ORIGINAL ARTICLE Diffusion Tensor Imaging in Cerebral Tumor Diagnosis and Therapy Aaron S. Field, MD, PhD* and Andrew L. Alexander, PhD† Abstract: Diffusion-weighted MR images and their analysis using the tensor model open up many new possibilities for tissue charac- terization, surgical planning, and treatment follow-up in patients with cerebral neoplasms. These possibilities are only just beginning to be fully explored. This article reviews the physical principles underlying diffusion tensor imaging (DTI); the various postprocessing methods available for DTI data in the context of tumor imaging; the commonly encountered patterns of tumor-related alteration to cerebral white matter, as depicted by directionally encoded color maps; and the current state of the art in DTI-based tumor diagnosis and treatment planning. Key Words: diffusion, diffusion MRI, diffusion tensor imaging, brain neoplasms (Top Magn Reson Imaging 2004;15:315–324) I n routine clinical neuroimaging, even the most anatomically detailed MR images do not allow an assessment of specific white matter (WM) fiber tracts. Diffusion tensor imaging (DTI) has recently enabled unprecedented, in vivo visualization of individual WM tracts and their relationships with cerebral neoplasms. While much of the work in this field remains ex- perimental, DTI is currently making its way into the clinical realm. It is no longer necessary to have a team of physicists and image-processing specialists on hand to obtain useful DTI in a clinical setting (although some of the more sophisticated techniques still have this requirement). As DTI makes further inroads into clinical practice, an understanding of basic prin- ciples and applications will become increasingly important to neuroimaging specialists. This article begins with a brief review of the basic principles underlying DTI, followed by a review of DTI postprocessing strategies relevant to tumor imaging. Space limitations prohibit a review of normal WM fiber tract anatomy, but several references to DTI-specific anatomic reviews are provided. Finally, the role of DTI in the diagnosis and therapy for brain tumors is reviewed, including discussions on typical patterns of WM tract alteration, quantitative approaches to DTI-based tumor characterization, and pre- operative planning issues. BASIC PRINCIPLES OF DTI MRI may be sensitized to the random, thermally driven motion (diffusion) of biologic water molecules in a specified direction by applying a matched pair of dephasing and rephasing magnetic field gradients in that direction. 1–3 This gradient pulse configuration is often referred to as Stejskal- Tanner diffusion weighting. 1 The ability of the rephasing gradient to refocus the signal from the prior dephasing gra- dient diminishes as protons diffuse to new positions in the gradient direction during the intervening time. Therefore, the measurement of signal loss or attenuation is a function of the diffusivity in a chosen direction. This signal attenuation is exponential with the level of diffusivity, S = S o e 2bD , where D is the estimated diffusivity or apparent diffusion coefficient (ADC), S is the diffusion-weighted signal, S o is the signal without diffusion-weighting gradients, and b is the diffusion weighting, which is a function of the amplitude and timing of the diffusion gradients. A useful measure of overall diffusivity that is independent of direction can be obtained by taking the mean of the ADCs for three orthogonal directions. This directionally averaged ADC is commonly known as mean diffusivity. The most widely used diffusion-weighted image acqui- sition method is single-shot echo-planar imaging (EPI) 4,5 be- cause it is fast, efficient, and insensitive to small motion, as well as readily available on most clinical MRI scanners. How- ever, the long readout time for EPI makes the images highly sensitive to magnetic field inhomogeneities 6 and eddy currents, 7 both of which cause distortions in the image data. Alternative scanning approaches include multiple-shot techniques, such as multi-shot EPI with navigator echo correction 8,9 or diffusion- weighted PROPELLER, 10 and parallel imaging methods, such as SENSE. 11,12 Descriptions of these and other methods may be found in a recent review article by Bammer. 13 In WM fiber tracts, organized bundles of axonal membranes and myelin sheaths present substantial barriers to diffusion of water molecules, especially in directions per- pendicular to that of the fibers. These barriers render the diffusion highly anisotropic (directionally dependent) in WM fiber tracts and the direction of maximum diffusivity has been shown to coincide with tract orientation. 14 This information is contained in the diffusion tensor, a mathematical model of diffusion in three-dimensional (3D) space. The diffusion tensor From the *Departments of Radiology and Biomedical Engineering, University of Wisconsin, Madison, WI; and †Departments of Medical Physics and Psychiatry, University of Wisconsin, Madison, WI. Supported in part by NIH grant no. EB002012. Reprints: Aaron S. Field, MD, PhD, Department of Radiology, Universityof Wisconsin Hospital and Clinics, 600 Highland Avenue, E3/311 CSC, Madison, WI 53792-3252 (e-mail: as.field@hosp.wisc.edu). Copyright Ó 2004 by Lippincott Williams & Wilkins Top Magn Reson Imaging  Volume 15, Number 5, October 2004 315 JOBNAME: tmri 15#5 2004 PAGE: 1 OUTPUT: Thu December 16 13:53:26 2004 lww/tmri/89799/V15N501