Diffusion-Weighted MR Imaging: A Useful Adjunct to Clinical Diagnosis or a Scientific Curiosity?

It has been appreciated from the earliest development of MR imaging that MR images are exquisitely sensitive to motion. This sensitivity remains the major limitation in the application of MR to the body. However, it also has been appreciated that motion is not simply a problematic source of artifact. Motion sensitivity can be used to advantage in sequences designed to highlight the flow sensitivity in blood, making possible MR angiography. Provided flow and pulsatile motion can be controlled satisfactorily, motions as small as those arising from perfusion and diffusion can be used to enhance contrast in specially sensitized MR imaging sequences. The pioneering work of Le Bihan et al. [1] in 1986 showed that MR pulse sequences can be sensitized to perfusion and diffusion. This sensitization showed specificity for certain neurologic disorders. The diffusion sensitivity of such images was based on straightforward physical principles , which had been elucidated two and three decades earlier [2, 3]. The MR acquisition records two interleaved images: (1) a standard spin-echo image and (2) an identical image with additional balanced diffusion-gradient pulses that have no effect on static spins but result in a loss of phase coherence for randomly diffusing spins. Subtraction of the second diffusionsensitive image from the first reference image results in a much noisier image in which the contrast is proportional to the diffusion coefficient at least for pure static liquids. In a patient, a large number of additional factors contribute to this different image. The contrast is therefore referred to as an