Visualization of wrist anatomy—a comparison between 7T and 3T MRI

The anatomy of the wrist, including the radiocarpal joint, the distal radioulnar joint, the intracarpal joints, and interconnecting ligaments, is complex. In addition, a large number of flexor and extensor tendons and three major nerves are located around the wrist. All wrist structures are prone to injury. The small and intricate anatomy of these structures makes them difficult to assess using MR [1]. Thus, wrist arthroscopy is still considered the reference standard for assessing intra-articular cartilage and ligament injuries in the radiocarpal and intercarpal joints [1‐3]. Improvement of a non-invasive method, such as MRI, in the evaluation of these structures could be valuable, as wrist arthroscopy is not only invasive but also expensive and comes with the risk of complications [4]. Additionally, some structures such as nerves and tendons are not possible to assess using an arthroscopic technique. Improved visualization of ligaments, nerves, tendons, and trabecular bone would, therefore, be of clinical value [5‐8].

Previous studies have reported 3T systems to be superior to 1.5T at depicting ligament structures of the wrist [3‐9], especially when using the improved signal-to-noise ratio (SNR) to increase spatial resolution [10]. 3T imaging is now widely used in clinical musculoskeletal practice. As ultra-high field (UHF) MR systems, 4T and above, have become increasingly available for research and clinical purposes, their potential for improving clinical diagnostics has increased. A higher field strength provides a higher intrinsic SNR which brings potentially higher resolution and potentially improved tissue contrast compared to lower field strengths [11]. Initial experiences in anatomical imaging of the wrist at 7T are promising [10, 12], but the number of published studies in this field is very limited [5‐7, 10, 12‐18]. In addition, the lack of commercially available, dedicated radio frequency (RF) wrist coils for 7T has earlier been a technical challenge for imaging at 7T [10]. No study has been published on the use of a commercially available wrist coil specifically developed for wrist imaging at 7T. The benefit in SNR at 7T compared to 3T has been established, but improved SNR does not necessarily improve the visibility of anatomic structures [15].

The purpose of this study was to investigate whether MR at 7T, using a dedicated, commercially available wrist coil, improves anatomical visualization of clinically important structures in the wrist such as ligaments, cartilage, nerves, tendons, and trabecular bone compared to imaging at 3T.

All evaluated anatomical structures, including ligaments, trabecular bone, cartilage, nerves, and tendons, were graded as better visualized at 7T compared to 3T (Table 3, Fig. 1). In evaluating image quality, 7T was significantly superior to 3T in the evaluation of edge sharpness and perceived tissue contrast. There was no significant difference in grading regarding artefacts. Examples of the superiority of 7T over 3T regarding anatomical visibility are given for the TFCC and the SLL in Fig. 2, for articular cartilage and the median and ulnar nerves in Fig. 3, and for the ECU tendon and trabecular bone in Fig. 4.

Regarding anatomical visibility, the difference in grading between 7T and 3T was most pronounced for the ulnar nerve (using 2D sequences), where 72% of the cases were graded as a 4 or a 5 (visualization of most anatomical detail or perfect visualization of anatomical detail) for 7T compared to 6% for 3T (Supplementary Table 3). Similarly, at 7T and 3T respectively, grade 4 or 5 was given in 99% versus 56% of cases for trabecular bone, in 79% versus 39% of cases for cartilage (using 2D sequences), in 75% versus 42% of cases for the foveal attachment of the TFCC (using 3D sequences), in 44% versus 19% of cases for the dorsal portion of the LTL (using 3D sequences), and in 69% versus 49% of cases for the dorsal portion of the SLL (using 3D sequences). Regarding image quality, the difference in grading between 7T and 3T was most pronounced for edge sharpness, where 90% of the cases were graded as a 4 or a 5 for 7T compared to 49% for 3T. Regarding anatomical visualization of ligaments, 3D sequences performed better than 2D sequences at both 7T and 3T, receiving a grade of 4 or 5 in a larger proportion of observations. Conversely, 2D sequences were superior to 3D sequences in the visualization of cartilage, nerves, and tendons at both field strengths. For the distribution of all grades given by the four observers, please see Supplementary Table 3.

The current study showed 7T to be superior to 3T in the visualization of anatomical structures of the wrist, as all structures were graded significantly higher for 7T compared to 3T in the VGC analysis. Edge sharpness and perceived tissue contrast were also graded higher at 7T. The proportion of cases with visualization of most anatomical detail, or with perfect visualization of anatomical detail (grades 4 and 5), was higher for all structures at 7T.

Ashman et al in 2002 compared UHF imaging of the wrist (at 8T) with imaging at 1.5T, reporting improved SNR, spatial resolution, and soft tissue contrast [23]. However, while clinically important structures were better visualized, magnetic susceptibility artefacts and chemical shift artefacts were increased. In the current study, there was no significant difference between 7T and 3T regarding artefacts. Commercially available coils for UHF imaging were not available in 2002, and it was suggested that using a phased array coil would increase SNR even further [23]. The technological development of 7T systems in the last decades has allowed wrist imaging at 7T to be significantly improved. It now shows great promise to improve diagnostic confidence and accuracy [10] by delivering excellent delineation of anatomical structures [14]. In contrast to the current work, a previous study revealed no significant difference in the visualization of anatomical structures in the wrist at 7T compared to 3T, despite an increase in SNR of up to 100%, with considerable variation between different anatomical structures [15]. However, Nordmeyer-Massner et al [15] used only one coronal 2D gradient echo sequence, not developed for clinical imaging, and they used a wrist coil array developed for 3T but replicated for operation at 7T, as no commercially available wrist coil for 7T was available at that time. This is in contrast to the current study comparing several sequences developed and optimized for clinical imaging and using commercially available dedicated wrist coils at both field strengths. Interestingly, a cadaver study published in 2011 [16] showed better visibility of articular cartilage surfaces with MR arthrography at 3T compared to 7T. The authors suggested that this difference was due to readers having more experience in evaluating 3T images than 7T images. In addition, sequences were not optimized for 7T because of limited knowledge about tissue and contrast media relaxation parameters [16].

Recently, a study compared MR of the knee at 7T and 3T, in 40 patients with pain of unknown etiology [24]. MR at 7T improved diagnostic confidence, mostly because of higher spatial resolution [24]. The current study demonstrates that anatomic structures in the wrist are better visualized at 7T compared to 3T, and a supposition is that better anatomical visibility and delineation will translate into better detection and definition of pathology. Future studies should be done in patients with wrist pain, to determine if an improvement in diagnostic confidence can be found at 7T compared to 3T, when pathology in the intricate structures of the wrist, such as ligaments and articular cartilage, is suspected. The wrist is a particularly challenging region to depict with MR, due to the small size of clinically important structures such as intercarpal ligaments, the TFCC and articular cartilage [5]. Utilization of a 3D sequence has been reported to enhance visualization of the SLL [25, 26] and the LTL [26], as it allows for MPR in any arbitrarily chosen plane, making it easier to visualize these small, complex, intercarpal structures that should be assessed in several imaging planes [27]. Although in depth comparison between 3D and 2D imaging is beyond the scope of the current study, the distribution of grades in Supplementary Table 3 points in the same direction, with 3D sequences more often receiving the highest grades (grades 4 and 5) regarding visualization of ligaments compared to the 2D sequences for both 7T and 3T.

The small size of the wrist makes it an auspicious area for UHF MR imaging, as it limits RF interference effects such as central brightening [15], which is a key problem in UHF MR brain and abdominal imaging [28, 29]. Tissue heating is generally of greater concern at higher field strengths due to higher transmitted RF energy [30]. This is less of a constraint in wrist imaging, as the extremities are less susceptible to RF power deposition than the head and trunk [30].

Evaluation of how different anatomical structures are visualized on a grading scale is highly subjective, as it relies on human senses and individual interpretation of data. Preference bias and recognition bias are two examples of why the human brain cannot be expected to be able to evaluate images truly objectively. Also, some observers will generally give higher grades than others, which may result in poor inter-observer agreement, even when all observers agree that one test condition is, for example, two steps superior to another. In order to increase conformity in grading, a training session was held with all the observers. To further alleviate these issues, VGC analysis was chosen as the statistical method for this study, as it provides a clear comparison between the two test conditions, without letting subjective interpretation of the grading scale or individual tendencies toward either side of the scale affect the results. Furthermore, it incorporates inter- and intra-observer variations in the results.

Limitations of the current study were the small number of subjects and the lack of pathology in the study population. Also, the 3D sequences were optimized for ligament visualization, which may have resulted in a less optimal visualization of other structures. This is particularly noticeable for tendons and nerves at 3T. However, 7T was superior to 3T in the assessment of these structures using the 2D sequences as well. Additionally, there were slight motion artefacts, affecting edge sharpness, present in several of the data sets at 7T and 3T. Taking into consideration that there was no significant difference in the presence of artefacts between 7T and 3T, these motion artefacts should not have affected the results.

In conclusion, the results show that 7T can improve visualization of anatomical structures of the wrist compared to 3T in healthy volunteers. Further studies are needed to assess if this superiority of 7T MR is evident also in patients with wrist injuries.