Published online before print April 19, 2002, 10.1148/radiol.2233010824
(Radiology 2002;223:870-876.)
© RSNA, 2002
Wrist: Improved MR Imaging with Optimized Transmit-Receive Coil Design1
Armen Kocharian, PhD,
Mark C. Adkins, MD,
Kimberly K. Amrami, MD,
Kiaran P. McGee, PhD,
Peggy A. Rouleau, MD,
Doris E. Wenger, MD,
Richard L. Ehman, MD and
Joel P. Felmlee, PhD
1 From the Magnetic Resonance Imaging Research Laboratory, Mayo Clinic, 200 First St SW, Rochester, MN 55905. From the 2000 RSNA scientific assembly. Received April 20, 2001; revision requested June 14; final revision received December 20; accepted January 4, 2002. Supported by a grant from the Whitaker Foundation. Address correspondence to A.K. (e-mail: kocharian.armen@mayo.edu).
 |
ABSTRACT
|
|---|
The optimized wrist coil was designed and built as a transmit-receive birdcage coil for 1.5-T magnetic resonance (MR) imaging. Phantom studies were conducted to compare the optimized design with custom-designed and commercially available phased-array wrist coils and showed a 50%–90% improvement in signal-to-noise ratio (SNR). Blinded review of wrist images obtained in six volunteers showed that the optimized birdcage coil was preferred in 75% of the comparisons. An optimized birdcage coil designed for wrist imaging has improved both SNR and uniformity compared with those with a phased-array coil with the same geometry.
© RSNA, 2002
Index terms: Magnetic resonance (MR), coils • Magnetic resonance (MR), technology • Wrist, MR, 43.121411, 43.121415
|
INTRODUCTION
|
|---|
Magnetic resonance (MR) imaging of the wrist can be technically demanding because high-spatial-resolution images of thin sections are required to adequately inspect the anatomy. Carpal tunnel syndrome, nerve compression, ligament injuries, tendon abnormalities, and scarring within the wrist can result in complex clinical scenarios that are difficult to interpret (1–7). To support the increasing demands of wrist imaging, this practice requires improved receiver coils with high signal-to-noise ratio (SNR) and uniformity.
To facilitate patient comfort during wrist MR imaging, the receiver coil can be placed at the side of the patient. Due to increased magnetic field inhomogeneity at the patient’s side, the resonance peaks of both tissue- and lipid-based protons broadens, which degrades the image quality (especially when fat saturation is used). To address this problem, patients can undergo imaging in the swimmer’s position by placing their arm over their head and into the receiver coil at the magnet isocenter. However, the extended imaging time associated with MR imaging increases the patient’s discomfort when the swimmer’s position is used, and it cannot be accomplished by all patients. Therefore, wrist coils should be designed to allow imaging at isocenter and at the patient’s side.
Recent advances in MR imaging involve the use of surface coil arrays to provide high SNR and an extended field of view (8). However, not all coils are capable of generating both the high SNR and high B1 field homogeneity required for optimal imaging (1). Other authors (9) have shown that phased-array coils (PACs) can provide higher SNR than that with Helmholtz coils but with dramatically reduced image uniformity. The spatial signal intensity variation at locations near surface coils or coil arrays causes image nonuniformity. Intensity variation reduces image reliability when high-spatial-resolution studies of the wrist are acquired.
Birdcage radio-frequency coils are a common head coil design and are known to provide a highly uniform B1 field, which results in high image uniformity, especially compared with that with PAC designs (10–12). To our knowledge, however, the transmit-receive designs have not been optimized for wrist imaging. Effective receiver coil design to balance both SNR and uniformity for wrist imaging remains an ongoing area of development in MR imaging (5,6). Unfortunately, the importance of both high SNR and high uniformity is not fully appreciated. While phased-array receiver coil designs offer improved SNR, the resulting images are inherently nonuniform and also require additional imager hardware and software. Uniform receiver coil designs often lack high SNR.
The optimized wrist coil that provides both high SNR and high uniformity without additional hardware requirements remains an engineering challenge. Our hypothesis was that birdcage receiver coil design can be optimized for wrist MR imaging at the patient’s side and can provide higher SNR and improved uniformity compared with those with the same-sized phased-array receiver coil.
The purpose of this study was to design and evaluate an optimized birdcage transmit-receive coil for the wrist that can provide higher SNR and higher uniformity compared with those with the same-sized PACs.
|
Materials and Methods
|
|---|
Coil Design and Construction
In our laboratory, we designed and built an optimized birdcage coil (BC10 [Birdcage Coil 10-cm diameter]) as a high pass circuit (Fig 1). Analysis of the range of clinical wrist MR images yielded a volume coverage for the coil of 10.5-cm diameter and 16.5-cm length. The 12-leg optimized birdcage coil was assembled with 12.7-mm-wide and 0.08-mm-thick copper for the end rings and 6.35-mm-wide copper for the legs. The design parameters for this coil included radio-frequency transmission and reception in quadrature mode.
The design was compared with two custom-designed four-channel PACs for the wrist, with sizes of 15 x 16 cm for each pair of coils (8). The first PAC was mounted on a cylinder (cylindric PAC) with a diameter of 10.5 cm, which exactly matched the diameter of the optimized birdcage coil. The second PAC was designed and built on flexible flat plates (flat PAC). Both cylindric and flat PACs were built for wrist imaging as receive-only PACs. Comparative studies were also carried out with a commercially available hand and wrist two-coil array (MRI Devices, Waukesha, Wis). This coil is roughly elliptic, with approximately 70-cm2 cross-sectional area (6 x 12 cm) and 10-cm length.
Data Acquisition
All experiments were performed with a 1.5-T MR imager (Signa Lx; GE Medical Systems, Milwaukee, Wis) with approval of our institutional review board. For the volunteer study, informed consent was obtained after the nature of the procedure had been fully explained. The coils were compared by performing phantom and volunteer studies. The phantoms contained 16 mmol/L CuSO4 solution. Two phantoms were fabricated (phantom A, 6-cm diameter, 18-cm length; phantom B, 9-cm diameter, 17-cm length) in response to the different aperture sizes of the coils being compared. The phantoms were imaged with the following spin-echo pulse sequences: axial plane, repetition time msec/echo time msec of 400/20, 10-cm field of view, 256 x 192 acquisition matrix, one signal acquired, 3-mm section thickness; and coronal plane with a 24-cm field of view, and identical remaining parameters.
To assess the receiver coil performance in the coronal and transverse planes, measurements of uniformity and SNR were performed on the phantom images. Uniformity (U) was calculated within a user-defined region of interest, according to the formula U = (Imax - Imin)/(Imax + Imin), where Imax and Imin are the maximum and minimum signal intensity values within the region of interest. Hence, a lower value of U corresponds to higher uniformity within the image. The SNR was determined with the subtraction technique specified by the National Electric Manufacturers Association (13); that is, two phantom images were acquired and subtracted to obtain the noise image. The SNR measurements were performed at the center of the phantom images by using a circular region of interest that was 10% of the phantom area. The uniformity measurements were performed within a rectangular region of interest that was 10% of the phantom area. The SNR and uniformity measurements were performed by one author (A.K.).
Fat saturation was analyzed (A.K., J.P.F.) by measuring the line width of the lipid spectra and lipid signal intensity on a T1-weighted image acquired with a fat-saturation pulse sequence (400/20, 10-cm field of view, 256 x 192 acquisition matrix, 3-mm section thickness). Two phantoms were used for this study. The first simulated the wrist and contained a center region (4 x 7 x 17 cm) filled with 16 mmol/L CuSO4 solution surrounded by a 1-cm-thick layer of canola oil. The second phantom was a 10-cm-diameter cylinder filled with canola oil that was placed outside the receiver coil to simulate adjacent body fat to the receiver coil.
The contrast-to-noise ratio (CNR) measurements were performed with phantom images, according to the formula CNR = (IW - IF)/N, where IW is a signal intensity for water solution, IF is a signal intensity for lipid, and N is noise. Noise was measured according to the technique described in reference 13. The fat to signal intensity ratio with fat saturation versus that without fat saturation was also determined with the phantom images. Figure 2 shows the phantom image (water solution of CuSO4 surrounded by canola oil).
View larger version (104K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. Axial T1-weighted MR image (400/20, 10-cm field of view, 256 x 192 acquisition matrix, 3-mm section thickness) depicts the phantom (water solution of CuSO4 surrounded by canola oil). Dashed circles indicate the regions of interest for signal intensity measurements.
|
|
Six studies were conducted with healthy volunteers (two women and four men; age range, 31–47 years; mean age, 41.5 years) with an imaging protocol designed for clinical wrist imaging and two coils, the optimized birdcage coil and the cylindric PAC. Examples of the in vivo wrist images are presented in Figure 3 (T1-weighted high-spatial-resolution images with 512 x 512 acquisition matrix), Figure 4 (T2-weighted transverse images obtained with our standard clinical protocol), and Figure 5 (coronal T2-weighted gradient-echo images). The following pulse sequences were used: T1-weighted transverse, coronal, and sagittal spin-echo MR imaging (500/14, 10-cm field of view, 3.0-mm section thickness with 1.5-mm spacing, two signals acquired, 256 x 256 acquisition matrix for the coronal and sagittal planes and 512 x 384 acquisition matrix for the transverse plane), T2-weighted transverse and coronal fast spin-echo MR imaging (3,000/48, 10-cm field of view, 3.0-mm section thickness with 1.5-mm spacing, two signals acquired, 256 x 256 acquisition matrix, echo-train length of eight).
View larger version (166K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a. Transverse in vivo T1-weighted MR images (500/14, 200-cm field of view, 512 x 512 matrix, 3-mm section thickness) of a normal wrist in a male volunteer were obtained with the (a) optimized birdcage wrist coil and (b) cylindric PAC. Arrows indicate the areas of the carpal tunnel, which appear sharper and with higher SNR in a.
|
|
View larger version (163K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b. Transverse in vivo T1-weighted MR images (500/14, 200-cm field of view, 512 x 512 matrix, 3-mm section thickness) of a normal wrist in a male volunteer were obtained with the (a) optimized birdcage wrist coil and (b) cylindric PAC. Arrows indicate the areas of the carpal tunnel, which appear sharper and with higher SNR in a.
|
|
View larger version (137K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a. Transverse in vivo T2-weighted MR images (3,000/48, 10-cm field of view, 256 x 256 matrix, 3-mm section thickness, echo train length of eight) of a normal wrist in a male volunteer were obtained with the (a) optimized birdcage wrist coil and (b) cylindric PAC. Arrow indicates the nonuniform fat saturation that can occur with the cylindric PAC.
|
|
View larger version (133K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b. Transverse in vivo T2-weighted MR images (3,000/48, 10-cm field of view, 256 x 256 matrix, 3-mm section thickness, echo train length of eight) of a normal wrist in a male volunteer were obtained with the (a) optimized birdcage wrist coil and (b) cylindric PAC. Arrow indicates the nonuniform fat saturation that can occur with the cylindric PAC.
|
|
View larger version (174K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a. Coronal in vivo T2-weighted gradient-echo MR images (33/15, 256 x 192 matrix, 10-cm field of view, 1-mm section thickness, 0° flip angle) of a normal wrist in a male volunteer were obtained with the (a) optimized birdcage wrist coil and (b) hand and wrist two-coil array. Arrows indicate the regions of signal intensity falloff and decreased z-axis coverage associated with the smaller coil.
|
|
View larger version (171K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b. Coronal in vivo T2-weighted gradient-echo MR images (33/15, 256 x 192 matrix, 10-cm field of view, 1-mm section thickness, 0° flip angle) of a normal wrist in a male volunteer were obtained with the (a) optimized birdcage wrist coil and (b) hand and wrist two-coil array. Arrows indicate the regions of signal intensity falloff and decreased z-axis coverage associated with the smaller coil.
|
|
Five observers (four board-certified radiologists [M.C.A., K.K.A., D.E.W., R.L.E.] and one board-certified medical physicist [J.P.F.]) reviewed the volunteer images in a blinded fashion. Images of the same volunteer and the sections acquired with each coil (optimized birdcage coil and cylindric PAC) were placed on a single film (14 x 17-inch film, 6:1 format). No annotation was indicated on the film except for a single code in the upper center for image tracking (eg, 1.1, 1.2). Images of the same three sections from each volunteer were presented by using the left side for one coil and the right side for the comparison coil. The left-to-right coil position on the film was randomly determined to prevent review bias and was unknown to the observers. The following image quality evaluation scale was used to grade each left-to-right comparison on the film: 1, unacceptable; 2, marginally acceptable; 3, acceptable quality; 4, good quality; 5, excellent quality. In general terms, SNR, uniformity, and artifacts were used to grade image quality and visualization (sharpness) of specific anatomic structures, such as tendons, ligamentous structures, soft tissues, and bones. The 180 images in six volunteers that were acquired with both the optimized birdcage coil and the cylindric PAC were analyzed.
|
Results
|
|---|
SNR and uniformity values are summarized in Tables 1 and 2. Coronal and axial MR images were acquired to study coil uniformity along the x, y, and z axes. Uniformity profiles of the coronal and axial images are shown in Figures 6 and 7. The data show that the optimized birdcage coil had improved uniformity along all three coordinate axes compared with that with the cylindric or flat PACs. SNR measurements at the phantom center showed a 50%–90% increase with the optimized birdcage coil compared with the cylindric or flat PACs.
View larger version (27K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a. SNR profiles of phantom images correspond to coronal plane. In a and b, BC10 = optimized birdcage wrist coil, MRID = hand and wrist two-coil array coil, PACC = cylindric PAC, and PACF = flat PAC. Uniform distribution is seen along (a) the z axis and (b) the x axis. In a and b, the optimized birdcage coil data demonstrate improved SNR and uniformity compared with data for the other coils.
|
|
View larger version (24K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b. SNR profiles of phantom images correspond to coronal plane. In a and b, BC10 = optimized birdcage wrist coil, MRID = hand and wrist two-coil array coil, PACC = cylindric PAC, and PACF = flat PAC. Uniform distribution is seen along (a) the z axis and (b) the x axis. In a and b, the optimized birdcage coil data demonstrate improved SNR and uniformity compared with data for the other coils.
|
|
View larger version (26K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7a. SNR profiles of phantom images correspond to axial plane. In a and b, BC10 = optimized birdcage wrist coil, MRID = hand and wrist two-coil array coil, PACC = cylindric PAC, and PACF = flat PAC. Uniform distribution is seen along (a) the x axis and (b) the y axis. In a and b, the optimized birdcage coil data demonstrate improved SNR and uniformity compared with data for the other coils.
|
|
View larger version (26K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7b. SNR profiles of phantom images correspond to axial plane. In a and b, BC10 = optimized birdcage wrist coil, MRID = hand and wrist two-coil array coil, PACC = cylindric PAC, and PACF = flat PAC. Uniform distribution is seen along (a) the x axis and (b) the y axis. In a and b, the optimized birdcage coil data demonstrate improved SNR and uniformity compared with data for the other coils.
|
|
Spectral measurements were performed with the wrist and body fat phantoms, and the data are summarized in Table 3. With the additional exterior oil phantom, the full width at half maximum of the lipid spectra obtained with the optimized birdcage coil was four times narrower than that obtained with the flat PAC. The full width at half maximum with the optimized birdcage coil was 2.3 times smaller in the absence of the exterior oil phantom, which simulated body-wall fat. Signal intensity measurements demonstrated that the contrast-to-noise ratio for the optimized birdcage coil was 1.8 times higher than that for the cylindric PAC when fat suppression was applied. The fat to signal intensity ratio (with fat saturation versus without fat saturation) was 3.6% for the optimized birdcage coil and 6.2% for the cylindric PAC.
Statistical analysis of the blinded review was performed with the Wilcoxon signed-rank test. Differences with a P value of .05 were considered statistically significant. The null hypothesis was that image quality with the optimized birdcage coil was equal to or inferior to that with the cylindric PAC; the alternative hypothesis was that the image quality with the optimized birdcage coil was superior to that with the cylindric PAC. Results of the blinded review are shown in Table 4, which indicates the median difference in image grades between the optimized birdcage coil and cylindric PAC, which was determined by averaging values for five radiologists and five imaging series for each volunteer. Positive values of differences in image grades demonstrate the preference for the optimized birdcage coil. The P values of .0313 show statistical significance against the null hypothesis in the analysis of SNR, image uniformity, and overall image quality. The presence of some artifacts on T1-weighted images changed the P value for only artifact analysis to .375. However, this change did not affect the overall impression of the performance of the optimized birdcage coil. Table 5 indicates the frequency of overall preference for each coil and shows that the optimized birdcage coil was considered the same as (16%) or preferred more than (75%) the cylindric PAC in 91% of the interpretations (P = .0313).
|
Discussion
|
|---|
Uniformity measurements showed substantial improvement in homogeneity for the phantom images acquired with the optimized birdcage coil. These data demonstrate a uniformity improvement of 3.5 between the optimized birdcage coil and the flat PAC and a uniformity improvement of 2.0–4.5 between the optimized birdcage coil and the hand and wrist two-coil array. Signal intensity profiles along the z axis (Fig 6a) show uniformity levels along the coil length. Comparison of uniformity profiles demonstrated that the optimized birdcage coil had a 1.8 greater coverage than that with the hand and wrist two-coil array and almost the same length of coverage as that with the flat and cylindric PACs. Uniformity profiles along the x axis for the coronal phantom images and along the x and y axes for the transverse phantom images demonstrate higher uniformity on MR images obtained with the optimized birdcage coil than that obtained with the comparison coils. Data for the optimized birdcage coil and the hand and wrist two-coil array show SNR that is almost twice as high in the center of phantom images than that with the cylindric and flat PACs. Note, however, that the hand and wrist two-coil array was approximately two times smaller than the optimized birdcage coil. This difference is relevant because an inverse proportional dependence of the SNR on coil surface area is expected, as shown with analytic calculation of SNR with single-turn circular and square coils (14). Hence, the increased coil area of the optimized birdcage coil should cause decreased SNR (in theory, by a factor of 2). However, the data analysis showed a mean of 7% difference in SNR.
Spectral measurements indicated a narrow lipid line when the optimized birdcage coil was used, presumably as a result of the localized B1 field. This spectral line width is particularly important when the coil is placed at the patient’s side and when the signal from the body is in an inhomogeneous magnetic field well away from the magnet isocenter. Spectral line data show that the B1 field transmitted by the optimized birdcage coil during automatic preimaging is less affected by external lipid signal than are the other coils.
In vivo images clearly demonstrate that the optimized birdcage coil is superior to the cylindric PAC. Figure 4 shows an example in which fat saturation did not work as well when the cylindric PAC was used. In vivo MR images obtained with the optimized birdcage coil and the hand and wrist two-coil array demonstrated comparable SNR in the center; however, the signal intensity decrease at the edges with the hand and wrist two-coil array allowed slightly less longitudinal coverage (Fig 5).
These data support our hypothesis that a small birdcage coil receiver that is optimized for wrist imaging can provide higher SNR and better uniformity compared with those with a PAC with the same geometry. The optimized birdcage coil has the same SNR and longer z coverage than those with a smaller PAC. The narrow lipid spectral line indicates improved coil performance when fat saturation is included as part of the acquisition. Our data contradict current trends in the industry to make use of phased arrays for high-spatial-resolution MR imaging of the wrist. Our results show that a small optimized birdcage coil offers significant improvement over phased arrays when it is optimized for wrist imaging.
|
ACKNOWLEDGMENTS
|
|---|
The engineering assistance provided by Phillip Rossman and Thomas Hulshizer of the MR Laboratory is gratefully acknowledged.
|
FOOTNOTES
|
|---|
Abbreviations: PAC = phased-array coil,
SNR = signal-to-noise ratio
Author contributions: Guarantor of integrity of entire study, J.P.F.; study concepts and design, A.K., R.L.E., J.P.F.; literature research, A.K., J.P.F.; clinical studies, A.K., J.P.F., R.L.E., M.C.A., K.K.A., K.P.M., D.E.W.; experimental studies, A.K., J.P.F.; data acquisition, A.K., J.P.F.; data analysis/interpretation, all authors; statistical analysis, A.K., J.P.F.; manuscript preparation, A.K., J.P.F., R.L.E.; manuscript definition of intellectual content, M.C.A., K.K.A., K.P.M., P.A.R., D.E.W.; manuscript editing, A.K., J.P.F.; manuscript revision/review and final version approval, all authors.
|
REFERENCES
|
|---|
-
Pretorius ES, Epstein RE, Dalinka MK. MR imaging of the wrist. Radiol Clin North Am 1997; 35:145-161.[Medline]
-
Siegel S, White LM, Brahme S. Magnetic resonance imaging of the musculoskeletal system. V. The wrist. Clin Orthop 1996; 332:281-300.
-
Farooki S, Seeger LL. Magnetic resonance imaging in the evaluation of ligament injuries. Skeletal Radiol 1999; 28:61-74.[CrossRef][Medline]
-
Zeiss J, Jakab E, Khimji T, Imbriglia J. The ulnar tunnel at the wrist (Guyon’s canal): normal MR anatomy and variants. AJR Am J Roentgenol 1992; 158:1081-1085.[Abstract/Free Full Text]
-
Radack DM, Schweitzer ME, Taras J. Carpal tunnel syndrome: are the MR findings a result of population selection bias? AJR Am J Roentgenol 1997; 169:1649-1653.[Abstract/Free Full Text]
-
Mongale K, Dai G, Chu A, Burnham RS, Snyder RE. Quantitative MR imaging of carpal tunnel syndrome. AJR Am J Roentgenol 1999; 172:1581-1586.[Abstract/Free Full Text]
-
Anderson MW, Kaplan PA, Dussault RG, Degnan GG. Magnetic resonance imaging of the wrist. Curr Probl Diagn Radiol 1998; 27:187-229.[Medline]
-
Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased arrays. Magn Reson Med 1990; 16:192-225.[Medline]
-
Hayes CE, Hattes N, Roemer PB. Volume imaging with MR phased array. Magn Reson Med 1991; 18:309-319.[Medline]
-
Hayes CE, Edelstein WA, Schenck JF, Mueller OM, Eash M. An efficient, highly homogeneous radiofrequency coil for whole-body NMR imaging at 1.5 T. J Magn Reson 1985; 63:622-628.
-
Jin J. Electromagnetic analysis and design in magnetic resonance imaging Boca Raton, Fla: CRC, 1999; 138.
-
Wong EC, Jesmanowicz A, Hyde JS. High-resolution, short echo time MR imaging of the fingers and wrist with a local gradient coil. Radiology 1991; 181:393-397.[Abstract/Free Full Text]
-
National Electric Manufacturers Association. Characterization of special purpose coils for diagnostic magnetic resonance images. NEMA Standard (MS 6-1991; R 2000) .
-
Wang J, Reykowski A, Dickas J. Calculation of the signal-to-noise ratio for simple surface coils and arrays of coils. IEEE Trans Biomed Eng 1995; 42:908-917.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
R. M. Szabo
Distal Radioulnar Joint Instability
J. Bone Joint Surg. Am.,
April 1, 2006;
88(4):
884 - 894.
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
