Practice Essentials
Diffusion-weighted imaging (DWI) is a well-established magnetic resonance imaging (MRI) method for diagnosing cerebral ischemia. Diffusion-weighted imaging is a routine protocol used in most institutions that perform neuroimaging; normal states and abnormal conditions are easily interpreted through the use of DWI in conjunction with apparent diffusion coefficient (ADC) imaging by correlating findings of hyperintensity on DWI images with findings of hypointensity on ADC images (see the images below). [1, 2] Imaging and interpretation of water diffusion have improved with the development of diffusion tensor imaging (DTI). Diffusion tensor imaging allows direct in vivo examination of aspects of tissue microstructure and takes advantage of diffusion anisotropy to provide excellent details of the brain; for example, it enables mapping of the orientation of white matter tracts.
Diffusion tensor imaging has shown strong momentum in research performed to evaluate the neural fibers of the central nervous system (CNS). This technique is used to study white matter (WM) microstructure in neurodegenerative disorders, including Parkinson disease (PD). [3] DTI has also improved the knowledge of brain microstructure in migraine headache, [4] as well as revealing opposing patterns of diffusion changes in frontal and cerebellar regions of COVID-19 patients, suggesting the 2 regions react differently to viral infection. [5]

Diffusion-weighted imaging has been shown to be useful in the investigation of brain disorders such as epilepsy, multiple sclerosis, brain abscesses, brain tumors, mild traumatic brain injury, and hypertensive encephalopathy (see the images below). [6, 7, 8, 9, 10, 11]


Diffusion, or brownian movement, denotes the random motion of molecules. All molecules exhibit such motion at temperatures greater than absolute zero. Diffusion is termed isotropic if the motion is equal in all directions. However, water diffuses asymmetrically in the white matter—that is, diffusion is restricted perpendicular to the long axis of the axons. By contrast, water diffuses faster along the Z axis (see the image below). This property, which is known as anisotropy, may be used to define the direction of axons in a particular voxel. [12]

Diffusion in structured tissue such as white matter is anisotropic. Diffusion tensor imaging can be used to measure anisotropy per voxel and provides directional information relevant for magnetic resonance tractography or fiber tracking in vivo. Diffusion tensor imaging allows direct examination of the brain microstructure and has become a useful tool for investigating brain disorders such as stroke, epilepsy, multiple sclerosis (MS), brain tumors, and demyelinating and dysmyelinating disorders. However, further improvements in this technique and in postprocessing analysis are needed to increase the widespread utility of DTI in research and in clinical applications. [13, 14, 15, 16]
Tensor and Diffusion Ellipsoid
On DWI commonly used to diagnose acute stroke (see the image below), diffusion is described by using the apparent diffusion coefficient (ADC). This is sufficient for pathologies in areas such as gray matter, where diffusion is usually isotropic.

To measure the presence of anisotropy in the white matter requires a tensor D, which describes the mobility of molecules in a particular direction and correlation between these directions. The tensor is symmetric; at least 6 elements are required to characterize it.
The diffusion ellipsoid defines the magnitude and direction of the diffusion of water molecules in each voxel in the brain. The tensor may be diagonalized such that 3 elements, called eigenvalues, remain along the diagonal. Three eigenvalues—lambda 1, lambda 2, and lambda 3—are derived.
Diffusion tensor imaging allows clinicians to look at anisotropic diffusion in white matter tracts, but it is limited in its ability to demonstrate spatial and directional anisotropy. Advanced methods such as color coding and tractography (fiber tracking) have been used to investigate directionality.
The eigenvector corresponding to the largest eigenvalue, termed the principal eigenvector, defines the main direction of diffusion of water molecules in that voxel (see the image below).

Mapping the principal directional eigenvectors in each voxel forms the basis for tractography (see the images below); the assumption is that the principal eigenvector is aligned with the direction of the fiber bundle. On these images, fibers are given different colors according to their direction of diffusion: blue for superior and inferior; green for anterior and posterior; and red for left and right.


Central nervous system tuberculomas often mimic tumors on conventional imaging; differentiation may not be possible without invasive tissue sampling. Diffusion tensor imaging, because of its ability to characterize molecular diffusion, may provide better lesion characterization, and tractography may enhance understanding of the pattern of white matter involvement by tuberculomas. [17]
Significant correlation is found between DTI parameters and various spinal injury scores. Furthermore, DTI can facilitate accurate lesion mapping and assessment of cord changes distant from an injury epicenter. [18]
Diffusion tensor tractography (DTT) has been shown to be an appropriate imaging technique for elucidating recovery mechanisms of aphasia in language-related neural tracts in stroke patients. [19]
MRI Technique
Brain magnetic resonance imaging (MRI) should be performed with a 1.5- or 3-T MRI machine. High gradient strength in the range of 20 to 60 mT/m with a slew rate of 120 T/m/s is ideal. Typical parameters for a single-shot spin-echo echo-planar imaging (EPI) sequence include repetition time (TR) of 6000 ms, echo time (TE) of 100 ms, and a field of view of 24 cm to obtain 3- to 5-mm axial or coronal sections with a 5-mm intersection gap. The acquisition matrix is 96 × 96 with a reconstruction matrix of 128 × 128. Images are obtained by using 4 linearly increasing b values in 6 to 7 noncolinear directions (bmax = 703-1000 s/mm2). In addition, a T2-weighted (T2W) image is obtained without diffusion weighting (b = 0 s/mm2).
Image Interpretation
A prudent approach to image interpretation is to use an image workstation different from the one used to acquire the images. Motion artifacts and image distortion may be corrected by using a coregistration program and filtration. [20]
Diffusion tensor measurements result in a rich data set. Diffusion anisotropy may be measured by applying simple or complicated mathematical formulas. However, an easy and common way to summarize diffusion measurements on diffusion tension images (DTIs) is to calculate parameters for overall diffusivity and for the degree of anisotropy. [21]
Imaging findings include the apparent diffusion coefficient (ADC), which is a measure of the magnitude of molecular motion divided by overall diffusivity; fractional anisotropy (FA), which is the measure of the portion of the diffusion tensor that results from anisotropy (ie, a measure of the directionality of the molecular motion of water); relative anisotropy (RA), or the ratio between anisotropic and isotropic portions of the diffusion tensor; and volume ratio (VR), which expresses the relationship between the diffusion ellipsoid volume and that of a sphere, the radius of which is the averaged diffusivity. [22]
Maps of both FA and RA may be presented as gray scale images. Maps of mean diffusivity and FA may be generated by using Pierpaoli and Basser's method on a pixel-by-pixel basis. Regions of interest (ROIs) are placed on both maps to calculate diffusivity and FA.
FA is sensitive to low values of diffusion anisotropy; VR is sensitive to high values of diffusion anisotropy; and RA is linearly scaled for different levels of anisotropy. Both FA and RA vary from 0 (isotropic) to 1 (anisotropic). The values of measurements in pediatric brains markedly differ from those in adult brains; values vary with increasing age. Mean diffusivity is approximately 0.7 × 10-3 in adults and 2 × 10-3 in neonates. Because anisotropy is greater in ordered structures such as myelinated axons, DTIs provide useful information regarding the myelination of white matter.
In a study conducted to measure developmental changes and sex and hemispheric differences of neural fibers in white matter, 52 healthy persons ranging in age from 2 months to 52 years underwent DTI for measurement of FA, ADC, axial diffusivity (AD), and radial diffusivity (RD). The tracts of interest (TOI) followed were the corpus callosum (CC), the cingulum hippocampus (CGH), the inferior longitudinal fasciculus (ILF), and the superior longitudinal fasciculus (SLF). Investigators found that for all TOIs, FA increased with age, whereas ADC, AD, and RD values decreased with age. In infants, growth rates of both FA and RD were larger than those of AD. According to study authors, developmental patterns differ by TOIs and by myelination, along with the development of white matter, which can be mainly expressed as an increase in FA and a decrease in RD. [23]
In many pathologic conditions, FA and ADC vary because of altered diffusivity and disorganization of white matter fibers, leading to decreased anisotropy. These measurements may become abnormal even before the lesion is morphologically apparent on conventional magnetic resonance imaging, and this may facilitate early detection and definition of lesion extent. [24]
Fractional anisotropy and ADC may vary independently. This may be explained by the fact that damaged or malformed brain has glia and neurons, respectively. Therefore, cell density is sufficient to prevent effects on ADC; however, as the result of disorganization, FA is reduced.
Artifact
The main artifacts in DTI data are associated with acquiring DWI data from which the diffusion tensor is estimated or measured. Artifacts include misregistration of data, which occurs as the result of eddy currents; ghosting, which is caused by motion artifacts; and signal, which reflects susceptibility variations. These artifacts may be minimized by using motion-corrected multishot echo-planar imaging (EPI) techniques, such as periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) and sensitivity-encoding EPI (SENSE-EPI). [25, 26, 27]
Clinical Applications
Specific qualitative features established for conventional magnetic resonance imaging (MRI) may be used to distinguish normal from abnormal brain development. Changes in the apparent diffusion coefficient (ADC) occur predominantly in the first 6 months of life and are believed to be related to decreasing total water content, myelination, and organization of white matter tracts. All of these processes decrease diffusivity. Because the diffusion tension imaging (DTI) technique allows improved objectivity and sensitivity in detection of subtle developmental changes, it may prove to be more useful than the relatively subjective evaluation attained with conventional MRI sequences. [28]
Stroke
There is ongoing research in identification of stroke recovery biomarkers, including those provided by structural neuroimaging techniques such as DTI and tractography for the study of white matter (WM) integrity. [29]
Diffusion-weighted imaging and DTI have been extensively used to detect acute ischemic brain injury (see the image below).

In the acute phase of ischemia, ADC is reduced and FA values are increased. In the chronic phase of ischemia, ADC is higher than normal. In contrast to the elevation of ADC seen in chronic stroke, diffusion anisotropy remains significantly lower in the infarcted area than in the similar contralateral region of the brain, even 2 to 6 months after an ischemic stroke. With combined ADC and anisotropy data, the severity of strokes may be assessed, and acute ischemic changes may be distinguished from chronic ischemic changes; this difference may affect treatment decisions. Also, ADC values are increased when purely vasogenic edema is present—for example, in the reversible posterior leukoencephalopathy syndrome or in high-pressure hydrocephalus. [30, 31]
Epilepsy
A common cause of epilepsy is mesial temporal sclerosis or hippocampal sclerosis in patients with chronic epilepsy. Findings seen on diffusion tensor imaging (DTI) include increased diffusivity and decreased anisotropy caused by loss of structural organization and expansion of the extracellular fluid space. Changes in DTI may extend to areas of the brain that appear morphologically normal on conventional MRI. In this way, DTI may define the true extent of pathology and may improve preoperative planning.
Refractory extratemporal neocortical epilepsy may be caused by malformations in cortical development (MCDs); these malformations may not be apparent on conventional MRI. However, differences in ADC in the affected region, as compared to the contralateral normal brain, may be seen; this helps in presurgical planning. Diffusion tensor imaging may erroneously depict regions of presumed seizure onset by showing subtle structural abnormalities caused by head injury or ischemia.
Kung et al explored the possibility of using DTI and neurite orientation dispersion and density imaging (NODDI) to discern microstructural abnormalities in the hippocampus indicative of mesial temporal sclerosis at the subfield level. The study results suggest that diffusion metric analysis at the subfield level, especially in dentate gyrus and CA1, may be useful for clinical confirmation of mesial temporal sclerosis. [32]
Brain tumors
Diffusion tensor imaging has shown potential for distinguishing gliomas and solitary metastases in the brain parenchyma. Significantly higher mean diffusivity and lower FA, as compared to levels in normal-appearing white matter, have been observed in the peritumoral regions of both gliomas and metastases.
Peritumoral mean diffusivity of metastasis and meningioma (see the image below) is significantly higher than that of glioma, whereas FA values are similar, confirming the infiltrative nature of glioma. Tractography combined with functional MRI may facilitate preoperative planning for brain tumor by mapping areas of active infiltration. [33, 34]

Multiple sclerosis
Various studies have demonstrated potential advantages of DTI in the diagnosis and follow-up of multiple sclerosis (MS) lesions. In MS, FA values are more sensitive than ADC values with regard to white matter abnormalities. Lesions with destructive pathology or acuity generally have increased diffusivity and decreased FA values. On conventional T2-weighted and fluid-attenuated inversion recovery (FLAIR) images, normal-appearing white matter adjacent to MS lesions may reveal abnormality; thus, the actual extent of the lesions becomes apparent. In some cases, the gray matter around white matter lesions is abnormal; this indicates that disease may not be isolated to the white matter. [35]
Alshehri et al compared DTI metrics in relapsing-remitting MS patients (RRMS) versus healthy controls (HCs) and explored correlations between DTI metrics, total brain white matter (TBWM), and white matter lesions (WML) with clinical parameters versus volumetric measures. They identified a correlation of DTI metrics with clinical symptoms of MS—in particular, cognition. [36]
Demyelinating versus dysmyelinating disorders
Diffusional anisotropy is present in dysmyelinating disorders such as Pelizaeus-Merzbacher disease; by contrast, it may be lost in demyelinating disease such as Krabbe disease or Alexander disease. [37] In contrast to relatively high signal intensity of lesions in Krabbe disease on DWI, lesions in Alexander disease have signal intensity. Therefore, DWI is clinically useful in differentiating dysmyelination from demyelination, as both have lesions of high intensity in the white matter, as shown on T2-weighted images.
Alzheimer disease
Diffusion tensor imaging is widely used in exploring the central nervous system mechanism and in finding appropriate potential biomarkers for early stages of AD. Diffusion features and structural brain cell connections can provide information for early recognition of AD. [38]
Tractography
Tractography potentially solves a problem for a neurosurgeon in terms of minimizing functional damage and determining the extent of diffuse infiltration of pathologic tissue to minimize residual tumor volume. In this way, tractography facilitates preoperative planning. Tractographic images (see the images below) may help clarify whether a tumor is compressing, abutting, or infiltrating contiguous white matter tracts. However, no consensus has been reached about an appropriate criterion standard for assessing the accuracy of DTI. [39]

-
Normal brain appearance in axial DWI (left) and ADC (right) images in a 35-year-old man.
-
Axial DWI image demonstrates a typical wedge-shaped, cortical-based, hyperintense lesion in the left temporoparietal lobes consistent with acute infarct.
-
Axial fluid-attenuated inversion recovery (FLAIR) image (top left), diffusion-weighted image (DWI) (top right), and axial and sagittal T1-weighted (T1W) images (bottom) in a 40-year-old man with a history of intravenous drug abuse and fever demonstrate multiple enhancing focal brain lesions in the gray-white matter junction (arrow) compatible with septic emboli. The lesions are hyperintense on both FLAIR and DWI images.
-
Axial diffusion-weighted image (DWI) (top left), T2-weighted (T2W) image (top right), fluid-attenuated inversion recovery (FLAIR) image (bottom left), and contrast-enhanced T1W (bottom right) image demonstrate a right convexity meningioma, which appears hypointense on DWI image. The perilesional brain edema (arrow) is hyperintense on T2W and DWI sequences.
-
Diffusion ellipsoid. Three eigenvectors are demonstrated, with the principal eigenvector along the Z direction. Courtesy of Dr Andrei I. Holodny, MD.
-
Axial tractographic image demonstrates white-matter tracts in the brain in the left-right (red), anterior-posterior (green), and superior-inferior (blue) directions. Courtesy of Dr Andrei I. Holodny, MD.
-
Coronal tractographic image demonstrates various nerve-fiber tracts. Courtesy of Dr Andrei I. Holodny, MD.