Evaluation of a MR-quadrupole imaging coil for spinal interventions in a vertical 1.0 T MRI



The in vivo pain treatment was successfully performed with the patient in a prone position. The PD-weighted TSE with echo time = 10 ms rendered contrast-to-noise-ratio values of 27 ± 10 for needle/fat, 1.6 ± 5 for needle/muscle, and 4 ± 4.7 for needle/nerve tissue. The mean diameter of the needle artifact was 1.2 ± 0.2 mm. In the T1-weighted gradient echo, the needle's artifact diameter was 6 ± 2 mm; the needle's contrast-to-noise ratio relative to muscle tissue was 4 ± 2, 7.6 ± 1.5 for needle/fat, and 5 ± 1 for needle/nerve tissue. With the PD-weighted TSE (echo time = 10 ms) and the T1-weighted gradient echo, the needle was imaged reliably throughout the intervention. The butterfly surface coil is feasible for the guidance of spinal interventions in a prone patient. Magn Reson Med, 2012. © 2011 Wiley Periodicals, Inc.


Magnetic resonance imaging (MRI)-guided spinal interventions have gained importance over the past years (1–7). Compared with computer tomography (CT), MRI especially simplifies a cranio-caudal approach due to a “real-time” adjustment of the imaging in arbitrary orientation without the need of postprocessing the image data (4). In consequence, the number of MR-guided interventions at the spine or near the spine has steadily grown. The introduction of more open MRI scanners with improved patient access has since enabled a variety of new therapeutic applications in MR-guided spinal surgery (8).

Two so-called open MRI scanner concepts are in use today (9). One has horizontally opposed pole shoes, creating a vertical static magnetic field, the other is a tunnel design, with a wider and/or shorter bore and horizontal main magnetic field (B0) orientation. Apart from the improved access provided by either of these “open” scanner designs, the imaging coil used with each system has a substantial influence on the actual patient access. Depending on the sort of intervention, vertical magnetic fields often require different or even new customized approaches on the design of imaging coils compared with tunnel systems (10, 11). Because of their flexibility and open design, surface single loop coils are frequently preferred for spinal interventions (12, 13). For an optimal access to the spine in open MR scanners with vertical magnetic fields, the patient is usually in a lateral decubitus position, and the single loop surface coil is placed on the patients' back. The lateral patient position is owed to the imaging characteristics of single loop coils; they reach an optimal signal output when the loop plane is parallel to B0. In general, the signal-to-noise ratio (SNR) in MR images is proportional to that of the left-hand circularly polarized component of the transverse magnetic field (B1) produced by the coil (14).

Today, the lateral decubitus position is well established with the use of single loop surface coils for MR-guided discographies, injection therapies, or percutaneous laser disc decompression. However, a prone position of the patient appears to be favorable for several other spine interventions. This is the case in procedures that require a shallower angle of needle/instrument insertion, e.g., mechanical percutaneous nucleotomy (15). In mechanical percutaneous nucleotomy, the protruded disc tissue is mechanically removed with long tenaculies. Geometrical restrictions of the scanner can hamper the insertion of long instruments if the patient is in a lateral decubitus position. To widen the scope of MR-guided interventions to procedures, which require the patient to be in a prone position, Philips introduced a custom designed surface coil with a so-called butterfly design. The theoretical makeup of quadrature surface coils is well known (16–19). Various designs for an optimal field of view have been previously discussed in detail. The designs include figure-8 coils, quadrature coils (butterfly design), and combinations of quadrature and single loop coils. A combination of quadrature coils to a phased array was also considered (14).

One advantage of butterfly coils is their reduced dependence of the SNR to their orientation to B0. Even with the least optimal orientation, their field has components perpendicular to B0, and therefore, these coils still receive a significant MR signal. The characteristics and handling of butterfly coils for field strengths at 1.0 T have not been analyzed, and thus the potential they bear for MR-guided interventions has not been explored. The purpose of our study was to compare the imaging characteristics of a custom butterfly coil with common coils used for abdominal interventions in a phantom study. We focused on imaging characteristics and geometric considerations and restrictions of spinal interventions. MR-guided pain therapy in a prone patient was performed with a butterfly coil as proof of concept.


Simulations of the magnetic field distribution of single loop coils and figure-8 coils in different orientations to B0 were calculated with MATLAB (Mathworks, Natick, MA) based on the law of Biot–Savart. The experiments were performed in an open 1.0 T Panorama MR scanner (Philips Healthcare, Best, The Netherlands). MR images of a thorax shaped gel phantom (Wirogel, Bremen, Germany) with a diameter of 30 cm along the long axis, a transverse diameter of 20 cm along the short axis, and a horizontal diameter of 16 cm were acquired. Three different imaging sequences, two turbo spin echo (20) sequences, and a gradient echo (GRE) sequence (see Table 1 for imaging parameters) were used to compare the four different coils' imaging characteristics: an abdominal coil with 37 cm diameter (vertical) was used as a reference sample and was compared with single loop surface coils with 15 and 20 cm diameters and a butterfly surface coil with two 20-cm apertures in short axis and a 22-cm aperture in long axis (Fig. 1). The outer diameter of the butterfly coil was 24 × 48 cm2. Ten axial image slices were acquired. To evaluate the dependence of signal quality and coil angulations relative to B0, both phantom and coil were tilted from 0° to 90° in 10° increments, resulting in a total of 1200 acquired MR images. The signal was measured in two circular regions of interest (ROI) with a 2 cm diameter at the center of the coil plane and at 20 cm from the surface of the phantom in three image slices (Fig. 2). The SNR was determined by the ratio of the signal of the phantom to the mean signal deviation in air. The SNR of the ROIs was averaged and used as a quantitative parameter to describe image quality; the standard deviation was taken as the error. The SNR of the surface coils was normalized by the SNR of the abdominal coil: SNR (surface coil)/SNR (abdominal coil). The compiled data were then compared with the theoretical signal distribution.

Figure 1.

Imaging coils used in the phantom study: (a) single loop coil with a 15 cm diameter, (b) single loop coil with a 20 cm diameter, (c) butterfly coil with 20 × 22 cm2 apertures, and (d) abdominal coil.

Figure 2.

Imaging characteristic of single loop and butterfly coil: (a) simulation of a single loop coil parallel to B0, (b) simulation of a butterfly coil perpendicular to B0, (c) experimental T2w TSE image of a single loop coil parallel to B0, and (d) experimental T2w TSE image of the butterfly coil perpendicular to B0 (sequence parameters as detailed in Table 1). The ROIs used for the SNR measurement are marked with black and white circles. The gray values are not normalized.

Table 1. Sequence parameters used in the phantom study and the in vivo intervention
ApplicationSequenceTR (ms)TE (ms)Flip angle (°)TA (s)FOV (mm)Resolution (mm)Slice (mm)SlicesTSE factor
PhantomT2w TSE300012090470350 × 3501 × 131024
PhantomPDw TSE600109042350 × 3501 × 251036
PhantomT1w GRE634022350 × 3501.8 × 1.851025
InterventionPDw TSE60010903250 × 1901 × 25136
InterventionPDw TSE60030903250 × 1901 × 25136
InterventionT1w GRE126351.2230 × 2001.7 × 271

The in vivo intervention performed was an MR-guided therapeutic injection in male (85 kg) suffering from lumbosacral pain. The local ethics committee approved the study. Informed consent was obtained from the patient in advance, and preprocedural studies confirmed an appropriate indication for interventional treatment. The patient was in a prone position during the procedure. The midline of the butterfly coil was aligned with the patient's spine in a horizontal orientation (90° to B0). Bilateral therapeutic drug injections were performed while the needle was inserted through the respective side's loop aperture: identification of target anatomy and needle guidance on the left was achieved with interactive proton density (PD)-weighted turbo spin echo (TSE) sequence with echo times (TE) of 10 and 30 ms (see Table 1). The right-sided intervention was performed under guidance of a T1-weighted gradient echo sequence (Table 1). After desired needle placement and subsequent aspiration, a mixture of a long-acting local anesthetic (Carbostesin™ 0.5%; AstraZeneca, Wedel, Germany) and a steroid (Triam™ 40 mg Lichtenstein; Winthrop, Fürstenfeldbruck, Germany) was injected. T2-weighted fat-saturated Spectral Presaturation with Inversion Recovery (SPIR) images (echo time [TE]/pulse repetition time 100/1500 ms, field of view 200 × 200 mm, number of signals (NOS) 6, echo train length (EL) 24, slice thickness (SL), and acquisition time (TA) 4.16 min) were acquired to visualize the mixture's distribution after the injection.

Contrast-to-noise ratio (CNR) of the needle versus spinal muscle tissue, the fatty tissue, and nerve tissue was measured five times to determine and describe needle visibility during the intervention for each sequence type. The diameter of the needle artifact was also measured for all used sequences.


The experimental images' SNR values acquired with the single loop and butterfly coils were in good agreement with the theoretically predicted dependence on the angle relative to B0 (Fig. 3). The SNR of the single loop coils decreased with decreasing the angle to B0 of the coil plane for almost all evaluated sequences. Except with the small single loop coil at the PD-weighted TSE, the characteristic of the SNR decrease was less distinctive. A reciprocal characteristic was found for the butterfly coil. The butterfly coil's SNR in the least optimal angulations was slightly worse than the optimal SNR of the single loop coil with a similar diameter at the T1-weighted GRE sequence (Table 2 and Fig. 3), but in a comparable range. The smaller single loop coil (15 cm diameter) received the highest signal in almost all angulations and sequences, with the exception of least optimal angulations, at which the butterfly coil exceeded the smaller loop coil's SNR or was in the same range. With T1-weighted GRE and T2-weighted TSE, the butterfly coil's SNR was higher than that of the solenoid abdominal coil in all evaluated angulations. With the PD-weighted TSE, the butterfly resulted in higher signals than the abdominal coil for angulations above 40° with respect to B0.

Figure 3.

Dependence of single loop coils' and the butterfly coil's mean SNR normalized by the SNR of the abdominal coil to respective angulations to B0 with the standard deviation as errors: (a) T1w GRE, (b) T2w TSE, and (c) PDw TSE with sequence parameters according to Table 1.

Table 2. SNR values of the different coils normalized by the SNR of the abdominal coil at 0° and 90° relative to B0 for the sequences used in this study
Sequence/coilButterfly coil Single loop coil 15 cm Single loop coil 20 cm 
Angle (°)090090090
T2w TSE1.07 (±0.1)1.6 (±0.1)2.6 (±0.2)2.0 (±0.4)2 (±0.1)0.6 (±0.02)
PDw TSE0.4 (±0.1)1.6 (±0.1)1.7 (±0.3)1.8 (±0.3)1.4 (±0.1)0.6 (±0.01)
T1w GRE1.2 (±0.2)13(±1.5)2.7(±0.3)1.8(±0.8)1.3(±0.4)0.7(±0.1)

With the butterfly coil orientated 90° to B0, dark stripes with lower signal occur at the very outer edge of the phantom due to contributions from Maxwell term (see Fig. 2d). Changing the orientation of the butterfly coil iteratively to 0°, the positions of the stripes changed, whereas the stripe at the upper part of the phantom was bended in field direction toward the middle of the phantom and broadened.

The in vivo pain therapy was executed successfully with the patient in a prone position. The angle of insertion was 74° ± 7° relative to the patient's back, resulting in angulations of 34° ± 5° relative to B0 (see Fig. 4). CNR in PD-weighted TSE images with a TE = 10 ms was 27 ± 10 for fat, 1.6 ± 5 for muscle, and 4 ± 4.7 for nerve tissue (Fig. 5). The mean diameter of the needle artifact was 1.2 ± 0.2 mm. Increasing the TE to 30 ms, increased the diameter of the needle artifact to 1.8 ± 04 mm. The corresponding CNR values changed to 23 ± 6 for fat, 0.1 ± 2.6 for muscle, and 5 ± 4 for nerve tissue. In T1-weighted GRE images taken during the right-sided intervention, the diameter of the needle artifact was 6 ± 2 mm. The CNR of the needle relative to muscle tissue was 4 ± 2, 7.6 ± 1.5 for fat, and 5 ± 1 for nerve tissue. The occurrence of dark stripes was not observed.

Figure 4.

Axial spinal MR images using the butterfly coil in vivo, perpendicular to B0 with the patient in a prone position: (a) PDw TSE with TE = 10 ms, (b) PDw TSE with TE = 30 ms, and (c) T1w GRE sequence. The position of the needle is marked with arrows. The gray values are not normalized.

Figure 5.

Parameters used to quantify the visibility of the needle in the used sequences during the in vivo intervention: (a) CNR values relative to muscle, fat, and nerve tissue and (b) plot of the artifact diameter for the three applied sequences.


For any safe and successful MR-guided intervention, sufficient imaging of instruments and the surrounding anatomy is crucial. Hence, interventional imaging coils should be optimized to the specific ROI in terms of signal intensity and depth of imaging (21). At the same time, rigid or bulky coils should not impair patient access. In MR-guided interventions with the patient in a prone position, like in breast biopsies, dedicated coils are needed (22–24). For most spine interventions, single loop surface coils have proven to be sufficient to monitor many kinds of procedures. The coil's position is generally adapted to the target region of the planned intervention. The disadvantages of single loop surface coils are their limited coverage with sufficient SNR and their dependency on their position plane's angle to B0. It is this dependence on angulations that restricts the patient positioning in open MR scanners with vertical magnetic fields for spinal interventions. In consequence, abdominal interventions can often only be performed on a patient in decubitus position. However, this position is not feasible for all patients or all interventions. Such restrictions have limited possible applications of MR guidance in interventional radiology. The butterfly coil has the potential to broaden the range of abdominal MR-guided interventions.

The small single loop coil with a diameter of 15 cm achieved the highest SNR in our experiments. The drawback of single loop coils with such small diameters is the small area for needle insertion and the limited expanse of imaging (field of view). This can cause difficulties when in obese patients or when the distance between the patient's spine and back exceeds the imaged area for other reasons (25). Moreover, angulated insertions of instruments may be restricted by the small diameter of the coil and the smaller distance between coil and insertion point is disadvantageous with regard to sterility. Nevertheless, single loop coils with small diameters can also be treated as an option for interventions with patients in a prone position.

The butterfly coil's SNR was less sensitive to different coil orientations. Although the position and shape of dark stripes at the outer parts of the field of view changed with lower angles of the butterfly coil relative to B0 toward the ROI, the area above the spine model in the phantom had SNR values only moderately fewer than the optimal SNR of the large loop coil. Nevertheless, the interventionalist should consider shifting the position of the butterfly coil on the patient's back depending on the path of insertion of instruments. In pain therapies analyzed in this study, these stripes stayed beyond possible paths of insertions. For other interventions with less steep insertions, these dark stripes should be considered and shifted accordingly by adjusting the position of the coil. A limitation of the study is the focus on interventional MR imaging. The diagnostic value of the imaging with a butterfly coil was not considered here. As preprocedural imaging, either CT or MRI, is needed to confirm the indication for interventional treatment, the potential use of butterfly coils in clinical routine for diagnostics should be analyzed in a separate study.

The imaging characteristics of the interventional imaging with the butterfly coil were similar to those of single loop coils with comparable loop diameters. Also the patient access was the same as with loop coils of similar size. With bigger patients, the restrictions of the scanner itself could limit the access due to the pole shoes. Even with steep angles to B0, imaging of the spine was still possible with the butterfly coil. This predestines the butterfly coil for interventions, which require a patient position other than decubitus.

The analyzed butterfly coil was not optimized to monitor interventions with instrument orientations parallel to the main magnetic field. Therefore, the CNR reached in the evaluated in vivo intervention was comparatively low, especially in muscle tissue, and the interventionalist did not always see the needle easily in the PD-weighted TSE sequence. Increasing the echo time to 30 ms increased the diameter of the needle artifact, and thus its visibility in fatty tissue. But due to the T2 time of muscle tissue, the CNR of the needle relative to muscle decreased, as did the visibility. The optimal angulations of a needle in MR-guided interventions were found to be about 45° (2).

The horizontally opposed superconducting magnetic pole shoes of the open MRI are about 40 cm apart from each other. With the patient in prone position and the patient table inside the scanner, the space to maneuver the instrument within the scanner was limited to about 30 cm, and an angle of 45° could not be achieved. The path from puncture site to the nerve root resulted in an angle of 34° to B0, and thus in a thinner than optimal needle artifact. Switching the sequence type to a gradient echo sequence, which is more prone to susceptibility artifacts (26), improved the visibility of the needle markedly. With other needles, which might induce bigger artifacts due to greater diameters or susceptibility, this reduction of the artifact due to its more parallel orientation to B0 could turn out to be an advantage.

Despite the thin needle artifact, the position of the needle was visibly displayed continuously throughout the entire intervention. The butterfly surface coil designed for the use in a high field scanner with a vertical field is sufficient for the MR guidance of spinal interventions in a prone positioned patient. In interventions, which require a steeper angle of instrument insertion, the butterfly coil will likely be more advantageous compared with single loop surface coils. Thus, enabling a wider range of applications of MR-guided interventions.