Full-wave electromagnetic modeling is often employed to characterize RF transmit coils and establish safety parameters restraining power to comply with IEC/FDA guidelines limiting SAR. Modeling SAR for prone breast imaging can be problematic, as the four publicly-available female whole-body models are oriented in the standing or supine positions.^1-4 Furthermore, applicability of breast modeling is complicated by the anatomical variability of lipid and fibroglandular tissues exhibiting disparate electrical and physical properties. This work presents simulations at 7T using a set of high-resolution, anatomically-correct breast phantoms exhibiting varying proportions of fatty and fibroglandular tissues, with ACR BI-RADS breast density classifications of a-d. Moreover, effects of altering the girth and length of the breast tissue, as well as the positioning within the RF coil, are modeled.Four breast phantom spanning the four density classifications and derived from 3D MRI acquisitions⁵ were converted for import into commercial FDTD software (XFdtd 7.4, Remcom, State College, PA). The frequency-dependent dielectric properties of breast fat, breast glandular tissue, and skin at 298 MHz, for 7T, were assigned from the IT’IS tissue properties database.⁶ A quadrature 1H breast volume coil gridded at 1-mm resolution (the maximum cell size validated against bench impedance measurements and scan data) was meshed to provide excitation.⁷ The phantom size was scaled for four different scenarios: 1) centered with 1-cm spacing and full coil loading of 8-cm depth interior of the coil; 2) 2-cm spacing with 8-cm depth; 3) 1-cm spacing with 4-mm depth; and 4) offset with 1-cm and 3-cm spacing to the coil on the L/R axis and 8-cm depth. Each of these phantoms was extruded and seamlessly fused to the Ella whole-body voxel model.⁸ The quadrature coil was driven by 298 MHz sinusoidal feeds with appropriate 90° phase shifts and amplitudes delivering equal power to each channel. Cell gridding was meshed between 0.5-1.0 mm for the coil and phantom, and all curved geometries utilized the software’s conformal meshing capabilities. The model was surrounded by a quarter-wavelength of free space padding cells, and the boundary comprised seven perfectly matched layers. Simulation convergence was determined by transients dissipating to -50 dB deviation from the pure sinusoidal wave. Steady state field data and SAR were calculated throughout the phantom. All results were scaled to achieve an average |B₁⁺| within the coil of 1 μT.Figure 1 displays calculated 10-g average SAR values and plots of the breast phantoms on the central sagittal plane. The SAR plots indicate increased power deposition in the higher-conductivity and permittivity fibroglandular tissue regions. At close 1-mm spacing and full 8-cm depth, the extreme fibroglandular breast phantom exhibited threefold the average SAR as compared to the fatty phantom. The effect was moderated with greater spacing between the phantom and the coil, but in all cases, the densest tissue phantom demonstrated the highest SAR.The results demonstrate variation among breast phantoms presenting varying levels of heterogeneity of healthy fat and glandular breast tissue. This example illustrates the discrepancy of modeling results at high fields for various breast densities. For studies requiring high power, a priori knowledge of the subject’s breast density classification allows for adjusting coil operation guidelines, ensuring safety limits are respected while maximizing power availability. Regardless, characterization and SAR analysis of a breast coil should include modeling with a breast phantom exhibiting extreme fibroglandular tissue.