Intersubject local SAR variation for 7T prostate MR imaging with an eight-channel single-side adapted dipole antenna array. Magn Reson Med 2014; 71 (4): 1559 – 1567.

[2] A.J.E. Raaijmakers, M. Italiaander, I.J. Voogt, P.R. Luijten, J.M. Hoogduin, D.W.J. Klomp, C.A.T. van den Berg. The fractionated dipole antenna: A new antenna for body imaging at 7 Tesla. Magn Reson Med. 2015 May 2 : 10.1002/mrm.25596. Published online 2015 May 2. doi: 10.1002/mrm.25596.

[3] M. Maspero, P.R. Seevinck, G.J. Meijer, J.J.W Lagendijk, M.A. Viergever, C.A.T. van den Berg. A Dixon Based Pseudo-CT Generation Method for MR-Only Radiotherapy Treatment Planning of the Pelvis and Head and Neck. Medical Physics, 42, 3316-3316 (2015), doi:http://dx.doi.org/10.1118/1.4924305.

[4] K. Caputa, M. Okoniewski, M. Stuchly. An algorithm for computations of the power deposition in human tissue. IEEE Antennas Propag Mag 1999;41:102–107.

Figures

Figure 1: Fractionated dipole antenna array for prostate imaging.

Table 1: Data of the volunteers.

Figure 2: Transverse section of the constructed body models and SAR10g distributions with 8x1W accepted power and phase settings to give the maximum average $$$B_1^+$$$ in the prostate region.

Figure 3: Maximum 10g averaged SAR as a function of BMI.

Figure 4: Transverse 10g averaged SAR distributions and maximum values moving the top and bottom antenna on the body.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3660