A flexible MRI coil based on a cable conductor and applied to knee imaging

Abstract

Flexible radiofrequency coils for magnetic resonance imaging (MRI) have garnered attention in research and industrial communities because they provide improved accessibility and performance and can accommodate a range of anatomic postures. Most recent flexible coil developments involve customized conductors or substrate materials and/or target applications at 3 T or above. In contrast, we set out to design a flexible coil based on an off-the-shelf conductor that is suitable for operation at 0.55 T (23.55 MHz). Signal-to-noise ratio (SNR) degradation can occur in such an environment because the resistance of the coil conductor can be significant with respect to the sample. We found that resonating a commercially available RG-223 coaxial cable shield with a lumped capacitor while the inner conductor remained electrically floating gave rise to a highly effective "cable coil." A 10-cm diameter cable coil was flexible enough to wrap around the knee, an application that can benefit from flexible coils, and had similar conductor loss and SNR as a standard-of-reference rigid copper coil. A two-channel cable coil array also provided good SNR robustness against geometric variability, outperforming a two-channel coaxial coil array by 26 and 16% when the elements were overlapped by 20-40% or gapped by 30-50%, respectively. A 6-channel cable coil array was constructed for 0.55 T knee imaging. Incidental cartilage and bone pathologies were clearly delineated in T1- and T2-weighted turbo spin echo images acquired in 3-4 min with the proposed coil, suggesting that clinical quality knee imaging is feasible in an acceptable examination timeframe. Correcting for T1, the SNR measured with the cable coil was approximately threefold lower than that measured with a 1.5 T state-of-the-art 18-channel coil, which is expected given the threefold difference in main magnetic field strength. This result suggests that the 0.55 T cable coil conductor loss does not deleteriously impact SNR, which might be anticipated at low field.

Figure 1

Coil schematics for Q measurements (top row) and S-matrix or MRI measurements (bottom row). A fuse was installed (not shown) in loops used for in vivo experiments. VNA: vector network analyzer. RFC: radiofrequency choke.

Figure 2

SNR maps (top) and profiles as a function of depth (bottom) in a phantom for Cu-FR4, cable, and coaxial coils. White lines in the maps indicate the position of the profiles.

Figure 3

Reflection coefficient (S₁₁) as a function of geometry for a cable coil (top) and coaxial coil (bottom). The loops were tuned with circular geometry and wrapped onto a 13.5-cm cylindrical dielectric phantom. The loops were then elongated into ellipses with major (minor) axes of 13-cm (8-cm) and 15-cm (6-cm) without retuning. Scaled representations of the coil contours are overlaid.

Figure 4

SNR profiles as a function of depth in a phantom for cable and coaxial coils arranged with circular (left), 13 × 8-cm elliptical (middle), and 15 × 6-cm elliptical geometry (right). Scaled representations of the coil contours are overlaid.

Figure 5

S-matrix measurements as a function of overlap for two-channel arrays based on cable and coaxial loops. Left: transmission coefficient (S₂₁). Right: maximum reflection coefficient (S₁₁ or S₂₂).

Figure 6

SNR as a function of depth and overlap along the main axis of one coil in two-channel arrays based on cable coils (a) or coaxial coils (b). Cable coil array SNR normalized by coaxial coil array SNR (c). Cable coil SNR normalized by cable coil SNR with 20% overlap (d), which is the approximate position that minimized the transmission coefficient (see Fig. 5). Coaxial coil SNR normalized by coaxial coil SNR with 20% overlap (e). Overlap is expressed as a percentage of the coil diameter (10-cm). Negative overlaps indicate gaps. Schematic of the setup illustrates the 20% overlap case and shows the SNR profile along the main axis of coil 1 (f). Each column of the ratio maps in (c–e) was smoothed to improve visualization using a median filter with 11-mm kernel.

Figure 7

Photograph of the proposed 6-channel flexible coil unrolled with (a), without (b) the protective cover, and wrapped around a cylindrical phantom (c).

Figure 8

The 6 × 6 scattering matrix of the cable coil wrapped around a phantom as shown in Fig. 7. Scale in dB.

Figure 9

Axial (top row) and sagittal (bottom row) flip angle maps measured with the 6-channel knee coil absent (left) and present (right) illustrate negligible interaction with the system body coil during transmission.

Figure 10

The six-channel cable array included MR-visible components. ABS plastic clamps (solid arrow) and foam cushioning (open arrow) were visible in an image acquired with TE = 0.7 ms (left). The components were not visible with TE = 2.5 ms (right).

Figure 11

Representative SNR maps acquired with the 0.55 T 6-channel cable coil, 0.55 T prototype coil, and 1.5 T clinical coil. SNR values overlaid in the bottom right are averaged over three volunteers in a 3-cm ROI in the distal femur (overlaid red circles). Coil photos are inset in the bottom left of each panel.

Figure 12

Inverse geometry factor (1/g) maps for twofold and threefold acceleration in the left–right direction in the axial imaging plane. Text overlays indicate the maximum geometry factors (g_max).

Figure 13

Representative turbo spin echo MR images of a 35-year-old man acquired using the 0.55 T 6-channel cable coil (left column), 0.55 T prototype coil (middle column) and 1.5 T clinical coil (right column). Mucinous degeneration with no meniscal tear is observed in the coronal images (arrows).

Figure 14

Turbo spin echo MR images of a 34-year-old man acquired using the 0.55 T cable coil (left column), 0.55 T prototype coil (middle column) and 1.5 T clinical coil (right column). The images reveal patellar cartilage defects (solid arrows) and a lateral femoral condyle bone contusion (open arrows). The bone contusion was not present at the time of the 1.5 T examination.

Figure 15

Turbo spin echo MR images from a 28-year-old woman acquired with the 0.55 T 6-channel cable coil (left column), 0.55 T prototype coil (middle column), and 1.5 T clinical coil (right column). The cable coil images have peripheral signal saturation (solid arrows). Images acquired with all coils had heterogeneous fat suppression (hollow arrows).

Figure 16

Turbo spin echo MR images from a 31-year-old man acquired with the 0.55 T cable coil during three separate scan sessions on the same day.