Intracranial chemical-shift artifacts on MR images of the brain: observations and relation to sampling bandwidth.

The purpose of this study was to evaluate the presence of chemical-shift artifacts on cranial MR and to illustrate the interrelationship among chemical-shift artifacts, variable acquisition parameters, and field strength. Measurements of chemical-shift artifacts were performed on scans obtained from a volunteer imaged in a 1.5-T General Electric system at bandwidths of 8, 16, and 32 kHz, using a 24-cm field of view and an 8-kHz bandwidth with a 48-cm field of view. Chemical-shift displacements at 8 kHz were 6.6 and 14.2 mm at the respective fields of view. Retrospective review was also performed in 77 cases of cranial MR performed on a 1.4-T Technicare unit for the presence and source of chemical-shift artifact on spin-density and T2-weighted images. Most data reviewed showed no significant interference of chemical-shift artifacts on cranial images. An artifactual subdural fluid collection was a common artifact (n = 30/77). When present, this was due to shift of fat signal from subcutaneous tissues onto the brain in patients younger than 10 years old (n = 4/10) and correlated with the distance between brain and subcutaneous fat of less than the linear value of the chemical shift. When this artifact was present in adults (n = 25/67), it was due to shift of the medullary fat signal across the inner table of the skull. The latter also occurred in one child under 10. Apparent location shifts, consistent with the displacement expected from the chemical-shift artifact, were noted in five of five cases of intracranial lipoma. In one of these, the chemical-shift artifact disguised the presence of a large associated vessel. The method of calculating the linear displacement of chemical-shift artifact is reviewed, and the interrelationship of machine parameters and chemical-shift artifact is illustrated. Chemical-shift artifact increases proportionally with field strength and field of view. Increasing the bandwidth to decrease chemical-shift artifact has a resultant penalty in signal to noise but allows a lower time to echo. A lower time to echo can also be accomplished without increasing the bandwidth if asymmetric sampling is used. Awareness of the relationships among chemical-shift artifacts, acquisition parameters, and field strengths can result in a more tailored examination when the chemical-shift artifact is going to be a significant factor. In addition, interpreter error can be avoided by awareness of these relationships when reviewing images from outside institutions.