Traveling Wave Imaging of the Human Head at 7 Tesla: Assessment of SNR, Homogeneity and B1+ Efficiency

Introduction: The traveling wave concept for 7T MR imaging was first demonstrated by Brunner et al [1]. In the most common RF transmit coil designs, a quadrature RF field is created within a cylindrical volume in the near-field regime. In contrast, the traveling wave approach creates a traveling wave mode down the entire bore of the scanner. This has the potential to provide more uniform excitation and whole body imaging, though reflections at dielectric boundaries such as between air and tissue typically create non-uniform excitations in large volumes even with the traveling wave system. We have constructed a traveling wave system based on a patch antenna and have assessed its performance for human head imaging, including homogeneity, B1 + efficiency and calibrated SNR comparisons to a CP head coil. Methods: A circular patch antenna was used for excitation and reception. The patch antenna consists of a square slab of 32mm thick acrylic (www.mcmaster.com) 400mm on a side. The back side is covered with self adhesive copper foil (3M St. Paul MN) to form a ground plane, and a 348mm diameter circular patch of copper foil is applied to the front side. 0 and 90 degree feed points are created 75mm from the center of the patch by drilling holes through the substrate and attaching the shield of an RG58 coax to the ground plane and the central conductor of the coax to the circular patch. RF power is split equally with a Wilkinson divider and phase between the two driving ports is controlled through cable length. One T/R switch and preamp assembly is used in each line (Stark Contrast, Erlangen Germany), and the signal from each port is recorded separately and the reconstructed images are combined as the sum of squares. The patch antenna was compared to a CP head coil (Invivo Corp, Gainesville FL) with 270mm diameter and 150mm length. Power deposition for the patch antenna was measured using a large rectangular gelatin phantom with physiological saline concentration (350 x 230 x 90mm dimensions) with a fluoroptic probe (Luxtron, Santa Clara CA) and an infrared thermometer (Fluke, Everett WA). Simulations were performed in XFDTD (Remcom, State College PA) to determine SAR distribution and radiative losses. Human images were acquired with a 2D GRE (TR/TE/Flip = 100/5/20, Slice=3mm, BW=300, 256x256, FoV 500mm and 280mm). B1+ was mapped in a coronal plane passing through the brightest part of the image for both the patch antenna and CP head coil by taking a series of 2D GRE images at a number of different RF pulse amplitudes (TR/TE = 1000/5, Slice=3mm, 64x64, FoV=200mm, BW=180). Based on the B1+ measure, SNR maps were obtained by taking a 2D GRE with RF pulse magnitude set to provide equal excitation at the brightest central region of the head for each coil (TR/TE/Flip = 200/4.07/50deg, 128x128, FoV 220x220mm, BW=300) and an additional scan with no RF excitation to provide noise statistics. All images were obtained on a Siemens 7 Tesla scanner (Siemens Healthcare, Erlangen, Germany).