Neuroradiologic Evaluation for Epilepsy Surgery

Published on 26/03/2015 by admin

Filed under Neurosurgery

Last modified 26/03/2015

Print this page

This article have been viewed 2152 times

CHAPTER 55 Neuroradiologic Evaluation for Epilepsy Surgery

Suzan Dyve, Leif Sørensen, Adam N. Mamelak, William W. Sutherling, Gregory D. Cascino

Computed Tomography, Magnetic Resonance Imaging, and Functional Imaging (Diffusion Tensor Imaging, Positron Emission Tomography, and Functional Magnetic Resonance Imaging)

Neuroimaging may point to the cause of epilepsy, aid in the planning of surgery, and serve as a guide during the neurosurgical procedure. The success of surgical resection for intractable epilepsy is highly dependent on presurgical delineation of the region responsible for generating the seizures. The aim of this chapter is to provide an overview of specific imaging modalities that can aid in providing the best surgical decision for epilepsy patients. The causes of epilepsy are varied, and each patient requires an individual approach. Seizures can occur in normal brain (hypoxia or hypoglycemia), in an apparently structurally normal brain (genetic, biochemical, microstructural), or in brains with a definite focal or general structural abnormality. Imaging can be classified broadly as anatomic or functional, and not all patients require use of the full repertoire that is available today. The decision to carry out epilepsy surgery depends on the surgeon’s assessment of the potential benefit versus risk for each patient. Traditionally, therefore, epilepsy surgeons have made use of all modalities that can aid in decision making before and during surgery. In earlier neurosurgical and epilepsy-specific textbooks, chapters on imaging focused on computed tomography (CT) and magnetic resonance imaging (MRI). The rapid development and expansion of new methods in structural and functional imaging modalities present a challenge to keep up with and make use of. This chapter describes the more recent imaging techniques that have been adopted by neurosurgeons performing epilepsy surgery.

Anatomic imaging modalities include CT, MRI, and more recently, diffusion tensor imaging (DTI) based on water diffusion, an MRI technique that allows imaging of axonal white matter tracts. Functional imaging involves neurovascular coupling, where an increase in blood flow is associated with increased oxygen delivery, to satisfy the needs for increased oxygen and glucose consumption that occur in neurons with increased activity. Magnetic resonance spectroscopy (MRS) allows in vivo analysis of neurochemicals and their metabolites.

Some of these imaging tools require the injection of single-photon– or positron-emitting radioactive nuclides (single-photon emission computed tomography [SPECT] and positron emission tomography [PET]). Functional magnetic resonance imaging (fMRI) uses “blood oxygenation level–dependent” (BOLD) signals that arise when blood flow increases more than the oxygen consumption of active neurons. Magnetoencephalography (MEG) uses a relatively new “super-cooled” quantum interference device (SQUID) that registers changing magnetic fields with an extremely sensitive recording technique. Clinical uses for MEG include the detection and localization of epileptiform activity and localization of eloquent cortex for surgical planning. Common to all these techniques is the potential for converting function to images on a screen.

Computed Tomography

CT is no longer the modality of choice, except in places with limited availability of MRI. Before the advent of MRI, CT had supplanted even earlier imaging methods. It was especially useful in detecting causes of focal epilepsy and revealing tumors, calcification, vascular anomalies, and atrophy. Its shortcomings include concern for ionizing radiation and problems caused by the presence of bone artifacts, especially (for epilepsy) in the temporal fossa. In addition, the spatial resolution of CT is on the order of millimeters lower than that of MRI. Nonetheless, CT still has a place in most centers as an acute screening method, for example, in the emergency department or in the event of postoperative complications.

Magnetic Resonance Imaging

Because of its high sensitivity and excellent tissue contrast, MRI should be the first step in screening an epilepsy patient because it detects underlying structural pathology in as many as 75% of patients with refractory focal seizures. MRI is the primary imaging modality when a patient experiences the first seizure or when the clinician suspects epilepsy. Most centers will have performed screening studies that usually include T1-weighted (with contrast enhancement if a space-occupying lesion is suspected), T2-weighted, and fluid-attenuated inversion recovery (FLAIR) sequences in the axial, sagittal, and coronal planes. These studies can be sufficient to reveal focal pathology. If focal pathology as shown by MRI is concordant with that noted on electroencephalography (EEG), a better surgical outcome is achieved, and identifying a focal lesion in patients with refractory epilepsy remains one of the most important factors in determining surgical outcomes.¹ However, if no focal lesion is seen and if EEG suggests a focal etiology, further MRI and other imaging studies should be performed.² In recent years, detection of lesions with MRI has improved by optimizing scan protocols with the use of FLAIR sequences, diffusion-weighted images, and volume acquisition with three-dimensional reconstruction.

MRI protocols specific for epilepsy patients can differ but often include variations of the following ³:

1 Axial and sagittal T2-weighted imaging.

2 Coronal T2-weighted FLAIR, which dampens the signal of free water (cerebrospinal fluid) and thereby highlights areas of increased tissue water (gliosis and low-grade tumors).

3 Coronal T1-weighted FLAIR, which is extremely sensitive in defining gray and white matter and therefore useful in revealing migration anomalies (Fig. 55-1).

4 Axial T1-weighted three-dimensional spoiled gradient echo (SPGR), a thin-slice volume scan (1 to 2 mm, no gap) that can be reformatted in all planes. Images can be fused with other modalities in the neuronavigational system (DTI, fMRI, functional PET [fPET] or 2-deoxy-2-[¹⁸F]fluorodeoxyglucose (FDG)-PET, and source localization from EEG and MEG) (Fig. 55-2). Functional or metabolic data can be coregistered to the three-dimensional anatomy to provide excellent separation and detail of gray/white matter boundaries.

5 Coronal T2*-weighted, gradient echo sequence, which is sensitive to paramagnetic substances such as blood and hemosiderin and useful in revealing cavernous angiomas.

6 Axial diffusion-weighted imaging (DWI), which can reveal infarct, abscess, or necrosis.

FIGURE 55-1 Coronal T1-weighted fluid-attenuated inversion recovery image showing a migration defect (arrow) adjacent to the anterior horn of the left lateral ventricle.

(Courtesy of Dr. Leif Sørensen, Århus University Hospital, Denmark.)

FIGURE 55-2 Cortical dysplasia in the right parietal region. A, T2-weighted fluid-attenuated inversion recovery (FLAIR) images in the sagittal, coronal, and axial planes. B, T2-weighted FLAIR image fused with T1-weighted three-dimensional spoiled gradient echo and diffusion tensor imaging (DTI). C, Images coregistered to the neuronavigational system (Medtronic Stealth Station) and fused with DTI and functional positron emission tomography (fPET). Lesion, thin arrow; fPET for the hand, thick arrow.

(Courtesy of J. Frandsen, K. Vang, and S. Dyve, Århus University Hospital, Denmark.)

Recent imaging improvements include the following:

1 Arterial spin labeling (ASL), which measures cerebral blood flow (CBF) and is therefore similar in utility to SPECT but advantageous in that it is noninvasive and can be coregistered to structural MRI. A disadvantage is that ASL assesses CBF at the time of scanning and is therefore usually interictal. ASL has been found to be useful in lateralization and determining epileptic foci.⁴

2 Susceptibility-weighted imaging (SWI), which is extremely sensitive to iron in the brain, including the cortical layers, and can therefore reveal structures that were not previously identified on MRI. SWI is beginning to be used clinically and in the context of epilepsy and has proved helpful in visualizing small cavernous malformations and hemosiderin from old infarcts.⁵ Improvements in the signal-to-noise ratio (SNR) are achieved with the use of phased-array surface coils with larger arrays. Phased-array coils are placed closer to the brain than earlier coil types and increase cortical signal. Each coil receives a signal from only a small area of cortex, thereby reducing noise. Increasing the number of arrays from 4 in the 1990s to 32 at 1.5 and 3 T at present, with up to 128 channels under development, will probably improve image quality in the future.⁶

3 Fetal MRI has enabled the diagnosis of cortical malformations and tuberous sclerosis in utero.⁷

3 T and 7 T versus 1.5 T

Some lesions, such as cortical dysplasia, can be difficult to detect, and there is ongoing interest in using new advances in MRI, such as high-field and multichannel technology. It remains to be tested whether multichannel 3- or 7-T imaging can provide additional useful information. Increasing field strength from 1.5 to 3 and 7 T for the clinical evaluation of epilepsy has required optimization of imaging protocols but has shown that increased field strength improves the SNR and may therefore prove useful in the detection of cortical dysplasia. However, some have found that 1.5 T appears to be more sensitive in cases of tissue loss and mesial temporal sclerosis.⁸

Diffusion Tensor Imaging

DTI, also known as “fiber tracking,” is an MRI modality that has allowed three-dimensional study of white matter fiber bundles at visible resolution (millimeters). Previously, white fiber (axonal) tracts were mapped post mortem by using specialized preparations or chemical techniques. DTI provides imaging of major white matter bundles connecting functional groups of neurons (Fig. 55-3). Computer analysis can connect fiber bundles in close relation to a planned surgical field such as the motor cortex to motor areas in the brainstem. DTI can therefore aid in planning surgery by avoidance of not only eloquent cortex but also the white matter tracts that connect functional areas.

FIGURE 55-3 Cavernous angioma near the left internal capsule in a 23-year-old man with several episodes of paresis in the right arm and leg. Based on imaging studies and navigation, a path was chosen through the left frontal lobe, above Broca’s language area, in front of the motor strip, and through the defect shown by tractography. Paralysis of the right arm and leg occurred postoperatively but resolved completely within 3 months. The patient had slight, permanent clumsiness in fine finger movement of his right hand. A, Axial T2-weighted fluid-attenuated inversion recovery image. B, Magnetic resonance imaging (MRI) and functional positron emission tomography (fPET) superimposed. fPET showed language activation in the left inferior frontal lobe (synonym production) (arrow). C, Preoperative superficial three-dimensional MRI reconstruction with fPET superimposed (right hand activation) in the left motor area to facilitate operative planning. D, Diffusion tensor imaging (DTI) in the corpus callosum (pink). The corticospinal tracts are green and the lesion is white. E, MRI with DTI showing disruption of the corticospinal tracts that was presumed to be due to a mass effect of the lesion and several episodes of hemorrhage. F, Postoperative 3-T MRI.

(Courtesy of S. Dyve, J. Frandsen, and A. Rodell, Århus University Hospital, Denmark.)

DTI visualizes the preferred movement of water molecules in the brain. Water molecules diffuse preferentially along white matter tracts (anisotropic movement). A diffusion-sensitizing magnetic field gradient is applied in multiple cartesian directions and yields representations of structures such as nerve fibers in three dimensions. The technique has the potential for elucidating the characteristics of tissue microstructure and therefore offers insight into the cause of the epilepsy and location, as well as serving as a guide during surgery.

Because white matter tract trajectories are complex, the use of DTI necessitates knowledge of the core or seed regions (voxels) of interest, specifically, where the tracts originate, where they pass through, and where they end. For example, when mapping the corticospinal tracts, the seed areas are the cortex in the motor area, the internal capsule, and the peduncles of the midbrain (Fig. 55-4). Placement of a seed area in the motor cortex is often preceded by fMRI to enable precise choice of the tracts of interest (Fig. 55-5).

FIGURE 55-4 Cavernous angioma. A, T1-weighted magnetic resonance image showing cavernous angioma in proximity to the right posterior insula and internal capsule. B, Seed area placed in the internal capsule.

(Courtesy of J. Frandsen and S. Dyve, Århus University Hospital, Denmark.)

FIGURE 55-5 Meningioangiomatosis in proximity to the left motor area and corticospinal tract. A, T2-weighted magnetic resonance imaging (MRI) showing the corpus callosum (green) and a mesial defect in the corticospinal tracts (arrow). B, Sagittal T1-weighted MRI showing the mesial extent of the tumor with diffusion tensor imaging (DTI) superimposed (arrow). C, T1-weighted MRI, DTI, and functional positron emission tomography (motor movement of the tongue, arrow). D, Preoperative three-dimensional MRI surface reconstruction showing the planned surgical path. Central sulcus (arrow). Postoperative paresis of the tongue and lower part of the face resolved within 1 month.

(Courtesy of K. Vang, J. Frandsen, and S. Dyve, Århus University Hospital, Denmark.)

In clinical practice, it is probably not necessary to study more than 20 prominent tracts in the cerebrum and brainstem. Regions of special interest to the epilepsy surgeon are tracts that connect areas of special functional importance and include the corticospinal tracts, corpus callosum, arcuate and uncinate fasciculi, and the inferior orbitofrontal tract, the latter including the optic tract and Meyer’s loop. In recent years, papers and atlases have been published on white tract anatomy, placement of seed areas, and comparison of DTI with traditional white matter dissection.^9,¹⁰ Recent advances include intraoperative DTI, in which real-time images of white matter tracts are generated during neurosurgical procedures and shifting of tracts as a result of surgery can be depicted.¹¹ The role of DTI in neurosurgical practice is currently being defined.

MRS is a noninvasive functional neuroimaging tool that has been especially useful in studying hippocampal pathology. Hippocampal sclerosis is found by pathologic examination in 65% of patients with temporal lobe epilepsy and is characterized by loss of neurons, atrophy, and replacement gliosis.¹² It can be seen by visual inspection or volumetric analysis and can be defined by MRI in 70% of cases.¹³ MRS allows in vivo analysis of neurochemicals and their metabolites. Attention has focused on proton [¹H] MRS, which has principally yielded data on N-acetylaspartate (NAA), choline, phosphocreatine, creatine, and lactate; MRS quantifies metabolites from brain regions that have underlying cellular abnormalities. NAA, which is an amino acid synthesized in mitochondria, is a neuronal and axonal marker that decreases with neuronal loss or dysfunction. Decreased levels of NAA can be interpreted as cell loss or neuronal damage. Total creatine, composed of phosphocreatine and its precursor creatine, is a marker of brain energy metabolism. Total choline is a marker for membrane synthesis or repair, inflammation, or demyelination and can reflect astrocytosis. Studies have shown that in comparison to controls, the temporal lobe ipsilateral to the seizure focus shows a reduction in NAA signal intensity and an increase in creatine and choline signal. Reduction of the ratio of NAA to choline, creatine, and phosphocreatine is a marker for neuronal loss and dysfunction. This method has shown promise for localizing epileptic foci with underlying pathology that is not visible with other imaging modalities. However, routine use of MRS for epilepsy is declining in many centers as a result of the increased application and improvement of other imaging modalities.

Functional Mapping

Human brain mapping has produced new data that have added considerable information on the anatomy of specific cerebral functions. As already noted, there are two basic types of brain mapping methods for measuring brain function: first, techniques that detect electromagnetic activity (EEG and MEG) measure the electromagnetic fields generated by neural activity, and second, techniques that are based on hemodynamic or metabolic signals (PET and fMRI) measure signs of neural activity. These two methodologies differ, including their temporal and spatial resolution. fPET and fMRI produce data from most of the brain with a spatial resolution of a few millimeters and a temporal resolution of minutes (fPET) or seconds (fMRI). The electromagnetic techniques, in contrast, produce data with limited spatial resolution but with a temporal resolution of milliseconds. High temporal resolution is important to resolve rapidly the changing patterns of brain activity that underlie cerebral function. However, traditionally, EEG and MEG have provided insufficient spatial detail to identify relationships between electrical events, structures, and functions, as visualized by MRI, PET, or fMRI. It is hoped that combining the two technologies will provide increasing information on processing in the human brain.¹⁴ The rationale behind fPET and fMRI is to determine the spatial relationship between active eloquent brain areas and to identify the least traumatic neurosurgical approach. In the case of fPET, structural MRI images are required for coregistration.

Functional Magnetic Resonance Imaging

In simplified terms, fMRI is based on images of deoxyhemoglobin. Deoxygenated hemoglobin is found in the veins and capillaries and is seen in a magnetic field because of the presence of iron in the hemoglobin molecule. The presence of oxygen “neutralizes” the effect of the visible iron because it is bound to the iron in the hemoglobin molecule. There is therefore a loss of MRI signal in the absence of deoxyhemoglobin, and this loss of signal can be visualized in the veins and capillaries; in an animal breathing 100% oxygen, venous structure signal is lost. The oxygen content in blood therefore acts as a contrast agent: BOLD contrast. In activation studies, finger movement, for example, results in increased blood flow in the contralateral motor cortex and is visualized as signal loss because of an increase in the concentration of oxyhemoglobin.

Applications of fMRI include mapping of eloquent cortex, particularly sensorimotor function and language. Baseline studies are carried out, followed by presentation of relevant tasks to the patient. It is hoped that in the future fMRI will replace the Wada test for memory and language and mitigate the need for intraoperative cortical mapping in certain cases.¹⁵ However, caution is necessary in comparing the localization of functions based on areas of brain activated, as with the use of fMRI, with areas inactivated by the Wada test or intraoperative stimulation mapping. There are also concerns of the accuracy of fMRI localization, which can be impaired by the low SNR (in comparison to fPET) and susceptibility to various artifacts, including movement during speech. Thus far, studies comparing fPET and fMRI in the same patient are rare.¹⁶

Positron Emission Tomography

PET records the accumulation of radioactively labeled compounds in regions of the brain. The radioactive compounds (tracers) follow the biochemical pathways of native molecules without altering the velocity of the reactions in these pathways. Although fMRI appears to be superseding fPET in terms of clinical use because of cost and availability, FDG-PET and fPET can assist in lateralization and localization of epileptogenic cortical areas (Fig. 55-6).

FIGURE 55-6 Low-grade ganglioglioma in the somatosensory area of the left parietal region. A, T2-weighted fluid-attenuated inversion recovery image of the tumor (arrow). B, Positron emission tomography ([¹⁵O]H₂O) shows decreased blood flow in the area of the tumor (arrow).

(Courtesy of P. Borghammer, Århus University Hospital, Denmark.)

Functional Positron Emission Tomography

During functional activity, fPET shows significantly altered signals in comparison to a reference or baseline state. Functional activity results in altered signals in the region of interest, and fPET highlights the neuroanatomic correlates of changes in these processes by comparing the active state with the reference or baseline state. As in fMRI, activation typically involves sequential tasks contralateral to the lesion, such as thumb opposition, flexion-extension of the foot, or language tasks (Fig. 55-7). fPET is generally believed to be an accurate localization method that has demonstrated good correspondence with intraoperative cortical mapping methods and a high SNR. In fPET, uptake of radioactive water into the brain is proportional to blood flow during the first few minutes after the intravenous injection of [¹⁵O]H₂O, and blood flow is assumed to change in proportion to functional activity.

FIGURE 55-7 Dysembryoplastic neuroepithelial tumor in the left motor cortex. Limited resection was previously performed at another center. A, T1-weighted magnetic resonance imaging (MRI) with contrast enhancement. B, Functional positron emission imaging (fPET) of the hand and leg area merged with neuronavigation (Stealth, Medtronic) during resection.C, (i) Preoperative three-dimensional MRI surface reconstruction with fPET (hand and leg) superimposed and the tumor “removed” was considered informative for the neurosurgeon and patient. (ii) Intraoperative comparison. a, fPET as shown by navigation; p and p1, intraoperative cortical mapping with movement of the anterior tibial muscle of the right leg and flexion of the fingers of the right hand (intubated, light anesthesia, and electromyographic electrodes placed in the relevant muscle groups); A, residual tumor, according to navigation, was close to the presumed supplementary motor area. D, T1-weighted MRI. Postoperative initial right-sided paralysis resolved within 3 months except for a slight decrease in fine finger movement.

(Courtesy of A. Rodell and S. Dyve, Århus University Hospital, Denmark.)

The energy metabolism variable most commonly recorded with PET for epilepsy is uptake of the fluorine 18–labeled glucose analog FDG, which is trapped in brain tissue as FDG-6-phosphate in proportion to the glucose phosphorylation rate. The accumulated FDG-6-phosphate stays in the brain as a semipermanent index of the metabolic rate because it is neither further metabolized in the glycolytic pathway because of incompatibility with the ensuing enzymes nor reconverted to FDG because brain tissue has negligible phosphatase activity.

FDG-PET has been found to be useful in nonlesional epilepsy by interictally locating areas of reduced glucose consumption (Fig. 55-8). This has been found to enhance the treatment of patients with cortical dysplasia.¹⁷ Others have found that FDG-PET asymmetry such as left temporal lobe hypometabolism can predict verbal memory after temporal lobectomy.¹⁸ Recent studies have indicated that FDG-PET can detect interictal hypometabolism in areas of cortical dysplasia in 81% of cases and has been found to be useful in identifying epileptogenic regions in patients with tuberous sclerosis.^19,²⁰ PET remains of value in the diagnostic work-up of patients with epilepsy because of its simplicity, speed of performance, accuracy, and comfort in patients. There are usually no concerns for claustrophobic or overweight patients or those who have metal implants. Radioactivity is not commonly an issue because of its low concentration and the short half-life of positron-emitting isotopes.

FIGURE 55-8 Fluorodeoxyglucose (FDG) and [¹⁵O]H₂O positron emission tomography (PET) in a patient with seizures and no focus seen on magnetic resonance imaging. (i) Interictal [¹⁵O]H₂O PET shows decreased blood flow in the left frontal region. (ii) Near ictal FDG-PET showing decreased FDG uptake in the left frontal area.

(Courtesy of A. Gjedde, Århus University Hospital, Denmark.)

In summary, functional brain-imaging techniques can support the presurgical diagnosis, especially in patients with nonlesional MRI findings or nonlateralizing or localizing scalp EEG recordings. Intraoperative cortical mapping is facilitated because eloquent areas are mapped preoperatively and coregistered to the surgical navigational system. They aid in planning surgery and can serve as a basis for discussion with patients and families (Fig. 55-9).

FIGURE 55-9 Large low-grade astrocytoma in the right hemisphere. A, Three-dimensional surface reconstruction from magnetic resonance imaging allowed informed consent for the patient and guidance for the neurosurgeon. B, m1 and m2, intraoperative cortical mapping showing the response of muscles in the left arm; p, positron emission tomography of the hand.

(Courtesy of S. Dyve and A. Rodell, Århus University Hospital, Denmark.)

Magnetoencephalography and Magnetic Source Imaging in the Presurgical Evaluation for Epilepsy

Fundamentals of Neuromagnetism

MEG measures the extracranial magnetic field activity arising from the electric currents produced in the brain. This activity arises largely from intracellular neuronal currents in the dendrites of tangentially oriented cortical pyramidal cells. MEG represents the magnetic signals corresponding to the brain’s electrical activity recorded with standard EEG.²¹^–²⁵ MEG is based on a fundamental principle of electromagnetism, namely, that for every electric current, there is a corresponding magnetic field. Applying the “right-hand rule” of electromagnetism, for every electric dipole, the corresponding magnetic field wraps around the dipole with the field flowing toward or away from the MEG detector in a counterclockwise fashion.²⁶ Much like its EEG counterpart, the majority of the magnetic signal produced by the brain arises from aligned groups of pyramidal cells in the six-layered cerebral cortex.²⁷ Because the detectors that measure the MEG signals are usually aligned perpendicular to the skull, MEG best detects the fields that arise from tangentially oriented current dipoles perpendicular to the cortical surface.^25,^26,^28,²⁹

The brain’s magnetic fields are incredibly small, typically in the pico-tesla (10⁻¹²) range. In contrast, the magnetic field activity of the earth itself is approximately 1 billion times (10⁹) larger, ambient environmental sources such as room lighting and electric power lines are 10⁵ to 10⁶ times larger, and even those of the human heart are 10² times larger. SQUID detectors ² must be used to measure these tiny magnetic fields (see elsewhere for further details^25,²⁷). These detectors are kept at very cold temperatures by bathing them in liquid helium in a magnetically shielded Dewar flask. Early MEG systems consisted of a single SQUID detector or small arrays of 3 to 7 detectors that were moved about the head to different positions during a recording session lasting several hours.³⁰ This would be equivalent to recording scalp EEG signals by moving a single EEG electrode from site to site on the scalp. Modern systems (Fig. 55-10A) consist of a whole-head helmet containing 36 to 240 or more detectors that simultaneously record MEG signals, again much the way that an EEG electrode montage records EEG signals from multiple brain sites simultaneously. These large sensor arrays allow recording of whole-brain activity, thereby providing excellent temporal and spatial resolution of brain activity.