Image distortion caused by gradient eddy currents is a major problem in the use of diffusion tensor imaging (DTI), as using the uncorrected images for calculation of apparent diffusion coefficient (ADC) and diffusion anisotropy will result in areas of artificially increased anisotropy and ADC at the edge of the images, as well as decreased spatial resolution and accuracy in ADC computations overall. This distortion may be substantially reduced by careful adjustment of the gradient preemphasis unit. A completely automatic method of adjusting the preemphasis unit is proposed which finds the optimal settings for all three gradient directions in approximately 15 min by estimating the magnitudes of the eddy currents at various delay times after a test gradient. The pixel shifts in a 64 ؋ 128 echo-planar diffusion-weighted image with a diffusion gradient strength of 30 mT/m were reduced to less than 0. Geometric distortion resulting from gradient eddy currents is a major source of image artifacts in diffusion-weighted echo-planar imaging (DW-EPI). Depending on the direction of the diffusion gradient relative to the image plane, the distortion will consist of shearing, stretching, or shift in the phase-encoding direction. Because the diffusionweighted images are misaligned relative to images acquired without diffusion gradients, computed parameter maps such as apparent diffusion coefficient (ADC) or diffusion anisotropy maps will suffer from degraded resolution, lower accuracy, and visible areas of artificially increased ADC and anisotropy around the edges of the object. While the use of actively shielded gradients has greatly reduced the magnitude of eddy currents, significant distortion often still remains in DW-EPI images, which becomes more severe at higher field strengths.Methods proposed for distortion correction include: postacquisition image warping (1,2), modifications of the DW-EPI pulse sequence (3-5), acquisition of calibration scans (6 -10), and gradient preemphasis calibration (11). Gradient preemphasis calibration offers the significant advantage of eliminating the need for time-consuming image postprocessing, thereby allowing the use of diffusionweighted imaging (DWI) and diffusion tensor imaging (DTI) in clinical settings. Additionally, by minimizing image distortion associated with gradient eddy currents during the image acquisition, postprocessing to remove any remaining distortion should result in near-perfect coregistration of the diffusion-weighted images.Very accurate preemphasis and B 0 corrections are necessary to adequately correct for all image distortions in DW-EPI. For a typical 3 Tesla DW-EPI single-shot acquisition time of 60 ms and a field of view (FOV) of 20 cm, 10 cm from the isocenter, a gradient eddy current of 4 T/m, or a B 0 offset of 0.4 T, will produce an image shift of one pixel. The pixel shift is equal to the number of revolutions of dephasing over the acquisition time according to the Fourier shift theorem. For a typical diffusion gradient strength of 40 mT/m, the preemphasis...