7 T Musculoskeletal MRI: Fundamentals and Clinical Implementation

Abstract

This review summarizes the current state-of-the-art of musculoskeletal 7 T magnetic resonance imaging (MRI), the associated technological challenges, and gives an overview of current and future clinical applications of 1 H-based 7 T MRI. The higher signal-to-noise ratio at 7 T is predominantly used for increased spatial resolution and thus the visualization of anatomical details or subtle lesions rather than to accelerate the sequences. For musculoskeletal MRI, turbo spin echo pulse sequences are particularly useful, but with altered relaxation times, B1 inhomogeneity, and increased artifacts at 7 T; specific absorption rate limitation issues quickly arise for turbo spin echo pulse sequences. The development of dedicated pulse sequence techniques in the last 2 decades and the increasing availability of specialized coils now facilitate several clinical musculoskeletal applications. 7 T MRI is performed in vivo in a wide range of applications for the knee joint and other anatomical areas, such as ultra-high-resolution nerve imaging or bone trabecular microarchitecture imaging. So far, however, it has not been shown systematically whether the higher field strength compared with the established 3 T MRI systems translates into clinical advantages, such as an early-stage identification of tissue damage allowing for preventive therapy or an influence on treatment decisions and patient outcome. At the moment, results tend to suggest that 7 T MRI will be reserved for answering specific, targeted musculoskeletal questions rather than for a broad application, as is the case for 3 T MRI. Future data regarding the implementation of clinical use cases are expected to clarify if 7 T musculoskeletal MRI applications with higher diagnostic accuracy result in patient benefits compared with MRI at lower field strengths.

FIGURE 1

7 T MRI knee protocol with turbo spin echo (TSE) sequences at the authors' institution. Although at lower field strength, TSE sequences are the mainstay of musculoskeletal MRI; at 7 T MRI, severe chemical shift artifacts and signal intensity inhomogeneities needed to be overcome for implementing TSE sequences. A, Coronal T1 sequence, (B) coronal proton density (PD) fat-saturated (fs) sequence, (C) sagittal PD fs sequence, and (D) axial PD fs sequence. Note that the small cartilage delamination (B) and superficial defects (C) of the femoral cartilage (arrowheads in inserts) are well depicted on high-resolution 7 T MRIs.

FIGURE 2

7 T spin echo MRI with matched excitation and refocusing RF bandwidths showing coronal PD-weighted 2D TSE images of a right knee (on the left) and sagittal reformations through the lateral (in the middle) and the medial (on the right) femoral condyles. The 1 mm coronal images were acquired without gaps and with RF bandwidths of 1000 Hz (A) and 250 Hz (B) of the excitation (90 degrees) and refocusing (130 degrees) pulses, respectively, and with identical RF pulse durations in A and B. The lower RF bandwidth in B resulted in an increased through-slice (anterior-posterior direction) chemical-shift displacement of bone marrow and adipose tissue, as indicated by the caliper arrows. The asterisk (*) in B highlights a corresponding artifactual absence of signal in the coronal image. Also note the higher overall signal and the blurrier image appearance in B due to a less sharp definition of the 2D slice profile associated with the lower RF bandwidth. The phase- and frequency-encoding directions were right-left and superior-inferior, respectively.

FIGURE 3

7 T spin echo MRI with unmatched excitation and refocusing RF bandwidths using coronal PD-weighted 2D TSE images of a right knee acquired with identical parameters, except for the nominal RF bandwidth of the refocusing RF pulses (130 degrees), which was set to 1000, 750, and 500 Hz in A, B, and C, respectively. The 90-degree RF excitation pulse bandwidth was 1000 Hz. Estimated specific absorption rate values, SAR, were 0.26, 0.20, and 0.16 W/kg, respectively. The decreasing overlap of the slices with excited and refocused fat protons decreased fat-signal intensity from A to C. A moderate mismatch in B allowed a significant SAR reduction with a fat-signal loss that may be tolerable in many applications while still producing sufficiently crisp images due to the slice profile being dominated by the higher-bandwidth excitation pulse. A sufficiently large mismatch, as in C can provide fat-signal suppression without increased SAR.

FIGURE 4

Intermediate weighted, fat-saturated sagittal finger MRI with identical TR and TE, acquired with a dedicated signal-receive hand coil at a field strength of 3 T (A and C) and a transmit-receive knee coil at 7 T (B and D) as off-label use. Acquisition parameters (A–D): TR, 3000 milliseconds; TE, 44 milliseconds; slice thickness, 2 mm; number of averages, 2. In-plane voxel dimensions: 0.19 mm² (A and B), 0.06 mm² (C and D); acquisition time: 2:26 minutes (A and B), 4:50 minutes (C and D); receive bandwidth: 116 kHz (A and B), 274 kHz (B and D); acquisition time: 2:26 minutes (A and B), 4:50 minutes (C and D). Note the lower signal-to-noise ratio of the 7 T MRIs despite the higher field strength. However, the receive bandwidths would not necessarily have had to be adapted for these fat-saturated images, and the TR should be longer at 7 T for comparable image quality.

FIGURE 5

Sagittal 3D spoiled gradient echo 7 T MRI using (A) (FLASH; TR, TE, flip angle = 17.2 milliseconds, 2.04 milliseconds, 10 degrees) and sagittal turbo spin echo images (B) (SPACE; TR, TE_eff, flip angle, echo-train-length = 1500 milliseconds, 42 milliseconds, 120 degrees, 43) of the knee without fat-signal suppression. The TE in A corresponds to the second in-phase echo time. Note the dark appearance of bone marrow and fat in the gradient echo image and the inhomogeneous signal brightness in the turbo spin echo image (eg, relatively dark Hoffa's pad [*] versus bright posterior subcutaneous fat [‡]).

FIGURE 6

7 T MRI of the left forefoot with transverse (A), sagittal (B), and coronal (C) 3D DESS reconstructions with partial maximum intensity projection in a 34-year-old asymptomatic male volunteer, demonstrating numerous Pacinian corpuscles (hyperintense ovoid nodules of 1–4 mm size in the plantar subcutaneous fat [arrows] and interdigitally arranged in chain-like formations), which are sensory receptors for vibration and deep pressure.

FIGURE 7

3D dual-echo steady-state (DESS) MRI of the cervical spine in a 34-year-old male healthy volunteer. Sagittal oblique reconstructions as a 5-mm maximal intensity projection visualizing the left posterior intraspinal nerve rootlets (arrows) of nerve roots C3–C7. A and B, 3 T MRI and (C and D) 7 T MRI. Note the higher contrast and higher number of rootlets that were identified at 7 T.

FIGURE 8

Spatially matched 3D dual-echo steady-state (DESS) images of a 66-year-old man. 7 T MRI (A–C) and 3 T MRI (D–F) of the left knee. At 7 T, the hypointense calcium crystal deposits in the femorotibial and patellar cartilage are clearly and sharply visible (arrows), whereas at 3 T MRI, these are much less visible, even without equivalent signal change (arrowheads).

FIGURE 9

Sagittal 3D dual-echo steady-state (DESS) 7 T MRI (A) of an asymptomatic 43-year-old amateur athlete depicting delamination (arrowhead) of the femoral cartilage in the medial compartment of the knee, and corresponding sagittal 3 T DESS MRI (B), where the cartilage defect is less well defined (arrowhead).