The design of a transmit-array RF coil for high field MRI aims at providing the best excitation homogeneity. More elements usually provide higher homogeneity and lower local specific absorption rate (SAR). A hybrid RF head coil, later referred as the ASTRE coil, consisting of 12 transceive elements ̶̶ 11 dipoles and one circularly-polarized (CP) patch ̶̶ and 10 receive-only loops has recently been proposed by our group¹ as an attempt to reach the best compromise between performance and complexity. Here we compare the performance of the ASTRE coil with 2 commercial coils, both in terms of signal-to-noise ratio (SNR) and excitation homogeneity.Our experiments were carried out on a 7T Magnetom scanner (Siemens Healthcare, Erlangen, Germany) equipped with an AC84 head gradient set and 8 independent RF channels for parallel transmission. The commercial RF head coils used for benchmarking are 8-transceiver-loop Rapid-Biomedical (Rimpar, Germany) and narrow-made 1-Tx/32-Rx Nova-Medical coils (Wilmington, MA, USA). The ASTRE coil components are presented elsewhere1. The 1-kW power amplifiers then drive the 12 Tx elements via an SVD box². All three coils are depicted in Fig. 1. The comparisons were based on two tests performed on a spherical 156-mm-diameter doped water phantom with the same conductivity as that of a human brain. First, the SNR was assessed for each coil by estimating the receive profile from a 3D FLASH sequence where the flip angle (FA) map was predicted from prior B0- and B1-mapping³. The FLASH image itself was produced from a standard sum-of-squares combination of the Rx-channels. On the transmit side, for the commercial coils, the default CP mode was used with a broadband RF pulse. For the ASTRE coil, the k_T-point method⁴ was applied to obtain a homogeneous excitation since no proper equivalent CP-mode could be derived from the ASTRE coil geometry. A Bloch equation simulator then provided the predicted FA-map for the FLASH sequence, which could yield the subsequent reception profile. For the SNR calculation, the signal was taken as the mean of the reception map in the phantom, while the noise was the standard deviation of a “0-Volt” FLASH acquisition. In a second step, an excitation homogenization experiment exploiting the ASTRE and Rapid coils’ pTx capability was performed to compare their Tx-efficiencies. The XFL sequence was used for multiple-channel B1-mapping in interferometric mode⁵. Subsequently a 5-k_T-point 30° pulse was designed with 0.9-ms total duration, letting an Active Set algorithm find optimal k_T-point locations given identical RF power constraints for both coils⁶. Because the preamplifier gain levels were different for each coil, all experiments were performed with Siemens “high gain” mode in reception, amplifying the signal by an extra 19 dB before digitization, which prevented noise to be dominated by the ADCs or external sources for the pTx coils.The 3D coil SNR profiles are depicted for each coil in Fig. 2. Compared to the Rapid coil, the ASTRE coil has 49% more SNR on average. Moreover, even though it has 10 fewer Rx-elements than the Nova coil, its mean SNR is also slightly better, by 7%. A deeper SNR analysis of the ASTRE coil reveals the gain brought by the loops and patch is 19% on average (28% in the phantom center) compared to the dipole-only coil. The patch alone accounts for a 4% gain on average, 15% in the central part. For the transmit efficiency comparison, the reference voltage to generate a given average B1-field in the phantom with a pseudo CP-mode was similar for both Rapid and ASTRE coils. However, for the k_T-point experiment, the ASTRE coil outperformed the Rapid coil: the Bloch simulator returned a 5.5 % rms FA spread (NRMSE) across the phantom (normalized to 30° target) versus 7.4 % for Rapid. As a reference, the 1-Tx Nova coil fixed CP mode yielded 33 % NRMSE. Moreover, the ASTRE coil seems to withstand higher duty cycle, with a maximum time-averaged input power on the order of 8W per channel according to preliminary results. We yet have to determine the SAR-predictive model of the ASTRE coil. This work is underway with FEM simulations and subsequent VOP compression⁷.The new ASTRE hybrid pTx coil¹ shows promising results in terms of intrinsic SNR and excitation homogenization efficiency. The improvements include z-segmentation of a higher number of Tx-elements (some of which preceded by SVD power splitters), a transceiver patch at the top of the head, and receiving loops decoupled from transceiving dipoles by their peculiar 3D geometry.