A Novel SENSE-Optimized 8-Channel Hybrid Transmit/Phased Array Receive Head coil for 3T and 4T Horizontal Systems

The quest towards higher field strengths such a 3.0 T and 4.0 T has its advantages due to higher SNR values, but also has its downfalls. For 3.0 T and 4.0 T imaging SAR plays a prominent role on determining many crucial parameters like number of slices per acquisition, duration of RF pulses, etc., which can affect the overall scan time of a sequence. Furthermore, when designing transmit and receive coils for 3.0 T and 4.0 T systems, it is important to ensure that radiofrequency (RF) fields remain below SAR levels that could cause significant tissue heating in human subjects. Commercial available software using a standardized Head Model [7] can assist towards this direction. Parallel Acquisition Techniques (PAT), such as SENSE [1] and GRAPPA [2], can be more useful at 3.0 T and 4.0 T field strengths, since they allow us to reduce sequence time, or increase spatial resolution that can help for obtaining artifact free images from single shot MR sequences, such as, diffusion EPI, BOLD, etc. Various 8-channel receive only head coils suitable for SENSE imaging at both 1.5 T and 3.0 T field strengths were presented [4], [5]. When the receiver coils use the RF body coil as a transmitter at 3.0 T frequencies, SAR limitations occur and pose additional restrictions on sequences when compared to the 1.5 T systems. These limitations might have adverse effects in overall performance for sequences with multi-slice and volumetric capabilities for scanners with 3.0 T frequencies or higher. In this paper, a novel 8-channel hybrid design suitable for head imaging at 3.0 T and 4.0 T whole body MRI systems is presented. It is an innovative design where the transmitter coil is electrically separated from the receive coils but they are mechanically integrated in a single package. The transmitter portion of the coil comprises from a 16-leg birdcage resonator, while the receiver part of the coil consists of 8 non-overlapping mutually decoupled elements conformed on an open dome design. SAR calculations regarding the transmitter coil at 3.0 T and 4.0T, using commercially available software, are presented with excellent agreement with published data at similar frequencies [6]. Prototypes of the Hybrid head coil at both 123.2 MHz and 168.2 MHz were constructed. A direct comparison at 123.2 MHz between the Hybrid head coil and the RF body coil regarding the power required for 1ms 180 pulse was also made. In vivo volunteer imaging using the Hybrid Head coil at 3.0 T incorporating GRAPPA and SENSE sequence protocols was performed for reductions factors of R=2, 3 and 4. MATERIALS and METHODS The PAT-Optimized Hybrid Quadrature Tx/8 channel array receive Head coil is shown in Fig. 1. The coil’s internal diameter is 26cm with an overall electrical length of 25cm. The separation between the adjacent loops for the receiver coil was chosen to be 1.0cm in order to minimize the g factor values inside a 30cm FOV [3]. The two hybrid head coils tunned to 123.2MHz and 168.2MHz. The transmitter is a 16-leg quadarture birdcage resonator, while the receiver part of the coil consists of 8 nonoverlapping inductively decoupled loops conformed on a dome former design. The matching of both ports of the transmitter coil to the head was better than -20dB, while the isolation between them was better than -25dB on 123.2 MHz and 168.2 MHz. The matching of each element of the receiver array to the head was better than -18dB for both frequencies. The isolation between the transmitter and the receiver channels was better than -50dB. The isolation between adjacent loops on the receiver part of the coil was –17dB or better without low impedance pre-amp decoupling. With pre-amp decoupling, the isolation between adjacent channels was better than 20dB. Table 1 illustrates all the technical electrical characteristics for the transmitter and receiver parts of the hybrid head coil at both frequencies. Furthermore, finite difference time domain(FDTD) method (3,4) was employed to solve Maxwell’s equations for calculation of SAR on the human head at both frequencies (123.2 MHz, 168.2 MHz). A digital body model with mesh size 5mm×5mm×5mm has been used. A ROI (region of interest) area is 25 and 25cm. The SAR in each cell was calculated from the normalized root mean square (rms) E field data as: ρ σ / | E | SAR 2 = (E: the normalized rms electrical field, σ: conductivity (S/m), ρ : Density to tissue(kg/m)). The intensity of a 3ms square RF pulse was adjusted to have the 90° flip angle at the target area. Then, the average SAR over the entire head, the maximum SAR in one cell, and the maximum SAR in approximately 1g of tissue for both Tx coils are calculated. The B1 field map and g-factor in Rx coils is calculated. RESULTS and DISCUSSION For our calculations, the average SAR over the entire head and the maximum SAR in approximately 1g of tissue (each cubic 3×3×3 cell region) in the transmitter coil are given in Table 2. Results are in very good agreement with published data [6]. A slight difference between the two calculations is attributed to the used a 12-leg birdcage resonator in [6], while on our calculations a 16-leg birdcage resonator was considered. Also our resonant frequency was slightly lower that 175 MHz. In head model, averaged SAR levels are 2.5 times higher at 168.2MHz than at 123.2MHz. Maximum local SAR levels in 1g of tissue are 22% higher at 168.2MHz than at 123.2MHz. The B1 homogeneity in the head model is 88.9% and 83% at 123.2MHz and 168.2MHz, respectively. The plots of the B1 field magnitude and SAR distributions on the central sagittal plane at 123.2MHz and 168.2MHz are shown in Fig. 2(a), 2(b). At 168MHz, B1 inhomogeneity within the head model is more apparent than at 123.2 MHz. The regions of highest SAR levels are near the outer surface of the head model at locations between which the greatest B1 flux passes. Table 2 illustrates the g-factor profiles for 8 channel SENSE coil at 123.2MHz and 168.2MHz. A comparison between these values indicates that at 4.0 T, the g-factors are slightly improved due to larger phase variations of the B1 field. Furthermore, for 123.2 MHz, a direct comparison between the Hybrid transmit head coil and the RF body coil for an 1ms 180 pulse revealed that the required power for the head transmitter coil is 4.5 times less than the power required from the body RF coil. Fig. 3(a), (b), (c), and (d) show GRAPPA images of human volunteer at 123.2 MHz utilizing the Hybrid head coil and with reduction factors R=2, 3 and 4. A comparison between conventional and PAT imaging revealed that for R=2 and R=3 no apparent artifacts are generated from the PAT algorithm with slight (∼10%) degradation on SNR, while for R=4 the artifacts from PAT reconstruction are more obvious with a further degradation on SNR (∼20%). f (MHz) Average SAR Max SAR in 1g of tissue f (MHz) g-factor R=2 g-factor R=3 g-factor R=4 123.2 0.87 7.11 123.2 1.15 1.69 4.09 168.2 2.09 9.15 168.2 1.09 1.59 3.19 Table 1. Average and maximum SAR Table 2. g-factor values at 3.0 T & 4.0T values at 123.2MHz and 168.2MHz for the Hybrid Tx/8-channel Rx Head