Chemical shift refers to small changes in resonant frequency due to different molecular environments of nuclei. The ¹H protons of fat, for example, are nestled within long-chain triglycerides and covered by electron clouds. These clouds partially shield the fat protons from the full effects of an externally applied magnetic field. The ¹H protons of water, however, are less shielded because their electron clouds are pulled away from them by the highly electronegative oxygen atom.
Because of this differential shielding by electron clouds, a fat proton experiences a slightly weaker local magnetic field and will resonate at a slightly lower frequency than a nearby water proton. The difference in frequencies is extremely small, on the order of 3.4 parts per million (ppm) or 3.4 x 10-6. At 1.5 T this corresponds to an absolute frequency difference of approximately 215 Hz. At 3.0 T the difference is twice as large, or about 430 Hz.
Reducing the bandwidth per pixel will accentuate this artifact. Narrow bandwidth techniques should therefore generally be avoided in locations where chemical-shift artifacts may obscure important interfaces (for instance, between the optic nerve and orbital fat).
Because the chemical-shift artifact is a spatial mismapping of MR signal based on frequency, it will typically be seen in the frequency-encode direction. (An exception occurs in echo-planar imaging, where chemical shift artifacts are commonly seen in the phase-encode direction. See the next Q&A).
Additionally, in 2DFT imaging, recall that slice selection is defined by variations in frequency. In this method of acquisition, chemical shift artifacts may occur not only within a plane of imaging, but also between slices (i.e., in the slice-select direction). Inter-slice chemical-shift artifacts may appear as dark or light "halos" around certain anatomic structures; others are less obvious and result in subtle degradation of image quality.
Finally, one should be aware that chemical-shift artifacts occur not only between water and fat, but at the interface of any two substances having different chemical shifts. For example, the chemical shift between silicone oil (used in ophthalmology) and water is ~4.4 ppm; between silicone gel (in breast prostheses) and water is ~4.7 ppm. Thus chemical shift artifacts may be seen in locations like the breast, head and neck, and extremities where silicone injections or implants are commonly used.
Advanced Discussion (show/hide)»
Babcock EE, Brateman L, Weinreb JC et al. Edge artifacts in MR images: chemical shift effect. J Comput Assist Tomogr 1985; 9:252-257. (One of the first descriptions of this artifact)
Birkefeld AB, Bertermann R, Eckert H, Pfleiderer B. Liquid- and solid-state high-resolution NMR methods for the investigation of aging processes of silicone breast implants. Biomaterials 2003; 24:35-46.
Dwyer AJ, Knop RH, Hoult DI. Frequency shift artifacts in MR imaging. J Comput Assist Tomogr 1985;9:16-18.
Hood MN, Ho VB, Smirniotopoulos JG, Szumowski J. Chemical shift: the artifact and clinical tool revisited. Radiographics 1999; 19:357-371 (Excellent review).
Mathews VP, Elster AD, Barker PB, et al. Intraocular silicone oil: in vitro and in vivo MR and CT characteristics. AJNR Am J Neuroradiol 1994; 15:343-7.
Smith RC, Lange RC, McCarthy SM. Chemical shift artifact: dependence on shape and orientation of the lipid-water interface. Radiology 1991; 181:225-229.
Soila KP, Viamonte M, Starewicz PM. Chemical shift misregistration effect in magnetic resonance imaging. Radiology 1984; 153:819-820. (First published description of the chemical shift artifact).
Doesn't the chemical shift between water and fat protons also result in a phase shift between them? If this is so, then why aren't chemical shift artifacts seen in the phase-encode direction also?