Magnetic resonance imaging (MRI) of the brain

Basic MRI physics^{135, 142, 235, 278, 351, 354, 385}

Magnetic resonance imaging (MRI) is the imaging modality of choice for the diagnosis of most neurologic diseases in human and veterinary patients. While an in-depth review of MR physics is beyond the scope of this chapter, an overview of basic principles and sequences will be provided to facilitate understanding of clinical applications and interpretation.

In general, any element with either an odd number of protons or an odd number of neutrons has a nuclear magnetic dipole moment and may therefore be suitable for MR imaging or spectroscopy. Hydrogen is the optimal element for MR imaging as (1) it is the most common element in the body, (2) its nucleus consists of a single proton and has the strongest magnetic dipole moment of any element suitable for MR imaging, and (3) most pathologic processes affecting the central nervous system (CNS) result in alteration of content, distribution, and ambient environment of hydrogen protons facilitating differentiation of diseased from normal tissue. This overview will focus on MR imaging of hydrogen protons.

Hydrogen protons in tissues are not static but spin around their axes, generating their own micromagnetic environments. In the absence of an external magnetic field the magnetic moments of the spinning protons are randomly oriented. However, when brought into a strong external magnetic field (i.e. an MR scanner), they rearrange under its influence. The magnetic field strength is denoted by the unit “tesla” (T). The strength of clinically used MR scanners ranges from approximately 0.2T (low-field) to 3T (high-field); higher-strength scanners (7–21T) are at this point limited to research institutions. The alignment of individual protons may be parallel or antiparallel with the external magnetic field. A slight majority of protons will align with the magnetic field, generating a magnetic vector (“net magnetization vector”) which is utilized during MR imaging. The main magnetic field is denoted by B₀, the tissue magnetization vector by M₀. As long as M₀ is parallel with the much stronger B₀ it cannot be easily separated out and cannot be used for imaging. The goal of MR imaging is to manipulate tissue magnetization in a way that it can be distinguished from the external magnetic environment.

In addition to a spinning motion around their individual axes, hydrogen protons wobble (or precess) under the influence of B₀, similar to a spinning top wobbling under the influence of gravity. The precession frequency of the spinning protons is dependent on the strength of the external magnetic field and is described by the Larmor equation:

where:

   ω₀ = Larmor or resonant frequency

   γ = gyromagnetic ratio specific for each MR active nucleus (42.56 MHz/T for hydrogen)

   B₀ = external magnetic field strength.

In order to manipulate tissue magnetization so it can be separated from the main magnetic field, radiofrequency (RF) pulses are applied. Once the nuclei are exposed to an RF pulse exactly matching their precessional frequency they gain energy and start resonating. As a result of the energy gain some protons change their alignment with the magnetic field, causing the magnetic net vector to move away from B₀, or “flip.” The most common flip angle of the tissue magnetization vector is 90° used in spin echo sequences (see below). Hydrogen protons flipped into this “transverse plane”—which is in perpendicular orientation to the main magnetic field (the “longitudinal plane”)—continue to precess. According to Faraday’s law, any change in the magnetic environment of a coil of wire will cause an electric signal. Strategic placement of receiver coils in the MR unit allows detection and measurement of magnetization in the transverse plane, which is the basis of image formation in MRI (Fig. 6.1).

After the RF pulse is switched off, the signal induced in the receiver drops off rapidly due to two concurrent processes:

• T1-relaxation (spin-lattice relaxation): Excited hydrogen protons return to lower energy states, and the magnetic net vector realigns with the main magnetic field (Fig. 6.2).
  
• T2-relaxation (spin-spin relaxation): The unified front of hydrogen protons quickly loses coherence, resulting in a signal drop-off. While true T2-relaxation is solely the result of micromagnetic inhomogeneities (spin–spin interactions, i.e. interference of one spinning proton’s micromagnetic field with its neighbors), extrinsic magnetic field inhomogeneities (magnet imperfections, disruption of the magnetic field by paramagnetic or ferromagnetic substances, etc.) contribute to an even more rapid loss of phase coherence. This process is called T2*-relaxation (Fig. 6.3).

Three important tissue parameters for MR imaging include T1- and T2-relaxation times and proton density (PD), i.e. the actual amount of hydrogen protons in a certain tissue volume. The goal of MR imaging is to translate these inherent tissue differences into image contrast, which is accomplished by using different MR sequences.

MR sequences

Basic spin echo (SE) sequences^{115, 179, 274, 279, 285–287, 293, 385}

These include T1-W, T2-W and proton density (PD) weighting, which are the most basic but also the most commonly used MRI sequences. Each SE sequence starts with a 90° RF pulse followed by a 180° pulse applied exactly halfway between the initial 90° pulse and the generation of the signal (echo) (Fig. 6.4). The 180° pulse is applied to cancel out external magnetic field inhomogeneities. It essentially reverses any effects an external disturbance (e.g. a nearby flowing vessel or paramagnetic methemoglobin in a hematoma) will have on proton alignment and resultant tissue signal. The small interactions occurring between individual protons and contributing to signal loss in the transverse plane cannot be reversed, resulting in true T2 relaxation contributing to image contrast. As the MR signal generated during a single episode of proton excitation is too small to create an image, the process is repeated many times until enough data have been collected. The time between the 90° pulse and the echo is called “time of echo” (TE) and the time between successive 90° pulses is called “time of repetition” (TR). The length of TR and TE determines weighting of an SE sequence (Table 6.1).

• T1-weighting (T1-W): A short TR is chosen to maximize the differences in T1 relaxation between tissues (Fig. 6.5). This is combined with a short TE to minimize T2 effects.       Fat has a short T1 relaxation time and is hyperintense, while fluid has a long T1 relaxation time and appears hypointense. Soft tissues have somewhat variable, intermediate T1 relaxation times and are medium in intensity. After uptake of administered paramagnetic contrast agents, physiologically contrast-enhancing tissues (e.g. pituitary gland) and contrast-enhancing pathologic lesions (e.g. certain brain tumors) are hyperintense (Fig. 6.6).
  
• T2-weighting (T2-W): A long TE is chosen to maximize differences in T2 relaxation between tissues, combined with a long TR to minimize T1 relaxation effects. A long TE ensures that tissues with a short T2 relaxation time will have completely lost their transverse magnetization and have a low signal at the time of image acquisition, while tissues with longer T2 relaxation times still maintain their transverse magnetization and appear brighter (Fig. 6.7). Fluid has a long T2 relaxation time and therefore is hyperintense on T2-W images. Soft tissues have intermediate T2 relaxation times. Fat has a short T2 relaxation time and appears hypointense on conventional T2-W images. However, as conventional T2-W SE sequences have largely been replaced with shorter fast spin echo (FSE) or turbo spin echo (TSE) sequences in which additional pulses are applied (see below), fat typically appears hyperintense on today’s T2-W MRI studies. A T2-W sequence can be considered a “pathology” scan, because abnormal fluid collections and tissues with abnormal increased fluid content (“juicy tissue,” e.g. edema, inflammation, neoplasia) will appear hyperintense compared to normal tissues (Fig. 6.8).
  
• Proton density weighting (PD-W): PD-W is achieved by choosing a long TR in combination with a short TE to minimize T1 and T2 effects on image contrast. PD-W images are characterized by excellent anatomic detail (see Fig. 6.41A below) and are very useful in orthopedic imaging. They also provide a good contrast between gray and white matter, and although their value in the neuroimaging of small animals is limited, anecdotal evidence suggests they may be helpful in the evaluation of patients with degenerative brain disease.

TR	TE	Weighting
Short (300–700ms)	Short (5–30ms)	T1-W
Long (2000–4000ms)	Long (60–150ms)	T2-W
Long (> 2000–4000ms)	Short (5–30ms)	PD-W

Modified spin echo (SE) sequences^{23, 50, 83, 94, 95, 131, 179, 217, 269, 293, 309}

• Inversion recovery sequences: These are characterized by an initial 180° pulse (inversion pulse). Dependent on the time elapsed between this 180° pulse and initiation of the regular SE sequence (time of inversion; TI), they result in selective suppression of fluid (fluid attenuated inversion recovery; FLAIR) or fat (short tau inversion recovery; STIR).
  
• FLAIR: A long TI prior to initiation of a SE sequence allows selective suppression of fluid (Fig. 6.9). Although FLAIR images are most commonly used for brain imaging, they may occasionally be helpful in the evaluation of spinal lesions. T2-W FLAIR images are useful in conjunction with regular T2-W images in characterizing T2 hyperintense lesions. Using FLAIR, pure fluid (cerebrospinal fluid and fluid in cystic lesions) is suppressed and becomes hypointense, while solid lesions remain hyperintense (Fig. 6.10). Additionally, this sequence increases conspicuity of small lesions bordering a fluid-filled ventricle or the subarachnoid space. Finally, FLAIR is helpful in differentiating true T2 hyperintense parenchymal lesions from pseudolesions created by inclusion of fluid-filled structures and brain parenchyma within the same slice thickness (volume averaging). Without modification of acquisition parameters, FLAIR is unable to suppress signal from fluids containing high-protein cell components or blood by-products, a potential pitfall when interpreting images. Postcontrast T1-W FLAIR images are very sensitive in the detection of contrast-enhancing lesions and may be used as an alternative to conventional postcontrast T1-W SE images.
  
• STIR: A short TI prior to initiation of an SE sequence allows selective suppression of fat (Fig. 6.11). This sequence is very valuable in orthopedic and spinal imaging as it allows differentiation of pathologic T2 hyperintense lesions within the vertebral canal, vertebrae, and surrounding paraspinal tissues from fat (Fig. 6.12).
  
• Fast (FSE) or turbo (TSE) spin echo techniques: In conventional SE imaging one 180° pulse is applied during each TR, and one echo (signal) is generated. In FSE and TSE, multiple 180° pulses are applied during each TR and multiple echoes are received, resulting in a decrease in scan time without compromising image quality. FSE/TSE techniques have essentially replaced conventional SE sequences in T2-W imaging. One potential disadvantage is strong hyperintensity of fat on T2-W FSE/TSE images as hyperintense epidural fat in the vertebral canal may obscure T2-hyperintense lesions within adjacent spinal cord, and hyperintense fat in bone marrow may obscure or mimic skull or vertebral lesions. This disadvantage can easily be compensated for by adding a fat-saturation technique or comparing T2-W images to other sequences (see Fig. 6.12).
  
• Single-shot techniques: These ultrafast techniques employ a single RF pulse, further decreasing scan time. The resultant images are characterized by very strong T2 contrast and are most beneficial in imaging of fluid-filled spaces. “Quick-brain” MR imaging was initially introduced as an alternative technique to CT scanning for assessing children with hydrocephalus. Other indications in humans include macrocephaly, Chiari malformation, intracranial cysts, screening prior to lumbar puncture, screening for congenital anomalies, and trauma. These sequences have gained popularity in veterinary medicine for spinal imaging due to their myelographic effect, which can be used to classify spinal lesions, identify sites of significant intervertebral disc herniation and diagnose spinal subarachnoid cysts (Fig. 6.13).

Gradient recalled echo (GRE)^{36, 81, 94, 104, 110, 131, 179, 208, 224, 259, 265, 279, 286, 289, 294, 319, 350, 352, 366, 383–385, 387}

While the generation of SE images relies on the application of pairs of RF pulses, GRE sequences utilize only one initial RF pulse in conjunction with gradient field reversals (Fig. 6.14). GRE sequences use smaller flip angles and shorter TRs than SE sequences (Table 6.2), resulting in shorter scan times. The lack of a 180° pulse has important implications for image weighting and quality: (1) conventional GRE sequences can be used to acquire T1-W, PD-W and T2*-W images—acquisition of truly T2-W images is not possible—and (2) GRE sequences are prone to susceptibility artifacts as there is no compensation for external field inhomogeneities. A plethora of GRE applications have been developed in recent years, including conventional, steady-state, coherent, incoherent (spoiled), steady-state-free precession, balanced, fast, single-shot and echoplanar imaging (EPI) sequences. A comparison between sequences developed by different vendors is difficult as a uniform nomenclature system does not exist. For example, vendor-specific names/acronyms used for a similar coherent gradient echo sequence are “FISP” (Siemens), “GRASS” (GE), “FFE” (Philips), “Rephased SARGE” (Hitachi) and “SSFP” (Toshiba). Rapid development in the field of GRE sequences led to numerous new and advanced applications of MR imaging, such as motion-free (breath hold) abdominal imaging, 3D volumetric imaging, and 3D MR angiography (MRA). Some GRE sequences more or less routinely used for the imaging of the CNS in small animals at this point include T2*-W GRE, T1-W 3D GRE, 2D/3D volumetric acquisition, and magnetic resonance angiography (MRA).

• T2*-W GRE sequence: Gas interfaces, soft-tissue mineralization, fibrous tissue, and certain blood degradation products (e.g. methemoglobin) cause magnetic field inhomogeneities which appear as a signal void (susceptibility artifact, see below) on T2*-W images. T2*-W is most commonly utilized to identify intracranial or spinal hemorrhage and to differentiate it from other lesions (Fig. 6.15; see also Figs 6.38, 6.41B and 6.50). Additional indications may include identification of intracranial mineralization (e.g. in meningiomas) or abnormal gas pockets (e.g. in brain abscesses).
  
• T1-W 3D GRE: These sequences may be beneficial to evaluate small structures (e.g. pituitary gland or cranial nerves; Fig. 6.16) after intravenous administration of contrast medium, as they allow acquisition of thin slices (< 1 mm) without interslice gap and permit multiplanar reconstruction of the 3D dataset in additional planes.
  
• 2D/3D volumetric acquisition (steady-state-free precession): These sequences are characterized by high contrast between fluid-filled structures and surrounding tissues and may be beneficial in evaluating small structures, such as the inner ear.
  
• Magnetic resonance angiography (MRA): MRA techniques maximize vascular contrast by enhancing the signal from spins in flowing blood and/or suppressing the signal from surrounding stationary tissues. Although this can be accomplished without contrast medium administration (digital subtraction MRA, phase contrast MRA, time-of-flight MRA) contrast-enhanced MRA (CE-MRA) is considered a superior technique due to improved image quality. In humans, evaluation of the intracranial circulation provides valuable information in the diagnosis and prognosis of various abnormalities such as aneurysms, arterial and venous steno-occlusive diseases, inflammatory arterial diseases, and congenital vascular abnormalities. Although intracranial vascular abnormalities are infrequently reported in the veterinary literature, MRA might be considered a quick and low-risk procedure to evaluate intracranial vessels in select cases.

Functional imaging^{18, 25, 37, 45, 48, 62, 126, 142, 143, 154, 194, 223, 238, 253, 268, 284, 297, 334, 352, 370, 381, 385, 399}

• Diffusion-weighted imaging (DWI): Diffusion describes the motion of molecules in tissues. This process is not truly random due to the presence of physiologic boundaries (cell membranes etc.) and is referred to as “apparent diffusion.” DWI utilizes opposing gradients to produce signal differences based on mobility and direction of water diffusion. Normal tissues have more water mobility, resulting in greater signal loss, while tissues with less water mobility experience restricted diffusion. In human as well as veterinary medicine DWI is most commonly used in the diagnosis of ischemic stroke. In acute cerebral ischemia, restricted diffusion occurs secondary to failure of the cell membrane ion pump and subsequent cytotoxic edema. An acute stroke is characterized by marked hyperintensity on a DWI and hypointensity on a synthesized apparent diffusion coefficient (ADC) map derived from two or more DWIs (Fig. 6.17). In humans DWI is also used to differentiate benign from malignant lesions and distinguish neoplasia from edema or infarction. While some initial studies in animals have yielded promising results, the value of DWI in the diagnosis of neurologic disorders other than acute stroke remains to be determined. Diffusion tensor imaging (DTI) is a specialized DWI technique which utilizes strong multidirectional gradients to map white matter tracts (Fig. 6.18). Initial studies proved feasibility of this technique in dogs, which may ultimately aid in the diagnosis of white matter disease and facilitate surgical planning.
  
• Perfusion-weighting imaging (PWI): PWI allows an estimate of blood volume passing through the capillary bed per unit of time. This is most commonly accomplished by tracing the passage of a bolus of contrast agent through the cerebral vasculature. Perfusion imaging is often used in combination with DWI in patients with acute ischemic stroke, where the difference between diffusion and perfusion abnormalities provides a measure of the ischemic penumbra (area of reversible ischemia that can be salvaged if blood flow is re-established promptly). Other potential applications include assessment of tumor malignancy based on metabolic activity and evaluation of tissue viability of vascular organs.
  
• Functional MR imaging (fMRI): This is a rapidly evolving area in human medicine which allows evaluation of brain activity during certain activities or following stimulation. Any activity/stimulus (e.g. viewing a picture or smelling food) results in activation of a specific area in the brain, which necessitates an increase in blood flow. The resultant focal increase in oxyhemoglobin and decrease in deoxyhemoglobin can be detected by means of MRI due to their inherent difference in magnetic susceptibility (blood oxygen level dependent, or BOLD, imaging). The result is a map of functional brain areas during a specific activity or after a specific stimulus (Fig. 6.19). Unfortunately, application of this technique in veterinary patients is thus far limited due to the need for immobilization or general anesthesia. However, initial attempts in conditioned awake dogs yielded promising results, and fMRI in animals is likely to gain importance with development of faster sequences and experience.
  
• Magnetic resonance spectroscopy (MRS): MRS is a method to measure tissue chemistry by recording signals from specific metabolites. In vivo MRS is most commonly performed using hydrogen protons (1H [proton] spectroscopy). Although other metabolites can be recorded using this technique, proton spectroscopy has the advantages of a high signal-to-noise ratio and a relatively short examination time as it can be added to conventional MR imaging protocols. Applications in neuroimaging include monitoring of biochemical changes occurring in tumors, metabolic disorders, inflammatory, and neurodegenerative diseases. With increasing availability of higher-strength magnets, MRS is gaining popularity in clinical veterinary medicine.

Technical modifications^{34, 56, 64, 65, 70, 99, 370, 373}

• Spatial presaturation: Spatial presaturation pulses are used to suppress undesired signals from anatomic areas within the imaging field of view. These pulses are not commonly applied in brain imaging but may be helpful in spinal imaging where they can be used to suppress signal from neighboring vessels or peristaltic bowel, thus minimizing motion artifacts.
  
• Fat saturation: Unlike STIR, which is an entirely separate MR sequence, selective fat saturation pulses can be applied to any sequence to suppress signal from fat without affecting signal intensity of other tissues. Fat saturation has proven especially beneficial when applied to postcontrast T1-W images of the brain and spine as it facilitates differentiation of contrast-enhancing lesions from adjacent fat and aids in the identification of meningeal enhancement (Fig. 6.20).
  
• Magnetization transfer imaging: Magnetization transfer pulses can be applied in SE or GRE sequences to produce additional signal suppression of tissue water. The technique may be used qualitatively to increase the visibility of lesions seen during MRA and following contrast administration or quantitatively to aid in the diagnosis of white matter disease.

Artifacts^{37, 57, 100, 101, 132, 307, 406}

• Motion: Motion artifacts are probably less common in veterinary than in human medicine as most patients are scanned under general anesthesia. However, even if the animal does not move during the scan, physiologically moving structures (e.g. the heart) will still result in artifacts. Motion artifacts always occur in the direction in which the phase encoding gradient was applied, regardless of the direction of motion. They manifest as “ghosts” of the moving structure at various locations along the phase axis, blurring and/or parallel bands. Remedies include better restraint of the patient, breath-hold techniques (limited in most systems due to duration of scan), cardiac/respiratory gating, presaturation pulses, flow compensation techniques, motion correction techniques, and flipping phase and frequency encoding gradients (if motion artifact interferes with evaluation of area of interest).
  
• Cerebrospinal fluid (CSF) flow artifacts: A CSF flow void artifact appears as artificial loss of signal from CSF, which is most commonly encountered on T2-W images. It is attributed to a rapid or turbulent flow of CSF, where flowing protons move so quickly that they are not exposed to both the initial 90° and the 180° refocusing RF pulse. It may occasionally be seen in normal dogs, but it seems more common in small-breed dogs with increased ventricular size and syringomyelia. As this artifact is more likely to occur with a smaller slice thickness and a longer TE, modification of imaging parameters will decrease severity. However, a decrease in TE will also result in an undesirable change in weighting and may not be feasible. A similar artifact known as “entry slice phenomenon” appears as an artificially high signal at a site of CSF flow when nonsaturated spins enter the imaging plane and generate a strong signal after application of the 90° pulse (Fig. 6.21). CSF flow associated artifacts can easily be identified by comparison with other sequences and image planes as they will not be consistent findings.
  
• Chemical shift artifact of the first kind: This is a misregistration artifact resulting from a minimal difference in precession frequency between fat and water protons. The MR unit is tuned to “listen” for hydrogen protons and expects them to precess at a certain frequency at a certain location. Hydrogen protons in fat precess ever so slightly slower than hydrogen protons in water. This minimal difference is enough to misplace the signal from fat on the resultant image. This artifact occurs in the direction of the frequency encoding (or “readout”) gradient and manifests at a strongly hypointense and an opposite strongly hyperintense crescent at any fluid-fat interface. This artifact is important to recognize as it occurs at the interface of subarachnoid space and epidural fat when imaging the spine and may mimic a lateralized lesion if the frequency encoding gradient is applied in a laterolateral (rather than dorsoventral or ventrodorsal) direction (Fig. 6.22). The effect of this artifact on diagnostic image quality can be minimized by swapping the phase and frequency encoding gradient, increasing the receiver bandwidth, and using fat-suppression techniques (see above).
  
• Susceptibility: Magnetic susceptibility is a term used to describe the magnetic properties of a material. Diamagnetic materials (e.g. soft tissues) have very low susceptibility and weaken a magnetic field. Paramagnetic materials (e.g. gadolinium and certain hemoglobin degradation products) have a slightly stronger susceptibility and focally enhance a magnetic field. Ferromagnetic materials (e.g. iron) become strongly magnetized and experience a large force when placed in an external magnetic field. Presence of materials with differing susceptibility in the field of view results in susceptibility artifacts. These range in severity from focal signal voids in case of hemorrhage (see Figs 6.15, 6.38, 6.41B, 6.50) to geometric image distortion, progressive or abrupt signal void in the area of the object, and areas of sharply defined high signal intensity adjacent to the object in case of ferromagnetic materials. Susceptibility artifacts may be beneficial (e.g. improved detection of small hemorrhagic infarcts) or detrimental (e.g. identification microchip immediately adjacent to area of interest). Possible remedies include the use of (F)SE rather than GRE sequences, decreasing TE, use of STIR rather than fat saturation pulses, swapping phase and frequency encoding gradients, and increasing receiver bandwidth.
  
• Partial volume averaging: This artifact occurs when materials of different intensity are included in the same slice thickness (or even the same voxel) and their intensities are averaged. This artifact is significant as hyperintensities adjacent to fluid-filled structures (subarachnoid space, ventricles) on T2-W images resulting from averaging of brain and CSF signal may be misinterpreted as parenchymal lesions. Remedies include decreasing slice thickness, verification of lesions on additional planes, and verification of any T2 abnormality on FLAIR.

Contrast media in MRI^{62, 64, 116, 119, 150, 186, 271, 275, 276, 310, 405}

MRI contrast agents most commonly used in veterinary medicine are gadolinium-based. Contrast medium is administered at a dose of 0.1 mmol/kg, which may be increased to improve detection of poorly enhancing lesions. Enhancement is seen if a lesion is vascularized and is located outside the blood–brain barrier. Gadolinium-based contrast media predominantly affect T1 relaxation, and enhancing lesions appear hyperintense on T1-W images (see Figs 6.6, 6.16, 6.20). Certain normal intracranial structures outside of the blood–brain barrier—such as pituitary gland, choroid plexus, trigeminal nerve, and blood vessels—show physiologic contrast uptake. Some disagreement exists as to the optimal timing of MR image acquisition following contrast medium administration (immediate vs. delayed). However, at least for the brain, postcontrast images are usually acquired in more than one plane, and immediate and delayed images are therefore acquired inadvertently. Dynamic studies monitoring contrast enhancement (wash-in and wash-out) over time are not commonly used in veterinary medicine at this point but may be useful for the evaluation of the pituitary gland, cerebral perfusion, and brain tumors. MR contrast agents are considered very safe for use in veterinary patients, and reports of adverse effects are limited to one publication describing suspected anaphylactoid reactions in three dogs.

MRI guided tissue sampling^{49, 91, 123}

MRI of the brain

Associated findings in intracranial disease^{12, 14, 15, 46, 68, 71, 72, 75, 82, 86, 129, 162, 163, 165, 180, 219, 229, 236, 304, 323, 333, 348, 351, 353, 371, 376, 380, 400}

MRI findings in hydrocephalus include dilation of one or more ventricles and/or dilation of the subarachnoid space. In most cases, abnormal CSF accumulation appears hyperintense on T2-W images, hypointense on T1-W images, and attenuates on FLAIR (Fig. 6.23). If CSF contains abnormal cells and/or protein (e.g. in cases of inflammation or intraventricular hemorrhage), altered signal intensity may be observed. Hydrocephalus may be associated with periventricular edema characterized by periventricular T2 hyperintensity, which is most obvious on FLAIR images. In extreme cases CSF may dissect along periventricular white matter, creating diverticula and clefts. Dependent on the etiology of hydrocephalus, potential concurrent findings include other congenital anomalies, an intracranial mass, or signs of trauma. Imaging diagnosis of pathologic hydrocephalus can be challenging. In one study describing ultrasound evaluation of canine ventricles, normal lateral ventricular height was reported to be 0–14% of dorsoventral height of the cerebral hemisphere, moderate ventricular enlargement was defined as 15–25%, and > 25% was considered indicative of severe ventricular enlargement. However, ventriculomegaly and ventricular asymmetry are common findings in asymptomatic animals, and this finding may or may not be clinically significant. Progressive dilation of ventricles and subarachnoid space are also anticipated findings with increasing age and may not be pathologic. Therefore, imaging diagnosis especially of mild ventricular and/or subarachnoid space dilation should be judged in the light of clinical presentation.

• Vasogenic edema: Vasogenic brain edema may be seen concurrently with a number of intracranial diseases. Under normal circumstances, exchange of substances between the blood and the brain is limited by the blood–brain barrier. Damage to brain capillaries results in leakage of fluid into the extracellular space (vasogenic edema). The edema migrates along the white matter fiber tracts and may create a mass effect. Vasogenic edema appears hyperintense on T2-W MR images and hypointense on T1-W images and often has the same signal intensity as the lesion causing the edema (see Figs 6.8, 6.10, 6.15, 6.34, 6.42, 6.46, 6.49). After contrast medium administration, gadolinium leaks out of damaged capillaries and may result in increased signal intensity of the underlying lesion, while edema remains isointense to hypointense (see Fig. 6.6).
  
• Mass effect: Space-occupying lesions within the cranial vault (e.g. tumor, abscess/granuloma, edema, hydrocephalus) are commonly associated with a mass effect, which is indicated by displacement of the falx cerebri or compression of the ventricular system (see Figs 6.6, 6.8, 6.10, 6.15, 6.24, 6.31, 6.32, 6.34, 6.41, 6.42, 6.46, 6.49).
  
• Brain herniation: Increase in intracranial pressure (e.g. due to an intracranial mass) can lead to compression and displacement of brain parenchyma. Foramen magnum herniation (herniation of the caudal portion of the cerebellum into and through the foramen magnum), subfalcine (herniation of part of the cerebral cortex across midline into the opposite half of the cranial vault), and caudal transtentorial herniation (displacement of portions of the cerebral cortex ventral to the tentorium cerebelli) may be encountered (see Fig 6.24, Fig. 6.42, Fig. 6.49).
  
• Seizures and cerebral necrosis: The relationship between seizures and evidence of structural brain damage seen on MRI remains diagnostically challenging. In humans, severe seizure activity causes reversible changes in certain areas of the brain, such as the neocortex, hippocampus, and amygdala. Brain parenchymal changes including edema, neovascularization, reactive astrocytosis, and acute neuronal necrosis have been reported in dogs with seizures and manifest as unilateral or bilateral T2 hyperintense and T1 hypointense foci with variable contrast enhancement associated with piriform and/or temporal lobes (Fig. 6.25). Resolution of these changes on recheck examination indicates that they most likely represent sequelae to rather than underlying cause of seizures. On the other hand, cerebral cortical necrosis (polioencephalomalacia) appearing as increased T1 and T2 signal intensity of the gray matter of temporal and parietal lobes with mild contrast enhancement reported in a dog with seizures and necrosis of the hippocampus and the piriform lobes in cats with seizures appearing as bilaterally symmetric T2 hyperintense lesions with variable contrast enhancement are thought to be the underlying cause rather than result of seizures. In some cases it may not be possible to differentiate whether brain changes found on MRI and/or histopathology represent the underlying cause or are the result of seizures.

Congenital brain disorders

Forebrain (telencephalon and diencephalon)^{66, 127, 145, 146, 158, 193, 212, 226, 229, 295, 303, 313, 314, 332, 333, 343}

Congenital hydrocephalus is most commonly seen in toy and brachycephalic breed dogs and appears as dilation of the ventricular system of variable severity as described above (see Fig. 6.23).

Hydranencephaly manifests as a near complete destruction and/or lack of development of the neocortex and has been described in two kittens following intrauterine parvovirus infection. Affected animals show a reduction of size of one or both cerebral cortices to a thin mantle surrounding a large, centrally located cavity.

Porencephaly appears as cystic cavities in the cerebrum due to cell destruction or failure of development. These cavities show signal typical of CSF on MRI and may communicate with ventricles or subarachnoid space (Fig. 6.26).

Lissencephaly is a disorder of cortical neuronal migration characterized by paucity, absence, and/or hypoplasia of cerebral gyri and thickening of the cerebral cortex. The disease has been reported in dogs and cats and appears to be hereditary in Lhasa Apsos. MRI findings include a smooth cerebral surface and a thick neocortex with absence of the corona radiata.

Holoprosencephaly (HPE) is a failure of the forebrain to bifurcate normally and is characterized by an absence or reduction in the size of midline prosencephalic structures (corpus callosum, septum pellucidum, septal nuclei, fornix, and optic nerves), incomplete separation of normally paired forebrain structures (lateral ventricles, cingulate gyri, and caudate nuclei), and hydrocephalus. Dependent on severity, HPE can be subdivided into alobar, semilobar, and lobar HPE.

Agenesis or dysgenesis of the corpus callosum is a feature of HPE but may also occur as an isolated abnormality. Miniature Schnauzers appear to be predisposed both for this condition as well as for lobar HPE. Agenesis or dysgenesis of the corpus callosum is best seen in midsagittal MR images. On transverse images, unusually upturned, pointed corners of the lateral ventricles are evident (Fig. 6.27).

Protrusion of meninges alone or meninges along with brain tissue through a calvarial defect are termed meningocele and meningoencephalocele, respectively. Superficial meningoencephaloceles may be diagnosed clinically; however, diagnosis of basal or ethmoidal meningoencephaloceles requires advanced imaging. Ethmoidal encephaloceles are characterized by a defect in the cribriform plate with rostral herniation of the olfactory bulb(s) (Fig. 6.28).

Rathke cleft cysts are pituitary cysts containing mucoid or, less commonly, serous fluid and cellular debris and appear as cystic lesions in the middle cranial fossa. They are hypointense on T1-W images, hyperintense on T2-W images, may show mild ring enhancement, and may not suppress on FLAIR due to composition of fluid.

Midbrain and hindbrain (mesencephalon, metencephalon, myelencephalon)^{33, 41, 43, 44, 53, 55, 60, 63, 67, 73, 76, 78, 105, 152, 164, 166, 167, 198, 209, 210, 212, 213, 220, 229, 230, 240, 264, 267, 272, 282, 291, 292, 299, 302, 313, 327, 333, 339, 361, 363, 368, 369, 401}

Chiari malformations are a group of structural defects involving brain stem, cerebellum, upper spinal cord, and surrounding bony structures in humans. Secondary formation of a cystic cavity within the cervical spinal cord parenchyma and/or dilation of the central canal (syringomyelia) are common. A disorder similar to Chiari type I malformation in humans termed “Chiari-like malformation and syringomyelia” has been reported in dogs. Cavalier King Charles Spaniels are most commonly affected, but the disease is seen in a variety of breeds and can be found in symptomatic and asymptomatic animals. The condition is characterized by crowding of the caudal fossa resulting in attenuation of the subarachnoid space surrounding the cerebellum, compression and, in severe cases, herniation of the cerebellum into or through the foramen magnum, which is best demonstrated on T2-W median-sagittal images. Additional findings include a focal bending of the cranial aspect of the spinal cord, hydrocephalus, and syringomyelia (Fig. 6.29).

Atlanto-occipital overlapping without prior trauma is a recently recognized anomaly in small- and toy-breed dogs, which is usually associated with other abnormalities such as Chiari-like malformation, syringomyelia, occipital dysplasia, or atlantoaxial instability. The abnormality is best identified on midsagittal images, which reveal craniodorsal displacement of the atlas and overlap of its lamina with the occipital bone (Fig. 6.30).

Cerebellar hypoplasia can occur as a primary developmental defect or secondary to viral infection in utero, most commonly parvovirus. The cerebellum may appear small, with increased CSF signal noted around and extending into the folia. Congenital cerebellar abnormalities may not always be apparent on MRI examination; and no abnormalities were detected in Coton de Tuléar dogs with neonatal cerebellar ataxia. Differentiation of true cerebellar hypoplasia from degenerative disease (cerebellar atrophy, abiotrophy, degeneration; see below) may not be possible solely based on imaging findings.

The Dandy–Walker malformation complex in human patients refers to a group of congenital CNS anomalies that primarily involve the cerebellum and adjacent tissues. The primary abnormality is partial or complete absence of the cerebellar vermis and cystic dilation of the fourth ventricle. Comparable cases of cerebellar vermian aplasia or hypoplasia and associated cystic dilation of the fourth ventricle have been reported in dogs. Possible concurrent findings include generalized ventricular enlargement, extension of the cystic fourth ventricle into the supratentorial space with displacement of the occipital lobes, reduced size of the cerebellar hemispheres, widening and irregular gyrification of cerebral sulci, and absence of the corpus callosum.

Polymicrogyria is a disorder of cerebrocortical development resulting in an increased number of small, disorganized gyri in the dorsal and lateral cerebral cortex. It has been reported in Standard Poodles in which a hereditary basis is suspected. On MRI abnormal gyri are best seen on dorsal T2-W images. Concurrent ventriculomegaly is common.

Intracranial epidermoid and dermoid cysts are benign space-occupying lesions that originate from remnants of ectodermal tissue due to defects of neural tube closure. They are often located in the cerebellopontine angle or the fourth ventricle. Signal intensity is variable and dependent on cyst content: cysts with a high lipid content appear hyperintense on T1-W and T2-W images, while cysts with lower lipid content appear hypointense on T1-W images and hyperintense on T2-W images. Dermoid cysts containing adnexa (e.g. hair) may show suspended low-intensity foci in all sequences. Cysts often contain protein and keratin and therefore do not attenuate on FLAIR (Fig. 6.31). They usually do not show contrast enhancement, although ring enhancement may occasionally be noted.

Intracranial intra-arachnoid cysts (more appropriately termed diverticula) arise from splitting/duplication of the arachnoidea and occur in close association with an intracranial arachnoid cistern. They are usually considered a primary malformation, although they may develop secondary to inflammation or trauma. Quadrigeminal cistern cysts dorsal to the quadrigeminal plate are most common, but cerebellomedullary cistern cysts have also been reported. Intracranial intra-arachnoid cysts contain fluid isointense to CSF, with attenuation on FLAIR and no evidence of contrast enhancement (Fig. 6.32). Hemorrhage into intracranial intra-arachnoid cysts may change the signal intensity. Quadrigeminal cysts are of variable significance and are frequently incidental. One study found that compression of the occipital lobe by the cyst greater than 14% on median-sagittal image was always associated with clinical signs, while no association was found between the degree of cerebellar compression and clinical signs.

Other cystic lesions documented in the caudal fossa in dogs include cerebellar ependymal cysts and choroid plexus cysts. On MRI these are well circumscribed, strongly hyperintense on T2-W, and of variable intensity on FLAIR images. After contrast medium administration they may be nonenhancing, ring-enhancing, or, surprisingly reported in a case of a choroid plexus cyst, strongly enhancing.

Acquired brain disorders

Inflammatory brain diseases^{27, 52, 64, 190, 237, 250, 344}

Inflammatory brain diseases can affect brain parenchyma (encephalitis), meninges (meningitis), or both (meningoencephalitis). Dependent on underlying etiology, involvement of the spinal cord (myelitis/meningomyelitis) may occur. Encephalitis may cause no detectable abnormalities on MRI or may manifest as multifocal (rather than focal) lesions associated with brain parenchyma which typically appear hyperintense on T2-W images and hypointense on T1-W images. FLAIR has higher sensitivity than conventional SE sequences in detecting subtle brain lesions in dogs with clinical signs of multifocal brain disease, and its use is encouraged in all of these cases. Meningitis may not be detected with MRI or may appear as meningeal enhancement following administration of contrast medium (see Fig. 6.20).

Infectious inflammatory brain diseases^{4, 13, 17, 19–21, 24, 29, 38, 59, 69, 71, 93, 97, 106, 114, 120, 133, 134, 148, 171, 174, 175, 191, 192, 199, 221, 228, 239, 249, 250, 255, 258, 308, 320, 328, 331, 335, 344, 384, 392, 404}

Canine distemper virus (CDV) and feline coronavirus (FCoV; previously called “feline infectious peritonitis virus,” FIPV) are the most common causes of viral encephalitis in dogs and cats, respectively. In acute CDV infection, T2 hyperintense lesions and loss of contrast between gray and white matter on T2-W images may be found in the cerebellum and/or brain stem, corresponding to areas of demyelination. T2 hyperintense areas are occasionally seen in the temporal lobes, which may be related to infection or postictal edema. MRI findings in chronic distemper meningoencephalitis include essentially bilaterally symmetric T2 hyperintensity of the cortical gray/white matter junction of the parietal and frontal lobes, T2 hyperintensity of the arbor vitae of the cerebellum with partial loss of cerebellar cortical gray/white matter demarcation, subtle focal T2 hyperintensity of the pons, and meningeal contrast enhancement. In cats with feline infectious peritonitis (FIP) affecting the central nervous system MRI may show T2 hyperintensity and contrast enhancement of the ventricular lining, choroid plexus, and meninges compatible with ependymitis, choroiditis, and meningitis (Fig. 6.33). Concurrent findings may include hydrocephalus, syringomyelia, and herniation of the cerebellum secondary to increased intracranial pressure. A study investigating MRI findings in dogs experimentally infected with rabies found parenchymal abnormalities affecting the hippocampus, hypothalamus, basal ganglia, brain stem, and white matter which were more pronounced in the paralytic than the furious stage of the disease.

Mechanisms of bacterial infection of the CNS in cats and dogs include hematogenous spread, contiguous infection from adjacent structures (middle/inner ear, nasal cavity, sinuses, orbit, skull, vertebrae), direct inoculation (trauma, bite wound, surgery), and migration of foreign bodies or aberrant parasites. In addition to meningitis, diffuse cerebritis, and meningoencephalomyelitis, CNS infection may result in focal parenchymal abscesses or empyema in the subdural or epidural space. MRI features of intracranial infection secondary to plant foreign body migration, hematogenous spread from a mediastinal abscess, bacterial endocarditis, local extension from orbital disease and from osteomyelitis of the sphenoid bone, and as complication of otitis media or interna have been described in small animals. Intracranial abscesses are typically hypointense on T1-W and hyperintense on T2-W images, with strong peripheral contrast enhancement and associated brain edema (Fig. 6.34). Concurrent meningitis appearing as meningeal enhancement and/or thickening is common. Dependent on lesion location and extent, additional findings might include hydrocephalus and/or brain herniation. Neurotuberculosis (tuberculous meningitis, cerebral tuberculoma) results from hematogenous spread of organisms of the Mycoplasma tuberculosis complex and is infrequently reported in animals. MRI findings in one affected dog included a large heterogeneous T2 isointense to hypointense extra-axial mass with patchy peripheral contrast enhancement and hyperostosis and erosion of adjacent skull bones.

Fungal infections (Cryptococcosis, Phaeohyphomycosis, Aspergillosis, Blastomycosis, Histoplasmosis) have been reported to affect the CNS in dogs and cats. MRI findings are variable and may include intra-axial or extra-axial masses of variable contrast enhancement, gelatinous pseudocysts, meningeal enhancement, enhancement of the ventricular lining, and intra-cranial extension of nasal or retro-orbital masses (Fig. 6.35). Associated findings may include edema, hydrocephalus, syringohydromyelia, and brain herniation.

Protothecosis is a rare infection with the achlorophyllic algae Prototheca, and only a few reports of CNS involvement are found in the veterinary literature. MRI findings in one dog included multifocal T2 hyperintense intra-axial lesions, ventriculomegaly, cerebellar herniation, syringomyelia, and meningeal enhancement.

Parasitic meningoencephalitis in dogs and cats is caused by aberrant migration of parasites such as Dirofilaria, Baylisascaris, Cuterebra, Taenia, Ancylostoma, Toxascaris, and Angiostronglylus. MRI features include focal or multifocal parenchymal lesions of variable signal intensity and peripheral parenchymal and/or meningeal contrast enhancement. Intraparenchymal hemorrhage is a common feature in parasite migration, and T2*-W images are especially useful in establishing a (presumptive) diagnosis.

Protozoal meningoencephalitis may be caused by a variety of organisms. MRI features in dogs infected with Neospora include cerebellar atrophy with T2 hyperintense material surrounding the cerebellum and extending into the sulci, loss of contrast between cerebellar gray and white matter, T2/FLAIR hyperintensities within the cerebellum, and meningeal contrast enhancement. In two dogs with systemic Leishmaniasis nonenhancing T2 and FLAIR hyperintense lesions in thalamus and brain stem found on MRI were attributed to ischemic infarcts secondary to systemic necrotizing vasculitis. Toxoplasma infection in cats may manifest as multifocal indistinct T2 hyperintense contrast-enhancing parenchymal lesions with associated brain edema.

Noninfectious inflammatory brain diseases/ meningoencephalomyelitis of undetermined etiology (MUE)^{30, 40, 51, 58, 89, 137, 138, 151, 155, 157, 168, 172, 188, 189, 204, 206, 298, 326, 356, 375, 396, 402}

Several inflammatory conditions unrelated to infectious agents have been identified in dogs which generally respond to immunosuppressive treatment suggesting immune mediated etiology. Attempts have been made to separate these into specific diseases based on breed (e.g. “Pug dog encephalitis”) and other criteria, however, due to overlap in clinical, diagnostic and pathologic findings a definitive premortem diagnosis is difficult. These disorders may be summarized under Meningoencephalomyelitis of undetermined etiology (MUE).

Granulomatous meningoencephalitis (GME) can affect any breed of dog but most often occurs in young to middle-aged toy breeds. The disease can affect the brain and/or spinal cord. On MRI, GME lesions can be focal or multifocal and commonly affect the brain stem. Although the disease has a predilection for white matter, it is not associated with distinct topography. Lesions are typically hyperintense on T2-W and FLAIR images, iso- to hypointense on T1-W images and variably contrast enhancing (Fig. 6.36). Meningeal enhancement may or may not be observed. In ocular GME, MRI may show enlargement of the optic chiasm and fairly symmetric contrast enhancement of the optic nerves, optic chiasm, and visual pathways.

Necrotizing meningoencephalitis (NME) is characterized by cavitary necrosis in the neuroparenchyma. The disease was initially described in the pug breed (“pug dog encephalitis”), but similar disorders have since been reported in other small breeds including Maltese, Chihuahua, Pekingese, French Bulldog, Shi Tzu, and Lhasa Apso. A distinct form of NME described mainly in Yorkshire Terriers has been termed necrotizing leukoencephalitis (NLE). Descriptions of imaging findings in this group of inflammatory brain disorders are not available for all breeds. NME in pugs is usually limited to the forebrain. MRI findings include diffuse asymmetric cerebral lesions which are nonuniformly hyperintense on T2-W images, isointense to hypointense on T1-W images, and affect gray and white matter resulting in loss of gray/white matter distinction. Additional findings include variable degrees of contrast enhancement of brain and leptomeninges, enlarged and asymmetric lateral ventricles, mass effect, brain herniation, and T2/FLAIR hyperintensity associated with the hippocampus and piriform lobes. MRI findings reported in Chihuahuas and French Bulldogs with NME are similar; however, brain stem involvement seems more common. In almost all cases of NLE reported in Yorkshire Terriers, lesions are located in the cerebrum and brain stem and appear iso- to hypointense on T1-W images, hyperintense on T2-W and FLAIR images, and show variable contrast enhancement. Concurrent hydrocephalus of variable severity is possible.

Other inflammatory intracranial diseases^{250, 306, 341, 392, 398}

Greyhound nonsuppurative meningoencephalitis manifests as T2/FLAIR hyperintense lesions with minimal or absent contrast enhancement associated with olfactory lobes, frontal and frontotemporal gray matter, and caudate nuclei bilaterally. MRI in dogs with idiopathic eosinophilic meningitis/meningoencephalitis may be normal or show patchy regions of T2 hyperintensity and contrast enhancement in the cerebral cortex, solitary or multiple masses, meningeal enhancement, or enlargement and contrast enhancement of cranial nerves. In a boxer with cerebral extension of steroid-responsive meningitis arteritis, MRI findings included symmetrical multifocal to diffuse T2/FLAIR hyperintense changes of the cerebral gray matter and ependymal lining, increased periventricular signal, and a large amount of intraventricular sediment. A case of lymphocytic meningoencephalitis in a cat was characterized by multifocal T1 isointense and T2 hyperintense contrast-enhancing lesions. No intracranial abnormalities were detected on MRI in a cat diagnosed with histiocytic encephalitis. Trigeminal neuritis in dogs is characterized by diffuse enlargement of the nerves and is usually bilateral. Affected nerves are isointense to brain parenchyma on T1-W and PD-W images, isointense or hyperintense on T2-W images and show contrast enhancement.

Cerebrovascular disease

Stroke (cerebrovascular accident)^{3, 5, 7, 11, 26, 28, 35, 39, 45, 62, 79, 104, 107–109, 112, 114, 117, 126, 141, 148, 149, 153, 160, 181, 182, 196, 208, 216, 232, 263, 300, 329, 333, 345, 347, 352, 354, 383, 384, 389}

A stroke is a suddenly developing focal neurologic deficit resulting from an intracranial vascular event. In ischemic stroke, blood flow to an area of tissue is compromised due to intracranial arterial or venous obstruction. Hemorrhagic stroke results from the rupture of intracranial blood vessels.

Ischemic strokes can be categorized according to anatomic site, size, age, type (pallid or hemorrhagic), pathology (arterial vs. venous), mechanism (thrombotic, embolic, hemodynamic), and etiology. Causes of ischemic stroke in dogs include primary hypothyroidism, hypertension and diabetes, embolic metastatic tumor cells, necrotizing vasculitis due to systemic leishmaniasis, chronic renal disease, hyperadrenocorticism, intravascular lymphoma, septic thromboemboli, fibrocartilaginous embolism, migrating parasite or parasitic emboli (Dirofilaria immitis), and hypercoagulable state associated with hyperadrenocorticism, protein-losing nephropathy (PLN), or neoplasia. In approximately 50% of dogs with ischemic stroke, no underlying condition is identified. In cats, hypertrophic cardiomyopathy, neoplasia (lymphoma), and liver disease including hepatic lipidosis have been identified as potentially predisposing factors. Territorial infarcts occur when one of the main arteries supplying the brain is occluded. Lacunar infarcts are subcortical infarcts limited to the vascular territory of an intraparenchymal superficial or deep perforating artery. Small-breed dogs are more likely to have territorial cerebellar infarcts, and large-breed dogs are more likely to have lacunar thalamic or midbrain infarcts. Cavalier King Charles Spaniels and Greyhounds may be predisposed to infarcts. Watershed infarcts are infarcts in the boundary zone between large artery territories and are uncommon in veterinary patients. Diffusion weighted imaging (DWI) is sensitive to alterations in brain parenchyma following stroke and is capable of demonstrating abnormalities within minutes of an event (see above). Restricted diffusion (impairment of normal Brownian motion) occurs secondary to failure of the cell membrane ion pump and subsequent cytotoxic edema. This appears as marked hyperintensity on DWI and hypointensity on a synthesized ADC map derived from two or more DWI (see Fig. 6.17

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Signalment and History: The First Considerations
2. Seizures and Narcolepsy
3. Head-Trauma Management
4. Myopathies: Disorders of Skeletal Muscle

Stay updated, free articles. Join our Telegram channel

Join