Imaging of the wrist at 1.5 tesla using isotropic three-dimensional fast spin echo cube

Imaging of the Wrist at 1.5 Tesla Using Isotropic Three-Dimensional Fast Spin Echo Cube

Kathryn J. Stevens, MD1,*, Charles G. Wallace, MD1, Weitian Chen, PhD2, Jarrett K.Rosenberg, PhD1, and Garry E. Gold, MD1

1Department of Radiology, Stanford University, Stanford, California, USA.2GE Healthcare Applied Sciences Laboratory, Menlo Park, California, USA.

AbstractPurpose—To compare three-dimensional fast spin echo Cube (3D-FSE-Cube) with conventional2D-FSE in MR imaging of the wrist.

Materials and Methods—The wrists of 10 volunteers were imaged in a 1.5 Tesla MRI scannerusing an eight-channel wrist coil. The 3D-FSE-Cube images were acquired in the coronal planewith 0.5-mm isotropic resolution. The 2D-FSE images were acquired in both coronal and axialplanes for comparison. An ROI was placed in fluid, cartilage, and muscle for SNR analysis.Comparable coronal and axial images were selected for each sequence, and paired images wererandomized and graded for blurring, artifact, anatomic details, and overall image quality by threeblinded musculoskeletal radiologists.

Results—SNR of fluid, cartilage and muscle at prescribed locations were higher using 3D-FSE-Cube, without reaching statistical significance. Fluid–cartilage CNR was also higher with 3D-FSE-Cube, but not statistically significant. Blurring, artifact, anatomic details, and overall imagequality were significantly better on coronal 3D-FSE-Cube images (P < 0.001), but significantlybetter on axial 2D-FSE images compared with axial 3D-FSE-Cube reformats (P < 0.01).

Conclusion—Isotropic data from 3D-FSE-Cube allows reformations in arbitrary scan planes,which may make multiple 2D acquisitions unnecessary, and improve depiction of complex wristanatomy. However, axial reformations suffer from blurring, likely due to T2 decay during the longecho train, limiting overall image quality in this plane.

KeywordsMRI; wrist; 3D-FSE-Cube; isotropic

MRI OF THE wrist can be challenging due to the small size and complex anatomy ofstructures such as the intrinsic and extrinsic intercarpal ligaments, and triangularfibrocartilage complex (TFCC) (1–4). Technological advances in MR scanner technologyand MRI pulse sequence design, together with the introduction of high field strengthmagnets and dedicated surface coils, have lead to improved visualization of the intricateanatomic structures in the wrist, thereby increasing diagnostic capabilities (3,5–11).However, the spatial resolution provided by standard two-dimensional (2D) fast spin echo(FSE) sequences is still a significant limitation in the evaluation of complex structures suchas the individual components of the TFCC, intercarpal ligaments and articular cartilage.

© 2011 Wiley-Liss, Inc.*Address reprint requests to: K.J.S., Department of Radiology, Stanford University School of Medicine, Room S-062A GrantBuilding, 300 Pasteur Drive, Stanford, CA 94305-5105. [email protected].

NIH Public AccessAuthor ManuscriptJ Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

Published in final edited form as:J Magn Reson Imaging. 2011 April ; 33(4): 908–915. doi:10.1002/jmri.22494.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Many of the ligaments and tendons in the wrist are obliquely oriented with respect to thestandard orthogonal planes acquired in a 2D MRI study of the wrist, and are, therefore,prone to artifacts such as magic angle and partial volume effects, which may furtherdecrease diagnostic accuracy. The articular cartilage in the wrist is relatively thin, and thearticular surfaces are curved and complex, which limits the diagnostic accuracy of MRI inthe evaluation of cartilage pathology in the radiocarpal and intercarpal joints (12–15). Theability to reformat the articular surface in any plane may, therefore, help to improvevisualization of cartilage defects.

Standard MRI protocols usually comprise multiple 2D-FSE acquisitions acquired inorthogonal planes. More recently, isotropic three-dimensional (3D) FSE pulse sequenceshave been developed (16,17). Data only need to be acquired once in a single scan plane, andcan subsequently be reconstructed into the standard orthogonal planes or any arbitrary scanplane to optimize visualization of complex anatomical structures. Fast spin echo contrast inthese sequences may provide a diagnostic advantage compared with gradient echo contrast,which also has the capability of isotropic 3D acquisition. The utility of isotropic 3D MRIsequences has so far been explored in the knee (18–20), ankle (21,22), and shoulder (23,24).The aim of our study was to compare 3D-FSE-Cube with conventional 2D-FSE sequences inthe evaluation of the wrist in normal volunteers.

MATERIALS AND METHODSAfter institutional review board approval, informed consent, and following HIPAA privacyguidelines, MR imaging was performed on the wrists of 10 healthy volunteers aged between27 and 43 years of age (mean, 31 years). Images were acquired on a GE Signa 1.5 Tesla (T)MRI scanner (GE Healthcare, Milwaukee, WI) using an eight-channel wrist coil (InvivoInc., Gainesville, FL) and an experimental version of 3D-Cube. Patients were imaged in theprone position, with the arm extended over the head, and the wrist in the isocenter of themagnet. The 3D-FSE-Cube images were acquired in the coronal imaging plane both withand without fat saturation using repetition time/echo time (TR/TE) 3000/35 ms, 256 × 256matrix, 12-cm field of view (FOV), 0.5-mm sections, echo train length (ETL) 60, andreceiver bandwidth ±83 kHz.

The 3D-FSE-Cube is a multiecho acquisition, and the equivalent TE of 35 ms is based uponthe echo position at the center of k-space. The use of auto-calibrated parallel imaging and anacceleration factor of 2.8 enabled the acquisition of 132 slices in only 4 min 28 s. ARC(Autocalibrating Reconstruction for Cartesian sampling) parallel imaging reconstruction(25) was performed online using host-based prototype software.

Conventional 2D-FSE images were acquired for comparison in the coronal plane both withand without fat saturation, and subsequently in the axial plane using TR/TE 3000/35 ms, 256× 256 matrix, 12 cm FOV, 2-mm slices and 0.5-mm gap, ETL 8, receiver bandwidth ±83kHz. acquisition time 3 min 18 s for 22 slices.

Both 2D-FSE and 3D-FSE-Cube were also acquired with the radiofrequency (RF) pulseturned off to allow measurement of noise.

Image EvaluationFor each method, a region of interest (ROI) was placed in muscle in the thenar eminence, influid in the triquetral–pisiform recess, and in cartilage in the radiocarpal joint. The regionsof interest measured approximately 0.5 cm2 for muscle, 0.03 cm2 for fluid, and 0.05 cm2 forcartilage respectively. Measurements were performed by a single radiologist (K.J.S.) with 13years of experience in interpretation of musculoskeletal MRI.

Stevens et al. Page 2

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Parallel imaging and image combination processes in phased array can alter the spatialdistribution of noise statistics. To address this problem, for each imaging data set, werepeated data acquisition without RF excitation to collect a noise-only data set. The noise-only data set is passed through a linear reconstruction pipeline to obtain the actual spatialdistribution of noise (26). Such a reconstruction pipeline consists of identical reconstructionsteps with respect to those used for reconstruction of anatomy images, but with modificationof reconstruction algorithms such that each step is effective in processing a noise-only dataset. To calculate the standard deviation of the noise, the same ROI used to measure thesignal in muscle, fluid and cartilage was placed on the corresponding noise image. Thesignal-to-noise ratio for each was calculated as the mean of the signal in each structuredivided by the standard deviation of the noise in the same location (Eq. [1]).

[1]

The fluid–cartilage contrast-to-noise ratio (CNR) was calculated by subtracting cartilageSNR from the fluid SNR (Eq. [2]).

[2]

A paired t-test was then used to compare both SNR and CNR (Microsoft Excel 2008).

Axial reformats of the isotropic 3D-FSE-Cube images were performed using Osirix(www.osirix.com) and compared with 2D-FSE images acquired in the axial plane. Imagesfrom 3D-FSE-Cube were also reformatted in multiple arbitrary imaging planes to bestdemonstrate the intrinsic and extrinsic intercarpal ligaments, triangular fibrocartilagecomplex, and tendons.

A central coronal slice through the carpal bones was selected on the 2D-FSE images foreach patient, and compared with a comparable slice from the 3D-FSE-Cube acquisition.Similarly, an axial 2D-FSE image in each patient was compared with an equivalent axialreformatted 3D-FSE-Cube image. The paired images were then randomized andindependently assessed for blurring, artifacts such as chemical shift and pulsation artifact,visualization of anatomic structures and overall image quality by 3 blinded musculoskeletalradiologists (K.J.S. 13 years, G.E.G. 13 years, C.G.W. 9 years of musculoskeletal radiology,respectively), who graded the images on a 5 point scale from –2 to +2: –2 = image A muchbetter than B, –1 = image A somewhat better than B, 0 = no difference between images, +1= image B somewhat better than A. and +2 = image B much better than image A.

Statistical AnalysisStatistical analysis of SNR and CNR measurements was performed using Excel 11.1.1(Microsoft, Redmond, WA). The 3D-FSE-Cube and 2D-FSE were both compared withrespect to SNR and cartilage-fluid CNR using a paired sample t-test. A P-value of <0.05 wasconsidered significant.

The ratings of the 3 readers for blurring, artifact, visualization of anatomical structures, andoverall image quality in the paired coronal and axial MR images in each of the 10 patientswere then compared and analyzed using a two-tailed paired Wilcoxon test. Inter-observeragreement was assessed by linear-weighted kappa. All statistical analyses were performedwith Stata release 9.2 (Stata Corp., College Station, TX).

Stevens et al. Page 3

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

RESULTSThe SNR of fluid, cartilage, and muscle measured in the prescribed locations (Fig. 1) werehigher using 3D-FSE-Cube, but this did not reach statistical significance (P = 0.05, P = 0.09,P = 0.53, respectively). Fluid–cartilage CNR was also higher with 3D-FSE-Cube (P = 0.77),manifesting as increased conspicuity of joint fluid on 3D-FSE-Cube images (Fig. 2). The3D-FSE-Cube images had much thinner slices, allowing improved visualization of smallanatomic structures such as the articular cartilage, intercarpal ligaments, and triangularfibrocartilage complex.

Both coronal 2D-FSE and 3D-FSE-Cube were acquired with and without fat suppression.Fat suppression was more uniform on the coronal 3D-FSE-Cube images. In 5 of the 10patients, there was much poorer fat saturation in the region of the ulnar styloid on 2D-FSEimages (Fig. 3).

The 2D-FSE and 3D-FSE-Cube images were compared for blurring, artifact, visualization ofanatomical structures, and overall image quality by three readers (Table 1). There wassignificantly less blurring on the 3D-FSE-Cube images in the coronal plane compared withthe 2D-FSE images (P < 0.0001). However, in the axial plane, blurring was significantlygreater on the 3D-FSE-Cube images (P < 0.0063) (Fig. 4).

Significantly less artifacts were seen on 3D-FSE-Cube images on both the coronal (P <0.0001) and axial (P < 0.0001) images, when compared with 2D-FSE.

Visualization of anatomical structures was significantly better on coronal 3D-FSE images (P< 0.0001). However, anatomical structures were significantly better visualized on the 2D-FSE images in the axial imaging plane (P < 0.0001).

When the two sequences were compared for overall image quality, the 3D-FSE-Cubeperformed significantly better than 2D-FSE in the coronal imaging plane (P < 0.0003).However, the situation reversed in the axial plane, with 2D-FSE performing significantlybetter than 3D-FSE-Cube (P < 0.0001).

Interobserver agreement was good for artifact ratings, with a mean kappa score of 0.783, butwas low for blurring, visualization of anatomical structures, and overall image quality(Table 2).

A significant advantage of 3D-FSE-Cube was the ability to reformat images in arbitraryplanes. Reformations of the 3D-FSE-Cube images were similar to the directly acquired 2D-FSE data, except that the 3D-FSE-Cube had much thinner slices. Images from 3D-FSE-Cube were reformatted in multiple oblique imaging planes to optimally depict the intrinsic(Figs. 5, 6), and extrinsic (Fig. 7) ligaments of the wrist, the triangular fibrocartilagecomplex (Figs. 6, 8), the oblique courses of some of the tendons around the wrist, andarticular cartilage (Fig. 9). Compared with images acquired directly in the reformation planewith 2D-FSE, the reformatted 3D-FSE-Cube images were similar in quality and depiction ofanatomy, only with much thinner slices. However, in some cases the 3D-FSE-Cube imagesalso appeared noisier than the 2D-FSE images, despite the higher SNR measured in specificanatomic structures, and this may reflect the spatially varying noise across the imagesintroduced by parallel imaging (Fig. 2).

DISCUSSIONMRI, with its exceptional soft tissue contrast and multi-planar imaging capabilities, isideally suited for evaluation of patients with wrist pain. MRI has a high sensitivity and

Stevens et al. Page 4

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

specificity in the diagnosis of both TFCC tears and ligamentous pathology, particularly ifhigh resolution coils (3,8,27), high field strength magnets (5,6,8–10,14,28) and MRarthrography (15,29–33) are used. Conventional MRI studies of the wrist usually includemultiple 2D-FSE sequences acquired in orthogonal scan planes. These sequences have beenshown to be relatively sensitive in the diagnosis of TFCC tears and ligamentous injury(5,10). However, 2D-FSE sequences are usually acquired with relatively thick slices and aninterslice gap, which may limit visualization of small anatomic structures such as intercarpalligaments and components of the triangular fibrocartilage complex, or may potentiallyobscure subtle areas of pathology such as fissures within the articular cartilage. Many of theligaments and tendons in the wrist are obliquely oriented with respect to the standardorthogonal planes acquired in a normal MRI study of the wrist, and are, therefore, also proneto artifacts such as magic angle, which may decrease diagnostic accuracy. In our study, theacquisition of isotropic data in 3D-FSE-Cube allowed us to reformat images in any desiredimaging plane, making it ideal for the interrogation of obliquely oriented anatomic structuresaround the wrist such as tendons or curved articular surfaces. Another advantage of 3D-FSE-Cube is the ability to average slices, so that both source and reformatted images can bedisplayed at any multiple of the 0.5-mm slice thickness. Images can, therefore, be directlycomparable to conventional 2D-FSE sequences if desired, although this also increasespartial volume effects.

A previous study comparing 2D-FSE with 3D-FSE in the knee concluded that the diagnosticperformance was similar between 2D-FSE and 3D-FSE sequences (20) However, imagequality was purportedly lower in the 3D-FSE images, largely due to the blurring seen onreformatted images, and both the SNR and CNR measurements were higher in the 2D-FSEsequence. In our study, the SNR and CNR measurements were all higher in the 3D-FSEsequence, although the differences were not statistically signifi-cant. We found that overallimage quality was significantly higher in 3D-FSE-Cube images in the coronal plane.However, in the axial plane, the image quality was significantly lower in the 3D-FSEimages, largely due to blurring from the long echo train readout on the axial 3D-FSEreformations.

Isotropic high-resolution 3D datasets contain a large number of voxels, and this combinedwith the relatively long repetition times (TR's) required for proton density and T2-weightedFSE sequences, would create extremely long scan times. Different techniques have beenused to decrease scan times, including modulating the flip angles of the refocusingradiofrequency (RF) train, which allows very long readout trains to be acquired with onlyminimal blurring (16,17,34). A combination of parallel imaging in both the phase encodingand slice encoding direction (35,36) and partial-Fourier acquisition (37) also greatlydecreases the number of phase encodes required to encode a large 3D data set. An earlyprototype combining flip angle modulation (34) and 1D-accelerated autocalibrating parallelimaging (36,38) was used to image the knee (16). More recently 3D-FSE-Cube (previouslycalled 3D-FSE-XETA) has been developed which combines a more recent flip anglealgorithm (39) with 2D-accelerated autocalibrating parallel imaging (16,40). The improvedacquisition efficiency of this sequence allowed us to acquire isotropic 3D-FSE-Cube of thewrist in only 4 min 28 s. Images were acquired with the subject in the prone position, andthe arm extended over the head, placing the wrist in the isocenter of the magnet. Analternative position is to scan patients with the arm at the side, but this places the wrist at theperiphery of the magnetic bore, making it prone to inhomogeneous fat saturation. All of oursubjects were healthy volunteers, who had no difficulties maintaining this position for theduration of the scan. However, many older patients, or patients in pain have difficulty lyingin this position for a long period of time, making the scan prone to motion artifact.Therefore, an imaging protocol with fewer sequences, ultimately leading to a shorter overall

Stevens et al. Page 5

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

scan time, would be advantageous in patients unable to tolerate the prone extended positionfor very long.

Limitations of our study include the small number of subjects. Only healthy volunteers wereincluded in the study, and all were asymptomatic. However, our aim was to optimize thestudy parameters, and demonstrate that the 3D-FSE-Cube images were comparable toconventional 2D-FSE sequences. A larger study is currently underway to compare 3D-FSE-Cube with 2D-FSE at 3T magnet in a symptomatic population. We also compared 0.5-mm-thick slices from the 3D-FSE sequence with 2-mm-thick slices from the 2D-FSE sequence.However, we wanted to show that depiction of anatomic detail was higher on the highresolution 0.5 mm 3D-FSE images.

In conclusion 3D-FSE-Cube is a promising high-resolution MR imaging technique whichimproves depiction of complex wrist anatomy, and may obviate the need to acquire multiplesequences in orthogonal scan planes, which in turn may lead to a decrease in overall scantime.

AcknowledgmentsThe authors acknowledge Reed F. Busse, PhD, Anja C.S. Brau, PhD, and Philip J. Beatty, PhD for providing theprototype version of 3D-FSE-Cube, and Charles Li, BS, for his assistance in calculation of noise measurements.

REFERENCES1. Sasao S, Beppu M, Kihara H, Hirata K, Takagi M. An anatomical study of the ligaments of the ulnar

compartment of the wrist. Hand Surg. 2003; 8:219–226. [PubMed: 15002101]2. Theumann NH, Pfirrmann CW, Antonio GE, et al. Extrinsic carpal ligaments: normal MR

arthrographic appearance in cadavers. Radiology. 2003; 226:171–179. [PubMed: 12511687]3. Yoshioka H, Ueno T, Tanaka T, Shindo M, Itai Y. High-resolution MR imaging of triangular

fibrocartilage complex (TFCC): comparison of microscopy coils and a conventional small surfacecoil. Skeletal Radiol. 2003; 32:575–581. [PubMed: 12942205]

4. Zlatkin MB, Rosner J. MR imaging of ligaments and triangular fibrocartilage complex of the wrist.Magn Reson Imaging Clin N Am. 2004; 12:301–331. vi–vii. [PubMed: 15172388]

5. Anderson ML, Skinner JA, Felmlee JP, Berger RA, Amrami KK. Diagnostic comparison of 1.5Tesla and 3.0 Tesla preoperative MRI of the wrist in patients with ulnar-sided wrist pain. J HandSurg [Am]. 2008; 33:1153–1159.

6. Ashman CJ, Farooki S, Abduljalil AM, Chakeres DW. In vivo high resolution coronal MRI of thewrist at 8.0 tesla. J Comput Assist Tomogr. 2002; 26:387–391. [PubMed: 12016368]

7. Behr B, Stadler J, Michaely HJ, Damert HG, Schneider W. MR imaging of the human hand andwrist at 7 T. Skeletal Radiol. 2009; 38:911–917. [PubMed: 19277647]

8. Bittersohl B, Huang T, Schneider E, et al. High-resolution MRI of the triangular fibrocartilagecomplex (TFCC) at 3T: comparison of surface coil and volume coil. J Magn Reson Imaging. 2007;26:701–707. [PubMed: 17729361]

9. Lenk S, Ludescher B, Martirosan P, Schick F, Claussen CD, Schlemmer HP. 3.0 T high-resolutionMR imaging of carpal ligaments and TFCC. Rofo. 2004; 176:664–667. [PubMed: 15122464]

10. Magee T. Comparison of 3-T MRI and arthroscopy of intrinsic wrist ligament and TFCC tears.AJR Am J Roentgenol. 2009; 192:80–85. [PubMed: 19098183]

11. Saupe N. 3-Tesla high-resolution MR imaging of the wrist. Semin Musculoskelet Radiol. 2009;13:29–38. [PubMed: 19235670]

12. Haims AH, Moore AE, Schweitzer ME, et al. MRI in the diagnosis of cartilage injury in the wrist.AJR Am J Roentgenol. 2004; 182:1267–1270. [PubMed: 15100130]

13. Mutimer J, Green J, Field J. Comparison of MRI and wrist arthroscopy for assessment of wristcartilage. J Hand Surg Eur. 2008; 33:380–382.

Stevens et al. Page 6

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

14. Saupe N, Pfirrmann CW, Schmid MR, Schertler T, Manestar M, Weishaupt D. MR imaging ofcartilage in cadaveric wrists: comparison between imaging at 1.5 and 3.0 T and gross pathologicinspection. Radiology. 2007; 243:180–187. [PubMed: 17312277]

15. Moser T, Dosch JC, Moussaoui A, Dietemann JL. Wrist ligament tears: evaluation of MRI andcombined MDCT and MR arthrography. AJR Am J Roentgenol. 2007; 188:1278–1286. [PubMed:17449771]

16. Busse, RF.; Brau, ACS.; Beatty, P., et al. Proceedings of the 15th Annual Meeting of ISMRM.Berlin: 2007. Design of refocusing flip angle modulation for volumetric 3D-FSE imaging of brain,spine, knee, kidney and uterus.. (abstract 1702)

17. Mugler JP III, Bao S, Mulkern RV, et al. Optimized single-slab three-dimensional spin-echo MRimaging of the brain. Radiology. 2000; 216:891–899. [PubMed: 10966728]

18. Gold GE, Busse RF, Beehler C, et al. Isotropic MRI of the knee with 3D fast spin-echo extendedecho-train acquisition (XETA): initial experience. AJR Am J Roentgenol. 2007; 188:1287–1293.[PubMed: 17449772]

19. Kijowski R, Blankenbaker DG, Klaers JL, Shinki K, De Smet AA, Block WF. Vastlyundersampled isotropic projection steady-state free precession imaging of the knee: diagnosticperformance compared with conventional MR. Radiology. 2009; 251:185–194. [PubMed:19221057]

20. Ristow O, Steinbach L, Sabo G, et al. Isotropic 3D fast spin-echo imaging versus standard 2Dimaging at 3.0 T of the knee--image quality and diagnostic performance. Eur Radiol. 2009;19:1263–1272. [PubMed: 19137309]

21. Bauer JS, Banerjee S, Henning TD, Krug R, Majumdar S, Link TM. Fast high-spatial-resolutionMRI of the ankle with parallel imaging using GRAPPA at 3 T. AJR Am J Roentgenol. 2007;189:240–245. [PubMed: 17579177]

22. Stevens KJ, Busse RF, Han E, et al. Ankle: isotropic MR imaging with 3D-FSE-cube--initialexperience in healthy volunteers. Radiology. 2008; 249:1026–1033. [PubMed: 19011194]

23. Jung JY, Yoon YC, Choi SH, Kwon JW, Yoo J, Choe BK. Three-dimensional isotropic shoulderMR arthrography: comparison with two-dimensional MR arthrography for the diagnosis of labrallesions at 3.0 T. Radiology. 2009; 250:498–505. [PubMed: 19188318]

24. Oh DK, Yoon YC, Kwon JW, et al. Comparison of indirect iso-tropic MR arthrography andconventional MR arthrography of labral lesions and rotator cuff tears: a prospective study. AJRAm J Roentgenol. 2009; 192:473–479. [PubMed: 19155413]

25. Brau AC, Beatty PJ, Skare S, Bammer R. Comparison of reconstruction accuracy and efficiencyamong autocalibrating data-driven parallel imaging methods. Magn Reson Med. 2008; 59:382–395. [PubMed: 18228603]

26. Li, CQ.; Chen, W.; Beatty, PJ., et al. Proceedings of the 18th Annual Meeting of ISMRM.Stockholm: 2010. SNR quantification with phased-array coils and parallel imaging for 3D-FSE..(abstract 552)

27. Tanaka T, Yoshioka H, Ueno T, Shindo M, Ochiai N. Comparison between high-resolution MRIwith a microscopy coil and arthroscopy in triangular fibrocartilage complex injury. J Hand Surg[Am]. 2006; 31:1308–1314.

28. Saupe N, Prussmann KP, Luechinger R, Bosiger P, Marincek B, Weishaupt D. MR imaging of thewrist: comparison between 1.5- and 3-T MR imaging--preliminary experience. Radiology. 2005;234:256–264. [PubMed: 15550374]

29. Berna-Serna JD, Martinez F, Reus M, Alonso J, Domenech G, Campos M. Evaluation of thetriangular fibrocartilage in cadaveric wrists by means of arthrography, magnetic resonance (MR)imaging, and MR arthrography. Acta Radiol. 2007; 48:96–103. [PubMed: 17325933]

30. Ruegger C, Schmid MR, Pfirrmann CW, Nagy L, Gilula LA, Zanetti M. Peripheral tear of thetriangular fibrocartilage: depiction with MR arthrography of the distal radioulnar joint. AJR Am JRoentgenol. 2007; 188:187–192. [PubMed: 17179363]

31. Zanetti M, Bram J, Hodler J. Triangular fibrocartilage and inter-carpal ligaments of the wrist: doesMR arthrography improve standard MRI? J Magn Reson Imaging. 1997; 7:590–594. [PubMed:9170047]

Stevens et al. Page 7

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

32. Maizlin ZV, Brown JA, Clement JJ, et al. MR Arthrography of the wrist: controversies andconcepts. Hand (N Y). 2009; 4:66–73. [PubMed: 19048349]

33. Scheck RJ, Romagnolo A, Hierner R, Pfluger T, Wilhelm K, Hahn K. The carpal ligaments in MRarthrography of the wrist: correlation with standard MRI and wrist arthroscopy. J Magn ResonImaging. 1999; 9:468–474. [PubMed: 10194719]

34. Busse RF, Hariharan H, Vu A, Brittain JH. Fast spin echo sequences with very long echo trains:design of variable refocusing flip angle schedules and generation of clinical T2 contrast. MagnReson Med. 2006; 55:1030–1037. [PubMed: 16598719]

35. Wang Z, Fernandez-Seara MA. 2D partially parallel imaging with k-space surrounding neighbors-based data reconstruction. Magn Reson Med. 2006; 56:1389–1396. [PubMed: 17063471]

36. Weiger M, Pruessmann KP, Boesiger P. 2D sense for faster 3D MRI. MAGMA. 2002; 14:10–19.[PubMed: 11796248]

37. Knoll DC, Nishimura DG, Macovski A. Homodyne detection in magnetic resonance imaging.IEEE Trans Med Imaging. 1991; 10:154–163. [PubMed: 18222812]

38. Brau ACS, Beatty PJ, Skare S, Bammer R. Comparison of computational accuracy and efficiencyin autocalibrating parallel imaging reconstructions. Magn Reson Med. 2008; 59:382–395.[PubMed: 18228603]

39. Busse RF, Brau AC, Vu A, et al. Effects of refocusing flip angle modulation and view ordering in3D fast spin echo. Magn Reson Med. 2008; 60:640–649. [PubMed: 18727082]

40. Beatty, P.; Brau, ACS.; Chang, S., et al. Proceedings of the 15th Annual meeting of ISMRM.Berlin: 2007. A method for autocalibrating 2D-accelerated volumetric parallel imaging withclinically practical reconstruction times.. (abstract 1749)

Stevens et al. Page 8

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 1.SNR comparison between 3D-FSE-Cube and 2D-FSE showing no significant difference forSNR values of fluid, cartilage and muscle

Stevens et al. Page 9

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 2.Coronal 3D-FSE-Cube (a) and 2D-FSE (b) images through the wrist demonstratingincreased contrast between the fluid and adjacent articular cartilage and carpal bones on the3D-FSE-Cube (arrows). Note the partial volume artifact on the 2D-FSE image over thedistal radius (open arrow).

Stevens et al. Page 10

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 3.Coronal 3D-FSE-Cube (a) and 2D-FSE (b) images with fat saturation demonstratenonuniform fat saturation over the ulnar styloids in both sequences (open arrows). Fatsaturation is less uniform on the 2D-FSE image, and in addition there is pulsation artifactadjacent to the radial artery (arrows).

Stevens et al. Page 11

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 4.Axial 3D-FSE-Cube (a) and 2D-FSE (b) images through the carpal tunnel. Increasedblurring is seen in the 3D-FSE-Cube images around the tendons and capsular ligaments.

Stevens et al. Page 12

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 5.Coronal 3D-FSE-Cube images through the central (a), and volar (b) and aspects of the wristdemonstrating the scapholunate (arrowhead) and lunotriqetral (small arrows) ligaments. InFigure 5a, the triangular fibrocartilage can be seen (open arrow), with the triangularligament insertions into the ulnar styloid and fovea (*). In Figure 5b, the ulnotriquetralligament can be seen arising from the volar aspect of the triangular fibrocartilage to insertinto the triquetrum (double line arrow).

Stevens et al. Page 13

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 6.a,b: Coronal 3D-FSE-Cube images through the volar (a) and dorsal (b) capsular ligaments.a: On the volar aspect, the deltoid (arrows) and scapholunate-triquetral ligaments (*) arevisualized. b: On the dorsal wrist, the dorsal radiocarpal ligament and dorsal intercarpalligament (double line arrows) are seen. The dorsal fibers of the scapholunate ligament canalso be visualized (*).

Stevens et al. Page 14

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 7.a–c: Sagittal 3D-FSE-Cube images through the triangular fibrocartilage (arrow). a: Theradial border is thinner than the more substantial ulnar portion. b: The ulnotriquetralligament is partially visualized (double line arrow). c: However, a sagittal oblique imagedemonstrates the entire ulnotriquetral ligament (double line arrow).

Stevens et al. Page 15

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 8.Sagittal oblique 3D-FSE-Cube image demonstrating the insertion of the flexor carpi radialis(arrow) and extensor carpi radialis brevis tendon (double line arrow) into the secondmetacarpal base.

Stevens et al. Page 16

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 9.Coronal (a) and sagittal (b) fat-suppressed 3D-FSE-Cube images showing details of thearticular cartilage within the radiocarpal, intercarpal, and carpometacarpal joints.

Stevens et al. Page 17

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Stevens et al. Page 18

Tabl

e 1

Rat

ing

of B

lurr

ing,

Arti

fact

, Vis

ualiz

atio

n of

Ana

tom

ical

Stru

ctur

es, a

nd O

vera

ll Im

age

Qua

lity

on C

oron

al a

nd A

xial

MR

Imag

es, C

ompa

ring

3D-F

SE-

Cub

e W

ith 2

D-F

SE Im

ages

for t

he T

hree

Rea

ders

(< =

Les

s Goo

d, >

= B

ette

r Tha

n)

Rat

ing

Para

met

erIm

agin

g pl

ane

3D>2

D3D

>2D

3D=2

D3D

<2D

3D<2

DT

otal

Ass

essm

ent a

nd si

gnifi

canc

e

Blu

rrin

gC

oron

al6

148

20

303D

-FSE

-Cub

e <

2D-F

SE, P

<0.0

001

Axi

al2

110

170

303D

-FSE

-Cub

e >

2D-F

SE, P

<0.0

063

Arti

fact

Cor

onal

1313

40

030

3D-F

SE-C

ube

< 2D

-FSE

, P<0

.000

1

Axi

al3

189

00

303D

-FSE

-Cub

e <

2D-F

SE, P

<0.0

001

Vis

ualiz

atio

n of

ana

tom

yC

oron

al12

107

10

303D

-FSE

-Cub

e >

2D-F

SE, P

<0.0

001

Axi

al0

010

200

303D

-FSE

-Cub

e <

2D-F

SE, P

<0.0

001

Ove

rall

imag

e qu

ality

Cor

onal

614

73

030

3D-F

SE-C

ube

> 2D

-FSE

, P<0

.000

3

Axi

al0

26

220

303D

-FSE

-Cub

e <

2D-F

SE, P

<0.0

001

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Stevens et al. Page 19

Table 2

Inter-observer Agreement for Grading Parameters

Parameter Imaging plane Kappa P 95% CIa

Blurring Coronal 0.4 0.0004 0.015 – 0.693

Axial 0.763 0.0001 0.426 – 0.999

Overall 0.634 0.0001 0.465 – 0.860

Artifact Coronal 0.78 0.0001 0.464 – 0.999

Axial 0.753 0.0001 0.316 – 0.999

Overall 0.783 0.0001 0.550 – 0.949

Visualization of Anatomic structures Coronal 0.307 0.0063 0.032 – 0.643

Axial 0.25 0.0855 0.118 – 0.814

Overall 0.452 0.0001 0.284 – 0.690

Overall image quality Coronal 0.459 0.0001 0.034 – 0.868

Axial 0.362 0.0075 0.135 – 0.639

Overall 0.546 0.0001 0.370 – 0.762

aBased on bias-corrected bootstrap, 1000 replications.

J Magn Reson Imaging. Author manuscript; available in PMC 2012 January 31.