Resonator design

The resonator was constructed from five ceramic discs, which are placed on a thin supporting plastic tube. Each disc was made from BaSrTiO3, including Mg-containing compositions32. The relative dielectric permittivity of the disc material is ɛ 1000 and tan δ 4 × 10−4 (at 1 MHz). Each disc has the following dimensions: the inner diameter of 101.5 mm, the outer of 124.4 mm, and the height of 20 mm. The weight of the resonator (five ceramic discs and plastic spacers) is equal to 2.45 kg. An operational frequency of the resonator TE01δ mode was tuned to 123.25 MHz (which corresponds to the proton Larmor frequency at the given 3 Tesla MR scanner) by changing the spacing between the ceramic discs. In general, spacers can be made from the low-permittivity and low-loss materials, such as plexiglas, plastic, or thick paper. Here we have used several thin discs made of plexiglas with relative dielectric permittivity around 3.5 and electrical conductivity of 0.02 S m−1, and the height of 1.5 and 3 mm, and several discs made of paper with permittivity around 2.3 and the thickness of 0.25 mm. All spacers had the inner and outer diameters as for ceramic one and were placed at the same support tube between the ceramic discs. To preserve the efficiency of the dielectric resonator with proximity to the chest, a thin metallic disc made of the 14 µm-thick aluminum foil was added (shown in Fig. 2a). The metallic disc was sufficiently thin to avoid distortion of the gradient magnetic fields necessary for a spatial encoding of the MR signal. The aluminum foil was fixed at a supporting plastic tube via a plexiglas disc with an inner diameter of 101.5 mm and the outer diameter of 200 mm. It is noteworthy that metal had no contact with ceramics. The resonator prototype was placed inside an extruded polystyrene case of weight around 1 kg with two cylindrical holes to fix the resonator and provide the proper location of the volunteer and the breast inside the ceramic resonator during the MRI study.

The resonance frequency of the TE01δ mode was measured using a small, non-resonant magnetic field probe placed above the dielectric resonator inside the MR system using a portable vector network analyzer (OBZOR TR1300/1, PLANAR LLC). Figure 2a illustrates that the minima of the probe S11 curve corresponded to the desired frequency of 123.25 MHz.

To investigate the feasibility of TE01δ mode frequency adjustment, we performed the electromagnetic numerical modeling in CST Microwave Studio 2017. The dielectric resonator was excited by a linearly polarized plane wave propagating along the x-direction with the magnetic field parallel to the y-axis (Fig. 2b). An operational frequency of the TE01δ mode as a function of the gaps widths between the adjacent ceramic discs (parameters and g in Fig. 2b) was calculated using a parametric sweep. The metric ∆ was varied from 0 to 2 mm, while g was changed between 0 and 5 mm. Thus, the total resonator height was varied within the range of 100–138 mm. Figure 2c demonstrates the feasibility to tune the operational frequency of the dielectric resonator (∆f = fres − 123.25 MHz) in the range ±5 MHz. The ability to shift the operational frequency of the dielectric resonator slightly makes them applicable to any 3 Tesla MR systems.

Numerical simulations

Electromagnetic simulations were performed using a female voxelized model placed in the center of a standard body birdcage coil in CST Microwave Studio 2017 (see Supplementary Fig. 1). The body model was initially made for a supine position and the breasts did not have proper shape for the prone position. Therefore, an approximated layered breast model was created on top of the body model consisting of the following tissue types: skin (ɛ = 72.93, σ = 0.49 S m−1), fat (ɛ = 5.6, σ = 0.03 S m−1), gland (ɛ = 67, σ = 0.8 S m−1). The dielectric resonator was placed around the right breast. For the reference case, the body birdcage coil was simulated with the same body model at the same central position, but without the resonator. The dielectric resonator coupled effectively to the birdcage coil and focused its B1+ field on the breast area (see Supplementary Fig. 2). The B1+ field and SAR averaged over 10 g tissue (SARav.10g) distributions were normalized to 1 W of total accepted power at the post-processing step.

A composite effect of the resonator on the birdcage coil transmit performance was evaluated by the ratio (left| {B_{1,{mathrm{rms}}}^ + } right|/sqrt {{mathrm{psSAR}}_{{mathrm{av}}.10{mathrm{g}}}}) (RF safety, rms: root-mean-squared) in the presence of a resonator to the one with the body birdcage coil alone—a so-called RF safety gain:

$$frac{{left| {B_{1,{mathrm{rms}}}^ + } right|/sqrt {{mathrm{psSAR}}_{{mathrm{av}}.10{mathrm{g}}_{{mathrm{with}},{mathrm{resonator}}}}} }}{{left| {B_{1,{mathrm{rms}}}^ + } right|/sqrt {{mathrm{psSAR}}_{{mathrm{av}}.10{mathrm{g}}_{{mathrm{without}},{mathrm{resonator}}}}} }},$$


where the (left| {B_{1,{mathrm{rms}}}^ + } right|) -field value was spatially averaged over the targeted area. The RF safety gain was calculated in the central axial slice (yx-plane) through the breast (Fig. 2d). The peak spatial SAR (10 g averaged) was increased by six times in the presence of the resonator (hot-spots depicted as black circles in Supplementary Fig. 2e, h), whereas an RF safety was increased by five- to ninefold across the breast (Fig. 2d). That means to obtain the same B1+-field value in the breast area and one should reduce the input power, thus reducing the peak values of SAR (see Supplementary Fig. 2f, i).

The International Electrotechnical Commission specifies peak SAR limits for normal and the first‐level controlled operating modes of an MRI examination (SARav.10g = 10 W kg−1 and SARav.10g = 20 W kg−1, correspondingly). The peak SAR values are directly defined by the power accepted by the system (Pacc) that, in turn, sets the maximum of the RF magnetic field (B1+) amplitude. In Supplementary Table 1, a comparison of the Pacc and corresponding mean B1+ values across the breast area for two operating modes without and with the resonator in place are presented. As could be seen in Supplementary Table 1, the mean B1+ values that could be reached with the resonator are more than sevenfold higher than with the birdcage coil alone. It means that (1) more efficient RF pulses could be used in DWI and CEST sequences; (2) repetition times of these sequences (that are often restricted by the RF safety regulations) could be shortened.

To improve the spatial coverage of the resonator, the design with a bigger inner diameter of ceramic discs (126 mm) was simulated. The results demonstrate a 1.6-fold higher field-of-view and a 1.13-fold higher RF magnetic field in the lateral areas (see Supplementary Fig. 4) in comparison with the proposed resonator with an inner diameter of 101.5 mm. However, it is worth noting that, in this case, a 30% loss in the RF magnetic field enhancement (mean value of the B1+-field) in the breast area was observed, i.e., the efficiency of the resonator operation is slightly decreased but still tenfold higher than compared with the body birdcage coil alone. Thus, it is worth noting that via engineering optimization of the design, one should compromise between the effectiveness of the resonator operation and field-of-view.

MR experiments

A written informed consent approved by the institutional review board was signed by volunteers before in vivo MR examinations. In vivo MR images of five healthy female volunteers of BMI range 17.1–21 kg m−3 (min–max), with breast volume 381.4 ± 68.5 ml (mean ± SD) were acquired on 3T Siemens Magnetom Verio whole-body system using T1-weighted 3D GRE sequence with Dixon-based fat suppression: FA = 15°, TR/TE1/TE2 = 7.8/2.4/3.7 ms, acquisition matrix = 640 × 380 × 208, and voxel size = 0.7 × 0.7 × 0.6 mm3. A noise level was calculated as a SD of pixel values in the background areas of MR images. For comparison, MR images were acquired using the body birdcage coil and a 16-channel Breast Coil from Siemens. In the presence of the dielectric resonator, the birdcage coil was used for both transmission and reception of the MR signal. All MR experiments with resonator were performed only after precise RF power calibration procedure, which ensures that used input power within RF safety limits.

To confirm the effectiveness of the manual RF power calibration procedure, B1+ maps were evaluated using the double flip angle method (α = 20°, 2α = 40°) with a small phantom filled with salted water, which imitated the breast, and a bigger one mimicking the human body (see Supplementary Fig. 3). These maps illustrate the actual flip angle distribution in the region of interest. B1+ stands for a circularly polarized RF magnetic flux field that is used to excite the MR signal and the flip angle depends directly onto B1+ amplitude. The GRE images to calculate a flip angle map were acquired with the following parameters: FA1/FA2 = 20/40, TR/TE = 6000/2.5 ms, acquisition matrix = 64 × 64, and a field of view = 320 × 320 mm2. The resulting average angle was in both cases close to the nominal flip angle, confirming the reliability of the manual transmitter calibration.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *