Adult Electroencephalography Artifact
William O. Tatum, IV
The scalp EEG is composed of a signal that is generated by a pool of cerebral neurons. It may be modified by the conductive properties of an electrode, the dipole orientation of the cortical generator, and properties conducting electrophysiological potentials from the brain to the recording electrode on the scalp (1). Artifact is an especially important feature to recognize in the adult EEG as it is present in virtually every recording. The ability to distinguish an artifact is subject to human interpretation. Artifact is identifiable when waveforms that do not conform to an expected electrophysiological field are generated by the brain (2). However, while artifact may produce “interference” of identifying the underlying electrocerebral activity, it is also an essential part of the recording to aid in identifying the levels of consciousness and sleep stages. Oftentimes however, artifact is a limiting feature that interferes with the interpreter’s ability to render a confident diagnosis to fully characterize the EEG (Figure 7.1). When present, artifact may act as a “contaminant” that interferes with proper interpretation of the EEG recording. Excessive artifact is quickly and readily apparent during review of the EEG when waveforms are obscured by movement or myogenic artifact sufficient to interfere with useful interpretation. Judging how much artifact is excessive remains a common problem and is subject to individual determination (3). Without guidelines to identify when and why excessive artifact is present, the author uses a cutoff of greater than 50% of the recording in more than 50% of the EEG channels before interpreting a “limited EEG” recording (patient factor). When technical issues are involved that are uncorrected (e.g., high electrode impedances), a “technically limited EEG” recording (technical factor) is reasonable to distinguish the different conditions. Quantifying the amount of artifact using a computer-based scoring system may become more advantageous in future applications of digital EEG reporting to overcome the subjectivity involved, for both clinical and research purposes (4).
The sheer prevalence of artifact underscores its potential to serve as a pitfall to EEG interpretation. Both physiological and nonphysiological sources of artifact may occur to fool the interpreter into believing an abnormality exists. The presence of many “suspicious” spiky waveforms exist (Figure 7.2) in the normal adult EEG (5). Maulsby noted almost 50 years ago, “Every spiky-looking wave is an artifact unless there are one or more good reasons for suspecting otherwise” (6). Spiky waveforms become problematic when normal features of the EEG are interpreted as abnormal interictal epileptiform discharges (IEDs) (7,8). Artifact is much more prevalent than benign (normal) variants of uncertain significance, yet the latter are commonly misinterpreted, leading to an epilepsy diagnosis (7). Early definitions of a spike (20–70 ms) and sharp wave (70–200 ms) identify the potential morphology but do not indicate whether it is a normal or pathological waveform, nor identify the source. Some artifact closely mimics abnormal IEDs and challenges the interpreter to separate artifact from abnormal features of the EEG (Figure 7.3). When the EEG is misinterpreted as abnormal, overinterpretation of pathological IEDs may lead to a misdiagnosis of epilepsy and result in mistreatment of patients with antiseizure drugs (ASDs) and miss the treatment of the true underlying condition (9,10). In a survey of 47 clinical neurophysiologists, 97.9% reported encountering an EEG read as an abnormal recording based upon artifact (10). Uncorrected and unrecognized misinterpretation of artifact may lead to inappropriate treatment with ASDs and potentially compromise patient care and overall quality of life (11).
112Difficulty recognizing artifact in the EEG stems from several factors. Waveform interpretation involves identifying brain signals derived from a three-dimensional source in the brain that is translated into two-dimensional EEG recordings. Pattern recognition alone, when applied to EEG, may serve as a pitfall to proper interpretation unless the principles of artifact recognition are applied (Table 7.1). Recognizing undesirable signals such as artifacts requires a basic understanding of polarity and localization in the orderly approach to EEG interpretation (12). Furthermore, standards for digital recording necessitate an inherent understanding of digital systems and technology before artifactual waveforms can be identified. Recognizing artifacts and identifying the source is a skill learned through familiarity of the interpreter provided by experience and exposure to expert teaching during residency or fellowship.
Most standard adult EEG artifact encountered is scalp-based. Ensuring the EEG signal reflects an electrophysiological potential or series of potentials derived from a cerebral source as opposed to artifact is required for determining an accurate interpretation for clinical purposes. However, artifact may intervene in any form of EEG recording including ambulatory EEG (aEEG) monitoring, video-EEG monitoring (VEM), intracranial EEG (iEEG) recording, and EEG recorded in the operating room (OR) (Table 7.2). All forms of EEG recording share similar susceptibility to artifact, though with increasing use of long-term EEG monitoring (LTM), a greater predisposition to artifact exists. Video recording as an adjunct to EEG has helped interpreters confirm artifact. Time-locked frequencies of a possible abnormality on the EEG linked to a specific clinical behavior with a similar frequency on the video recording increases the likelihood of identifying artifact. During VEM in epilepsy to characterize seizures during a presurgical evaluation, artifact may impair the ability to localize seizure onset (Figure 7.4).
A team approach is essential. By applying basic concepts of recording, manipulating instrument parameters post hoc, and having an experienced technologist modify the environment during the recording, the technologist and clinical neurophysiologist serve as crucial members of the team to obtain an interpretable record. Ensuring the minimum technical requirements are met, standards for electrode position and montages are used, and guidelines for recording EEG are adhered to, help ensure an optimal outcome for interpretation (13). When these standards are met, interpretation of the EEG reduces the risk of overinterpretation and facilitates proper management of patients (7,11). Dr. Chadwick noted the “Routine interictal EEG recording is one of the most abused investigations in clinical medicine and is unquestionably responsible for great human suffering” (14). In this chapter, we highlight common and uncommon examples of EEG artifact obtained during routine clinical practice to highlight the impact it may have on patient management.
Test impedance, regel, or replace the electrode until resolved.
If the above action is ineffective, extra electrodes applied around the eyes/mouth may reveal the generator.
3. Spatial distribution
Noncontinuous activity and nonepileptiform generalized potentials
Eliminate extraneous movement and electromagnetic fields that could lead to nonphysiological distribution.
Adjacent double/triple phase reversals
Reapplication of electrodes involved in the affected region.
Complex waveforms or discrete periodicity
Troubleshoot for an extracerebral source (e.g., pumps, ventilator, tubing).
Very fast (>30 Hz) and very slow (<1 Hz)
Unless iEEG is used, these frequencies require troubleshooting for a source.
iEEG, intracranial EEG.
Implanted electrical devices
PRINCIPLES OF ARTIFACT
The principles of artifact recognition apply to virtually every EEG recording (15). EEG interpretation is based upon the inverse problem of localizing a three-dimensional generator from a two-dimensional graphic projection (16). Current flow through the tissues is volume conducted from deep and surface-based generators recording EEG at the scalp–electrode interface. The principles of artifact are predicated on the same principles involved in localizing a physiological source. Extracerebral artifact must be recognized before the cerebral source is identified. Artifacts from overfiltering, electrode pops, and myogenic artifact are common types of artifact during standard EEG recording. Elimination of artifact is attempted when the waveforms disrupt an accurate interpretation of the underlying EEG. Principles involved in identifying artifact are essential rules to arrive at an accurate interpretation of the EEG recording (see Table 7.1). Artifact is present when extracerebral electrophysiological potentials are recorded. Identifying the mismatch between potentials, which are generated by the brain from potentials that are not generated by the brain, form the basis of artifact recognition (8). Some simple concepts for recognizing artifact are important to understand for both the EEG technologist and the electoencephalographer alike before considering a waveform abnormal. Activity that is confined to a single electrode (Figure 7.5) with a restricted spatial distribution should be considered artifact until proven otherwise. To this end, when a technologist is recording the EEG and single-channel potentials are encountered, especially if they are in question, an electrode test may reveal a high impedance, demonstrating a predisposition to artifact. Another concept involved with artifact occurs when EEG activity is present at the end of an electrode chain within a montage (Figure 7.6). This limits the ability to appreciate an electrophysiological field of a waveform or series of waveforms that is generated by the brain. Changing montages may help delineate a restricted field of artifact. Alternatively, application of a movement monitor (e.g., ocular or perioral electrodes) may be the electrical source as artifact generated by ocular or glossopharyngeal movement. Artifact should also be considered when a noncontinuous or nonphysiological spatial distribution or atypical generalized potentials are encountered (Figure 7.7). Intravenous (IV) mechanical pumps, feeding tube delivery systems, electrical beds, and so on should be sought by the technologist to look for any extraneous movement or interference from electrical generators in an effort to resolve the artifact. Diffuse artifact involving all channels of the EEG occurs when a common reference (Figure 7.8), the ground, or cable connections between the patient and machine have compromised integrity. Similarly, when periodic patterns are precise without variation (Figure 7.9) this suggests a mechanical as opposed to a biological source. Identifying a competing electrical source for EEG waveforms may lead to quick resolution by transiently unplugging nonessential equipment. Additionally, when periodicity is present in the EEG, using a slow display speed (e.g., 10–15 mm/sec) may help demonstrate the precision necessary to recognize artifact. Adjacent double- and triple-phase reversals are also common manifestations of artifact (Figure 7.10). In this case, a negative phase reversal lies immediately adjacent to a positive phase reversal (double-phase reversal) in an electrode chain of a montage. This is best seen in a bipolar montage and results from repetitive movement. The nonphysiological origin is evident when negative–positive–negative (triple) phase reversals appear. The field speaks for artifact and becomes increasingly likely over time as the scalp–electrode interface deteriorates. It is the responsibility of the technologist during the recording to recognize and identify whether a waveform is due to an artifact, and eliminate it from the recording when it is possible (17).
The vast majority of adult EEG interpretation is based upon standard scalp EEG recorded in the “Berger band” for clinical purposes. The bandwidth for interpreting clinical EEG focuses on a range of frequencies between 1 and 30 Hz with digital filtering between 1 and 70 Hz. Frequencies that appear outside this bandwidth suggest artifact (e.g., very slow; sweat, very high; myogenic or 60 Hz) (Figure 7.11). The cause of an artifact is subdivided into physiological and nonphysiological sources (Table 7.2). Physiological artifacts arise from active bioelectric properties generated by tissue other 114than the brain. Physiological artifacts such as eye movement and myogenic artifact may facilitate interpretation of the EEG and does not require elimination. Nonphysiological artifacts may be extrinsic arising from electrodes, equipment, and those within the environment or intrinsic and arise from internal electrical devices (e.g., pacemakers, stimulators). Nonphysiological morphology, polarity, or spatial field of distribution may be recognized as artifact, but defy identification of the source or isolation to attempt elimination of its presence in the EEG (Figure 7.12).
Physiological sources of artifact present during routine EEG recording occur from biological generators that produce an electrical dipole. Some are routinely present in the adult EEG. Eye movements, myogenic and body movements, and cardiac sources are among the most common types of artifact that are frequently encountered by clinical electroencephalographers during interpretation of the EEG. Factors that underlie electrical fields and polarities derived from the brain and living tissue define a physiological source of artifact (18). Common physiological sources of artifact are listed as follows.
Eye movement artifact is perhaps the most common physiological form of artifact identified in the adult EEG. It occurs in essentially every waking individual during standard EEG recording. The artifact produced by yoked movement of the eyes is detected because of an inherent corneal–retinal resting electrophysiological potential (Figure 7.13). The cornea is relatively electropositive compared with the electronegativity present in the retina, generating a DC potential difference measurable by EEG. The potential difference is easily measured in the vertical (Figure 7.14) and horizontal (Figure 7.15) plane. The benefit of recognizing eye movement artifact lies in the ability to correlate the clinical level of arousal and of consciousness. In the normal waking individual, vertical eyeblink artifact is present. Upon closing the eyelid, there is a concomitant upturning of the eyeball (Bell’s phenomenon). Slow, rolling eye movements are characteristic of drowsiness (Figure 7.16), while rapid eye movement (REM) artifact defines REM sleep. Sleep cycling is more obvious during prolonged recordings (and those who are sleep-deprived). Unless eye movements are monitored (see Figure 7.14), they may mimic abnormalities such as frontal intermittent rhythmic delta activity (Figure 7.17), or combine with other forms of artifact to provide the false appearance of an epileptiform abnormality (see Figure 7.3).
Eye movements are differentiated from cerebral waveforms with the use of an external occulogram (or electrooculogram [EOG]). An EOG requires application of surface electrodes above and below the eye (referred to a reference or each other). The EOG is an essential means of recording eye movement artifact during EEG as well as to delineate the level of arousal in polysomnography (19). In-phase deflections with a surrounding electrophysiological field are characteristic of a cerebral dipole. The exception is when the generator lies below the EOG electrodes. Out-of-phase deflections characterize a generator that lies between the recording EOG electrodes typical of an eye movement (see Figure 7.17). Horizontal and skew eye movements are identified using a bipolar montage if opposite polarities occur in homologous anterior temporal derivations of the EEG (e.g., F7/F8). Phase reversals that are “positive” (i.e., out of phase) reflect the location of the eyes closest to the cornea. The “negative” phase reversal reflects the polarity associated with the retina in the contralateral eye during a lateral eye deviation.
Muscles are another common source of artifact in the EEG capable of producing “spikes” that may be mistaken for IEDs. The temporalis and frontalis muscles are the principal myogenic sources of artifact on EEG that produce a characteristic temporal (Figure 7.18) or frontal predominance (see Figure 7.3). Chewing is a common source for myogenic artifact (Figure 7.19). Unless the principles of artifact recognition are followed, chewing artifact may mimic the appearance of polyspike-and-waves. Myogenic spikes, also referred to as “muscle spikes,” are high-frequency signals with a duration less than 20 milliseconds. The technologist may ask the patient to relax their jaw when jaw clenching occurs or request the patient open their mouth during the recording to eliminate bitemporal artifact, which occurs with tonic or phasic muscle contraction from the masseter and muscles of mastication. The presence of spikes may also occur from movement of the eye muscles of the globe, such as the lateral rectus muscles (Figure 7.20). These are differentiated from pathological spikes due to the intimate association with REMs. They represent motor unit potentials best seen in the F7 and F8 derivations of the scalp EEG. Frontalis muscle contraction during eye movement may produce a superimposition of frequencies to mimic a generalized periodic discharge (see Figure 7.3). When this occurs during intermittent photic stimulation, vertical eyeblink or eye flutter artifact, in combination with frontalis myogenic spikes, can produce a response that may be mistaken for a photoparoxysmal response. A high index of suspicion is required in this case to prevent overinterpretation. Eliminating myogenic artifact may occur 115by restricting appropriate movement, though this may necessitate medication promoting muscle relaxation when significant enough to interfere with interpretation.
Similar to the artifact generated by eye movement, artifact generated by the tongue is based on the principal of a bioelectric dipole. The tip of the tongue is relatively electronegative compared with the root, which is electropositive. Glossokinetic artifact may occur during tongue and oropharyngeal movements due to the DC potential produced. By elevating and depressing the tongue during swallowing or speaking, motions are generated that often produce a brief series of diffuse, frontally distributed waveforms that appear abnormal due to the presence of intermittent bursts of delta on the EEG (Figure 7.21). The spatial distribution of the electrophysiological field produced by the tongue is a frontaloccipital broad-based artifact compared with artifact produced by eye movements. A simple bedside technique used by technologists to identify glossokinetic artifact is simulated by asking the patient to repeat “Tom Thumb” or “lilt.” This action uses tongue and glossopharyngeal muscles to recreate the artifactual pattern when intermittent bursts are in question as an abnormal finding in the EEG. Like eye movements, motions of the mouth may also be monitored. Similarly, validation is obtainable by applying an electrode above the upper lip, and below the lower lip around the mouth. Using bipolar EEG recording, a tongue movement monitor will identify artifact with deflection of the waveforms that are out of phase. During movement of the mouth such as speaking, teeth may touch one another. When the teeth contain metallic implants or amalgam to fill dental caries, brief spikes may be produced due to electrical properties generated. The field of dental artifact is typically diffusely distributed, and the single spikes are very rapid (Figure 7.22).
Monitoring cardiac function is performed during EEG using a rhythm strip recording the ECG. Due to the unique information incorporated during EEG, it is arguably the single most important channel recorded but, like EEG, may be prone to artifact (Figure 7.23). There are multiple EEG channels dedicated to recording brain function, but there is only one used to represent the heart. However, cardiac function may significantly influence artifact that appears in the EEG (Figure 7.24). Contamination of the EEG by the ECG is more likely to occur when referential montages use ear references. The A1 and A2 electrode derivations accentuate ECG artifact. This is augmented in patients with short necks, those who are overweight, and when higher sensitivity or long interelectrode distances are used. These conditions facilitate recording electrophysiological signals from the left ventricle to override the EEG signals during the process of amplification. Left lateralization of ECG artifact may occur due to the direction of the QRS vector associated with ventricular contraction. The QRS complex reflects the spread of electrical current through the myocardium. When distributed over a broad field, ECG artifact may mimic periodic discharges. These may have a lateralized or generalized distribution (Figure 7.25). When present, ECG artifact may demonstrate opposite polarities of the R-wave in the left and right hemispheres when an ear reference montage is used. The left ear reference may transmit an electronegative potential, while the right ear electrodes appear as an electropositive deflection. When periodic potentials are visible in the EEG and time-locked to the QRS complex, the precise intervals make ECG artifact readily recognizable and the source easily identifiable. Eliminating ECG artifact can be accomplished by repositioning the patient’s head and neck to alter detection of the cardiac signal, or by changing the montage (e.g., using an average reference or linking the ear reference). Eliminating or reducing ECG artifact can help optimize interpretation when it obscures the EEG.
Pulse artifact occurs as a periodic waveform that is synchronized to the QRS complex of the ECG. Unlike ECG artifact, the mechanical pulsation of blood produces artifact that involves a single channel (Figure 7.24). Like ECG artifact, pulse artifact may be identified as periodic waveforms set in time-locked fashion to the ECG 200 to 300 ms following the QRS complex. Elimination of pulse artifact requires moving the electrode to a distance away from a scalp arterial source. The contraction of the myocardium can also produce a forceful mechanical movement of the head to produce a ballistocardiographic artifact that appears as though the patient is repetitively rocking his head, resulting in a regional or lateralized slow wave component. Elimination may require stabilizing the head with towels to minimize head movement.
Bone is inactive compared with other soft tissues in the body; however, the skull may create the appearance of artifact when a breach rhythm is encountered. This is due to the high frequencies that appear in a focal or region of the EEG following head trauma or neurosurgery (Figure 7.26). A bipolar montage is the best recording technique to recognize a breach effect. Recording EEG this way accentuates the spatial contrast of voltage differences between beta frequencies normally attenuated by the skull to make a breach rhythm more obvious. Myogenic artifact may also occur 116when electrodes for recording EEG are applied adjacent to a skull defect. The patterns of myogenic artifact may exist as a repetitive motor unit potential (similar to lateral rectus spikes) or as a continuous pattern. Recording surface electromyogram (EMG) produces similar patterns created without a skull breach by the temporalis and frontalis muscles. Individual motor unit potentials (Figure 7.6) or continuous musculoskeletal artifact surrounding a breach rhythm may be reduced by sleep or when muscles are relaxed.
The skin acts as an electrical source with measurable electrodermal–electrophysiological activity. When the skin produces sweat, sodium chloride and lactic acid may react with the metallic component of the EEG electrodes to produce ultraslow frequency artifact (e.g., 0.25–0.5 Hz). When the electrode impedance is greater than 10 kΩ, this provides the foundation for artifact at the electrode sites responsible for sweat artifact. These are typically maximal in the front of the head (over the forehead), where sweat tends to occur, and deteriorate the scalp–electrode interface. The alternating polarity when bilaterally distributed is without a physiological spatial field of distribution. The sweat produces a sway of the baseline, which produces a notable waveform that can interfere with an interpretable recording (Figure 7.27).
A “salt bridge” is formed when adjacent channels are connected by a smear of electrolytic gel. It is recognized when a single channel lacks similar electrocerebral activity to other nearby electrode derivations, producing an artifact with the appearance of a near isoelectric or low-amplitude EEG signal (Figure 7.28). The formation of a salt bridge occurs because similar or undetectable voltage differences are identified by the differential amplifier of the EEG machine. The result is that there is cancellation of the electrophysiological potentials from the two sites to result in the false appearance of a focal, near-isoelectric EEG present in the affected channel.
Movements of respiration may produce a physiological respiratory artifact (Figure 7.29). Respiration artifact is common in the ICU during mechanical ventilation. Periodic potentials may be identified as respiratory artifact if the technologist annotates the record when respiration is synchronizing with body movement. Very slow synchronous frequencies are characteristic during the thoracic movements of respiration at approximately 0.5 Hz. This regularity creates an artifact that appears rhythmic, periodic (e.g., 20 breaths/minute), and occipital-predominant. Faster frequencies during rapid inspiration and exhalation during hyperventilation may produce movement-induced periodic potentials, which are especially prominent in high-impedance electrodes in the standard EEG (Figure 7.30). Additional respiratory monitors may be applied with uncertainty. Artifact may be monitored using a chest belt or nasal thermistor to help determine sleep stages, though application of a movement monitor over the site of involvement may also be performed to discern artifact when breathing compromises the interpretability of EEG. Sleep-related movements such as snoring, periodic limb movements, myoclonic jerks, or arousal patterns may be suggested by the pattern of myogenic or movement artifact that appears on the EEG (20).
Movement artifact in the EEG is time-locked to patient movements and may be evident on video recording. Certain movement disorders produce unique patterns of movement-induced artifact on the surface EMG. For example, characteristic patterns of artifact may help identify disorders such as essential tremor and Parkinson’s disease based upon frequency, similar to actigraphy. For example, essential tremor-precipitated artifact may appear as a bilateral, 7- to 11-Hz frequency in the EEG when compared with a unilateral, fluctuating, higher amplitude 4- to 7-Hz tremor artifact often encountered in Parkinson’s disease (Figure 7.31). On the other hand, rhythmic sinusoidal tremor artifact may be confused with electrocerebral activity. The artifact may appear rhythmic, intermittent or continuous, and lateralized or generalized depending upon the source of movement. The movement itself may create an artifact with a field involving opposite polarity in adjacent channels (Figure 7.32) including double- and triple-phase reversals. Regardless of the pattern, the presence of a nonphysiological field should suggest artifact (Figure 7.33). Monitors will have higher amplitudes compared to the scalp EEG unless the sensitivity of the recording channel is adjusted (Figure 7.34).
Some movements are complex and may require simultaneous monitoring of several functions (Figure 7.35). Video recording may be helpful in these cases. Spontaneous movement, as well as interactions with a medical examination, nursing assessment, or other health care professional may produce complex movement artifacts. By applying extra electrodes on or nearby a suspected source, muscle/movement artifact that is present in electrodes placed to monitor motion can characterize certain movements when they occur.
The majority of adult EEG artifact is generated by sources and electrical fields that are external to the patient. External artifact usually arises from extrinsic sources in the environmental nonphysiological generators. Often, this is produced by the EEG equipment with the most unstable connection associated 117with the electrode interfacing with the patient. Extrinsic artifact may be associated with simple (Figure 7.36) and complex (Figure 7.37) waveforms. When activity is confined to a single channel without an electrical field, electrode artifact should be suspected irrespective of its complexity (see Figure 7.36). Activity that is generated in the brain has a field of distribution that radiates to neighboring electrodes (21). Eliminating single-electrode artifact first requires assessment of electrode impedances. Securing or replacing the involved electrode may be necessary without resolution. Electric motors in nearby machines produce electromagnetic fields that may intermittently create “noise” in the EEG or become continuous depending upon the type of machine. The ubiquitous nature of artifact is such that it may arise from multiple electrical generators and “contaminate” the EEG at any place between the patient’s scalp and the EEG machine. Nonphysiological artifact that is recognized as extrinsic to the patient requires isolating the source and eliminating it from the environment, modifying the EEG parameters, or annotating and monitoring the source during the recording. Some extrinsic artifacts are difficult to eliminate despite isolating the source due to necessity (e.g., medication pumps). A technologist experienced in artifact detection is crucial when called to “troubleshoot” and eliminate or minimize the source. A heightened awareness should be maintained with repeat impedance checks if the integrity of EEG becomes compromised (22).
Environmental artifacts produced outside an EEG laboratory are common. Medication infusion pumps, mechanical beds, feeding delivery systems, and electrical monitoring devices produce electromechanical and electromagnetic sources of interference with EEG recording. Nearby telephones (Figure 7.38), portable computers and tablets, and gaming devices can also produce artifacts in the adult EEG. Electrostatic, capacitive, and inductive artifacts may occur. IV machines are commonly used in the hospital setting outside the EEG laboratory. A transient electrostatic force is produced when a drop of fluid or medication is conducted to nearby electrodes and wires. This creates an intermittent nonphysiological potential in the EEG that may lead to the appearance of a spiky artifact. However, the precise periodicity and generalized spatial distribution distinguishes the potentials as IV drip artifact (see Figure 7.9). Similarly, static electricity may be introduced into the nearby wiring or recording electrode. The cables from the patient to the EEG machine containing electrode wires can produce capacitance artifact because the cable acts like a capacitor. Multiple insulated wires in the cable are capable of generating conductive artifactual waveforms due to movement or compression of the cable (Figure 7.39). Sometimes the appearance of a paroxysmal or epileptiform discharge may result, underscoring the importance of the technologist’s role in annotating the record during recording. As a general rule, the polarity and nonphysiological field will be evident to suggest artifact.
Environmental artifact due to alternating current (AC) is one of the most common environmental artifacts encountered during EEG recording. Magnetic fields generated by electrical motors near the EEG recording can result in electrical potentials visible as artifact. This may interfere with the identification of the brain’s biological signals. Artifact in the EEG due to AC current is induced by nearby power mains and electrical outlets introducing electrical interference. This is often produced by the equipment used in recording EEG.
The most common artifact encountered in the EEG is due to an insecure connection between the electrodes and the patient scalp or connections to the machine (23). Artifact in the EEG may arise from many different combinations of electrode types and configurations (Table 7.3). Scalp EEG is performed most frequently. During standard EEG recording, electrodes must be secured to the patient’s scalp to limit artifact. This is accomplished by applying an adhesive compound to the electrode to conduct the EEG. High impedances of electrode pairs greater than 10 kΩ increase the risk for electrode artifact. 118During LTM, collodion provides a more secure and longer-lasting contact of the electrode to the patient’s scalp. Electrodes are composed of conductive metals such as silver, silver–silver chloride, and gold plating, though nonferromagnetic or plastic electrodes are available for MRI compatibility. Abrading the skin enhances the contact with the scalp. The epidermis has high electrical impedance. Use of an electrolyte gel enhances electrical conductance and reduces the scalp–electrode biological resistance (impedance). High-impedance electrodes predispose to artifact, which may be evident during intermittent photic stimulation. When activation with photic stimulation is performed, an artifact known as the photoelectric (or photovoltaic) effect may be present (Figure 7.40). This occurs when photic flashes produce a transitory electrochemical response to the light in susceptible electrodes. The morphology may appear as a train of brief, spiky potentials time-locked to the flash frequency. The photoelectric effect can be eliminated by covering the involved electrodes with a cloth or piece of paper to block the reaction produced by light. Electrode artifacts are usually easy to recognize, though complex discharges should raise the suspicion of artifact (Figure 7.41).
• Standard noninvasive
– Cup (10-20 system)
– Other (e.g., nasoethmoidal, nasopharyngeal)
– Needle (ICU and OR)
– Sphenoidal (EMU)
– Foramen ovale (OR)
– Intracranial (OR)
• Noninvasive plus
– 10-10 system
– High density
• Special uses (combinations)
– Subdural strips/grids (stereo-EEG)
– Intracortical depths
– High density
– Hybrid grids/depths
EMU, epilepsy monitoring unit; OR, operating room.
When an electrode mismatch of greater than 5 kΩ is present or when poor electrode contact with the scalp exists, this facilitates a greater risk for 60-Hz artifact. When artifact is unresponsive to simple troubleshooting techniques such as securing or replacing the electrode, changing to a referential montage may prove a cerebral source is present when more than a single electrode is involved. When electrode artifact remains in question, switching the involved electrode pin to another site on the jack box supports a cerebral source when the abnormality remains. A very focal abnormality will not switch to the new electrode derivation with a properly functioning jack box. Similar to high-impedance electrodes, electrical noise caused by EEG amplifiers are also capable of producing 60-Hz artifact (50 Hz in Europe). This is due to the internal electronic components and is easily eliminated when a 60-Hz notch filter is applied during the EEG recording (Figure 7.42).
When diffuse 60-Hz artifact is encountered (Figure 7.43A), then the integrity of the ground should be reassessed. In addition, ensuring proper patient–machine connections ensures that the integrity of patient safety has not been compromised. Supplemental use of filters, including a 60-Hz notch filter (Figure 7.43B), can be applied during postacquisition review. One earth ground is important for patient safety. The ground electrode directs the flow of electrical current through the EEG machine instead of conducting current to the patient. A ground loop recording (Figure 7.44) can also be seen when unsecured electrodes default to the ground electrode instead of the normal electrode derivation. Occipital electrodes are particularly vulnerable to this type of artifact due to their location. Like the frontopolar electrodes, occipital electrodes may become dislodged. When this occurs, the resultant morphology is due to the electrode–ground comparison of the differential amplifier.
Intrinsic electrical nonphysiological potentials may also occur during EEG. Intrinsic artifact caused by electromagnetic generators may produce recognizable artifacts in the adult EEG (Figure 7.45). Subcutaneous, implanted electrical devices may produce artifact in patients during EEG recording when an electrical signal produced by the device is recorded. The internal electronic circuitry generates the electrical signal that is detected by the EEG electrodes. The electrodes function like an antenna to detect stimulator-induced electrical currents. Similar to the effect observed when 60-Hz artifact is present in the EEG, internal power sources from devices depolarize EEG electrodes through magnetic fields created by electrical current flow (24). Intrinsic artifact due to both cardiac (Figure 7.46) and neurological devices, such as the vagus nerve stimulator (Figure 7.47), may be encountered during EEG recording. Permanent cardiac pacemakers and implantable cardiac defibrillators are common medical devices that produce intermittent artifact in the EEG. Intermittent artifactual potentials are composed of a similar morphology and precise periodicity, and strongly suggest artifact. Newer implantable electrical devices such as the responsive neurostimulator may also produce intrinsic artifact during EEG recording, recognizable as a brief series of spiky potentials (Figure 7.48). The internal electronics of the battery-operated stimulator can be detected when a seizure is detected, and the device discharges a burst of electrical stimulations.
LONG-TERM EEG MONITORING
There are many different techniques to record adult EEG (Table 7.4). When standard scalp EEG is unrevealing and diagnostic concerns exist, LTM may be considered. LTM refers to a prolonged duration of EEG recording 119beyond the standard recording time of 20 to 30 minutes. Otherwise known as continuous EEG (cEEG) monitoring, it typically refers to VEM of a patient for days during hospitalization. VEM and cEEG monitoring are routinely performed in patients when extending EEG recording to detect, quantify, or characterize seizures is desirable (25). Extending the EEG recording time may also be performed outside the hospital while the patient is ambulatory and identify unexpected findings (Figure 7.49). Either form has a higher recovery rate of interictal and ictal abnormalities, but also introduces a greater likelihood of artifact contaminating interpretable EEG (26).
• Standard scalp (routine) clinical EEG
• Ambulatory EEG
• Video-EEG monitoring
• Critical care continuous EEG monitoring
• Quantitative EEG/trend analysis
• Intracranial EEG
– Wideband intracranial EEG
Computer-assisted LTM systems that record, analyze, process, and store data digitally are widely available to clinical practitioners both in the hospital as well as when the patient is in an outpatient ambulatory setting (27). An experienced technologist is essential to ensure LTM is not limited by artifact. The American Board of Registration for EEG Technologists and Evoked Potentials has standards set by examination to provide credentials reflecting competency for technical performance of EEG (www.abret.org). The American Clinical Neurophysiology Society (www.acns.org), in conjunction with the American Board of Clinical Neurophysiology (www.abcn.org), has established minimum standards and guidelines to aid physician interpretation and reporting. When excessive artifact intervenes in the EEG, a technologist who is able to “troubleshoot” and eliminate the source of artifact to facilitate interpretation of a recording is critical. Applying additional extracerebral monitors, obtaining control of patient position, noting IV drip and the medication delivery system, or temporarily disabling nearby electrical generators help identify the source and can often lead to elimination of artifact that would otherwise impair interpretation.
Prolonging outpatient LTM offers a distinct advantage in recording signals of interest when compared with routine scalp EEG (28). Monitoring aEEG typically occurs over 1 to 3 days depending upon the indications for use (29). Advantages of AEEG include diagnostic use when the clinical history and routine EEG are nondiagnostic. Spike and seizure detection software may be helpful but may capture artifact mimicking an epileptiform discharge (Figure 7.50). Recording paroxysmal neurological events and seizures is much more likely to occur when aEEG recordings are performed compared with routine scalp EEG, though artifact increases relative to the length of the recording (30). In addition, quantifying recurrent seizures and monitoring other physiological parameters are important uses of aEEG, which may also incur artifact. aEEG may be an alternative to definitive diagnosis in the epilepsy monitoring unit (EMU) (Table 7.5). It may also bridge VEG when seizure capture fails to occur in the EMU to extend recording after discharge from the hospital (31). aEEG monitoring is portable and records episodic behaviors in one’s “natural” environment where multiple sources of artifact exist.
• Very young or very old
• Diagnosis in a patient with frequent events
• Hospital-based LTM has been ineffective
• Barriers exist for hospitalization
• Medication withdrawal
• Ancillary testing (e.g., ictal SPECT*) necessary
• Presurgical evaluation and invasive EEG used
• Outpatient LTM has been ineffective
LTM, long-term monitoring.
Adapted from Tatum WO. Artifact and ambulatory EEG. In: Tatum WO, ed. Ambulatory EEG Monitoring. New York, NY: Demos Medical; 2017:41-73.
The primary disadvantage of aEEG is the inherent tendency for movement and routine activities of daily living to introduce artifact into the recording. Artifact becomes increasingly obscured by artifact in parallel with the duration of the recording. In addition, there is limited technical support to ensure proper ongoing integrity of the recording system. It may be inconclusive for patients requiring differential diagnostic evaluation. It is inappropriate for safety reasons to withdraw ASDs and is therefore limited for detailed behavioral analysis associated with a presurgical evaluation. Despite improvements in technology with increasing memory capability and artifact reduction algorithms, instrumental problems occur that defy correction of artifact until the technologist is able to physically interact with the patient (32). Some artifact correction and reduction is able to be performed by some aEEG companies when real-time monitoring is able to intervene. Greater convenience, lower cost, improved access, and greater sampling of natural sleep and assessment of normal circadian rhythms are key advantages of aEEG (26).
Epilepsy Monitoring Unit
Certain artifacts have a distinctive appearance that helps in identification and elimination, whereas others may be unavoidable and limit interpretation of 120the EEG. Movement artifact, common during LTM, is able to be determined using the same principles used to identify IEDs (see Table 7.1). The lack of an electrophysiological spatial distribution and polarity, such as the presence of double- and triple-phase reversal in a bipolar montage in conjunction with video analysis of the semiology, will often suggest artifact. While artifact may serve as a contaminant (Figure 7.1), artifact may also be an important component that could add to the common problem of a failed diagnosis (33). When rhythmic movements produced by the patient occur in nonepileptic conditions such as psychogenic nonepileptic attacks (PNEAs), resultant EEG artifact can mimic a seizure and lead to misdiagnosis (34) (Figure 7.51). With movement-induced artifact in the EEG, the movement frequency is identical to the artifactual frequency. Another clue to movement artifact is, oftentimes, ipsilateral location of the artifact in relation to the movement. However, lateralization is dependent upon associated head movement that may at times be in a different plane than the primary visible movement of the patient. The combination of EEG with video analysis increases the yield of EEG and helps separate artifact when analysis of the EEG alone is inadequate (35). If the technologist is present during the recording, they will be able to identify movement artifact by annotating the record. Certain standards are necessary to assist the interpreter to avoid misinterpretation due to artifact in the EMU (36). When technical challenges arise, an experienced LTM technologist will be able to minimize artifact by ensuring appropriate parameters (Figure 7.52) and integrity of the recording system is appropriate and intact. Scalp-EEG-based seizure monitoring is often contaminated by myogenic artifact with high-frequency discharges in the range of 30 to 100 Hz and even higher when vigorous movement is involved, and may interfere with a presurgical evaluation as a result (Figure 7.53). However, artifact has also been helpful in monitoring epilepsy. The spatiotemporal evolution of movement and myogenic artifact may also impart diagnostic information. It has been used to successfully differentiate epileptic seizures from PNEA (37). For example, time–frequency mapping using the fast Fourier transform assessing the dominant ictal frequency with a coefficient of variation was successfully used to differentiate an epileptic convulsion from a nonepileptic event (38). Variation was greater with epileptic seizures due to the evolving frequencies absent in patients with PNEA despite a similarity of the median frequencies involved. This dissimilarity has been successfully utilized in seizure detection software algorithms to provide monitoring systems that use non-EEG-based signal detection in diagnosis of patients with paroxysmal behavioral events (39).
Invasive EEG recordings are produced from surgically placed intracranial electrodes. There is a wide array of contact configurations (e.g., strips, grids, and depths) which may be implanted in the brain parenchyma or placed on the cortical surface. The electrodes connect to a cable through a stab wound in the scalp to attach to the EEG equipment. Digital standards for EEG recording systems exist to ensure optimal acquisition, analysis, management, transfer, and storage of information. Visual inspection is the gold standard in iEEG interpretation, which emphasizes the importance of ictal recording. Human screening is the standard to rely on artifact recognition, identification, and elimination. Seizure monitoring has diagnostic implications when a clear ictal recording on iEEG is present, though it may be similar to scalp EEG also containing artifact (Figure 7.54). During invasive, `EEG-based seizure monitoring, high-frequency waveforms representing surface-based motor unit potentials are typically absent unless a contaminated, extracerebral reference (e.g., mastoid) is utilized. However, external to the invasive electrodes outside the skull, inductive and capacitance artifact is still possible (Figure 7.55). Despite the differences between scalp EEG and iEEG, the principles of artifact in EEG recording continue to apply (see Table 7.1).
Identification of high-frequency oscillations (HFOs) greater than 30 Hz present in epileptogenic tissue has been a major discovery in the surgical treatment of epilepsy (40). When brief intermittent high-frequency artifact is present during iEEG, it may appear similar to HFOs. HFOs have been recorded using commercially available equipment and, similar to scalp EEG artifact, may interfere with interpretation in the iEEG (Figure 7.56). When surgical resection of normal-appearing brain tissue is performed for drug-resistant epilepsy based upon HFOs, it is critical to separate artifact from HFOs generated by the brain (41).
The OR is an area in the hospital that is electrically complex (hostile environments) and is predisposed to forming artifacts (42). In the OR, nonphysiological artifacts are common. Multiple sources are encountered that introduce AC from nearby electromechanical devices. This produces 60-Hz artifact in the EEG from multiple sources requiring an electrical outlet and power supply. Like 60-Hz artifact, some forms of artifact are easily recognizable due to the pattern (Figure 7.57). Multiple machines with electrical motors are present in the OR. Motors produce magnetic fields and generate artifact that limits visualization of the underlying brain signals in the EEG (20). Furthermore, artifact may be present in the EEG to mimic electrophysiological activity (Figure 7.58). Physiologic artifacts are infrequent due to the anesthesia and neuromuscular blocking medication routinely utilized. However, when an awake craniotomy is performed during electrocorticography, a mixture of environmental and instrumental 121artifact may coexist (Figure 7.59). In this case, low-voltage, high-frequency components that are not normally seen in scalp EEG become visible in the electrocorticogram, requiring separation from other high-frequency sources such as myogenic artifact.
In the OR, highly unusual artifacts may be encountered. During surgery, the technologist must exert supervisory control of essential life-support machinery to perform an artifact-free recording. When elimination is not feasible, the relationship to the source generating the artifact may be identified by application of additional monitors or annotating the chart to delineate the generator. When intervention is possible (e.g., through suction of nasal respiratory passage and repositioning of head), this may lead to resolution of the artifact. Some highly complex waveforms with unusual morphologies from artifactual patterns may be encountered in the OR (43). In this case, the possibility of artifact requires a high index of suspicion to guide the dialogue between the surgeon, the anesthesiologist, and the neurophysiologist.
Automated Artifact Elimination
Artifact reduction techniques include several methods. Artifact removal manually extracts artifact from the EEG. Technologists identify and eliminate artifacts during visual review of the EEG recording. This method is good for detecting multiple and complex artifacts. However, it is performed after the EEG has been recorded, is time-consuming, cost-ineffective, and impractical for LTM. Furthermore, subjectivity and human bias for artifact rejection is present to create variability in quality and reliability of its use. If a segment of EEG contains artifact, it is then eliminated from final analysis. Therefore, a limited representation of the entire EEG recording results, potentially eliminating useful information from final interpretation. Automated artifact rejection is another artifact rejection technique. This is able to identify most artifacts. Filtering, montage selection, adjustment in sensitivity, and alteration of the display speeds are methods to eliminate the source of artifact. However, digital filters alter the amount of artifact visible to the interpreter, distorting waveforms that may lead to misdiagnosis (Figure 7.52). There can be a significant loss of physiological signal that is lost using automated artifact rejection software. Automated artifact subtraction uses software algorithms to reject short segments of EEG when certain segments exceed preset parameter thresholds. These sections are then eliminated from the EEG. Some techniques use additional extracerebral channels (e.g., EOG, EMG, ECG) to monitor artifact. When these channels exceed preset thresholds, EEG containing the artifact will be rejected. Artifact regression is another technique to limit the appearance of artifact in the EEG. It is a computerized, software-based method that selectively subtracts artifact based upon a temporal or frequency-based domain “regressing” common sources of artifact from the final EEG recording. Similar to rejection and subtraction techniques, artifact regression includes subtracting physiological signals from the EEG. Common sources of artifacts (e.g., eye movements and ECG) are most amenable to reduction techniques (44). Multiple artifacts and complex artifacts often associated with movement frequently are not amenable to reduction method isolation and elimination.
Improved software algorithms are being designed for artifact reduction involving the most common types of artifacts generated by the eye, skeletal muscles, and ECG (45). Methods of artifact reduction that separate electrical sources increasingly use independent component analysis and methods based upon principal component analysis (46). Breaking down the EEG into individual frequency components can identify, preselect, and selectively eliminate artifact present in scalp EEG (47). Blind source separation is a new technique (48) that separates the original signals detected from various sources and subsequently eliminates common types of artifact, such as eye movements, myogenic potentials, single-electrode artifact, and ECG (49). Eliminating several simultaneous sources of artifact is the goal for artifact reduction techniques, and as a result limit associated interpretive error that may lead to patient harm (50).
The adult EEG recording contains a variety of physiological and nonphysiological artifacts that may challenge the interpreter’s expertise (2), and visualizing the patterns may be a helpful learning tool (51,52). Some artifacts in the adult EEG reflect physiological function that is essential for proper interpretation of EEG. However, many nonphysiological sources obscure the EEG and limit interpretation. Worse yet, some artifact can mimic abnormality, leading to EEG misdiagnosis and subsequently to mistreatment of patients. Qualitative interobserver agreement in EEG interpretation has been only moderate according to some studies (4). LTM increases the yield in seizure monitoring yet so does the risk of artifacts. Some are simple and easy to recognize; however, others are more complex and difficult to discern. It is impossible to eliminate all diagnostic errors from the decision-making process (52,53). However, certain principles are available to help recognize artifact though recognition should be followed by source identification in an effort to eliminate it from the EEG recording when it interferes with interpretation. Specific inpatient and outpatient characteristics of recording exist, though artifact is omnipresent and some site-sensitive specificity can be present for “hostile environments.” Performance and interpretation of EEG is a large component of clinical neurophysiological procedures performed by neurologists and neurophysiologists. It is through understanding these patterns that optimal patient management will be achieved.