Healthy or not? How to control chemical-shift artifacts in 7-Tesla MRI

Chemical-shift displacement in the slice-selection direction can cause artifacts in magnetic-resonance images acquired at higher field strengths. The artifacts present with a surprising variety. To learn how to keep them under control, check-out the recent publication of Constantin von Deuster et al. here or here.

Chemical-shift artifacts

In magnetic-resonance imaging (MRI), hydrogen atoms in both fat and water molecules give rise to a bright signal. However, since they are bound to different atoms (carbon and oxygen, respectively), they generate signals with different frequencies.

This causes fat (such as in bone marrow) and water (such as in skeletal muscle) to show up at slightly different positions in a MR-image – even if they were located at the same position in the body.

Radiology professionals are familiar with an apparent fat displacement within the image plane, along the readout (or frequency-encoding) direction. When reading MR-images, radiologists routinely account for it. When doing so, many among them likely take for granted, however, that the fat- and water-signal sources depicted in a single image are still located in the same slice within the examined body.

Through-slice chemical-shift artifacts

At lower field strengths, this assumption holds well enough. But at a field strength of 7 Tesla, the fat and water components shown in an image may physically be located in two quite different slices. Their spatial separation can easily reach the 2-fold of the prescribed slice thickness – when using similar radio-frequency pulses as at lower fields.

Consider the above images from the same image series. In both images, structures that predominantly generated a water signal, e.g., muscles and nerves, in a given slice are depicted together with fatty (e.g., bone-marrow) tissues from another slice, which is located further in the back of the knee, by more than 2 slice widths.

In the representation of the fibular head on the left, this ended up looking almost as it should, since the fibular head extended far enough beyond the imaged slice to the back for bone marrow to show up at approximately the "right" location. In contrast, since the slice on the right is closer to the rear end of the fibular head, there was no more corresponding bone marrow even further back in the knee. Thus, no bone-marrow signal shows up in the fibular head, where we would expect it.

Interpreted in isolation, the right image could suggest, e.g., a bone-marrow pathology, while, in fact, it is only an artifactual result of the imaging process.

• Can you spot associated peculiarities (artifacts) in the left image?
• And how about on images acquired at 3 Tesla without fat saturation?

Minimizing chemical-shift artifacts

MRI technology allows us to minimize such undesired effects. However, there is no free lunch:

• At all field strengths, protocols are routinely optimized for smaller fat shifts in the readout direction by increasing receive bandwidths. This comes at the cost of a reduced signal-to-noise ratio (SNR) of the images and a careful tradeoff has to be made.
• In contrast, through-slice chemical-shift artifacts can be minimized by increasing the spectral width of the applied radio-frequency pulses. Luckily, this may not have a strong effect on the signal-to-noise ratio. Here, the trade-off is between an improved image quality and an increased amount of radio-wave energy that is deposited in the examined body parts (increased specific absorption rate, SAR).

Optimizing the sequence and parameters as described in Constantin's publication, high-quality concurrent fat and water 2D spin-echo imaging of the knee at 7 Tesla is possible.

You can get it done on our open-access Siemens MAGNETOM Terra system.