Artifacts and Noise
L. John Greenfield Jr
One of the challenges in EEG interpretation is separating out signals that appear to be of cerebral origin but are in fact derived from other sources. One example of how a reader can be misled by EEG appearances occurred during a long-term epilepsy monitoring session. The EEG showed a slow rhythmic discharge with an apparent field involving a single electrode, which then increased in frequency, spread to adjacent electrodes, gradually increased in amplitude over the course of 15-20 seconds, and then abruptly ceased. This might have been the classic appearance of a partial-onset seizure. It was not until reviewing the video taken during the event that the explanation became clear—the patient had been scratching an itchy scalp electrode! Sometimes, the causes are less obvious, and it is important to be aware of the possible sources of noise and artifact in EEG recordings.
Artifacts can be divided into several different types: (1) electrode-related noise and potentials, (2) noncerebral biological potentials, (3) electric device- and power supply-related signals, and (4) patient movement and extraneous physical artifacts. It is important for the technician to seek out and eliminate these problems as much as possible and, when not possible, to note when they occur and their presumed etiology on the EEG record. For greater detail, readers are referred to the excellent chapter on this topic in Ebersole and Pedley’s Clinical Practice of Electroencephalography (3rd ed).1
ELECTRODE-RELATED NOISE
Electrode “Pops”
One of the most common sources of noise is a sudden increase in the impedance of the electrode known as an electrode “pop,” due to a head movement or the drying of the conductive paste. These appear as sudden disconnects or jumps in the voltage potential with a drift back to baseline. Electrode pops are distinguished by their absence of a field, with the disturbance affecting only a single electrode. In a bipolar montage, they are made obvious by “mirror image” activity on adjacent channels within a chain, due to the electrode being connected to the negative input of one channel and the positive input of the next. The hallmark of an electrode artifact is the absence of involvement of other electrodes or, in EEG parlance, the lack of a “field,” though artifactual electrode problems can involve more than one electrode in a scalp region giving the false impression of that a cerebral field exists (Fig. 5.1). The technician should attend to such electrodes as soon as the problem is identified, often by mildly abrading the scalp through the hole in the electrode while adding conductive gel, which can lower impedance and fix the problem. Until the situation is corrected, the information from that electrode should be viewed with caution and skepticism.
Salt Bridges
As previously mentioned, low impedance connections can result from sloppy application of conductive gel, sweating, or other situations, resulting in a slow undulating potential (usually 0.5 Hz or slower) involving the electrodes affected by the low impedance connection.
High Impedance
Increased electrode impedance causes several different kinds of noise. The mismatch of impedances between inputs of the differential amplifier prevents suppression of 60-Hz AC line noise and can amplify pickup of other extraneous signals including EKG. These problems can be minimized by ensuring that the impedance of all electrodes is <5000 Ω.
Photoelectric Response
During flash photic stimulation, electrodes with high impedance will sometimes generate an electrochemical response in which the flash stimulates a very brief photoelectric potential in the Fp (and sometimes other frontally placed) electrodes. This potential has virtually no delay and extremely fast onset and termination, as it is the direct result of the action of light on the electrodes, which helps distinguish it from the photomyoclonic response generated by reflex blink activity of the orbicularis oculi in response to the flashing light.
REVIEW
5.1: What are the four main categories of EEG artifacts?
View Answer
5.1: The four main types of EEG artifact are (1) electrode-related, (2) noncerebral biological signals, (3) noise from nearby electric devices or power lines, and (4) patient movement and physical artifacts.
5.2: How can you tell an electrode pop from an epileptic spike?
View Answer
5.2: Electrode pops involve a single electrode and have no “field,” that is, no comparable signal in other leads. They also show up as mirror images on bipolar montages in which the affected electrode is represented on two adjacent derivations.
5.3: What is the difference between a photoelectric response and a photomyoclonic response?
View Answer
5.3: A photoelectric response is created by light acting directly on the electrode. The photomyoclonic response is due to muscle twitches in response to the flash.
NONCEREBRAL BIOLOGICAL SIGNALS (ARTIFACTS)
The high gain amplification required to record EEG potentials can pick up other patientgenerated biological potentials. Many of these are obvious to experienced EEGers, but some can pose interpretive problems. All are more prominent at high sensitivity settings (high gain) and with long interelectrode distances as occur with referential montages or double-distance bipolar montage recordings. Some can be reduced or eliminated by appropriate filtering, while others must simply be recognized and noted.
Electrocardiogram
Electrocardiogram (ECG/EKG) potentials generated by ventricular contraction (the QRS complex) may be large enough in amplitude to be detected by cerebral electrodes. These usually cause a regular spiky-appearing potential (see Fig. 5.2) that may be more prominent at some electrodes than others (particularly those near the skull base or over the left hemisphere) and may vary in amplitude and frequency due to intrinsic cardiac problems like atrial fibrillation or sinus arrhythmia.
Patients with hypertension-induced myocardial hypertrophy may generate larger voltage potentials with greater likelihood of contaminating the EEG. This activity occurs independently from cerebral potentials, and does not alter cerebral activity, but occasionally the coincidental occurrence of an ECG-related sharp wave with an “aftergoing” slow wave can lead to the erroneous diagnosis of a spike-and-wave complex. Concurrent recording of a single channel of ECG allows comparison of cerebral and ECG-generated potentials. It should be standard procedure to examine the timing of the QRS complex when evaluating possible spike-and-wave activity, and when these are simultaneous, the cerebral origin of such a complex must be questioned. It should be noted that the presence of an ECG lead in the EEG montage serves an additional purpose by allowing the detection of arrhythmias, which may help distinguish between epilepsy and syncopal disorders with similar presentations. If an ECG anomaly is found during the recording, its presence and potential significance should be noted in the EEG report.
Patients with hypertension-induced myocardial hypertrophy may generate larger voltage potentials with greater likelihood of contaminating the EEG. This activity occurs independently from cerebral potentials, and does not alter cerebral activity, but occasionally the coincidental occurrence of an ECG-related sharp wave with an “aftergoing” slow wave can lead to the erroneous diagnosis of a spike-and-wave complex. Concurrent recording of a single channel of ECG allows comparison of cerebral and ECG-generated potentials. It should be standard procedure to examine the timing of the QRS complex when evaluating possible spike-and-wave activity, and when these are simultaneous, the cerebral origin of such a complex must be questioned. It should be noted that the presence of an ECG lead in the EEG montage serves an additional purpose by allowing the detection of arrhythmias, which may help distinguish between epilepsy and syncopal disorders with similar presentations. If an ECG anomaly is found during the recording, its presence and potential significance should be noted in the EEG report.
Pulse Artifact
A related issue is artifact due to pulse. These potentials are generated by subtle physical movement of the scalp or the brain due to cardiac pulsations (Fig. 5.3). They are thus slower than the QRS waveform and occur at a fixed delay after the QRS complex. Since pulsation artifacts are rhythmic and often sinusoidal in character, they can be mistaken for cerebral potentials. They can be very prominent in patients who have undergone neurosurgery, in whom the cranium is no longer intact. A tight correlation with the ECG waveform is usually sufficient to make the diagnosis.
Eye Movements
The eyes, like several other body parts, have an associated electric potential that causes electric field fluctuations with eye movements. Due to the electric activity of retinal neurons, there is a DC potential associated with the globe of the eye in which the retinal/posterior end is negative and the cornea is positive (see Fig. 5.4). Since this is a DC potential, it does not affect the EEG when the eyes are still, but eye movements create moving or changing fields that generate signals detected by nearby EEG electrodes, predominantly Fp1/Fp2 and F7/F8, which sit above the medial and lateral orbit, respectively. At times, the eye muscles also generate fast “spiky” potentials known as “rectus spikes” at the onset of eye movements. The potentials generated by eye movements depend on the direction, nature, and speed of the movements. Fast saccadic movements to a target, both during wakefulness and in rapid eye movement (REM) sleep, cause sudden stepwise movements with a decay to baseline consistent with the time constant of the recording. Lateral saccades are sometimes accompanied by a brief spiky myogenic potential from the lateral rectus muscle (on the side to which the eyes are deviated), seen best in F7- or F8-linked derivations.
In drowsiness, slow lateral eye movements cause drifting or “rolling” potentials that shift gradually from side to side over several seconds. This is an important clue regarding the state of the patient and should not be mistaken for focal slowing. The distinction becomes obvious due to the opposite polarity of slow waves over the left and right frontal regions.
In the longitudinal bipolar montage, eye movements are easily discerned due to the bipolar connections between Fp1-F7 and F7-T3 on the left and Fp2-F8 and F8-T4 on the right. In Figure 5.4A, we see that either a saccadic or slow drifting eye movement to the left causes the front part of the eye, with its positive charge, to move toward F7 and away from Fp1, while the negatively charged retina moves toward Fp1 and away from F7, both making F7 relatively positive compared to Fp1. The opposite occurs at the right eye. The cornea moves toward Fp2 and the retina toward F8, making Fp2 positive to F8. These opposite polarities cause mirror image effects in the temporal chains. The negativity at Fp1 relative to F7 causes an upward deflection, while the positivity at F7 relative to T3 (too distant to be involved) causes a downward deflection. The positivity at F7 forms another kind of phase reversal, this time positive with waveforms pointing away from each other. This is easy to remember as you can imagine a plus sign fitting in between the oppositepointing waveforms. On the right side, the positivity at Fp2 compared to F8 causes a downward deflection, while the negativity of F8 relative to uninvolved T4 causes an upward deflection, revealing a phase reversal with negativity at F8. Again, a mnemonic for the negative potential at the shared F8 electrode is that only a “minus sign” will fit in the narrow space between the two waveforms pointing at each other. These two events will occur simultaneously since eye movements are tightly coordinated. Examples of conjugate left, right, up, and down eye movement potentials in the anteroposterior bipolar montage are shown in Figure 5.5.