Prior work has demonstrated SAR reduction with a high dielectric constant (HDC) ring surrounding the human brain [1]. This SAR reduction was achieved while maintaining the same mean excitation flip angle throughout the brain as compared to a baseline case without the dielectric insert, with the associated interpretation that for the same B field, the E field (and SAR) has been reduced in the brain. This result was supported with a simulation of SAR in the brain and global power measurements experimentally. Global power measurements do not offer insight into the behavior of the electric fields locally within the region enhanced by HDC. It is possible to examine this behavior experimentally, with PRFS temperature mapping in an agar phantom [2]. In this experiment, the local heating is directly proportional to the local SAR and electric field.A spherical Siemens quality assurance phantom (diameter 160 mm) was drained and then filled with a mixture of water, agar, and NaCl, yielding material with an approximate permittivity of 78 and conductivity of 1.7 S/m. The phantom was then wrapped to provide thermal insulation. Heating experiments were conducted without (Fig. 1a) and with (Fig. 1b) a ring of HDC material present. The experiment with HDC was repeated twice, altering the configuration of material to generate a different enhancement pattern the second time. This material has an approximate permittivity of 900 and block dimensions of 105 x 88 x 20 mm. B1 mapping was first performed with the Bloch-Siegert technique [3,4] to acquire a transmit efficiency map (units uT/Volt) over the transverse and coronal orientations. This was followed by a multi-echo GRE acquisition to acquire baseline phase data, a heating scan that utilized the volume body coil to apply heating, and another multi-echo GRE acquisition to acquire phase data. The multi-echo GRE acquisition utilized 6 monopolar echoes ranging from 6 – 29 ms, a TR of 35 ms, 2 averages, flip angle of 15 degrees, a resolution of 2 x 2 x 10 mm, scan time of 6.3 seconds, and were acquired in the coronal orientation. Temperature maps were formed as described in Ref 2, using an off-resonance field map instead of a phase image at a single echo time. The heating scan applied a series of hard pulses (duration 1 ms) with a TR of 55 ms, for a duration of 1 minute and 53 seconds. The two cases (with and without HDC) were normalized such that the heating sequence achieved the same mean B1+ over the whole central transverse slice of the phantom. All scans were performed on a Siemens 3T PrismaFit. Reception utilized a Siemens 20 Ch head matrix. Transmission utilized the 2 Ch Body coil driven in a circular polarization mode.Measured temperature maps and B1+ maps are displayed in Figure 2. The B1+ maps of the case with HDC exhibit more inhomogeneity due to the strong and uneven field enhancement near the edges, but the mean field value over the transverse slice in all cases is 14 uT. This normalization lead to a heating pulse voltage of 379.09 volts for the case without dielectrics and 232.06/225.14 volts for the two cases with dielectrics, and a measured system power reduction of 57.4/59.4%. The temperature map of the case without HDC exhibits a strong asymmetry along the upper right quadrant. The first case with HDC (“Case 1”) exhibits a heating pattern similar to the case without HDC, but the overall level of heating is significantly reduced. In the second case with HDC (“Case 2”), the B1+ enhancement in the coronal slice becomes asymmetric. The temperature map shows a similar pattern, with the peak heating rotated slightly relative to the peak B1+ enhancement. Overall heating levels are also significantly reduced (by 50 – 66%) relative to the case without HDC.The results appear to support the conclusion that a ring configuration of dielectric material can reduce the E field and the SAR for the same nominal B field, observed through a reduction in the heating. These results also suggest that the E and B fields may not be fully coupled, as they are enhanced unequally by the dielectric material. The measured global power reduction, taken as the peak forward – reflected power, was 57.4/59.4% for the two cases with HDC. Normalizing over the central portion of the transverse B1+ map only would lessen the overall power reduction and heating reduction compared to the case without HDC. An understanding of these temperature & SAR patterns could help guide development of dielectric inserts to further reduce SAR in the brain.