Historically, low channel count scanners have necessitated arrays of coil elements whose dimensions were based on coverage and SNR at a specific target depth, while SNR was sacrificed at other depths. Today, MRI receive chains with high channel counts have been exploited in general purpose phased array coils, where SNR can be maximized over a wider range of depths by covering the target region with an array of small coils. The strategy is typically to reduce the coil size while maintaining sufficiently high unloaded-to-loaded Q ratio to guarantee that the composite array maintains performance at depth^1. Our approach aims to improve upon the SNR of a high element count array by introducing coils whose fields are orthogonal to those of traditional elements. This strategy immediately brings to mind multi-lobed (i.e. “butterfly”), stripline, or orthogonal loops^{2, 3,4,5,6}. We propose an alternative “loopole” element that has a non-uniform current distribution which exhibits both loop and electric dipole behavior ⁷. Orientating the loopole with its main axis orthogonal to that of traditional array elements provides local quadrature behavior; this advantage can be accentuated by placing the “high-current” arm of the loopole adjacent to the sample. In this work, the 3-D loop/loopole array was applied to 3T spine imaging.

A six element 3D loop/loopole spine array was constructed consisting of four planar loops 12cm along z and 12 cm wide with 6 capacitors, and two perpendicularly oriented loopole elements concentric with the central loops 15cm along z and 5 cm high with 8 capacitors (Fig.1A). Planar loops had a balanced capacitance distribution and the loopoles had an unbalanced capacitance distribution (Fig. 1B). One active and one passive de-tuning circuit were integrated to each element. An in-house built pre-amplifier interface was used to receive signals from the spine array on a 3.0T scanner (Siemens, Germany). The array was analyzed in various modes A) 2 central loops compared against 2 central loops plus 2 loopoles to demonstrate quadrature benefit and B) 4 planar loops compared against 6 element 3-D Spine array (Fig.2) to evaluate performance in the presence of more elements. SNR was calculated with the Kellman method8 from GRE acquisitions with and without RF excitation (TR/TE/Flip/BW = 1500ms/4.07ms/90/300Hz per pixel, Matrix =64, FoV = 350mm, Slice = 8mm) with flip angle calibrated in an elliptical phantom (50cmL x 30 cmW x 20 cmH , Єr = 40, σ = 0.58)

The Unloaded to loaded Q value of the planar loop elements and loopole elements were 275 & 45 and 290 & 100 respectively, with Q ratios of 6 and 2.9 indicating sample noise dominance. Better than -20dB match was achieved for all elements. Isolation between concentric planar loops and the orthogonal loopoles was -20dB indicating a high level of orthogonality, while the isolation between loopoles and the planar side loops in the 6-element array was -8 dB. Central sagittal SNR maps normalized by the excitation flip angle (Fig.3) indicate that the addition of orthogonal loopoles to 2 central loops provided a SNR improvement of approximately 20% throughout the imaging plane. The 6 element 3-D spine array exhibited enhanced SNR at shallow and deep regions, while the arrays performed similarly in central regions (Fig 4).

The addition of orthogonal loopole elements to central balanced loops provided the expected quadrature SNR boost near the elements as well as deeper regions, which may be attributed to the deliberate imbalance imposed on the loopole current distribution. One might expect this SNR advantage to be eliminated with a dense receive array that, through weighted signal combination, can emulate optimal element structures at different depths. However, our preliminary results showed improved SNR performance when loopole elements were added to 4 planar loops. The SNR boost observed in deep regions by the addition of loopole elements to planar array is somewhat unexpected and indicates that complimentary fields were collected. This observation may be related to the loopole’s unique current pattern that includes both curl-free and divergence free components that is predicted to improve performance in ultimate SNR calculations⁹, whereas balanced loops capture only divergence free components. Recent works on electric dipole antennas have also suggested potential SNR gains for deep tissue at 3T^10,11. The addition of orthogonal loopole elements could be a practical means for improving SNR, for example, in the posterior section of general-purpose spine/torso arrays where the loopoles could occupy unused space in the patient table. A “3-D loop/loopole” array also has the potential to improve parallel imaging performance due to more diverse spatial profiles.