7 Tesla whole body MRI systems equipped with knee and extremity coils are seeing increasing application in musculoskeletal MRI. The higher magnetic polarization yields an SNR increase that can be exploited to achieve higher resolution 3D scans than practical in clinically acceptable scan times at 1.5 and 3 Tesla. These very high-resolution 3D scans can potentially be used to make even more precise measurements of cartilage morphology. High SNR combined with high contrast between cartilage and adjacent musculoskeletal tissues (such as synovial fluid and muscle) can also potentially improve semi-automated and automated segmentation techniques, helping reduce the need for painstaking (and sometimes error-prone) manual segmentations.

Balanced SSFP (bSSFP), DESS, and 3D SPACE sequences are all common for morphological evaluation of cartilage at 1.5 and 3 Tesla, but they have not yet been commonly used at 7T. Balanced SSFP further suffers from sensitivity to off-resonance, typically resulting in bands of signal null across the image (“banding artifact”). High field strengths compound the banding artifact problem with bSSFP, making some kind of banding artifact removal essential for application of bSSFP at 7T. One of the most common techniques for banding artifact removal is the acquisition of multiple phase-cycled bSSFP acquisitions, followed by a root sum-of-squares (SOS) combination of the individual phase-cycled images [1].

In this study, we (1) implemented a two-acquisition 3D phase-cycled bSSFP protocol at 7 Tesla that achieves 0.31mm isotropic resolution in under 9 minutes of scan time, (2) implemented a 3D DESS protocol at 7 Tesla that achieves 0.36mm isotropic resolution in just under 7 minutes, (3) performed a contrast optimization to identify flip angles that maximize both cartilage/muscle and cartilage/synovial fluid contrast, and (3) compared to a 3D SPACE acquisition at 7T that achieves 0.55mm isotropic resolution in a scan time of 11:37.

Phase-Cycled bSSFP Protocol: Two phase-cycled acquisitions were performed followed by a root sum-of-squares combination with the following parameters: TR/TE = 4/35/1.88ms, 25 degree flip angle (max due to SAR constraints), 0.31mm isotropic resolution over a 15.8cm x 15.8cm x 12cm FOV, sagittal acquisition slab, total scan time = 8:50.

DESS Protocol: The 3D DESS protocol utilized the following parameters: TR/TE = 8.19/2.26ms, 25 degree flip angle (max due to SAR constraints), 0.36mm isotropic resolution over a 15.8cm x 15.8cm x 12cm FOV, sagittal acquisition slab, total scan time = 6:54.

Contrast Optimization: Signal equations for bSSFP and DESS were used to perform a numerical optimization of cartilage/muscle contrast-to-noise ratio (CNR) and cartilage/synovial fluid CNR as a function of flip angle. Actual DESS scans were also performed across a range of flip angles, and the CNRs measured from the scans for both to compare with the theoretical predictions.

All scans were performed on a 7 Tesla Siemens whole body scanner equipped with a 32-channel knee coil. The phase-cycled bSSFP and DESS scans were then compared with a 3D SPACE protocol that achieved 0.55mm isotropic resolution over a similar FOV in a total scan time of 11:37.

Results from the numerical contrast optimization are shown in Figure 1. The measured DESS CNR values are in excellent agreement with those predicted by the numerical simulation up to a flip angle of 25 degrees. (Flip angles above 25 degrees could not be validated due to SAR constraints.) Cartilage/fluid CNR is optimized for DESS at ~25 degrees and for bSSFP at ~35 degrees, so a 25 degree flip angle (the maximum achievable due to SAR constraints) was used achieve the highest cartilage/fluid CNR possible.

The phase-cycled bSSFP protocol at 0.31mm resolution yielded exquisite results. Native sagittal images along with axial and coronal reformats are shown in Figure 2. A comparison between the 3D phase-cycled bSSFP images, the 3D DESS images, and the 3D SPACE images is shown in Figure 3. All of the sequences achieved beautiful high resolution images, although the bSSFP images provided the best combination of high resolution (0.31mm isotropic) and excellent cartilage/synovial fluid contrast at a scan time of under 9 minutes.

Finally, both the SNR (normalized by voxel volume) and SNR efficiency (SNR normalized by square root of scan time) were measured in cartilage in the patella, the anterior femur, and the posterior femur for both the bSSFP and DESS scans. These are reported in Table 1.

All three of the tested protocols provide excellent delineation of cartilage morphology, high resolutions and SNR, and short scan times. While the 3D DESS scan provides superior cartilage SNR than bSSFP at the flip angle achievable, bSSFP provides superior cartilage/fluid contrast while achieving higher resolution at acceptable SNR.