CHAPTER 4

ARTIFACT AND AMBULATORY EEG

WILLIAM O. TATUM, IV, DO

INTRODUCTION

Ambulatory EEG (aEEG) is an important diagnostic tool for patients with epilepsy and is well suited for assessing patients with seizures and seizure-like episodes in the outpatient environment. There are many conditions that can be evaluated with different forms of EEG recording including scalp-based EEG, intracranial EEG, video-EEG, and continuous EEG monitoring in addition to aEEG for adult and pediatric patients. Similar to other forms of EEG, aEEG serves as an adjunct to clinical diagnosis. Like routine scalp EEG, common problems may arise during recording, including artifact (1). The most specific use of aEEG is for neurophysiological identification, classification, and localization of epileptiform discharges (EDs) in patients with seizures (2). When compared with routine EEG, aEEG demonstrated a higher yield and diagnostic sensitivity (3). Like inpatient video-EEG monitoring (VEM), aEEG can provide support for the clinical diagnosis of epilepsy, assist in classifying the underlying epilepsy syndrome, quantify EDs and seizures, and be used, albeit rarely, to characterize seizures for epilepsy surgery (4–6).

Greater convenience, lower cost, improved access and a greater sampling of natural sleep and circadian rhythms are advantages of aEEG that can provide a high yield (Table 4.1). One major advantage of aEEG is the ability to prolong EEG recording in the natural environment beyond a 20- to 30-minute routine scalp EEG. Most systems are not cumbersome but instead are light and easily worn. Some patients find aEEG socially embarrassing when performing activities of daily life; however, others do not find it limiting (Figure 4.1).

Artifact is present in virtually every type of EEG recording and becomes increasingly more likely as the duration of the recording increases (7). Outside the controlled environment of a hospital neurophysiology laboratory, multiple physiological and nonphysiological sources of artifact exist and may contaminate the record (Figure 4.2). Recording very small signals (microvolts) in aEEG systems typically have fewer electrodes and reduced spatial resolution. In addition, there is less control over external sources. This can challenge the interpreter by the greater degree of interference from lower signal-to-noise ratios present in the ambulatory setting leading to a greater number of artifacts. When artifact appears in the record, it may be subtle or obvious and may challenge the reader to differentiate physiological waveforms from nonphysiological ones. The importance of artifact in the aEEG is reflected by the greater frequency of occurrence associated with prolonged unsupervised recording time. Incorrect recognition of artifact as an abnormality can lead to overinterpretation and adversely affect treatment. Artifact impacts patient care when it is misinterpreted as waveforms that mimic EDs. The development of automated spike and seizure detection algorithms was instrumental in seizure identification when clinical signs are absent, subtle, or occur without the patient’s awareness. Prior to automated seizure detection programs, subclinical seizures were only identifiable with complete manual review of all the data accumulated. Algorithms designed to detect interictal EDs and seizures have allowed for time-efficient seizure detection at any point during the recording session and in clinical research (8). However, most systems are designed to overidentify suspicious waveforms leading to false positives (Figure 4.3). Artifact may be detected as an ED at the expense of false negative detections that miss true EDs. The accuracy of computer-based detection programs that falsely detect artifact as EDs (Figure 4.3) are inherent in most proprietary computer-assisted aEEG (CAA-EEG) systems. The basis for detection algorithms use human identification of EDs as the gold standard for comparison to ensure adequate presence, frequency, and duration (Figure 4.4) have been properly represented. The use of detection algorithms occur with less than 100% accuracy to identify spikes and seizures. In one study, correct identification of abnormalities was found in less than 30% of aEEGs without additional review (9). Intermittent EEG abnormalities including nonepileptiform (Figure 4.5) changes and EDs (Figure 4.6) may appear in the recording so infrequently they limit the usefulness of routine scalp EEG to detect abnormalities. The effect of natural sleep, circadian rhythm variation, and the natural lifestyle of the patient is able to be duplicated during prolonged aEEG. The influence of circadian rhythms associated with EDs extends beyond N2 sleep, and hence, the yield of prolonged aEEG recording is therefore not limited by time (10). Circadian rhythms in genetic generalized epilepsy were seen to influence the yield of aEEG; Fittipaldi et al. (11) found that EDs were identified within the first few waking hours that otherwise would be missed without overnight EEG recording. In another study, Koepp et al (12) found EDs appearing exclusively during sleep in 83% of patients evaluated with aEEG that would not have been as readily detected by a routine scalp EEG recording. Prolonged video-EEG is now able to be performed in and out of the hospital environment to allow full mobilization of patients during recording sessions (13). Ambulatory EEG as an outpatient has emerged as a cost-effective alternative to inpatient VEM in the diagnosis of epilepsy and paroxysmal events (10). In the past several decades, aEEG has evolved from a rudimentary beginning as a “Holter monitor for the brain” to a highly sophisticated, engineered, computer-assisted technology incorporating video with high-fidelity synchronized with EEG for home monitoring. In selected cases, aEEG has been used to evaluate patients for epilepsy surgery (14).

TABLE 4.1  Ambulatory EEG: Advantages and Disadvantages

FIGURE 4.1 A 72-year-old female with “personalized” computer-assisted ambulatory EEG monitoring being performed while shopping at Bealls department store.

FIGURE 4.2 Multiple artifacts including 60 Hz, electrode, myogenic, ECG, and movement artifact. Note the appearance of spike-like electrode artifact that could be misinterpreted as epileptiform discharges.

FIGURE 4.3 Spike detections by software that are due to artifact. Note concomitant ECG artifact.

FIGURE 4.4 Incomplete representation of a burst of generalized spike and waves (GSWs) during aEEG. Note the line that interrupts the burst of GSW from myogenic artifact created by intermittent sampling.

FIGURE 4.5 Spike detection algorithm detecting a generalized rhythmic burst of delta during the first 2 seconds. Note the “spike” detected in the second detection in a channel determined to contain single-electrode artifact (arrow).

FIGURE 4.6 GSW and polyspike-and-wave EDs in the fourth and fifth spike detections. Also note the C3 single-electrode artifact detected by the algorithm in the first through third detection.

ARTIFACT AND AEEG

Potentials that occur in the aEEG during recording that do not possess the appropriate polarity and electrophysiological field that is generated by the brain reflect the features of an extracerebral signal. It is incumbent on the aEEG interpreter to identify the mismatch between potentials generated by the brain from electrical activity that does not conform to a realistic head model; this underlies the skill to recognize artifact. There are multiple types of artifact that can arise from a variety of extracerebral sources present in the outpatient environment that may coexist with EDs, making temporal distinction difficult (Figure 4.7). It may become a clinical problem due to diagnostic error if the EEG is not clear or if artifact is misinterpreted as an abnormality (15). The essential technique to identify artifact is recognizing a mismatch between activities that does not conform to a realistic head model. Pathological spikes generated by the brain may become difficult to separate from spikes due to artifact (Figures 4.6 and 4.8). Artifacts that occur with increasing frequency mimic evolution and simulate a seizure (Figure 4.9) during the process of recording long-term EEG (16). In the ambulatory environment, the potential for artifacts to become introduced into the EEG is ubiquitous. Multiple sources can contaminate the recording (Figure 4.2) and obscure interpretation of an abnormality (false negative) or beguile the interpreter into misidentifying waveforms that simulate EDs (false positive). Pathological EDs may be difficult to separate from physiological (e.g., myogenic) sources or nonphysiological sources (e.g., electrode “pop”) of artifact. Spikes and sharp waves are defined by duration rather than the pathological nature of the source. They also do not portend the capability to generate seizures. Combining the use of video and EEG has significantly enhanced our ability to differentiate cerebral and extracerebral sources by correlating behavior that is time locked to the electrophysiological potentials seen on the EEG (17). When video recording is not used, this poses a significant challenge for identifying artifact, and differentiating them from EDs may become difficult. The importance of separating nonepileptic behavioral episodes with aEEG that contains video is crucial to assist in identifying artifact manifested by pseudoepileptiform patterns derived from nonepileptic sources.

FIGURE 4.7 Single-electrode artifact that mimicked occipital spike-and-waves on aEEG that led to treatment for focal epilepsy.

FIGURE 4.8 Spike detection by the computer amidst myogenic artifact in the same region raising the issue of an abnormality due to the presence of a field of an artifact challenging to the interpreter.

Common artifacts (Figure 4.10) may be misinterpreted during any form of continuous EEG that would routinely be identified with the addition of extracerebral monitors or noted by a skilled technologist. During aEEG recordings, technologists, nursing, and medical support staff who would otherwise ensure appropriate acquisition of the EEG before final interpretation are absent (Figure 4.11). Recognizing artifact, identifying the source, and eliminating nonessential nonphysiological activity are key roles of the technologist that are notably absent (Figure 4.12) during aEEG recording (18). Guidelines for prolonged EEG monitoring have been developed (19); however, despite pitfalls that are known to exist in hospitalized patients during long-term EEG, using similar standards seems reasonable for aEEG. Artifact may occupy the majority of the tracing to contaminate the aEEG by such a degree that it obstructs visual analysis and limits the ability to perform a meaningful interpretation. In this case, the EEG may be considered technically limited. Identifying artifact first requires identification among other suspicious waveforms (20). These include atypical variations of normal background activity, benign variants of uncertain significance, and artifacts that mimic epileptiform activity. Similarly, while there are several types of physiological spikes or sharp waves, particularly during drowsiness and light sleep, the presence of an electropositive spike or sharp wave in an adult should raise the suspicion for an artifact in the absence of a skull defect.

FIGURE 4.9 Rhythmic artifact detected by the computer software as a focal seizure. Paroxysmal “seizure” termination ended during the end of a movement artifact (arrow). No video was associated to identify the source. Note the nonphysiological spatial distribution and similar involvement of the ECG (oval) to indicate an extracerebral source.

FIGURE 4.10 Repetitive vertical eye blink artifact that may be misconstrued as frontal intermittent rhythmic delta activity without eye movement monitors as in this case.

FIGURE 4.11 Sweat sway artifact during aEEG.

FIGURE 4.12 Chewing artifact present on CAA-EEG. Note the bitemporal electrographic maximum and superimposed myogenic polyphasia during chewing motions.

TECHNICAL CONSIDERATIONS

The era of computer-based aEEG has allowed for quality recording of EEG signals on digital media outside the hospital setting (21). Recording, analyzing, and storing large quantities of information are now easily facilitated for several days of aEEG recording with a variety of montages, filters, and display speeds available. Computer-assisted EEG has advanced high-quality, high-fidelity acquisition of brain signals similar to inpatient VEM systems, overcoming many limitations previously imposed by recording EEG on paper media (22). The combination of video with aEEG has improved the technical ability to separate artifacts in the outpatient setting, such as eating (Figure 4.13), from cerebral potentials by coupling alterations in behavior with similar corresponding frequencies identified on aEEG (17). Still, technical problems remain the limiting factor in aEEG recording. Electrode artifact due to deterioration of the electrode paste is common with prolonged aEEG recordings stemming from faulty contact of the electrodes. Other types of instrumental artifacts may alter the recorded signal and be mistaken for abnormalities (Figure 4.14). By observing out of phase deflections, the lack of a credible physiological field can validate the presence of artifact and help the interpreter separate the waveform(s) in question from abnormality (Figure 4.15). The majority of artifact in the EEG appears as extraneous high-frequency noise. Even single-electrode artifact may saturate the amplifiers to limit the usefulness of an aEEG unless the channel is “hidden” from view. Post hoc filtering and montage manipulation are now possible with most aEEG systems to clarify polarity and electrical fields that may suggest an extracerebral source. In addition, the technical specifications of most commercially available aEEG systems typically have adequate analog–digital converters to limit aliasing that will misrepresent the signal. The physical and functional components of EEG are represented by a few critical parameters of recording (Table 4.2). The initial limitations in the number of aEEG channels available to record and in the limited storage capability first present for long-term EEG (23), have been overcome in modern day recording systems (24). In the last two decades, digital technology has evolved to the point that CAA-EEG can record up to 32 channels, including video and software applications utilizing spike and seizure detection. Aliasing based on undersampling a signal in time due to low sample rates or in space due to a small number of electrodes falsely recording the true signal is now unlikely with the present-day specifications of most aEEG amplifiers. A minimum of 16 channels including a single channel dedicated to ECG should be used. Push-button activation aids in identification of events observed by witnesses or caregivers similar to inpatient VEM. Minimum guidelines exist for performing EEG through the American Clinical Neurophysiology Society (25). Recording is tailored to meet the needs of the individual patient, though is usually performed from hours to several days and, therefore, requires daily technical review to ensure optimal integrity of the recording system.

FIGURE 4.13 CAA-EEG demonstrating vertical eye blink artifact and chewing artifact while eating.

FIGURE 4.14 Electrostatic artifact during cable manipulation by the technologist during troubleshooting to correct the ECG channel. Note the similarity to atypical spike and waves.

FIGURE 4.15 Artifact in aEEG that is manifest as clear double phase reversals of opposite positive and negative polarity in adjacent electrodes signifying a nonphysiological field for cerebral activity.

TABLE 4.2  Common Sources of EEG Artifact

Related posts:

Reimbursement Issues in Ambulatory EEG Instrumentation and Polygraphic EEG The History of Ambulatory EEG—A Personal Perspective Ambulatory EEG Cases Technical Aspects of Ambulatory Electroencephalography Clinical Use of Ambulatory EEG in Adults

Stay updated, free articles. Join our Telegram channel

Join