Authored date:2006-03-22

SNR Advantage:

Orthopedic imaging is greatly enhanced by improvements in resolution. However, this reduction in voxel size is ultimately limited by the signal-to-noise ratio (SNR) of the images. The development of 3T systems affords a significant increase in SNR, compared to widely-used 1.5T imagers. The improved SNR gives images of better quality; alternatively, the SNR currency may be spent to obtain higher resolution or shorter imaging times.
Sites that routinely acquire high-resolution scans in two or three planes at 1.5T may want to save time by using a single 3-D isotropic voxel protocol at 3T, displaying the other two planes by reformatting.
Even using a single acquisition, the imaging time for scans with large matrix sizes (e.g. 1024) can be prohibitive at 1.5T, and parallel imaging techniques (iPAT) result in a loss of SNR. At 3T, the additional SNR permits both small voxels and the use of parallel imaging techniques to make these studies feasible. Doubling the field strength to 3T doubles the initial magnetization of the nuclear spins. This can lead the unwary to think that the SNR of every protocol will be doubled.
In fact, great care must be exercised to take full benefit of the higher field strength, and even this does not ensure that every examination is significantly improved at 3T.

Fig. 1  Proton density turbo spin echo with fat-sat. Matrix 512, 0.3x0.3x0.3 mm.

Fig. 2  High resolution knee imaging with 8 channel tx/rx knee coil. 0.3x0.3x03 mm, PATx2.

More Spatial Shift between the Fat and Water Components of the Image

In nearly all MR scans, the fat and water components of the image are not in perfect registration. Orthopedic images often contain fatty bone marrow or fat pads in close proximity to important structures such as cartilage. A poorly-chosen protocol may cause the bright, shifted fat signal to obscure these structures. Many orthopedic protocols suppress the fat signal to avoid this problem.
During the data acquisition (readout) period, the fat spins precess (and the signal oscillates) slightly slower than the water spins. If, during this period, the fat spins precess two cycles less than the water spins, the image reconstruction will shift the position of the fat by two pixels. Even though the precessional frequencies of the water, the fat, and the difference between them all depend on the magnetic field, the strength of the readout gradient has almost no effect on this frequency difference. If a gradient increases the field of a voxel by 1 mT, it changes the precessional frequencies by about 42,600 Hz, but changes the frequency difference between fat and water by a miniscule 0.15 Hz. The shift between the fat and water components of the image is essentially unaffected by the readout gradient strength. What will change the spatial shift between these components is shortening the duration of the readout: if the duration in this example is halved, the fat and water spins now dephase by only one cycle, and the fat-water shift following reconstruction is reduced from two pixels to one pixel.
This is referred to as “doubling the receiver bandwidth”, but this phrase is shorthand for four changes: the readout period is halved (this is what we care about), the readout gradient is doubled (this maintains resolution), the voltage is sampled twice as often (the “sampling bandwidth” is doubled) so that the same number of points are acquired, and the bandwidth of the analog or digital filtration of the signal is made wider (reducing the SNR by 30%).
If the orthopedic protocol uses fat suppression, a low receiver bandwidth can be used, and it may be possible to realize the full SNR enhancement of the 3T scanner. If fat suppression cannot be used, the receiver bandwidth should be approximately doubled relative to the 1.5T protocol. The theoretical 3T SNR gain of a factor of two will be reduced to 
1.4-fold gain. A receiver bandwidth of 220 Hz/pixel gives a fat-water shift of one pixel at 1.5T; a bandwidth of 440 Hz/pixel gives the same shift at 3T.
This shift of the fat and water image components does not occur in the phase encoding direction. The phase changes caused by the incrementation of the phase encoding gradient affect the fat and water spins almost equally. It should be noted that the dark border that completely surrounds some organs in the abdomen when gradient echo images are acquired with an “opposed phase” echo time (the “India ink effect”) can ordinarily be attributed more to partial volume effects rather than to the fat-water spatial shift.

Better Fat Suppression

The increased separation of the fat and water resonant frequencies at 3T makes it easier to saturate the fat without affecting the water signal. It also permits the use of a shorter FATSAT radiofrequency (RF) pulse. Similarly, the waiting time between the components of water excitation pulses may be halved. Fat suppression is simply easier at 3T than at 1.5T.

Fig. 3  High resolution shoulder imaging. Proton density turbo spin echo with fat-sat. 
0.4x0.4x2mm PATx2.

Longer T1

The T1 of many tissues of orthopedic interest increases about 20% as the field doubles to 3T [1]. This has little effect on proton density- or T2-weighted scans, but will ordinarily require some increase in TR for T1-weighted scans. Whether this leads to a less time-efficient scan will depend on the protocol. If the corresponding 1.5T protocol had less than the desired number of slices, or if the slices were thicker than desired to reduce their number, then a 20% increase in TR could lead to a more desirable result at 3T. If the 1.5T protocol had multiple averages, then fewer averages could be used, trading some of the SNR advantage of 3T for a faster scan. Similarly, if the 1.5T scan were a single acquisition, parallel acquisition techniques (iPAT) could result in a faster scan A side effect of the elongation of T1 is somewhat better saturation of blood. This may improve the effectiveness of saturation bands used to reduce the pulsatility artifact of arteries such as the popliteal.

Fig. 4  VIBE arthrography. 0.5x0.5x1mm.

iPAT

There is a natural fit between 3T orthopedic imaging and parallel imaging techniques (iPAT). High resolution joint imaging with matrices of 512 and greater is desirable, but this is limited by both SNR and acquisition time. The extra SNR at 3T maintains the image quality, while iPAT keeps the imaging time within reason.

Shorter T2

Doubling the field strength to 3T shortens the T2 of various tissue by 10-37% [1], suggesting that excessive echo train lengths for turbo spin echo pulse sequences should be avoided to prevent blurring. It does not appear that the changes in T2 will produce significant changes in tissue contrast for most protocols.

Fig. 5  Isotropic 3D wrist imaging. VIBE, 0.4x0.4x0.4 mm.

Fig. 6  3D isotropic ankle imaging. DESS, 0.4 x 0.4 x 0.4 mm.

Courtesy: Prof. Fukatsu, University of Nagoya, Japan

Fig. 7  Visualization of anterior cruciate ligament with isotropic 3D DESS dataset.

Courtesy: Prof. Fukatsu, University of Nagoya, Japan

Increased Specific Absorption Rate (SAR)

For RF pulses of the same duration, tip angle and shape, a doubling of the main magnetic field leads to approximately four times the SAR. Fortunately, many orthopedic exams employ smaller transmit/receive coils that use much less power than the body coil, and most protocols can be run without alteration. T1-weighted spin echo sequences with many slices and turbo spin echo sequences with many 180º refocusing pulses are the most likely to be troublesome.When SAR must be reduced, low SAR RF pulses can be selected, the number of slices can be reduced, or alternative sequences may be employed. These alternatives may be as commonplace as low angle FLASH or FISP sequences, or new techniques such as hyperechoes. Sites that are willing to accept these changes in contrast relationships relative to their 1.5 T protocols will adjust quickly to imaging at 3T.

Increased Artifact from Orthopedic “Hardware”

Nonferromagnetic orthopedic implants have a magnetic susceptibility that differs from tissue, and can distort the local magnetic field. This distortion increases linearly with the main magnetic field, and can be considered an unwanted, misshapen gradient. Since the slice selection process and the readout depend directly on the linearity of the gradients, these field errors cause artifacts. By contrast, phase encoding uses the changes that occur as the phase gradient is incremented, and is less affected by constant local field errors. The solution, then, is to ensure that the applied slice and read gradients overwhelm the gradients created by the implant. As the main magnetic field is doubled to 3T, it is desirable to double the slice and read gradients for implant imaging. The readout gradient and associated pulses along the same axis can be doubled by doubling the receiver bandwidth on the Sequence card of the protocol. This reduces the SNR by 30%. Some of this SNR loss may be recovered by averaging multiple high bandwidth echoes, as in the MEDIC pulse sequence. The slice selection gradient can be increased by selecting the “fast RF mode” on the Sequence card. This will increase the SAR and may increase the minimum slice thickness.
Additional artifacts are caused by RF current flow that is induced in a metallic implant. The appearance of these artifacts is complex [2], but is not dramatically worse at 3 T compared to 1.5T. However, the change in this artifact may cause implants to have a different appearance at different field strengths.

RF Artifacts

At 3T, the dielectric properties of the body reduce the wavelength of the RF such that it is comparable to body dimensions, leading to the formation of standing waves and degrading the RF homogeneity. In addition, signal inhomogeneities caused by tissue conductivity are more noticeable at 3T than at 1.5T. These effects are minimal for small joint imaging, but may become significant for hip protocols.

Conclusions

The improved signal-to-noise performance at 3T relative to 1.5T makes this the preferred field strength for the majority of orthopedic exams. At the same time, one must resist the temptation to consider increased field strength as a simple amplifier of image quality. Those users who recognize the unique aspects of 3T imaging, and adjust their protocols accordingly, will be rewarded with superior results. In general, fat suppressed protocols used for patients without metallic hardware will give results greatly superior to 1.5T. When the receiver bandwidth must be increased to reduce the chemical shift artifact or to reduce metal artifacts, the SNR will still be better than a 1.5T exam, and this improvement may be traded for higher resolution. With careful selection of parameters, only a few examinations will have results that show no improvement compared to 1.5T.

References

[ 1 ] Garry E. Gold, Eric Han, Jeff Stainsby, Graham Wright, Jean Brittain, and Christopher Beaulieu, “Musculoskeletal MRI at 3.0 T: Relaxation Times and Image Contrast,” AJR 183 (2004) 343.
[ 2 ] Ulrike A. Lauer, Hansjorg Graf, Alexander Berger, Claus D. Claussen, and Fritz Schick, “Radio frequency versus susceptibility effects of small conductive implants – a systematic MRI study on aneurysm clips at 1.5 and 3 T,” Magnetic Resonance Imaging 23 (2005) 563.
[ 3 ] Garry E. Gold, Brian Suh, Anne Sawyer-Glover, and Christopher Beaulieu, “Musculoskeletal MRI at 3.0 T: Initial Clinical Experience,” AJR 183 (2004) 1479.