Magnetic Resonance Imaging for Epilepsy Surgery

Epilepsy Protocol

Three-tesla MRI scanners provide high-resolution images and whenever possible should be preferred over 1.5T scanners. Images should be acquired in an oblique coronal orientation perpendicular to the long axis of the hippocampus. The entire brain should be included in the field of view to avoid missing subtle peripheral findings. Routine MRI sequences ^, include the following:

• Volumetric T1-weighted data set with slices of 1- to 2-mm thickness

• Axial and sagittal high-resolution T2-weighted data set with slices of 2- to 3-mm thickness

• Coronal fluid-attenuated inversion recovery (FLAIR) sequence in the same plane

• Three-dimensional FLAIR sequences

• Coronal T2∗-weighted gradient echo sequence (sensitive to paramagnetic substances such as hemosiderin, useful for ruling out hemorrhagic lesions such as cavernomas)

The routine MRI acquisition protocol in the diagnosis of epilepsy should be standardized to provide a widespread standard of care and also to encourage multicenter collaboration in clinical practice and research.

Optional Sequences

Many advanced MRI techniques are available and in use in different institutions with protocols available from multiple associations including the ILAE. ^{, ,}

• Magnetization transfer imaging: images are obtained with and without magnetization transfer to derive a magnetization transfer ratio. This ratio is a measure of the magnitude of exchange of magnetization between free protons in bulk water and tightly bound protons on macromolecules such as myelin or membrane. A low magnetization transfer ratio, caused by gliosis, is a sensitive marker of mesial temporal sclerosis.

• T2-weighted relaxometry imaging: dual-echo T2-weighted imaging is performed, and the voxelwise T2 relaxation time is calculated according to the formula

• in which SI1 and SI2 are the signal intensities in the early and late echo images, with echo times TE1 and TE2, respectively. The T2 relaxation time is a useful identifier of hippocampal abnormality. The T2 relaxation time is higher in patients with mesial temporal sclerosis than in patients with a normal hippocampus, and an intermediate value can be observed in patients without qualitative MRI evidence of hippocampal sclerosis. In comparison with the normal hippocampus, in mesial temporal sclerosis the normal anterior-to-posterior T2 relaxation time gradient is accentuated (from higher to lower value).

• Diffusion-weighted imaging (DWI): areas of restricted diffusion, such as infarcts, abscesses, or necrosis, are highlighted. In cases of persistent seizures, restricted diffusion is also seen in the seizure onset zone and along the areas of propagation of discharges that cause the seizures. The restricted diffusion represents seizure-induced cytotoxic edema.

• Magnetic resonance spectroscopy (MRS; described later).

The use of gadolinium is not recommended on a routine basis unless images are suggestive of a tumoral lesion.

Ultra–high-field scanning offers the advantage of improved signal-to-noise ratio, and this technique is emerging as an additional option for detecting semi-occult epileptogenic lesions and providing higher-resolution images of the hippocampus. Although access to the technology remains a significant barrier to routine employment, other drawbacks include reports of nausea, light flashes, metallic taste, and noise during image acquisition. 7T has been demonstrated to detect increased foci of epileptogenic potential; however, clinical relevance of these lesions must be interpreted with respect to electrophysiology concordance. 7T has also been useful in detecting cortical dysplasias, vascular malformations, and polymicrogyria.

Hippocampal Sclerosis

Mesial temporal lobe epilepsy is the most common form of epilepsy in adults, and hippocampal sclerosis or mesial temporal sclerosis is the main pathologic process associated with it. When hippocampal sclerosis is correctly identified and surgically treated, the rate of success in achieving seizure freedom is high (60% at 5 years).

An essential requirement in a dedicated MRI protocol in a patient with mesial temporal lobe epilepsy is thin-slice (1- to 3-mm) images (T1- and T2-weighted) perpendicular to the long axis of the hippocampus. Additional sequences such as 3D double-inversion recovery can be helpful in identifying abnormal white matter changes in the anterior temporal lobe.

Hippocampal sclerosis is demonstrated as hippocampal atrophy on coronal T1-weighted images and as increased signal intensity on T2-weighted images ( Fig. 84.1 ).

Features associated with hippocampal sclerosis are as follows:

• Temporal horn dilation caused by volume loss of the hippocampus

• Atrophy of the structures that form the outflow tracts from the hippocampus (inferiorly, the parahippocampal gyrus; posteriorly, the ipsilateral fornix connected to the ipsilateral mammillary body)

• Atrophy of the structures indirectly connected with the hippocampus (the remainder of the temporal lobe, the thalamus, and the caudate nucleus)

Qualitative visual inspection of MRI by neuroradiologists expert in the field is sufficient to detect hippocampal sclerosis in 80% to 90% of cases. A careful assessment of the whole brain is mandatory because a second disease process can be found in 8% to 22% of patients, represented mainly by focal cortical dysplasia (FCD).

The limitation with sole visual inspection is that hippocampal sclerosis can be missed if it is mild or bilateral. The assessment of hippocampal sclerosis can be greatly improved by quantitative measures of hippocampal volumes and hippocampal T2 signal.

Jack and colleagues showed that manually measuring the hippocampal volume can improve sensitivity, enabling detection of atrophy in 75% to 90% of the hippocampi on the side that is congruent with the electrical onset of the seizures.

A significant difference in hippocampal volume (expressed as difference or ratio) between the two sides is a reasonable indicator of unilateral hippocampal sclerosis. However, there are disadvantages to comparing the volumes of the two hippocampi, including the inability to detect bilateral hippocampal sclerosis and the possibility of false lateralization in patients with an epileptogenic lesion that expands the hippocampus. These issues can be partially addressed by establishment of hippocampal volume reference values in a normal control population.

In clinical practice, the manual segmentation to assess the volume of the hippocampi is a tedious, time-consuming process that is dependent on the expertise of the examiner, with some interrater variability. There is much interest in the use of automated analyses that can reproduce similar results, making the process less time-consuming and more reliable. Coan and associates compared visual analysis, volumetry, and T2 signal of the hippocampus with 3T MRI in a population of 203 patients affected by mesial temporal lobe epilepsy. They concluded that quantitative techniques were advantageous, with new detection of hippocampal sclerosis in 28% of the patients with mesial temporal lobe epilepsy. Emerging technology such as 7T MRI has the potential to provide histologic information about the hippocampus such as the ability to visualize its internal architecture. 7T MRI has even been reported to illustrate volume loss and signal intensity that correlates with histologic grading in individual CA regions.

Malformations of Cortical Development

Malformations of cortical development (MCDs) can be divided into three groups:

• Group I: malformations resulting from abnormal proliferation of neuronal and glial cells. This group includes microcephaly and macrocephaly, FCDs without and with balloon cells (Taylor dysplasia types IIa and IIb, respectively), hemimegalencephaly, tuberous sclerosis, dysembryoplastic neuroepithelial tumors (DNETs), gangliogliomas, and gangliocytomas.

• Group II: malformations resulting from abnormal neuronal migration. This group includes lissencephaly and heterotopias.

• Group III: malformations resulting from abnormal cortical organization. This group includes polymicrogyrias, schizencephaly, type I FCDs, and mild MCDs.

In this chapter, we first describe the MRI characteristics of the less frequent MCDs in accordance with the classification by Barkovich and colleagues, and then we dedicate a specific section to FCDs.

Group I

Tuberous sclerosis can manifest with multiple tubers or hamartomas containing balloon cells, visible on MRI as a cortical thickening with a hyperintense T2 signal ( Fig. 84.2 ). Most affected patients have subependymal nodular tubers that can show calcifications and occasionally enhancement. Subependymal giant cell astrocytomas can also be present, and they must be differentiated from subependymal tubers because of their less benign natural history. For this purpose, a contrast-enhanced scan is recommended because subependymal giant cell astrocytomas typically show uniform enhancement. Other distinctive features are that they are generally located near the foramen of Monro, and they tend to grow over time, which leads to obstructive hydrocephalus in 15% of cases.

In hemimegalencephaly, hemispheric enlargement is associated with ipsilateral ventriculomegaly and may be associated with changes in signal intensity of white matter, heterotopia, and cortical thickening (see Fig. 84.2 ).

Group II

Lissencephaly literally means “smooth brain,” and the condition is characterized by a lack of development of gyri and sulci. On MRI, the lissencaphalic brain appears agyric, and the sulci and sylvian fissures are shallow. The cortex is usually thickened (pachygyric).

Heterotopia is the presence of gray matter in abnormal sites and may be localized (nodular) or generalized (band; Fig. 84.3 ). Nodular heterotopias produce a signal that is isointense in relation to gray matter and should be distinguished from tubers that can be isointense or hypointense in relation to gray matter on T1-weighted images and hyperintense on T2-weighted images. Subependymal nodular heterotopias, which are commonly bilateral and most frequently located in the occipital horn of the lateral ventricle, are found predominantly in female patients. The associated seizures are usually partial in nature.

Generalized or band heterotopias are often described as a double cortex. They consist of a ribbon of gray matter within white matter, which runs parallel to the overlying cortex, which may be normal or macrogyric.

Focal Cortical Dysplasia

FCDs are the most common group of MCDs in patients presenting with intractable epilepsy and the most common cause of epilepsy in children. ^,

Imaging findings of FCDs can be subtle. Therefore images must be reviewed taking into consideration a full knowledge of the patient’s epilepsy, semiologic findings, and neurophysiologic characteristics. In cases of a negative MRI, other advanced imaging modalities—such as fluorodeoxyglucose positron emission tomography (FDG-PET), single photon emission computed tomography (SPECT), and magnetoencephalography—should be considered to try to locate any subtle regional abnormality. 7T MRI has also been demonstrated to improve detection of FCD that is not seen at conventional MRI strengths. Protocols including FLAIR and gradient echo at 7T are reported to improve sensitivity.

Coregistration of structural MRI with advanced imaging tools allows the user to assess the spatial concordance of positive findings and also to reevaluate a “negative” MRI with the benefit of this new information.

A number of researchers have attempted to identify the different imaging characteristics of the various subtypes of FCD ( Fig. 84.4 ). Colombo and associates reported the following findings:

• Type I FCD can be completely cryptogenic. In this subtype, common findings are hypoplasia or atrophy at a lobar or sublobar level. They are frequently associated with subcortical white matter volume loss, which has an increased signal on T2-weighted images and T2-weighted FLAIR images and a decreased signal on T1-weighted images. Cortical thickness is usually preserved, although there can be mild blurring at the junction between gray and white matter. Type I FCD is most frequently located in the temporal lobe, is associated with hippocampal sclerosis in more than 70% of cases, ^, and is the FCD most commonly associated with developmental tumors. If structural MRI shows mass effect, cystic components, or calcifications, a tumor should be suspected, and the MRI should be repeated with gadolinium contrast material.

• Type II FCD is characterized by increased cortical thickness and blurring of the junction between gray and white matter on T1- and T2-weighted images. The subcortical area can show an increased signal on T2-weighted and FLAIR images and a decreased signal on T1-weighted images. This abnormality creates a “funnel” track between the cortex and the ventricle, which is typical of type II FCDs and explains why the condition is also termed transmantle dysplasia. Other associated findings are abnormal cortical gyration and sulcation with increased T2 signal of the affected cortex (which remains hypointense in comparison with the underlying white matter). It is not possible to distinguish type IIa (FCD with dysmorphic neurons) from type IIb (FCD with dysmorphic neurons and balloon cells) on MRI, although type IIb is often better delineated. In contrast with type I, type II FCD often occurs in extratemporal areas, with a predilection for the frontal lobe.