Invited Review Clinical Applications of Diffusion Tensor Imaging Qian Dong, MD, Robert C. Welsh, PhD, * Thomas L. Chenevert, PhD, Ruth C. Carlos, MD, MS, Pia Maly-Sundgren, MD, PhD, Diana M. Gomez-Hassan, MD, PhD, and Suresh K. Mukherji, MD Directionally-ordered cellular structures that impede water motion, such as cell membranes and myelin, result in wa- ter mobility that is also directionally-dependent. Diffusion tensor imaging characterizes this directional nature of wa- ter motion and thereby provides structural information that cannot be obtained by standard anatomic imaging. Quantitative apparent diffusion coefficients and fractional anisotropy have emerged from being primarily research tools to methods enabling valuable clinical applications. This review describes the clinical utility of diffusion tensor imaging, including the basic principles of the technique, acquisition, data analysis, and the major clinical applica- tions. Key Words: diffusion tensor; MRI; DTI; brain; apparent diffusion coefficient; diffusion anisotropy J. Magn. Reson. Imaging 2004;19:6 –18. © 2003 Wiley-Liss, Inc. DIFFUSION TENSOR IMAGING (DTI) relies on thermally- driven random motion of water molecules to supply microscopic structural information in vivo (1,2). Ran- dom motion of water molecules, also known as Brown- ian motion, can be quantified and reflects intrinsic features of tissue microstructure in vivo (3). In uncon- strained water molecules in a pure liquid environment free of impediments or in a sample where the barriers are not coherently oriented as in a cyst, diffusion is equal in all directions. This situation is referred to as “isotropic.” In brain tissue, however, water diffusion is substantially reduced by impediments placed by struc- tures such as myelin sheaths, cell membranes, and white matter tracts. In general, the diffusion of the water molecules is less restricted along the long-axis of a group of aligned tissue fibers (such as those of white matter) than perpendicular to it. The condition of direc- tionally-dependent diffusion is referred to as “anisotro- pic.” Three descriptive levels are commonly used to por- tray tissue diffusion properties. First, the apparent diffusion coefficient (ADC) can be quantified to provide information on the degree of restriction of water mole- cules. Second, the degree of directionality is often de- scribed via an index such as fractional anisotropy (FA). Highly-directional axonal fibers, such as white matter, are revealed as hyperintense on an FA map. Third, the predominant diffusion direction can also be deter- mined, which is used as an input to fiber tracking algorithms. In general, the more unrestricted the water molecules are in a given tissue, the higher the ADC will be and the lower the anisotropy will be. Directionally- encoded color (DEC), another method to identify major fiber direction, recently developed by Pajevic and Pier- paoli (4), derives additional information from DTI. The location and orientation of major white matter fiber tracts can be revealed by hue with these full tensor- based color methods. With recent improvements in MR hardware, DTI acquisition times have been reduced to allow complete brain coverage in a clinically acceptable period. These features of the DTI technique provide a sensitive means to identify different components of brain tissue and evaluate the integrity and direction of the fiber tracts in various pathological conditions. Tis- sue maladies studied by DTI include cerebral ischemia, multiple sclerosis, epilepsy, metabolic disorders, and brain tumor. PHYSICS, ACQUISITION, AND DATA ANALYSIS Diffusion, Scalar, and Anisotropic Tensor imaging is predicated on the self-diffusion of water in vivo and how free-isotropic self-diffusion may be affected by the properties of the tissue. Diffusion, also referred to Brownian Motion, is the inherent ran- dom motion of a molecule due to its thermal energy. Self-diffusion was first observed by Brown in 1827 (5), and quantified by Einstein in 1905 (6). In a liquid, diffusion is determined by the size and temperature of the molecule and the viscosity of the medium. On av- erage, there is no net change in position over an ensem- ble of water molecules since there is no preferred mi- gration direction. Each molecule, however, will randomly move about, and will have a net root mean square displacement (RMS). Along any given direction, the RMS displacement is given by RMS =  2D (1) Department of Radiology, University of Michigan, Ann Arbor, Michigan. *Address reprint requests to: R.C.W., University of Michigan Medical Center, Department of Radiology-MRI Division, 1500 E. Medical Center Drive, Ann Arbor, MI 48109-0030. E-mail: rcwelsh@umich.edu Received May 13, 2003; Accepted August 22, 2003. DOI 10.1002/jmri.10424 Published online in Wiley InterScience (www.interscience.wiley.com). JOURNAL OF MAGNETIC RESONANCE IMAGING 19:6 –18 (2004) © 2003 Wiley-Liss, Inc. 6