Pituitary gland enlargement secondary to dural arteriovenous fistula in the cavernous sinus: appearance at MR imaging

PURPOSE

To evaluate pituitary gland size in cases of dural arteriovenous (AV) fistula in the cavernous sinus and to correlate the size with the degree of AV fistula.

MATERIALS AND METHODS

Magnetic resonance images in 21 patients with angiographically proved dural AV fistula of the cavernous sinus were retrospectively reviewed. In four patients, findings obtained before and after embolization therapy were compared.

RESULTS

The 21 patients were divided into group 1 (severe AV fistula) and group 2 (moderate or mild AV fistula). The superior contour of the pituitary gland was convex or flat in group 1 and flat or concave in group 2. The mean height of the pituitary gland was 9.4 mm +/- 1.5 (standard deviation) in group 1 and 6.7 mm +/- 1.2 in group 2 (P = .0002). After embolization therapy, pituitary gland height decreased and signal intensity voids improved in the four patients and no AV fistula was observed in three.

CONCLUSION

In cases of severe dural AV fistula in the cavernous sinus, the pituitary gland is enlarged, which should not be misdiagnosed as pathologic.

The effect of variation in slice thickness and interslice gap on MR lesion detection

Lesion detection by MR imaging depends on the contrast-to-noise ratio of the voxels containing the lesion relative to those containing the background. When the lesion voxels are less than completely filled, the inherent contrast between lesion and background is modified by the filling factor. Lesion detection thus depends on lesion size, slice thickness, lesion position relative to slice, thickness of gap between slices, and inherent contrast between lesion and background. Using computer simulation, the effect of variation in the slice thickness and the interslice gap on lesion detection is modeled as a function of lesion size, filling factor, and inherent contrast. Detection of small, low-contrast lesions is shown to be most sensitive to partial volume effects and to be greatest with thin slices. Detection of high-contrast lesions is shown to be limited primarily by the presence of a gap between slices. For patients with diffusely distributed disease-e.g., the small, low-contrast lesions of multiple sclerosis-lesion detection is greater for thin slices, even with a gap, than for thick, contiguous slices.