Functional Neuroradiology: Principles and Clinical Applications

By Scott H. Faro and Meng Law

Functional Neuroradiology: Principles and Clinical Applications, is a follow-up to Faro and Mohamed’s groundbreaking work, Functional (BOLD)MRI: Basic Principles and Clinical Applications.  This new 49 chapter textbook is comprehensive and offers a complete introduction to the state-of-the-art functional imaging in Neuroradiology, including the physical principles and clinical applications of Diffusion, Perfusion, Permeability, MR spectroscopy, Positron Emission Tomography,  BOLD fMRI  and  Diffusion Tensor Imaging.

With chapters written by internationally distinguished neuroradiologists, neurologists, psychiatrists, cognitive neuroscientists, and physicists, Functional Neuroradiology is divided into 9 major sections, including: Physical principles of all key functional techniques, Lesion characterization using Diffusion, Perfusion, Permeability, MR spectroscopy, and Positron Emission Tomography, an overview of BOLD fMRI physical principles and key concepts, including scanning methodologies, experimental research design, data analysis, and functional connectivity, Eloquent Cortex and White matter localization using BOLD fMRI and Diffusion Tensor Imaging, Clinical applications of BOLD fMRI in Neurosurgery, Neurology, Psychiatry, Neuropsychology, and Neuropharmacology, Multi-modality functional Neuroradiology, Beyond Proton Imaging, Functional spine and CSF imaging, a full-color Neuroanatomical Brain atlas of eloquent cortex and key white matter tracts and BOLD fMRI paradigms.

By offering readers a complete overview of functional imaging modalities and techniques currently used in patient diagnosis and management, as well as emerging technology, Functional Neuroradiology is a vital information source for physicians and cognitive neuroscientists involved in daily practice and research.

• Language: English
• Publisher: Springer
• Release date: Sep 8, 2011
• ISBN: 9781441903457

Book preview

Scott H. Faro, Feroze B. Mohamed, Meng Law and John T. Ulmer (eds.)Functional NeuroradiologyPrinciples and Clinical Applications10.1007/978-1-4419-0345-7_1© Springer Science+Business Media, LLC 2011

1. Physical Principles of Diffusion Imaging

Thinesh Sivapatham¹   and Elias R. Melhem²  

(1)

Department of Radiology, Hospital of the University of Pennsylvania, 3400 Spruce Street, 2 Dulles, Philadelphia, PA 19104, USA

(2)

Department of Radiology, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104, USA

Thinesh SivapathamFellow, Diagnostic Neuroradiology (Corresponding author)

Email: [e-mail]

Elias R. MelhemVice-Chair of Academic Affairs

Email: [e-mail]

Abstract

Diffusion-weighted imaging (DWI) utilizes the constant random motion of water molecules, called Brownian motion, to depict the movement or diffusion of water in tissue structures. The diffusion of water molecules in the brain provides us with a sensitive window to its underlying physiology and structure. DWI of the brain was introduced into clinical use in the early 1990s, primarily in the detection of acute ischemic stroke. Since that time, advances in technology have resulted in significant improvements in image quality, allowing the application of DWI to the evaluation of a variety of intracranial disease processes, such as infections, neoplasms, demyelinating processes, and trauma. In this chapter, we review the physical principles of DWI, starting with a description of Brownian motion and its relevance to molecular diffusion. We then describe the application of these principles to DWI of the brain using magnetic resonance imaging (MRI). We discuss basic imaging techniques in DWI of the brain, as well as limitations of current techniques, and newer imaging sequences that have been developed. The clinical applications of DWI are discussed in the following chapters.

Introduction

While a still body of water may appear static to the eye, water molecules are in constant random motion at a microscopic level. This is termed Brownian motion, and is a result of the thermal agitation of the water molecules. While Brownian motion is a microscopic phenomenon, it results in a macroscopically observable phenomenon known as diffusion. The diffusion of water molecules in the brain provides us with a sensitive window to its underlying physiology and structure. Diffusion-weighted imaging (DWI) of the brain was introduced into clinical use in the early 1990s, primarily in the detection of acute ischemic stroke [1–4]. Since that time, advances in technology have resulted in significant improvements in image quality, allowing the application of DWI to the evaluation of a variety of intracranial disease processes, such as infections, neoplasms, demyelinating processes, and trauma. The development of diffusion tensor imaging (DTI) has allowed mapping of white matter tracts in the brain and is discussed elsewhere in this book. In this chapter, we review the physical principles of DWI, while the clinical applications of DWI are discussed in the following chapters.

Brownian Motion and Molecular Diffusion

Robert Brown, a Scottish botanist, discovered in 1827 the random and constant motion of pollen grains suspended in water while studying them under a light microscope [5]. Brown initially believed this motion to be related to the male sexual cells of plants, but after further investigation of both organic and inorganic materials, he found this motion to be a general property of small particles suspended in solution. We, currently, know that this motion is attributed to the constant motion of the water molecules that the particles are suspended in. This motion is known as Brownian motion, or diffusion, and is a result of constant random microscopic molecular motion due to thermal agitation. The molecular motion is related to the thermal kinetic energy (E kin) of the molecules, which is proportional to the temperature, T: E kin = (3/2)k B T; where, k B = 1.38 × 10−23 J/K is the Boltzmann constant. This Brownian motion also depends upon the size of the particles, density, and viscosity of the medium.

Diffusion is a naturally occurring transport process at the molecular level that describes the spread of particles through random motion. This mixing or mass transport occurs as a result of collisions between molecules in liquids and gasses rather than by bulk motion, as is necessary for other transport mechanisms such as convection or dispersion. When there is a concentration gradient, particles spread from areas of higher concentration to areas of lower concentration until their distribution becomes equilibrate. This results in a net flux of particles with a magnitude that is proportional to the concentration gradient and to the diffusion coefficient, as described by Fick’s First Law and represented by the ­following equation:

$$ J=D*\nabla C$$

(1.1)

where J is the flux density, D is diffusion coefficient, and ÑC is change in concentration. The diffusion coefficient is constant for a given substance in a given medium of known ­viscosity at a given temperature. Diffusion can also occur in the absence of a concentration gradient, but in this scenario the diffusive fluxes cancel each other out, resulting in no net flux.

The concept of molecular diffusion was first formally described by Einstein in 1905 [6]. In a contained volume of water, each water molecule undergoes random motion as part of the diffusion process. This phenomenon of thermal motion of water molecules in a medium that itself consists primarily of water is termed self-diffusion. The path of any single water molecule is completely random and would be impossible to predict, limited only by the boundaries of the container. Einstein was able to prove that the squared displacement of molecules from their starting point could be described by the equation:

$$ <{r}^{2}>=6Dt$$

(1.2)

where refers to the mean squared displacement of the molecules from their original location, t is the diffusion time, and D is the diffusion coefficient for the particular substance being studied. The diffusion coefficient is typically expressed in units of mm²/s and relates the average displacement of a molecule over an area to the observation time; higher diffusion constants infer increased mobility of water molecules. The diffusion coefficient of water at body temperature (37°C) is 3 × 10−3 mm²/s. The distribution of squared displacements of free water molecules is typically a Gaussian (bell-shaped) function with peak at zero displacement, indicating that most molecules have the same position at the starting point and at time t [7, 8]. The probability of molecular displacement by a given distance from the starting position is the same regardless of the direction of measurement, with standard deviation proportional to the diffusion coefficient and time measured.

Diffusion-weighted (DW) magnetic resonance imaging (MRI) utilizes Brownian motion to study the movement of water in vivo, thereby garnering information about the underlying tissue structure of the brain. DW MRI does not measure the diffusion coefficient directly, but rather the effect of the mean displacement of water molecules within each three-dimensional (3D) volume element, or voxel, on the nuclear magnetic resonance (NMR) signal. One might ask that if the diffusion coefficient of water at body temperature is a constant, then how we can use this information to evaluate tissue structure. We must remember that in vivo, water molecules do not freely diffuse as they would in a medium of water alone. In tissues, the movement of water occurs largely in the extracellular space, and their movement is modified by interactions with cell membranes and macromolecules, i.e., the underlying tissue structure of the brain. Additionally, the movement of water molecules in vivo due to diffusion cannot be distinguished from the movement of water molecules from other sources such as active transport, pressure gradients, ionic interactions, and changes in membrane permeability. As a result, the overall movement of water molecules in the brain is reduced as compared to their movement in free water, and only the apparent diffusion coefficient (ADC) can be calculated [7–9]. The average ADC in the brain is approximately 0.7 × 10−3 mm²/s [10], which is about four times smaller than the diffusion coefficient in free water.

Diffusion Encoding and the Stejskal–Tanner Equation

Shortly after Bloch and Purcell discovered the NMR ­phenomenon [11–13], Hahn reported his findings that the NMR spin echo was sensitive to the effects of diffusion [14]. He noted that the random thermal motion of the spins would reduce the amplitude of the observed spin echo signal as a result of the dephasing that occurs in the presence of magnetic field inhomogeneity, which results in local magnetic field gradients. Building on these observations, Carr and Purcell proposed NMR sequences using constant gradients to sensitize the NMR spin echo to the effects of diffusion and developed a mathematical framework to measure the diffusion coefficient from these sequences [15]. In 1956, Torrey modified Bloch’s magnetization equations to include the effects of molecular diffusion [16]; the Bloch–Torrey equations describe how net magnetization depends on several factors, including longitudinal and transverse magnetization, as well as diffusion.

In their seminal paper on the spin diffusion experiment in 1965, Stejskal and Tanner described the theory of the pulsed gradient spin echo (PGSE), which replaced the steady state gradients used by Carr and Purcell with short duration gradient pulses [17]. This resulted in much improved sensitivity to diffusion by distinguishing the encoding time (pulse duration) and the diffusion time. This also allowed the direct measurement of the diffusion function and paved the way for quantitative measurements of the diffusion coefficient. In the presence of a magnetic field, static spins accumulate phase shifts according to the equation:

$$ \Phi (t)=\gamma B0t+{\displaystyle \int G(t)X(t)\text{d}t}$$

(1.3)

The first term (γB0t) in the equation represents the phase accumulation owing to the static magnetic field, while the second term (∫G(t).X(t)dt) reflects the effect of a magnetic field gradient. The phase accumulation owing to the gradient is proportional to the strength of the field gradient, spatial location of the spin, and the duration of the gradient pulse.

In their experiment, Stejskal and Tanner sensitized a standard spin-echo T2-weighted pulse sequence to diffusion by adding a pair of diffusion gradients along the same axis both before and after the 180° refocusing pulse [17]. The basic idea is to excite the spins with a 90° radiofrequency (RF) pulse, encode the spin position with a time-constant magnetic field gradient of duration δ, invert the spin phase with a 180° refocusing pulse, apply a second magnetic field gradient equal in intensity and duration to the first gradient at a time Δ after the first gradient, then acquire the echo at time TE (echo time). The application of a linear gradient to a homogeneous magnetic field results in phase shifts of the spins along that axis, which is position-dependent; this phase dispersion leads to signal loss. However, the application of a second gradient equal to the first in magnitude and duration but opposite in direction can refocus the phase changes. The first gradient is called the dephasing gradient, while the second is called the rephasing gradient [18].

If a spin is stationary between the applications of the two gradient pulses, the net effect of the gradients is zero and the spin maintains its signal intensity. This explains the relatively high signal intensity on DWI seen in the setting of cytotoxic edema, where there is a relative increase in intracellular water content and water molecules are trapped inside the cells (i.e., unable to diffuse freely) and relative decrease in the size of the extracellular compartment [19–25]. For a spin that moves along the axis of the diffusion gradient, phase accumulation due to the two gradients is no longer equal, so the rephasing is incomplete. As a result, intravoxel dephasing occurs and there is a loss of signal intensity; the greater the distance the spin moves along the axis of the gradient, the greater is the signal loss [8, 14, 16, 17]. This explains the relatively low signal intensity seen on DWI in the normal brain parenchyma, and even lower signal intensity in the cerebrospinal fluid (CSF) spaces, where water can diffuse the most freely.

The effects of signal loss due to diffusion can be explained by the Stejskal–Tanner equation:

$$ S={S}_{0}\mathrm{exp}(-bD)$$

(1.4)

where S is the signal intensity observed in a given voxel with a diffusion gradient applied, S 0 is the signal intensity of the same voxel in the absence of a diffusion gradient, b is the diffusion sensitivity factor, and D is the diffusion coefficient. The diffusion sensitivity factor (b) is a function of the strength, duration, and temporal spacing of the diffusion-sensitizing gradients and can be expressed by the following equation:

$$ b={g}^{2}{G}^{2}{d}^{2}(\Delta -d/3)$$

(1.5)

In this equation, γ is the gyromagnetic ratio (the ratio of magnetic moment to angular moment of a nuclear spin, a constant), G is the amplitude of the diffusion gradient (usually measured in milliteslas per meter), δ is the duration of the diffusion gradient pulse (measured in milliseconds), and Δ the time interval between the dephasing and rephasing gradient pulses (also measured in milliseconds). The diffusion sensitivity factor (b) is measured in units of seconds/mm². Raising the b-value increases the degree of diffusion weighting (i.e., increases the signal loss caused by the diffusion of water molecules along the direction that the gradient was applied; Fig. 1.1). The most effective ways to increase diffusion sensitivity, as seen in (1.3), are to increase gradient amplitude (G) and gradient duration (δ), as these parameters have a quadratic effect on b. Typical b values in clinical DWI range from 0 to 1,500 s/mm². However, it should be emphasized that higher diffusion weightings increase exponentially the contrast between tissues with different diffusion coefficients, but at the same time they decrease the overall signal intensity and signal-to-noise ratio (SNR) [26, 27].

A183863_1_En_1_Fig1_HTML.jpg

Fig. 1.1

Diffusion-weighted images acquired using diffusion sensitivity factors (b) of increasing strength. The b = 0 images (top row) demonstrate the T2-weighting inherent in DWI. As the diffusion weighting is increased (b = 500 and b = 1,000, second and third rows, respectively), the diffusion of water molecules results in signal loss, most apparent in the CSF where water can diffuse freely. Corresponding ADC map is shown in the bottom row

Diffusion Anisotropy

Isotropic diffusion refers to a condition where molecular motion is the same in all directions. Since the diffusivity is independent of direction in this scenario, the displacement distribution is Gaussian and can be conceptualized as a sphere (i.e., if the center of the sphere is the starting point of a water molecule, the probability of diffusing any given distance from the center is equal in all directions) [7, 8, 28]. Biological tissues are heterogeneous in their structure and consist of various compartments and barriers of various permeabilities. Neuronal tissue, in particular, is highly fibrillar, with tightly packed and coherently aligned axons surrounded by a network of glial cells that are often themselves organized in bundles. The movement of water molecules in these tissues is not isotropic, but rather has a propensity to be greater in a direction parallel to the fiber tracts than perpendicular to it. Diffusion in this scenario varies with direction, and is described as anisotropic [29–33]. The associated displacement distribution of anisotropic diffusion is not Gaussian or spherical, but rather ellipsoid (or even more complex if the underlying tissues contain fibers with various orientations). This anisotropy can be explained by the diffusion tensor model, and is discussed further in the chapter on DTI.

While isotropic diffusion can be seen in the CSF spaces and in the gray matter of the adult cerebral cortex, diffusion in the white matter is primarily anisotropic. Multiple explanations have been suggested for the mechanisms underlying diffusion anisotropy of white matter, including the myelin sheath, axonal direction, axonal transport, and local susceptibility gradients [34–36]. Myelin itself does not appear to be necessary for diffusion anisotropy, as experiments have shown anisotropic diffusion in the absence of myelin [37, 38]. A series of experiments in the 1990s excluded the effects of susceptibility-induced gradients, axonal cytoskeleton, and fast-axonal transport as the etiology of anisotropy in white matter [39–41]. Current evidence suggests that the presence of intact cell membranes in the tissue component is predominantly responsible for the anisotropy of molecular diffusion in white matter, while the degree of myelination and cellular density serve to modulate anisotropy [42].

Calculating the ADC

Since diffusion in the brain is approximated by the ADC rather than by direct measurement of the diffusion coefficient, (1.2) might be better represented as:

$$ {S}_{i}={S}_{0}\mathrm{exp}(-b{\text{ADC}}_{i})$$

(1.6)

where S i is the signal intensity in a given voxel with the ­diffusion sensitization gradient applied along direction i, and ADC i is the ADC in the i direction. In an isotropic environment, the direction of the gradient pulse is irrelevant, since the ADC i would be identical for all directions i, and a single (scalar) diffusion gradient application is sufficient to calculate the ADC. Higher ADC values result in lower signal intensity S i in the DWI, while reduced ADC values result in higher signal intensity. In addition to contrast due to differences in ADC, the long TE of a DWI pulse sequence makes it inherently sensitive to T2 contrast as well. T2 prolongation in pathologic tissues can elevate the DWI signal intensity in the absence of reduced ADC values [43]. This T2 shine-through effect results in the DWI hyperintensity being less specific than reduced ADC values on ADC maps in the measurement of true restricted diffusion in tissues (Fig. 1.2).

A183863_1_En_1_Fig2_HTML.jpg

Fig. 1.2

Images from a patient who presented with an acute stroke demonstrate the T2 shine-through effects of DWI. The FLAIR image (d) shows multifocal areas of T2 prolongation in the periventricular and deep white matter (arrow and arrowhead). The lesion marked by the arrow in (d) also demonstrates increased signal on the b = 0 (a), b = 1,000 (b), and ADC (c) images, indicating that this is not an area of restricted diffusion, but rather T2 shine-through on DWI. The lesion marked by the arrowhead in (d), on the contrary, also demonstrates marked signal intensity on the b = 0 and b = 1,000 images, but is dark on the ADC map, indicating that this is an area of restricted diffusion

In order to remove the T2 effects, the diffusion experiment is repeated for each gradient direction with a low b-value and high b-value (commonly 0 s/mm² and 1,000 s/mm² in clinical practice). The low b-value sequence essentially yields a T2-weighted image. The DW image S i (with b = 1,000 s/mm²) can then be divided by the T2-weighted image S 0 (with b = 0 s/mm²) to remove the T2-weighting effects and produce an exponential image, also known as an attenuation coefficient map, exponential DWI, or the attenuation factor map (shown in Fig. 1.3). Areas of restricted ­diffusion on these maps demonstrate increased signal intensity, similar to DWI.

A183863_1_En_1_Fig3_HTML.jpg

Fig. 1.3

Diffusion anisotropy. The top three rows show ADC maps derived from diffusion gradients applied in three orthogonal directions. Variations in brightness between the three sets of images demonstrate the anisotropic nature of diffusion in the brain; diffusion of water molecules is greater along nerve bundles than perpendicular to them. When the gradient is applied in the x direction (Dxx; diffusion gradient in the left to right direction), diffusion is higher (brighter signal) in fibers that course in the left to right direction (i.e., splenium) than perpendicular to that direction (i.e., posterior limb internal capsule, and anterior-to-posterior oriented white matter tracts of the cerebral hemispheres). With gradient applied in the y direction (Dyy; anterior-to-posterior), diffusion is greatest in the anterior-to-posterior oriented cerebral white matter tracts, and lowest in the perpendicularly oriented splenium and posterior limb internal capsule. With gradient applied in the z direction (Dzz; cranial to caudal), diffusion is greatest in the posterior limb internal capsule oriented in the same direction, and lowest in the perpendicularly oriented splenium and anterior-to-posterior cerebral white matter tracts. The trace ADC images shown in the fourth row are the average of the ADCs from each of the three directions, and show diffusion magnitude information alone with directional information removed. The fifth row shows the corresponding isotropic DWI images. The exponential DWI images (Exp, bottom row) were derived by dividing the b = 1,000 images by the b = 0 images (with b measured in seconds/mm²) to remove the T2-weighting effects

ADC values can also be calculated by solving for ADC in (1.7):

$$ {\text{ADC}}_{i}=-\mathrm{ln}({S}_{i}/{S}_{0})/b$$

(1.7)

where, ln is the natural logarithm. Instead of deriving the ADC value mathematically, however, the ADC is typically determined graphically by obtaining two image sets (again, low b-value and high b-value image sets) and plotting the natural logarithm of S i versus b for the two b values; the ADC is then determined from the slope of that line.

Biological tissue can be considered a combination of intra- and extracellular compartments in which the water molecule is in a state of continuous exchange between these two compartments. The observed signal attenuation in the diffusion experiment, therefore, depends on the rate of exchange and the diffusion time. In the limit of slow exchange, water spins remain within their respective compartments during the diffusion time and signal attenuation follows a biexponential behavior. This biexponential behavior is given by the following equation:

$$ S={S}_{0}[{f}_{1}\mathrm{exp}(-{b}_{1}{\text{ADC}}_{1})+{f}_{2}\mathrm{exp}(-{b}_{2}{\text{ADC}}_{2})]$$

(1.8)

where, f 1 and f 2 are the volume fractions of water spins within each of the two compartments and f 1 + f 2 = 1; ADC1 and ADC2 are the ADCs in the two compartments.

On the contrary, in the limit of fast exchange with complete redistribution of water between the two compartments during the diffusion time, the signal attenuation follows a single exponential behavior given by the equation:

$$ S/{S}_{0}=\mathrm{exp}(-b*\text{ADC})$$

(1.9)

The observed ADC for a two-compartment system includes contributions from both the intracellular and extracellular environments and is approximated by the following equation:

$$ \text{ADC}={f}_{1}{\text{ADC}}_{1}+{f}_{2}{\text{ADC}}_{2}$$

(1.10)

In most practical situations, the biexponential behavior of signal attenuation is not observed. Therefore, it is a common practice to fit the DWI data to the single exponential decay model.

Isotropic DWI and Trace ADC

Diffusion gradient pulses are applied in one direction at a time, with the resulting DW image giving both directional and magnitude information about the ADC. This would be sufficient if imaging an isotropic tissue, since the ADC would be the same regardless of the direction of the diffusion gradient. When imaging an anisotropic tissue such as the white matter of the brain, however, interpretation of the DWI would be challenging if only a single gradient direction were probed due to the variable signal intensity of white matter tracts depending on their orientation relative to the direction of the diffusion gradient. While the diffusion anisotropy discussed in the previous section can be exploited to image fiber tracts in DTI, the directional effects of anisotropy are undesirable in routine clinical DWI.

In order to overcome the directional influences of the diffusion gradient in anisotropic tissues, four separate scalar image acquisitions are required: one without a diffusion-sensitizing gradient (S 0, where b = 0 s/mm²), and three with the diffusion gradients applied in three orthogonal directions x, y, and z (which are called S x, S y, and S z, respectively). To create an image related only to the magnitude of the ADC, the DW images acquired with the gradient pulses applied in three orthogonal planes can be multiplied, and the cube root of that product yields a DW image weighted with diffusion magnitude information alone and directional information removed, called the isotropic DWI (S DWI):

$$ {S}_{\text{DWI}}={({S}_{x}{S}_{y}{S}_{z})}^{1/3}$$

(1.11)

The ADC derived from application of the gradients in three orthogonal planes, called the trace ADC (simply called ADC below), is the average of the ADCs from each of the planes (Fig. 1.3):

$$ \text{ADC}=({\text{ADC}}_{x}+{\text{ADC}}_{y}+{\text{ADC}}_{z})/3$$

(1.12)

The images submitted to the radiologist for interpretation in the clinical setting are typically the isotropic DWI and trace ADC images.

DWI of the Normal Brain

As mentioned previously, the average ADC in the brain is approximately 0.7 × 10−3 mm²/s; more specifically, mean diffusivities in the adult brain range from 0.67 to 0.83 × 10−3 mm²/s in gray matter, to 0.64–0.71 × 10−3 mm²/s in white matter [44]. Because the ADC values for gray and white matter are so similar, there is essentially no contrast between gray and white matter on the ADC map or exponential image. The subtle gray–white matter contrast that can be seen on the DWI can be attributed to T2 effects, which makes it important to concomitantly view the ADC map or exponential image to evaluate for true restricted diffusion.

In neonates and young children, the ADC is initially much higher in rapidly developing gray matter structures such as the thalamus and basal ganglia, and can be more than twice as high in the slowly developing white matter regions [45, 46]. This can be attributed to the relatively high water content in the immature brain, with relative paucity of myelinated neurons. ADC values decrease rapidly over the first two years of brain development and continue to decrease gradually through childhood, adolescence, and into young adulthood [47–53]. Owing to the higher ADC values in neonates and infants compared to adults [46–49], it is common to reduce the diffusion sensitivity factor (b) in this age group to 600 or 700 s/mm², compared to the 1,000 s/mm² typically used in adults. In the aging but otherwise healthy brain, mild increases can be seen in the ADC, particularly in white matter [54–59]. Engelter et al. found a significant correlation between the average ADC of white matter and age, with patients 60 years of age or older having increased ADC compared to those under the age of 60; a similar trend was seen in the average ADC of the thalamus [56]. Chen et al. found that the average ADC increases by 3% per decade after the age of 40 [57]. This has been attributed to loss of myelinated neurons and structural changes seen with aging.

Current DWI Techniques

The movement of water molecules detected by DWI occurs on a length scale of micrometers that is significantly larger than intracellular distances, but much smaller than the millimeter scale of voxel size in typical MR images. The original PGSE T2-weighted sequence described by Stejskal and Tanner was sensitive to even minimal bulk patient motion, which was enough to obscure the much more miniscule molecular motion of diffusion. As MRI technology has improved and high-performance gradients have been developed, DWI is now typically performed using spin-echo single-shot echo-planar imaging techniques (SS-EPI). With this type of pulse sequence, an entire 2-D image can be formed from a single RF excitation pulse (hence the term single-shot). Images can be acquired in a fraction of a second, minimizing artifacts related to patient motion. Additionally, the SS-EPI technique has a relatively high SNR; this is an important advantage in DWI, as high b-value diffusion gradients cause a significant loss of signal intensity [see (1.4)]. Using the SS-EPI technique, DW imaging of the brain can typically be completed in 1–2 min, beneficial in clinical settings where acutely ill and often uncooperative patients (i.e., hyperacute stroke patients) are being imaged.

Limitations of the SS-EPI technique include low spatial resolution, blurring effects of T2* and T2 decay, and sensitivity to artifacts. Matrix size using the single-shot technique is limited to 128 × 128 for a typical DWI using current MRI hardware and software, compared to matrix sizes of 256 × 192 or larger for standard T1- and T2-weighted sequences. Blurring of T2*- and T2-weighted contrast occurs during image readout due to the extended echo-train length. SS-EPI is also sensitive to artifacts due to Nyquist ghosting, chemical shift, and particularly magnetic field inhomogeneities caused by local susceptibility differences between adjacent structures. As stronger and faster gradients are developed to improve DWI, problems such as eddy currents and mechanical vibration may be exacerbated, resulting in additional artifacts. These artifacts are explained in detail in reviews by Le Bihan et al. [60] and by Mukherjee et al. [61].

The development of multichannel coils and parallel imaging has also led to improvements in DWI. Compared to standard head RF coils, which have a uniform sensitivity throughout their imaging volume, the newer multichannel phased-array RF coils have increased sensitivity around their periphery than at their center, resulting in improved SNR in the cerebral cortex and, therefore, improved DW images. These coils also allow parallel imaging on modern MR scanners due to their multiple independent receiver channels. Parallel imaging techniques can be used to shorten the echo-train length of EPI, thereby reducing the susceptibility-induced and blurring artifacts that typically occur with extended echo trains. The shorter readout may also increase SNR by decreasing the TE.

Alternatives to SS-EPI DWI include variations of fast spin-echo (FSE) or turbo spin-echo (TSE) imaging, multishot EPI, spiral imaging, and line-scan imaging techniques [61, 62]. SS-FSE and half-Fourier acquired single-shot turbo spin-echo are ultrafast sequences that limit artifacts related to bulk patient motion similar to SS-EPI, but with reduced susceptibility and chemical shift artifacts compared to SS-EPI. However, they require longer scan times due to their low SNR compared with SS-EPI and, therefore, have not become popular techniques for brain DWI imaging. Multishot techniques also confer a reduction in susceptibility artifacts compared with SS-EPI, but scan times are longer, which makes these methods more sensitive to bulk patient motion. One multishot FSE technique reduces motion artifacts by continually oversampling the center of k-space, and is called periodically rotated overlapping parallel lines with enhanced reconstruction. This technique improves detection of small acute infarcts, particularly at the skull base and in the posterior fossa, where SS-EPI techniques have significant susceptibility-induced distortion. Owing to much longer scanning times, however, the technique has not surpassed SS-EPI for routine clinical DWI.

Conclusion

From the observation of Brownian motion in 1827 to the description of molecular diffusion by Einstein in 1905, and the application of these principles to DW MRI, a great deal has been learned about the movement of water molecules in the brain (Table 1.1). DWI is a powerful tool that provides a great deal of information about the underlying structure and physiology of the brain that is not provided with standard T1- and T2-weighted imaging techniques. DWI was first routinely used in brain imaging for the detection of acute ischemia and is now an essential part of the imaging evaluation of patients with acute stroke. The applications of DWI in neuroimaging are extensive and include the evaluation of intracranial infections, neoplasms, demyelinating processes, and trauma. These are discussed in the following chapter. DWI techniques can also be used to evaluate the fiber tracts of the brain, and these techniques and applications are discussed elsewhere in this book. As MRI technology advances and improvements are made in imaging hardware and software, newer techniques and sequences may provide more sensitive DW images with reduction of the artifacts inherent in current DWI techniques.

Table 1.1

Summary of articles reviewing the basic principles of DWI discussed in this chapter

References

1.

Moseley ME, Cohen Y, Mintorovitch J, et al. Early detection of regional cerebral ischemia in cats: comparison of diffusion- and T2-weighted MRI and spectroscopy. Magn Reson Med. 1990;14:330–46.PubMedCrossRef

2.

Chien D, Kwong KK, Gress DR, et al. MR diffusion imaging of cerebral infarction in humans. AJNR Am J Neuroradiol. 1992;13:1097–102.PubMed

3.

Warach S, Chien D, Li W, et al. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology. 1992;42:1717–23.PubMed

4.

van Everdingen KJ, van der Grong J, Kappelle LJ, Ramos LM, Mali WP. Diffusion-weighted magnetic resonance imaging in acute stroke. Stroke. 1998;29(9):1783–90.PubMedCrossRef

5.

Brown R. A brief account of microscopical observations made in the months of June, July and August 1827 on the particles contained in the pollen of plants, and on the general existence of active molecules in organic and inorganic bodies. Philosophical Magazine. 1828;4:161.

6.

Einstein A. In: Furthe R, Cowper AD, editors. Investigations on the Theory of Brownian Motion. New York: Dover; 1956. (Collection of papers translated from German, first published in 1905).

7.

Jones DK. Studying connections in the living human brain with diffusion MRI. Cortex. 2008;44(8):936–52.PubMedCrossRef

8.

Hagmann P, Jonasson L, Maeder P, Thiran JP, Wedeen VJ, Meuli R. Understanding diffusion MR imaging techniques: from scalar ­diffusion-weighted imaging to diffusion tensor imaging and beyond. Radiographics. 2006;26 Suppl 1:S205–23.PubMedCrossRef

9.

Schaefer PW, Grant PE, Gonzalez RG. Diffusion-weighted MR imaging of the brain. Radiology. 2000;217(2):331–45.PubMed

10.

Pierpaoli C, Jezzard P, Basser PJ, et al. Diffusion tensor MR imaging of the human brain. Radiology. 1996;201:637–48.PubMed

11.

Bloch F. Nuclear induction. Phys Rev. 1946;70:460–74.CrossRef

12.

Bloch F, Hansen WW, Packard M. Nuclear induction. Phys Rev. 1946;69:127.CrossRef

13.

Purcell EM, Torrey HC, Pound RV. Resonance absorption by nuclear magnetic moments in a solid. Phys Rev. 1946;69:37.CrossRef

14.

Hahn EL. Spin echoes. Phys Rev. 1950;80:580–94.CrossRef

15.

Carr HY, Purcell EM. Effects of diffusion on free precession in nuclear magnetic resonance experiments. Phys Rev. 1954;94:630–8.CrossRef

16.

Torrey HC. Bloch equations with diffusion terms. Phys Rev. 1956;104:563–5.CrossRef

17.

Stejskal EO, Tanner JE. Spin diffusion measurements: spin echoes in the presence of a time-dependent field gradient. J Chem Phys. 1965;42:288–92.CrossRef

18.

Mori S, Barker PB. Diffusion magnetic resonance imaging: its principle and applications. Anat Rec. 1999;257(3):102–9.PubMedCrossRef

19.

Sevick RJ, Kanda F, Mintorovitch J, et al. Cytotoxic brain edema: assessment with diffusion-weighted MR imaging. Radiology. 1992;185(3):687–90.PubMed

20.

Ebisu T, Naruse S, Horikawa Y, et al. Discrimination between different types of white matter edema with diffusion-weighted MR imaging. J Magn Reson Imaging. 1993;3:863–8.PubMedCrossRef

21.

Schaefer PW, Buonanno FS, Gonzalez RG, et al. Diffusion-weighted imaging discriminates between cytotoxic and vasogenic edema in a patient with eclampsia. Stroke. 1997;28:1082–5.PubMedCrossRef

22.

Schwartz R, Mulkern R, Gudbjartsson H, et al. Diffusion-weighted MR imaging in hypertensive encephalopathy: clues to pathogenesis. AJNR Am J Neuroradiol. 1998;19:859–62.PubMed

23.

Ay H, Buonanno FS, Schaefer PW, et al. Posterior leukoencephalopathy without severe hypertension: utility of diffusion-weighted MRI. Neurology. 1998;51:1369–76.PubMed

24.

Mukherjee P, McKinstry RC. Reversible posterior leukoencephalopathy syndrome: evaluation with diffusion-tensor MR imaging. Radiology. 2001;219:756–65.PubMed

25.

Provenzale JM, Petrella JR, Cruz Jr LCH, et al. Quantitative assessment of diffusion abnormalities in posterior reversible encephalo­pathy syndrome. AJNR Am J Neuroradiol. 2001;22:1455–61.PubMed

26.

Chen PE, Simon JE, Hill MD, et al. Acute ischemic stroke: accuracy of diffusion-weighted MR imaging – effects of b value and cerebrospinal fluid suppression. Radiology. 2006;238(1):232–9.PubMedCrossRef

27.

Kim HJ, Choi CG, Lee DH, Lee JH, Kim SJ, Suh DC. High-b-value diffusion-weighted MR imaging of hyperacute ischemic stroke at 1.5T. AJNR Am J Neuroradiol. 2005;26(2):208–15.PubMed

28.

Le Bihan D. Looking into the functional architecture of the brain with diffusion MRI. Nat Rev Neurosci. 2003;4(6):469–80.PubMedCrossRef

29.

Cleveland GG, Chang DC, Hazlewood CF, Rorschach HE. Nuclear magnetic resonance measurement of skeletal muscle: anisotrophy of the diffusion coefficient of the intracellular water. Biophys J. 1976;16(9):1043–53.PubMedCrossRef

30.

Moseley ME, Cohen Y, Kucharczyk J, Mintorovitch J, Asgari HS, Wendland MF, et al. Diffusion-weighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology. 1990;176(2):439–45.PubMed

31.

Moseley ME, Kucharczyk J, Asgari HS, Norman D. Anisotropy in diffusion-weighted MRI. Magn Reson Med. 1991;19(2):321–6.PubMedCrossRef

32.

Doran M, Hajnal JV, Van Bruggen N, King MD, Young IR, Bydder GM. Normal and abnormal white matter tracts shown by MR imaging using directional diffusion weighted sequences. J Comput Assist Tomogr. 1990;14(6):865–73.PubMedCrossRef

33.

Chenevert TL, Brunberg JA, Pipe JG. Anisotropic diffusion in human white matter: demonstration with MR techniques in vivo. Radiology. 1990;177(2):401–5.PubMed

34.

Thomsen C, Henriksen O, Ring P. In vivo measurement of water self diffusion in the human brain by magnetic resonance imaging. Acta Radiol. 1987;28(3):353–61.PubMedCrossRef

35.

Hong X, Dixon WT. Measuring diffusion in inhomogeneous systems in imaging mode using antisymmetric sensitizing gradients. J Magn Reson. 1992;99:561–70.

36.

Lian J, Williams DS, Lowe IJ. Magnetic resonance imaging in the presence of background gradients and imaging of background ­gradients. J Magn Reson Series A. 1994;106:65–74.CrossRef

37.

Wimberger DM, Roberts TP, Barkovich AJ, Prayer LM, Moseley ME, Kucharczyk J. Identification of premyelination by diffusion-weighted MRI. J Comput Assist Tomogr. 1995;19(1):28–33.PubMedCrossRef

38.

Prayer D, Roberts T, Barkovich AJ, Prayer L, Kucharczyk J, Moseley M, et al. Diffusion-weighted MRI of myelination in the rat brain following treatment with gonadal hormones. Neuroradiology. 1997;39(5):320–5.PubMedCrossRef

39.

Beaulieu C, Allen PS. Determinants of anisotropic water diffusion in nerves. Magn Reson Med. 1994;31(4):394–400.PubMedCrossRef

40.

Beaulieu C, Allen PS. Water diffusion in the giant axon of the squid: implications for diffusion-weighted MRI of the nervous system. Magn Reson Med. 1994;32(5):579–83.PubMedCrossRef

41.

Beaulieu C, Allen PS. An in vitro evaluation of the effects of local magnetic-susceptibility-induced gradients on anisotropic water ­diffusion in nerve. Magn Reson Med. 1996;36(1):39–44.PubMedCrossRef

42.

Beaulieu C. The basis of anisotropic water diffusion in the nervous system – a technical review. NMR Biomed. 2002;15(7–8):435–55.PubMedCrossRef

43.

Burdette JH, Elster AD, Ricci PE. Acute cerebral infarction: quantification of spin-density and T2 shine-through phenomena on ­diffusion-weighted MR images. Radiology. 1999;212(2):333–9.PubMed

44.

Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G. Diffusion tensor MR imaging of the human brain. Radiology. 1996;201(3):637–48.PubMed

45.

Hüppi PS, Maier SE, Peled S, Zientara GP, Barnes PD, Jolesz FA, et al. Microstructural development of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging. Pediatr Res. 1998;44(4):584–90.PubMedCrossRef

46.

Neil JJ, Shiran SI, McKinstry RC, Schefft GL, Snyder AZ, Almli CR, et al. Normal brain in human newborns: apparent diffusion coefficient and diffusion anisotropy measured by using diffusion tensor MR imaging. Radiology. 1998;209(1):57–66.PubMed

47.

Mukherjee P, Miller JH, Shimony JS, Conturo TE, Lee BC, Almli CR, et al. Normal brain maturation during childhood: developmental trends characterized with diffusion-tensor MR imaging. Radiology. 2001;221(2):349–58.PubMedCrossRef

48.

Mukherjee P, McKinstry RC. Diffusion tensor imaging and tracto­graphy of human brain development. Neuroimaging Clin N Am. 2006;16(1):19–43. vii.PubMedCrossRef

49.

Mukherjee P, Miller JH, Shimony JS, Philip JV, Nehra D, Snyder AZ, et al. Diffusion-tensor MR imaging of gray and white matter development during normal human brain maturation. AJNR Am J Neuroradiol. 2002;23(9):1445–56.PubMed

50.

McGraw P, Liang L, Provenzale JM. Evaluation of normal age-related changes in anisotropy during infancy and childhood as shown by diffusion tensor imaging. AJR Am J Roentgenol. 2002;179(6):1515–22.PubMed

51.

Schneider JF, Il’yasov KA, Hennig J, Martin E. Fast quantitative diffusion-tensor imaging of cerebral white matter from the neonatal period to adolescence. Neuroradiology. 2004;46(4):258–66.PubMedCrossRef

52.

Snook L, Paulson LA, Roy D, Phillips L, Beaulieu C. Diffusion tensor imaging of neurodevelopment in children and young adults. Neuroimage. 2005;26(4):1164–73.PubMedCrossRef

53.

Hermoye L, Saint-Martin C, Cosnard G, Lee SK, Kim J, Nassogne MC, et al. Pediatric diffusion tensor imaging: normal database and observation of the white matter maturation in early childhood. Neuroimage. 2006;29(2):493–504.PubMedCrossRef

54.

Gideon P, Thomsen C, Henriksen O. Increased self-diffusion of brain water in normal aging. J Magn Reson Imaging. 1994;4(2):185–8.PubMedCrossRef

55.

Chun T, Filippi CG, Zimmerman RD, Uluğ AM. Diffusion changes in the aging human brain. AJNR Am J Neuroradiol. 2000;21(6):1078–83.PubMed

56.

Engelter ST, Provenzale JM, Petrella JR, et al. The effect of aging on the apparent diffusion coefficient of normal appearing white matter. AJR Am J Roentgenol. 2000;175:425–30.PubMed

57.

Chen ZG, Li TQ, Hindmarsh T. Diffusion tensor trace mapping in normal adult brain using single-shot EPI technique. A methodological study of the aging brain. Acta Radiol. 2001;42(5):447–58.PubMed

58.

Nusbaum AO, Tang CY, Buchsbaum MS, Wei TC, Atlas SW. Regional and global changes in cerebral diffusion with normal aging. AJNR Am J Neuroradiol. 2001;22(1):136–42.PubMed

59.

Moseley M. Diffusion tensor imaging and aging – a review. NMR Biomed. 2002;15(7–8):553–60.PubMedCrossRef

60.

Le Bihan D, Poupon C, Amadon A, Lethimonnier F. Artifacts and pitfalls in diffusion MRI. J Magn Reson Imaging. 2006;24(3):478–88.PubMedCrossRef

61.

Mukherjee P, Chung SW, Berman JI, Hess CP, Henry RG. Diffusion tensor MR imaging and fiber tractography: technical considerations. AJNR Am J Neuroradiol. 2008;29(5):843–52.PubMedCrossRef

62.

Bammer R. Basic principles of diffusion-weighted imaging. Eur J Radiol. 2003;45(3):169–84.PubMedCrossRef

63.

Mukherjee P, Berman JI, Chung SW, Hess CP, Henry RG. Diffusion tensor MR imaging and fiber tractography: theoretic underpinnings. AJNR Am J Neuroradiol. 2008;29(4):632–41. Epub 2008 Mar 13. Review.PubMedCrossRef

Scott H. Faro, Feroze B. Mohamed, Meng Law and John T. Ulmer (eds.)Functional NeuroradiologyPrinciples and Clinical Applications10.1007/978-1-4419-0345-7_2© Springer Science+Business Media, LLC 2011

2. Clinical Applications of Diffusion

Reza Forghani¹, ²   and Pamela W. Schaefer³  

(1)

Department of Radiology, Sir Mortimer B. Davis Jewish General Hospital, Montreal, QC, Canada

(2)

McGill University, 3755 Cote Ste-Catherine Road, Room C-210.2, H3T 1E2 Montreal, QC, Canada

(3)

Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Gray 273A, 55 Fruit Street, Boston, MA 02114, USA

Reza ForghaniAssociate Chief (Corresponding author)

Email: [e-mail]

Pamela W. SchaeferAssociate Director of Neuroradiology, Associate Professor of Radiology

Email: [e-mail]

Abstract

Diffusion-weighted magnetic resonance imaging (DWI) is a technique based on diffusion of water molecules in tissues, with clinical applications to a wide array of pathological conditions. Currently, DWI is the most reliable method for detection of early and small ischemic infarcts in the brain and the gold standard for determination of the infarct core. DWI is also an important sequence for characterization of various neoplastic conditions such as epidermoid tumors, lymphomas, and high-grade astrocytomas and enables distinction of pyogenic abscesses from ring-enhancing intracranial neoplasms. Additional applications of DWI include differentiation of vasogenic edema syndromes from acute ischemia, identification of acute demyelinating lesions, and characterization of encephalitides, toxic and metabolic lesions, and diffuse axonal injury. This chapter provides an overview of current clinical applications of DWI as well as potential future applications of DWI currently under investigation, including its use for prediction of complications and outcomes of ischemic strokes and distinction of tumor progression from treatment-related changes.

Introduction

Diffusion-weighted magnetic resonance imaging (DWI) is a technique based on diffusion of water molecules in tissues with clinical applications to a wide array of pathological conditions. In vitro measurements of diffusion coefficients for liquids based on their nuclear magnetic resonance signal using the pulse field gradient method were originally described in the 1960s [1] and improved in the early 1970s [2]. These were followed by in vivo measurements of molecular diffusion of water and applications to various pathologic states in human and animal studies, including cerebral ischemia and neoplasms in the mid-1980s and early 1990s [3–14]. However, because of significant technical requirements of DWI, there has only been widespread clinical application of this technique in the last decade. Currently, DWI is the most reliable method for detection of early and small ischemic infarcts in the brain. However, applications of DWI extend far beyond the work-up of acute ischemia. DWI is increasingly used for the evaluation of a large variety of neoplastic and nonneoplastic conditions affecting the brain, head, and neck. This chapter begins with a description of the basic mechanisms underlying DWI changes on magnetic resonance imaging (MRI) scans, focusing on the extensive work using experimental models of brain ischemia. This is followed by a discussion of currently established and experimental clinical applications of DWI, starting with the use of DWI for imaging acute ischemic stroke. The physics of DWI is reviewed in a separate chapter and is not discussed here.

Basic Mechanisms Underlying Diffusion Changes in Pathologic States

Much of our understanding of the mechanisms underlying DWI changes in pathologic states is based on in vitro and in vivo experimental models of cerebral ischemia in animals. Cerebral ischemia results in diminished diffusion of water molecules within the infarct territory with a rapid decline in apparent diffusion coefficient (ADC) values. Based on current data, the decline in the ADC seen during acute ischemia can be attributed to a combination of complex biophysical factors resulting from disruption of normal cellular metabolism (Table 2.1). Disruption of normal cellular metabolism and depletion of adenosine triphosphate (ATP) stores result in failure of Na+/K+ ATPase and other ionic pumps with loss of ionic gradients across cellular membranes. This in turn leads to a net shift of water from the extracellular (EC) to the intracellular (IC) compartment with a change in the relative volume of these compartments as well as alterations in the IC and EC microenvironments. These alterations ultimately result in the restricted diffusion seen on MR imaging. Although there is some controversy regarding the biophysical determinants of diffusion changes seen on MRI, three major mechanisms are believed to account for the restricted diffusion seen during acute ischemia; they are (1) changes in relative volumes of the intracellular and extracellular spaces, (2) increased extracellular space tortuosity resulting in increased impedance to diffusion of water, and (3) diminished energy-dependent cytoplasmic circulation/microstreaming in the intracellular compartment. Some of the key investigations and controversies pertaining to mechanisms of DWI changes are summarized in the paragraphs that follow.

Table 2.1

Theories for decreased diffusion in acute stroke

One widely proposed mechanism for the diffusion changes during cerebral ischemia is that disruption of energy metabolism results in failure of the Na+/K+ ATPase pump. This in turn leads to a net shift of water from the EC compartment, where the diffusion of water is relatively unimpeded, to the IC compartment where the diffusion of water is postulated to be relatively restricted [15]. This model is supported by animal studies demonstrating that ischemic infarction is associated with a reduction of the Na+/K+ ATPase pump activity [16], that pharmacologic inhibition of the Na+/K+ ATPase pump with ouabain in the absence of ischemia results in decreased ADC values [17, 18], and that nonischemic cytotoxic edema secondary to acute hyponatremia is associated with decreased ADC values [19]. However, the concept of relatively impeded diffusion of water in the IC compared to the EC compartment in not universally accepted, with some experiments supporting the theory [15] and others disputing it [20, 21].

Multiple investigations have demonstrated that alterations in the EC compartment can result in changes in diffusion seen on MRI [20–24]. Early DWI changes following transient ischemia in vivo correlate with shrinkage and reexpansion of the extracellular space [22]. In addition, based on observations on osmotically driven changes in compartment volume in ex vivo rat optic nerve preparations, it has been argued that assuming the IC and EC water constitute approximately 80% and 20% of total water in the brain, respectively, moderate fractional changes in cell volume can result in large changes in the contribution of the EC to the ADC [23]. Using these assumptions, it has been postulated that changes in the extracellular fraction are the main factor accounting for the ADC changes observed during ischemia [23]. Changes in relative compartment volume and associated cellular swelling may result in diffusion changes by increasing the tortuosity of the EC space, which in turn results in greater impedance to diffusion of water and the decreased ADC seen on MRI [20, 24].

Additional experiments suggest that changes in the intracellular microenvironment also contribute to the decreased ADC associated with acute ischemia. In experiments in which 2-[¹⁹F]luoro-2-deoxyglucose-6-phosphate was used as a compartment-specific marker for the IC and EC compartments by manipulation of routes of administration, the ADC in both compartments were similar at baseline and both were similarly reduced following cerebral ischemia [20]. In addition, experiments using diffusion-weighted MR spectroscopy evaluating predominantly intracellular metabolites such as N-acetyl-aspartate and phosphocreatine (Cre) [25] or the potassium analog Cesium (Cs) [26] have also demonstrated a decrease in the ADC of the IC compartment following ischemia. In light of investigations suggesting that the diffusion in the IC and EC spaces may not be significantly different, it has been proposed that the changes in IC ADC are secondary to disruption of the energy-dependent cytoplasmic motion of molecules or microstreaming [20].

Increased cytoplasmic viscosity, such as from breakdown of intracellular organelles or cytoskeleton, is an alternate potential mechanism for changes in IC ADC. However, the increased IC water and cellular swelling in isolation would be expected to facilitate diffusion [27], and dissociation of the cytoskeleton likewise appears to facilitate diffusion of water in experimental models [27, 28], making this hypothesis less attractive. A number of other potential mechanisms of ADC change during acute ischemia have been discussed in the literature including changes in tissue temperature and membrane permeability. Although theoretically valid, they probably play a minor and insignificant role under typical circumstances in vivo during an acute stroke [20, 29].

Basic Principles of Diffusion-Weighted Imaging Map Interpretation

All DWI images (linearly T2 weighted and exponentially diffusion weighted) should be processed at the MR console with generation of ADC maps (linearly diffusion weighted, without a T2 component) and exponential images (exponentially diffusion weighted, without a T2 component) (Fig. 2.1). The use of the ADC map is essential for proper interpretation of DWI images since both areas of diminished and increased diffusion can appear bright on DWI images. In lesions with restricted diffusion such as acute ischemic strokes, the T2 and diffusion effects both cause increased signal on DWI, and the DWI images consequently have the highest contrast-to-noise ratio (Fig. 2.2). Areas of restricted diffusion also appear hyperintense on exponential images, although less bright, compared to the DWI image because of absence of the T2 effect (Fig. 2.2). Lesions with restricted diffusion appear dark on the ADC map (Fig. 2.2). Areas of increased diffusion, on the contrary, may appear hyperintense, isointense, or hypointense to normal brain parenchyma on DWI images depending on the strength of the T2 and diffusion components, but are hyperintense on the ADC map (Figs. 2.3 and 2.4). When a lesion is hyperintense on both the DWI image and the ADC map and hypointense on the exponential image, the phenomenon is referred to as T2 shine-through and may be seen with late subacute infarcts or chronic ischemic lesions (Fig. 2.4). Lesions with T2 shine-through may be misinterpreted as an acute infarct if findings on the ADC map or exponential image are not taken into account.

A183863_1_En_2_Fig1_HTML.jpg

Fig. 2.1

Normal diffusion MR maps. (a) Axial DWI, (b) ADC, and (c) exponential images demonstrate normal diffusion MR maps from a 24-year-old man (b = 1,000 s/mm²; TR, 5,000 ms; minimum echo time; matrix, 128 × 128; field of view, 22 × 22 cm; section thickness, 5 mm with 1 mm gap). DWI are linearly T2-weighted and exponentially diffusion-weighted, whereas exponential maps are linearly diffusion-weighted without a T2 component. Note how the normal cerebral cortex and deep gray nuclei have mildly increased intensity on DWI compared to adjacent white matter. Normal CSF spaces appear dark on DWI and exponential maps and bright on the ADC map given the lack of physical barriers to free motion of water. Abbreviations: SI Signal intensity, SI(T2W) Signal intensity on T2-weighted images

A183863_1_En_2_Fig2_HTML.jpg

Fig. 2.2

Acute infarction. Diffusion MR maps from a 31-year-old man with embolic occlusion of the left M1 segment, imaged 3 h after developing aphasia and right hemiparesis, demonstrate a large area of signal abnormality in the left MCA territory that is (a) hyperintense on DWI, (b) hypointense on the ADC map, and (c) hyperintense on the exponential map consistent with restricted diffusion secondary to an acute infarction. (d) There is mild corresponding hyperintensity on the FLAIR image, although the infarct is more conspicuous on the DWI image

A183863_1_En_2_Fig3_HTML.jpg

Fig. 2.3

Chronic infarction on DWI. Axial MR images of a chronic left MCA territory infarction demonstrate (a) an area of heterogenous signal abnormality within the left MCA territory with local volume loss and associated ex-vacuo dilation of the left lateral ventricle on the FLAIR image. There is corresponding signal abnormality on the diffusion maps that is heterogenous, but (b) predominantly hypointense on DWI, (c) hyperintense on the ADC map, and (d) hypointense on the exponential image consistent with elevated diffusion

A183863_1_En_2_Fig4_HTML.jpg

Fig. 2.4

T2 shine-through. (a) Axial FLAIR image demonstrates nonspecific periventricular FLAIR hyperintense lesions. (b) On DWI, mildly increased signal raises the possibility of acute ischemia. However, the lesions are (c) hyperintense on the ADC map and (d) hypointense on the exponential map, consistent with elevated diffusion secondary to chronic ischemic changes

Diffusion-Weighted Imaging of Ischemic Arterial Infarcts

Diffusion Characteristics of Ischemic Lesions at Different Stages

DWI is highly sensitive (81–100%) and specific (86–100%) for detection of acute ischemia within the first 12 h after stroke onset [30–34], with sensitivities and specificities approaching 100% at specialized high volume stroke centers (Tables 2.2 and 2.4, Figs. 2.2 and 2.5). DWI may demonstrate infarcts as early as 11 min after symptom onset [35] and is superior to conventional MRI or computed tomography (CT) in the first 3–6 h, at which time there is frequently insufficient accumulation of tissue water for reliable detection of hypoattenuation on CT and hyperintensity on T2 weighted and fluid-attenuated inversion recovery (FLAIR) MR images [30, 31, 34] (Fig. 2.5). In a recent prospective evaluation, MRI with DWI had a sensitivity of 73% and specificity of 92% for identification of ischemic infarcts within 3 h of symptom onset compared to a sensitivity of 12% and specificity of 100% for CT [30]. In another study, DWI had a sensitivity of 97% and specificity of 100% for identification of acute ischemia within 6 h of stroke symptom onset compared to 58% and 100%, respectively, for conventional MRI sequences and 40% and 92%, respectively, for CT [34]. Although most useful in the first 6–8 h when infarcts are frequently not identifiable on T2 or FLAIR images, DWI can also be valuable at later time points because of its higher contrast-to-noise ratio compared with CT and conventional MRI images. DWI has superior sensitivity for identification of small infarcts that may be overlooked on FLAIR or T2 images by increasing their conspicuity and enables distinction of small recent white matter infarcts from nonspecific T2 hyperintense white matter lesions [36, 37] (Fig. 2.6).

Table 2.2

Diffusion MRI findings during different stagesa of human stroke

a The provided timelines are approximate and can vary between patients and depending on infarct type

b Pseudonormalization, a period during which the ADC of nonviable ischemic tissue is similar to normal brain ADC, most commonly occurs between 4 and 10 days post ictus in humans, and can overlap the early and late subacute stages

A183863_1_En_2_Fig5_HTML.jpg

Fig. 2.5

Superiority of DWI imaging for detection of early ischemic change compared to FLAIR images. (a) Axial DWI and (b) ADC images from an 89-year-old man with a history of acute aphasia and right hemiparesis of 4-h duration demonstrate a large area of restricted diffusion, (c) without any significant corresponding signal abnormality on the FLAIR image

A183863_1_En_2_Fig6_HTML.jpg

Fig. 2.6

Distinction of acute white matter infarctions from chronic white matter change based on diffusion characteristics. (a) There is a focus of hypoattenuation in the left corona radiata on computed tomography (CT) with (b) corresponding hyperintensity on the FLAIR image that is age indeterminate and indistinguishable from the ­remainder of white matter abnormalities. However, (c) the DWI and (d) ADC images clearly distinguish this acute infarct from nonspecific white matter changes by demonstrating restricted diffusion in the acute infarction and elevated diffusion in the nonspecific white matter changes

Following the onset of acute ischemia, there is a rapid decrease in water diffusion that is markedly hyperintense on DWI and hypointense on the ADC map and is generally postulated to represent cytotoxic edema (Table 2.2, Figs. 2.2, 2.5, and 2.7). After the initial period of decreased ADC values, there is a gradual increase in the ADC values secondary to cell lysis and increasing vasogenic edema, with a transient return to baseline known as pseudonormalization, a period during which the ADC of the nonviable ischemic tissue is similar to normal brain ADC (Fig. 2.7). In animal models of stroke, ADC values are reduced for a very short period of time and return to baseline between approximately 24–48 h [38, 39]. In humans, the peak signal reduction on ADC occurs between 1 and 4 days, with a return to baseline at approximately 4–10 days following the ictus [40–44]. The DWI at this stage is typically hyperintense secondary to T2 effects, although less intense than during the acute phase, while on ADC and exponential images the infarct is isointense to normal brain parenchyma. Subsequently, as the infarct progresses to a more chronic stage, there is a progressive increase in the ADC of an ischemic lesion. This is secondary to increased water content as well as emerging gliosis and cavitation, which result in breakdown of the normal tissue structure and barriers to free diffusion of water molecules. During the chronic stage, an infarct can be mildly hyperintense, isointense, or hypointense to normal brain parenchyma on DWI depending on the strength of the T2 effects and diffusion, but should be hyperintense on ADC and hypointense on exponential images (Figs. 2.3 and 2.7).

A183863_1_En_2_Fig7_HTML.jpg

Fig. 2.7

Temporal evolution of infarction on diffusion maps. (a–e) Axial DWI images and (f–j) ADC maps demonstrate the appearance of different infarct stages on diffusion maps. At 6 h, the right middle cerebral artery (MCA) territory infarction is mildly hyperintense on DWI images and hypointense on ADC maps secondary to early cytotoxic edema. By 30 h, the DWI hyperintensity and ADC hypointensity are pronounced secondary to increased cytotoxic edema. This is the ADC nadir. By 5 days, the ADC hypointensity is mild and the ADC has nearly pseudonormalized due to cell lysis and the development of vasogenic edema. The lesion remains hyperintense on the DWI images because the T2 and diffusion components are combined. At 3 months, the infarction is hypointense on DWI and hyperintense on ADC maps due to increased diffusion secondary to the development of gliosis and tissue cavitation

Although separation of infarct stages into hyperacute, acute, subacute, and chronic (Table 2.2) provides a useful general framework for dating an infarct on MRI, there is variability in the time course of the DWI and ADC signal