Lung imaging using MRI is challenging, due to the low proton density in the tissue, short T2* values due to multiple air-tissue interfaces, and respiratory and cardiac motion [1]. T2* values decrease with increasing magnetic field strength, thus making lung parenchyma visualization even more challenging in small animals imaged at ultra-high field [2]. With the introduction of ultra-short TE imaging techniques (UTE), the visualization of lung parenchyma was rendered feasible, but is still limited by the rather low SNR, and dedicated receive coils optimized for thoracic imaging of the target animals may be beneficial.

In this work, we present a four-channel receive-only mouse phased-array coil optimized for the thoracic anatomy of mice, and its performance is compared with conventional transmit/receive quadrature volume coil at 11.7 T.

The four-channel mouse thorax phased-array (RAPID Biomedical GmbH, Rimpar, Germany) is built with a very close fitting to the mouse body in order to achieve the highest possible filling factor and SNR (figure 1)[3] . The coil consists of an anterior and posterior element each having 2 coil elements. The coil array housing has an elliptical shape (left-right: 26 mm, anterior-posterior: 19 mm) mimicking very closely the anatomical shape of the mouse thorax in supine position. The coil elements on top are slightly larger (28 x 26 mm²) than the bottom elements (28 x 20 mm²). All neighboring coil elements are decoupled by overlapping the next-neighbor elements decoupled by capacitive network. The Q-factors (Q_unload/Q_load) for each channel are more than 3.

In order to investigate the improvement in SNR in lung parenchyma, UTE datasets of three BALB/c mice (weight 25 - 30 g) were acquired with the four-channels coil and on the subsequent day with a transmit/receive quadrature coil of 40 mm inner diameter (Bruker, Ettlingen, Germany) on an 11.7 T MR system (Bruker Biospec 117/16, Ettlingen, Germany). Mice were positioned supine in the coil array and prone in the quadrature coil, and anesthetized with 2% isoflurane in a mixture of N2:O2 (80:20). Two 3D UTE datasets with different field-of-view (FOV) of 45x45x45 mm³ (170 µm isotropic resolution) and 30x30x30 mm³ (117 µm isotropic resolution) were acquired with either coil. Acquisition parameters were as: TR=4 ms, TE = 0.008 ms, matrix size = 256x256x256, acquisition time 13 min, no respiratory or cardiac gating.

In order to calculate SNR, regions of interest (ROIs) were manually drawn in lung parenchyma (ROI_lung), carefully avoiding main vessels, and outside the animal (ROI_noise), in three coronal slices. SNR was calculated as $$$\frac{mean(ROI_{lung})-mean(ROI_{noise})}{std(ROI_{noise})}$$$

Resulting image quality for the 3D UTE acquisitions, for both coils and both FOVs, is shown in figure 2 for axial images and figure 3 for coronal images. Images acquired with the coil array clearly show less noise for both FOVs. The calculated SNR resulted being for FOV = 45x45x45 mm³ 10.47 ± 2.6 for the coil array and 5.96 ± 2.34 for the quadrature coil. In the acquisitions with FOV = 30x30x30 mm³, SNR resulted 12.11 ± 0.17 for the coil array and 3.7 ± 0.6 for the quadrature coil.

In this work a shape-optimized four-element phased-array coil optimized for thoracic imaging in mice has been presented and evaluated at 11.7 T. For small field of view / high-resolution imaging, the optimized coil yielded an average SNR improvement by a factor of almost three for 3D UTE data.

The gain in SNR enables high-resolution high-fidelity anatomy imaging of the lung parenchyma in reasonable short acquisition times.