In Vivo Sodium Imaging of Kidney Using 3d Ultrshort Echo Time Sequence

Introduction: Local Tissue Sodium Concentration (TSC) levels could serve as a biomarker for tissue viability and integrity in stroke [1] and tumor [2]. It is therefore anticipated that TSC measurements can also help to diagnose renal diseases, since vascular sodium concentration is regulated by the corticomedullary sodium concentration gradient [3,4]. However, the SNR in Na Magnetic Resonance Imaging (Na-MRI) is low due to the Na-nuclei’s low in vivo concentration, low gyromagnetic ratio, fast signal decay, and the required small voxel sizes (< 4 μl) in the rat kidney. The aim of this study was to adapt a 3D ultrashort echo time (3D-UTE) sequence for quantitative Na-MRI of the rat kidney and to elucidate the benefits of the UTE over the commonly used gradient-echo (GRE) sequence [4,5]. Methods: Na transceiver surface coil was developed and was used in conjunction with a custom built H transceiver volume resonator for phantom tests and in vivo imaging. The saddle-shaped transceiver surface coil was geometrically adapted to the size of the target object in order to maximize the measured signal intensity in both kidneys. For balancing purposes at 105 MHz, the resonance circuit was designed with two fixedvalue (C1,C2) capacitors and a variable capacitor (ΔC3) (see Fig. 1). Remotely variable matching and tuning was achieved on a separate circuit board to which the resonance coil was connected via a coaxial cable (7cm length). The unloaded to loaded Q-factor ratio was measured to be 170/130. A concentration phantom containing gels of various NaCl and agarose concentrations (40-160 mM, and 0-5% resp., Fig 2a) was used to compare both MR sequences. It is well-known that the fast Na transverse relaxation time (approx. 0.7 to 3ms) causes fast signal decay and limits the ability to quantify the TSC in vivo [6]. Nevertheless, spin-density Na signal can be recorded with fast imaging sequences, such as the 3D-UTE (ParaVision 5.1, Bruker BioSpin GmbH, Ettlingen, Germany). The 3D-UTE sequence is an FID readout technique which samples k-space in a radial manner starting from the k-space center [7]. Minimum TE is then only limited to half of the rf pulse length and the delay between the end of the rf excitation and the beginning of the data acquisition window. For comparison purposes a standard 3D-GRE sequence was used which suffers from long TE due to refocused read-out gradient and hence delayed sampling of the k-space centre. However, when radial read-out techniques are applied, deviations of practically achieved from theoretically expected gradient magnitudes cannot be neglected during ramp-up times. The 94/20 Bruker BioSpec’s gradient system provided 740 mT/m maximal amplitude and 6900 T/m/s slew rate. To correct for gradient imperfections radial k-space trajectories were measured using a spherical homogeneous phantom before the experiments in order to avoid reconstruction-related artifacts [8]. Due to the limited available SNR in the Na channel it was decided to use scaled H trajectories for reconstruction of the 3D-UTE Na images. SNR was calculated as the difference between the signal intensity in each voxel subtracted by the noise mean and divided by the standard deviation of the noise. Noise and signal Regions-of-Interests (ROIs) are shown as yellow and green ROIs (see Figure 2a and 3). SNR maps were calculated in order to compare the two sequences based upon their achieved SNR values in both phantom and in vivo measurements. SNR maps of the concentration phantom (Fig. 2b and c) and in vivo Na MR images of a healthy rat (~350 g) (Fig. 3) were acquired with 3D-UTE and 3D-GRE sequences using identical sequence parameters as follows: isotropic 1mm voxel size, TR = 20 ms, Tacq = 13 min, BW = 5 kHz, 0.1 ms block pulse length, FOV = 6.4 cm and TEUTE/GRE = 60μs / 3.16ms. H-reference scan was recorded with a 3D-Flash sequence using TE/TR = 3 / 20 ms, anisotropic resolution of 0.5x0.5x1mm, and Tacq = 3 min. Results and Discussion: Na SNR maps for the concentration phantom shown in Fig. 2a are presented in Fig. 2b-c and for the in vivo comparison in Fig. 3. According to the calculated SNR values shown in Table 1, the 3D-UTE sequence outperformed the 3D-GRE sequence by factor 1.6 in phantom and by factor 3 in the in vivo measurements. SNR benefit is most likely a result of the lower TE achieved with 3D-UTE (0.06 ms) compared to 3D-GRE (3.16 ms). The SNR gain was more evident in in-vivo kidney measurements due to the fact that radial 3D-UTE sequence was less prone to motion artifacts than the 3D-GRE sequence (Fig 3). Recently published in vivo studies [4,5] used 3DGRE technique with long TE > 1.7 ms to acquire axial slices through the rat’s kidney. High 1x1mm2 in-plane resolution could only be achieved in those studies through anisotropic sampling schemes. However, 3D-UTE enabled to acquire isotropic image data with high SNR and the ability to additionally present coronal and sagittal cross-sections through the rat’s kidney (Fig. 4b-d). Despite the inhomogeneous B1-profile of the surface coil, TSC quantification will now be possible with 3D-UTE, when a dual rf resonator system is used [9].