Physiological Principles of Electroencephalography

The cerebral cortex generates EEG signals. Spontaneous EEG activity reflects the flow of extracellular space currents generated by the summation of excitatory and inhibitory synaptic potentials occurring on thousands or even millions of cortical neurons. Individual action potentials do not contribute directly to EEG activity. A conventional EEG recording is a continuous graph, over time, of the spatial distribution of changing voltage fields at the scalp surface that result from ongoing synaptic activity in the underlying cortex.

EEG rhythms appear to be part of a complex hierarchy of cortical oscillations that are fundamental to the brain’s information processing mechanisms, including input selection and transient “binding” of distributed neuronal assemblies (Buzsaki and Draguhn, 2004). In addition to reflecting the spontaneous intrinsic activities of cortical neurons, the EEG depends on important afferent inputs from subcortical structures including the thalamus and brainstem reticular formation. Thalamic afferents, for example, probably are responsible for entraining cortical neurons to produce the rhythmic oscillations that characterize normal patterns like alpha rhythm and sleep spindles. An EEG abnormality may occur directly from disruption of cortical neural networks or indirectly from modification of subcortical inputs onto cortical neurons.

A scalp-recorded EEG represents only a limited, low-resolution view of the electrical activity of the brain. This is due in part to the pronounced voltage attenuation and “blurring” that occurs from overlying cerebrospinal fluid (CSF) and tissue layers. Relatively large areas of cortex have to be involved in similar synchronized activity for a discharge to appear on the EEG. For example, recordings obtained from arrays of microelectrodes penetrating into the cerebral cortex reveal a complex architecture of seizure initiation and propagation invisible to recordings from the scalp or even the cortical surface, with seizure-like discharges occurring in areas as small as a single cortical column (Schevon et al., 2008). Furthermore, potentials involving surfaces of gyri are more readily recorded than potentials arising in the walls and depths of sulci. Activity generated over the lateral convexities of the hemispheres records more accurately than does activity coming from interhemispherical, mesial, or basal areas. In the case of epileptiform activity, estimates are that 20% to 70% of cortical spikes do not appear on the EEG, depending on the region of cortex involved. Additionally, although the scalp-recorded EEG consists almost entirely of signals slower than approximately 40 Hz, intracranial oscillations of several hundred hertz may be recorded and, of clinical importance, have been associated with both normal physiological processes and seizure initiation (Schevon et al., 2009).

Such considerations limit the usefulness of electroencephalography. First, surface recordings are not useful to unambiguously determine the nature of synaptic events contributing to a particular EEG wave. Second, the EEG is rarely specific as to cause because different diseases and conditions produce similar EEG changes. In this regard, the EEG is analogous to findings on the neurological examination—hemiplegia caused by a stroke cannot be distinguished from that caused by a brain tumor. Third, many potentials occurring at the brain surface involve such a small area or are of such low voltage that they cannot be detected at the scalp. The EEG results then may be normal despite clear indications from other data of focal brain dysfunction. Finally, abnormalities in brain areas inaccessible to EEG recording electrodes (some cortical areas and virtually all subcortical and brainstem regions) do not affect the EEG directly but may exert remote effects on patterns of cortical activity.

Normal Electroencephalographic Activities

Spontaneous fluctuations of voltage potential at the cortical surface are in the range of 100 to 1000 mV, but at the scalp are only 10 to 100 mV. Different parts of the cortex generate relatively distinct potential fluctuations, which also differ in the waking and sleep states.

In most normal adults, the waking pattern of EEG activity consists mainly of sinusoidal oscillations occurring at 8 to 12 Hz, which are most prominent over the occipital area—the alpha rhythm (Fig. 32A.1, A). Eye opening, mental activity, and drowsiness attenuate (block) the alpha rhythm. Activity faster than 12 Hz beta activity normally is present over the frontal areas and may be especially prominent in patients receiving barbiturate or benzodiazepine drugs. Activity slower than 8 Hz is divisible into delta activity (1 to 3 Hz) and theta activity (4 to 7 Hz). Adults normally may show a small amount of theta activity over the temporal regions; the percentage of intermixed theta frequencies increases after the age of 60 years. Delta activity is not present normally in adults when they are awake but appears when they fall asleep (see Fig. 32A.1, B). The amount and amplitude of slow activity (theta and delta) correlate closely with the depth of sleep. Slow frequencies are abundant in the EEGs of newborns and young children, but these disappear progressively with maturation.

Fig. 32A.1 Samples of normal electroencephalographic recordings from two patients. A, Waking activity is characterized by a 9-Hz alpha rhythm that attenuates when the eyes are opened (EO) and resumes when the eyes are closed (EC). B, Stage 2 sleep is characterized by 2- to 5-Hz background activity, on which are superimposed vertex (V) waves and sleep spindles.

Common Types of Electroencephalographic Abnormalities

Recording Techniques

The EEG recording methods in common use are summarized in the following discussion. Details can be found in the American Clinical Neurophysiology Society’s Guidelines (2006).

A series of small gold, silver, or silver–silver chloride disks are symmetrically positioned over the scalp on both sides of the head in standard locations (the International Ten-Twenty system). In practice, 20 or more channels of EEG activity are recorded simultaneously, each channel displaying the potential difference between two electrodes. Electrode pairs are interconnected in different arrangements called montages to permit a comprehensive survey of the brain’s electrical activity. Typically, the design of montages is to compare symmetrical areas of the two hemispheres, anterior versus posterior regions, or parasagittal versus temporal areas in the same hemisphere.

A typical study is about 30 to 45 minutes in duration and includes two types of “activating procedures”: hyperventilation and photic stimulation. In some patients, these techniques provoke abnormal focal or generalized alterations in activity that are of diagnostic importance and would otherwise go undetected (Fig. 32A.2). Recording during sleep and after sleep deprivation, and placement of additional electrodes at other recording sites are useful in detecting specific kinds of epileptiform potentials. The use of other maneuvers depends on the clinical question posed. For example, epileptiform activity may occasionally activate only by movement or specific sensory stimuli. Vasovagal stimulation may be important in some types of syncope.

Fig. 32A.2 Intermittent stroboscopic light stimulation at 13 flashes per second elicited generalized bursts of 4- to 5-Hz spike-wave activity, termed a photoparoxysmal (photoconvulsive) response. The spike-wave paroxysm was associated with a brief absence, as documented by the patient’s (P) inability to respond to a tone given by the technologist (T). Normal responsiveness returned immediately on cessation of the spike-wave activity. The remainder of the electroencephalogram was normal.

In the past, EEG recording instruments were simple analog devices with banks of amplifiers and pen-writers. In contrast, modern EEG machines make use of digital processing and storage, and the electroencephalographer interprets the EEG from a computer display rather than from paper. Technological advances have not fundamentally changed the principles of EEG interpretation, but they have facilitated EEG reading. Early paper-based EEG systems required that all recording parameters—display gain, filter settings, and the manner in which scalp-recorded signals were combined and displayed (montages)—be fixed by the technologist at the time of recording. In contrast, digital EEG systems permit the electroencephalographer to adjust these settings at the time of interpretation. A given EEG waveform or pattern can be examined using a number of different instrument settings, including sophisticated montages (e.g., Laplacian montages), that were unavailable using traditional analog recording systems. Topographic maps can be useful to depict spatial relationships, displaying features of the EEG in a graphical manner similar to that for functional MRI (fMRI) or PET. For example, topographical maps can illustrate EEG voltage distributions over the scalp at a particular point in time (Fig. 32A.3) as well as the distributions of particular frequencies contained within the EEG. Although this flexibility does not change the interpretive strategies used to read an EEG, it does allow the electroencephalographer to apply them more effectively.

Fig. 32A.3 Frequency-domain topographical brain map obtained from a 32-channel bipolar electroencephalographic (EEG) recording. The patient was a 53-year-old man with hemodynamically significant left carotid stenosis. This map demonstrates an asymmetry over the occipital regions during eye opening, reflecting relative failure of left hemisphere alpha activity to attenuate normally. The color scale at the right reflects percentage change in EEG activity on going from the eyes-closed to the eyes-open state.

(Courtesy Dr. Bruce J. Fisch.)

In addition to facilitating the standard interpretation of EEGs, mathematical techniques can also be used to reveal features that may not be apparent to visual inspection of raw EEG waveforms. For example, averaging techniques, useful in improving the signal-to-noise ratios of spikes and sharp waves, can reveal field distributions and timing relationships that are not otherwise appreciable. Dipole source localization methods have been used to characterize both interictal spikes and ictal discharges in patients with epilepsy and may contribute to localization of the seizure focus (Ebersole, 2000). Such methods are based on a number of critical assumptions that, if applied without recognition of their limitations, can result in anatomically and physiologically erroneous conclusions (Emerson et al., 1995), so caution is warranted in their use.

For patients undergoing long-term EEG recordings as part of the diagnosis or management of epilepsy, a time-locked digitally recorded video image of the patient is recorded simultaneously with the EEG. EEG data are often processed by software that can automatically detect most seizure activity. Similar systems are finding increased use in intensive care units (ICU), where EEG monitoring has become increasingly important in the management of patients with nonconvulsive seizure activity, threatened or impending cerebral ischemia, severe head trauma, and metabolic coma (Drislane et al., 2008; Friedman et al., 2009). In this setting, compressed spectrograms, which graphically summarize the frequencies present in several hours of EEG on a single screen, can help the electroencephalographer to rapidly pinpoint important changes in the EEG and sometimes spot patterns or trends that otherwise might go unnoticed (Fig. 32A.4) (Scheuer, 2002). It is important to emphasize that fully automated robust systems analogous to those employed for cardiac monitoring are not now available for EEG, and while various automated methods can be very useful, their proper use in clinical practice should be as adjuncts to standard EEG recording and interpretation. False positives and negatives are commonplace; indeed, the very data reduction that makes such methods useful also makes them unsuitable for stand-alone application.

Fig. 32A.4 Frequency spectrograph depicting 2 hours of electroencephalographic (EEG) recording in a patient in focal status epilepticus. A, Upper two panels show activity in right and left parasagittal electrode derivations, respectively; lower two panels depict activity in temporal derivations. Time is represented on the horizontal axis. The vertical axis on each panel corresponds to frequency, from 1 Hz on the bottom to 20 Hz at the top. Colors indicate EEG voltage at a given frequency, from black for 0 mV/Hz to purple for 10 mV/Hz. This spectrograph shows bursts of relatively high-voltage, high-frequency (green) activity recurring approximately every 5 to 7 minutes, most prominent in the left parasagittal panel. This electroencephalogram (EEG) demonstrates a pattern of recurring electrographic seizures, as illustrated by representative EEG traces B and C, corresponding to points a and b on the spectrograph. Additionally, two quiescent periods with no activity (black) occur at most frequencies, corresponding to periods of postictal voltage depression, also illustrated in trace D.

Clinical Uses of Electroencephalography

The EEG assesses physiological alterations in brain activity. Many changes are nonspecific, but some are highly suggestive of specific entities (epilepsy, herpes encephalitis, metabolic encephalopathy). The EEG also is useful in following the course of patients with altered states of consciousness and may, in certain circumstances, provide prognostic information. It can be important in the determination of brain death.

Electroencephalography is not a screening test. It serves to answer a particular problem posed by the patient’s condition, so providing sufficient clinical information helps design an appropriate test with meaningful electrographical clinical correlation. The request for this study should specifically state the question addressed by the EEG.

EEG interpretation should be rational and based on a systematic analysis that uses consistent parameters that permit comparisons with findings expected from the patient’s age and circumstances of recording. Accurate interpretation requires high-quality recording. This depends on trained technologists who understand the importance of meticulous electrode application, proper use of instrument controls, recognition and (where possible) elimination of artifacts, and appropriate selection of recording montages to allow optimal display of cerebral electrical activity.

Epilepsy

The EEG usually is the most helpful laboratory test when a diagnosis of epilepsy is considered. Because the onset of seizures is unpredictable, and their occurrence is relatively infrequent in most patients, EEG recordings usually are obtained when the patient is not having a seizure. Fortunately, electrical abnormalities in the EEG occur in most patients with epilepsy even between attacks.

The only EEG finding that has a strong correlation with epilepsy is epileptiform activity, a term used to describe spikes and sharp waves that are clearly distinct from ongoing background activity. Clinical and experimental evidence supports a specific association between epileptiform discharges and seizure susceptibility. Only about 2% of nonepileptic patients have epileptiform EEGs, whereas as many as 90% of patients with epilepsy show epileptiform activity, depending on the circumstances of the recording and the number of studies obtained.

Nonetheless, interpretation of interictal findings always requires caution. Correlating most epileptiform discharges with the frequency and likelihood of recurrence of epileptic seizures is poor (Selvitelli et al., 2010). Furthermore, a substantial number of patients with unquestionable epilepsy have consistently normal interictal EEGs. The most convincing proof that a patient’s episodic symptoms are epileptic is obtained by recording an electrographical seizure discharge during a typical behavioral attack. Although ictal EEG tracings greatly increase the sensitivity of the study in assessing the pathophysiology of specific behavioral episodes, the clinician must still be aware of limitations inherent in such recordings. (Videos showing actual EEG recordings obtained during seizures [Videos 32A.1 to 32A.3] are available at www.expertconsult.com.)

In addition to epileptiform patterns, EEGs in patients with epilepsy often show excessive focal or generalized slow-wave activity. Less often, asymmetries of frequency or voltage may be noted. These findings are not unique to epilepsy and are featured in other conditions such as static encephalopathies, brain tumors, migraine, and trauma. In patients with unusual spells, nonspecific changes on EEG should be weighed cautiously and are not to be considered direct evidence for a diagnosis of epilepsy. On the other hand, when clinical data are unequivocal, or when epileptiform discharges occur as well, the degree and extent of background EEG changes may provide information important for judging the likelihood of an underlying focal cerebral lesion, a more diffuse encephalopathy, or a progressive neurological syndrome. Additionally, EEG findings may help determine prognosis and aid in the decision to discontinue antiepileptic medication.

The type of epileptiform activity on EEG is helpful in classifying a patient’s seizure type correctly and sometimes in identifying a specific epileptic syndrome (see Chapter 67). Clinically, generalized tonic-clonic seizures may be generalized from the outset or may be secondary to spread from a focus. Lapses of awareness with automatisms may be a manifestation either of a generalized nonconvulsive form of epilepsy (absence seizures) or of focal epileptogenic dysfunction (temporal lobe epilepsy). The initial clinical features of a seizure may be uncertain because of postictal amnesia or nocturnal occurrence. In these and similar situations, the EEG can provide information crucial to the correct diagnosis and appropriate therapy.

In generalized seizures of nonfocal origin, the EEG typically shows bilaterally synchronous diffuse bursts of spikes and spike-and-wave discharges (Fig. 32A.5). All generalized EEG epileptiform patterns share certain common features, although the exact expression of the spike-wave activity varies depending on whether the patient has pure absence, tonic-clonic, myoclonic, or atonic-astatic seizures. The EEG also may distinguish between primary and secondary generalized epilepsy. In the former instance, no cerebral disease is demonstrable, whereas in the latter, evidence can be found for diffuse brain damage. Typically, primary (idiopathic) generalized epilepsy is associated with normal or near-normal EEG background rhythms, whereas secondary (symptomatic) epilepsy is associated with some degree of generalized slow-wave activity.

Fig. 32A.5 Examples of generalized spike-wave patterns from different patients with primary generalized (idiopathic) epilepsy. The patient in A had mainly tonic-clonic seizures, with occasional absence attacks. The patient in B had juvenile myoclonic epilepsy.

Consistently focal epileptiform activity is the signature of partial (focal) epilepsy (Fig. 32A.6). With the exception of the benign focal epilepsies of childhood, focal epileptiform activity results from neuronal dysfunction caused by demonstrable brain disease. The waveform of focal epileptiform discharges is largely independent of localization, but a reasonable correlation exists between spike location and the type of ictal behavior. Anterior temporal spikes usually are associated with complex partial seizures, rolandic spikes with simple motor or sensory seizures, and occipital spikes with primitive visual hallucinations or diminished visual function as an initial feature.

Fig. 32A.6 Focal right anterior temporal spikes occurring on the electroencephalogram of a 69-year-old woman with complex partial seizures after a stroke involving branches of the right middle cerebral artery.

In addition to distinguishing epileptiform from nonepileptiform abnormalities, EEG analysis sometimes identifies specific electroclinical syndromes. Such syndromes include hypsarrhythmia associated with infantile spasms (West syndrome) (Fig. 32A.7); 3-Hz spike-and-wave activity associated with typical absence attacks (petit mal epilepsy) (Fig. 32A.8); generalized multiple spikes and waves (polyspike-wave pattern) associated with myoclonic epilepsy, including so-called juvenile myoclonic epilepsy of Janz (see Fig. 32A.5, B); generalized sharp and slow waves (slow spike-and-wave pattern) associated with Lennox-Gastaut syndrome (Fig. 32A.9); central-midtemporal spikes associated with benign rolandic epilepsy (Fig. 32A.10); and periodic lateralized epileptiform discharges (PLEDs) associated with acute destructive cerebral lesions such as hemorrhagic cerebral infarction, a rapidly growing malignancy, or herpes simplex encephalitis (Fig. 32A.11) (Pohlmann-Eden et al., 1996). PLEDs may also reappear in patients with chronic structural lesions in the context of new metabolic derangements.

Fig. 32A.7 Electroencephalographic pattern termed hypsarrhythmia in a recording obtained in an 8-month-old boy with infantile spasms. Background activity is high-voltage and unorganized, with abundant multifocal spikes.

Fig. 32A.8 A 3-Hz spike-and-wave paroxysm on the electroencephalogram of a 9-year-old boy with absence seizures (petit mal epilepsy). During this 12-second discharge, the child was unresponsive and demonstrated rhythmic eye blinking.

Fig. 32A.9 Generalized sharp-wave and slow-wave discharges on the electroencephalogram (EEG) of a 9-year-old child with mental retardation and uncontrolled typical absence, tonic, and atonic generalized seizures. This constellation of clinical and EEG features constitutes the Lennox-Gastaut syndrome.

Fig. 32A.10 Electroencephalogram obtained during drowsiness in a 10-year-old boy with benign rolandic epilepsy. Stereotypical diphasic or triphasic sharp waves occur in the right central-parietal and midtemporal regions.

Fig. 32A.11 Two examples of periodic lateralized epileptiform discharges (PLEDs) on the electroencephalogram. A, Right parasagittal PLEDs in a 69-year-old man with severe brain damage caused by meningitis with multiple cerebral infarctions. B, A 65-year-old woman with herpes simplex encephalitis. PLEDs are bilateral but have a right-sided predominance.

The increased availability of special monitoring facilities for simultaneous video and EEG recording and of ambulatory EEG recorders has improved diagnostic accuracy and the reliability of seizure classification. Prolonged continuous recordings through one or more complete sleep/wake cycles constitute the best way to document ictal episodes and should be considered in patients whose interictal EEGs are normal or nondiagnostic and in clinical dilemmas that are resolvable only by recording actual behavioral events. Although EEG documentation of an ictal discharge establishes the epileptic nature of a corresponding behavioral change, the converse is not necessarily true. Sometimes muscle or movement artifacts so obscure the EEG recording that it is impossible to know whether any EEG change has occurred. In these circumstances, postictal slowing usually is indicative of an epileptic event if similar slow waves are not present elsewhere in the recording and if the EEG recording subsequently returns to baseline. In addition, focal seizures not accompanied by alteration in consciousness occasionally have no detectable scalp correlate. On the other hand, persistence of alpha activity and absence of slowing during and after an apparent convulsive episode are inconsistent with an epileptic generalized tonic-clonic seizure.