Artifact on EEG is anything that does not arise from the brain itself, and can generally be broken down into physiologic, electric, and environmental in etiology. Identifying artifacts on EEG can be challenging for several reasons. First, artifacts are everywhere--they don't follow the rules of localization, can be very disorganized, and often are intermixed with important nonartifact signal. Second, some artifacts can look similar to cerebral signals, and some can even take on rhythmic properties that could be mistaken for seizures. However, these points of difficulty are also what allow you to differentiate artifact from actual cerebral activity, which tends to behave in more predictable ways.
Lateral eye movements are marked by opposing polarities in the F7 and F8 leads. This is due, again, to the positive charge of the cornea and negative charge of the retina. When you look to the right, the right cornea gets closer to the F8 electrode, which sees a positive charge; the left retina gets closer to the F7 electrode, which thus sees a negative charge. So, with lateral eye movements you look to the positive side.
Recall that in bipolar montages you are comparing the first and second electrode in every chain, with the calculation basically being the first electrode minus the second electrode in every lead, So, with the Fp2-F8 lead if F8 is more positive than Fp2, you subract a larger number from a smaller one and wind up with a negative value and thus an upward deflection on the tracing, while in the next lead of F8-T4 if F8 is more positive than T4 as well, you subtract a smaller number from a bigger one, so you get a positive value and a downward deflection. So, you get a sort of "opposite" phase reversal and the leads open up away from one another. The opposite happens on the contralateral side because F7 is negative, and the leads move toward one another.
Recall that EEG helps you keep a positive attitude, so you always look to the positive side on EEG. As such, when considering F7 and F8--the electrodes maximally affected by lateral eye movements--when you see a positive phase reversal (the leads move away from one another) in the F8 leads, the patient is looking to the right, and should have a complementary, simultaneous negative phase reversal (the leads move toward one another) at F7. Other notable things on this tracing are a good PDR of around 11 Hz, a normal AP gradient, some myogenic frontal muscle artifact, and likely a drowsy state given the slow somewhat undulating frontal eye movement artifact in the temporal chains.
Chewing and tongue (hypoglossal) artifact are rather hard to miss on EEG. Chewing artifact is really just muscle artifact from the temporalis muscle, and is marked by sudden onset, intermittent bursts of generalized very fast activity (muscle artifact). It is quite easy to characterize with video studies, as you can just look at the video to correlate, but even without video studies chewing artifact does not usually share a close morphology with any other important physiologic activity. Just be careful not to mix it up with generalized periodic fast activity, which tends to be slightly slower (beta frequency) and lower in amplitude.
Hypoglossal artifact is often but not requisitely seen with chewing artifact. It arises from the movement of the tongue, and appears as slower, diffuse delta frequency activity. Notice how, on the example below, all the tracings have essentially the same movement--this is much too organized for something like slow wave sleep, and while ictal patterns can certainly lead to such rhythmicity, there is no evidence on this single page of the evolution across time or location that characterize seizures.
One notable point about tongue movement artifact is that it is reproducible--you can ask the patient to move it, or to say "la la la" (aka a lingual phoneme) to see if the same pattern is brought out on the EEG.
Don't mistake this tracing for hypoglossal artifact! This page shows rhythmic 2.5 Hz bifrontal spike and wave activity. Notice first that the waves are not quite generalized, as they taper off into the occipital region in the temporal chains (hypoglossal artifact tends to be more generalized). Second, and even more importantly, notice the morphology--each waveform is composed of a small spike and an after going slow wave, which is not typical for hypoglossal artifact either.
ECG artifact is marked by waveforms that are time locked to the QRS complex on the ECG tracing. They tend to be present more so or entirely on the left side, because the heart is in the left half of the chest, and tend to be relatively low amplitude. However, as the example below shows, they can at times be quite prominent, but don't mistake them for posterior discharges or POSTS (which are not time locked to the QRS).
A much less commonly seen cardiac artifact is cardioballistic artifact, in which the EEG electrode is placed just above an artery, and each pulsation of the artery is picked up as motion artifact on the EEG.
This page from a normal tracing shows many artifact types typical for the awake state. Most prominent are the early eye blinks and the aggressive chewing in the back half of the page with some hypoglossal movement seen in between the bursts of myogenic artifact from the chewing. Of note, there is also myogenic artifact elsewhere, most prominent frontally, and there are higher amplitude, disorganized movement artifacts also seen best frontally. Not marked on the page are a few lateral eye movement artifacts interspersed among everything else.
Electrical artifact most commonly arises from interference from the 60 Hz electrical activity (in the USA; 50 Hz in Europe and elsewhere) that that runs through wires; so, it can be caused by anything from an electrical appliance to a cell phone charging, though most modern EEG equipment is quite good at minimizing it. Electrical artifact is a very fast, very monotonous activity, and you can use the notch filter to selectively remove all the EEG activity at 60Hz (this won't affect your interpretation of the signal, as no cerebral activity is that fast on scalp EEG). In the example below, the notch filter is not on, and the Fp1 electrode is likely not on quite right, leading to a lot of electrical interference.
Electrodes themselves can give artifact as well, usually in the setting of them becoming loose over time or getting bumped by things in the environment. A common such finding is electrode pop, caused most often by a loose electrode. On EEG pop is marked by a single electrode showing a very sudden, steep upslope with a slower downslope and absolutely no field. On the example below, F7 is involved; if this continues, you should take a look at the electrode to make sure it's not falling off.
This tracing shows prominent hypoglossal artifact, marked by diffuse slow activity. There is also a lot of myogenic artifact from the lateral temporal leads, likely from the temporals muscle. Towards the end of the page (the last three seconds), there is some movement artifact seen as "sloppy" appearing and disorganized slow activity more prominent frontally the temporal chains.
Myogenic artifact comes from muscle movements, and is most commonly found in the frontal or lateral temporal regions, due to the frontalis and temporalis muscles. It is marked by high frequency, often low amplitude activity overlying the normal cerebral rhythms, and is usually most prominent in the awake state. Of note, there's typically only minimal myogenic artifact near the vertex so if you see fast activity there, be slightly more suspicious (although, realistically, myogenic activity is much faster than the cerebral activity that can be picked up on scalp EEG).
This page has four eye blinks, seen as high amplitude frontal positive deflections. More pertinent, however, is something perhaps less apparent to early learners: diffuse excess beta activity. Notice all the low amplitude fast activity overlying the activity in all the leads--this is excess beta. It is most commonly seen in the setting of benzodiazepine use, and is considered an abnormal background finding. Do not mistake this excess beta for myogenic activity, which tends to be even faster in frequency.
There is also an "electrode pop" at T4, marked by a sudden discharge with very steep upslope in a single electrode without any field; this is just a sign of perhaps the electrode being touched, or some other physical issue (if these are persistent, take a look at the electrode to ensure its firmly in place).
Sweat artifact is marked by very slow (typically less than 0.5 Hz), relatively low amplitude activity that arises because the sodium chloride in sweat carries a charge, which is picked up by the EEG electrodes. Sweat artifact doesn't have to follow a particular pattern in terms of localization, and can be bilateral, unilateral, or even focal to just a few electrodes. The example below also has electrode artifact at P3.
In the hospital, particularly in patients intubated and sedated in the ICU, you'll commonly come across chest physiotherapy artifact. This is important to know about because it can appear rhyhthmic similar to a seizure, although you can easily distinguish the two (even without your keen electrophysiologic skills) by looking at the video, if available. On the EEG, it also typically differs from a seizure by its lack of a field or evolution. Chest PT tends to be more prominent posteriorly, but depending on the patient's position and where the PT is being applied, that can vary widely.
Movement artifact has a plethora of appearances, usually of chaotic looking, high amplitude activity that doesn't mimic any actual cerebral patterns. However, head shaking artifact can be slightly trickier and is marked by slow, low amplitude activity that is usually more prominent posteriorly if the patient is resting their head on a pillow. It can look similar to the roving eye movements of drowsiness, but eye movements are in the anterior leads.
While the above artifacts are perhaps the most commonly seen, you'll come across many other kinds of artifact as you continue to read EEG. On this example below, for example, in an intubated patient a small bubble on the tip of the nose fluctuated with breathing, leading to a sort of periodic burst of activity that is, in fact, just artifact.
Similar to eye blinks, note that the prominent positive deflections across this page do not extend posteriorly beyond the frontal leads. Given this constricted frontal area involved and the lack of any underlying spike waves to suggest epileptiform activity, these waveforms are most consistent with eye flutter. Perhaps the patient needs some eye drops.
Eye blinks are a large positive frontal deflection, due to Bell's Phenomenon. When you blink, the eye rolls slightly up and the negative retina moves away from the Fp1 and Fp2 electrodes, which thus become relatively positive.
Eye closing corresponds to a large frontal positive deflection, and eye opening to an opposing negative deflection. Note that eye movement artifact doesn't have a significant field posteriorly.
Lateral eye movements are marked by a frontal (F7/F8) positive deflection on the side to which you look with a contralateral negative deflection, due to the cornea's positive charge.
With lateral eye movements, you look to the positive side. Lateral eye movements are most easily seen in REM sleep, but in drowsiness you also see slower, roving lateral movements.
Chewing artifact arises from the myogenic activity of the temporalis muscles while you chew. Its marked by bursts of generalized fast activity, often with underlying hypoglossal artifact.
Hypoglossal, or tongue, artifact is seen as diffuse slow, synchronized activity. It arises from the mechanical movement of the tongue itself.
Tongue and chewing artifact often, but don't always, come together. Obviously, they are seen more during eating.
Muscle, or myogenic, artifact is seen as very fast activity often overlying the normal cerebral rhythms. It tends to be more prominent frontally, and is minimal at the vertex.
If you're not sure if something is tongue artifact, ask the patient to say "la la" or push their tongue into their cheeks to see if the EEG changes.
Bruxisms, or tooth grinding artifact, is seen as a broad checkerboard pattern of alternating, lateralized myogenic activity that arises from the alternating contraction of masticatory muscles.
Chest PT is done by repetitive tapping on the chest, and can appear as rhythmic movement artifact on EEG. Don't confuse it for seizure, which requires evolution in time and location.
Head shaking is just a type of movement artifact, but because we often shake our head yes or no in a rhythmic fashion, it appears rhythmic on EEG.
This burst of right frontally predominant rhythmic activity at first glance may appear like a lateralized rhythmic delta activity (LIRDA). However, note the significant ECG strip artifact in the same period, suggestive of artifact. Moreover, the video of this tracing showed the patient rubbing their head--no mystery there.
In patients intubated and sedated, mucous bubbles in the nose may vibrate with exhalation and inhalation, leading to bursts of activity that are time locked with their breaths.
Note here how the sudden onset, diffuse bursts of myogenic fast activity coincide with similar artifact in the ECG strip, and does not have any clear evolution as you'd expect with a seizure.
When an electrode gets loose or has insufficient conduction gel, you can see many artifacts, including breakthrough of 60 Hz electrical activity. The notch filter would remove this.
Electrode pop arises most commonly due to a loose electrode, and is marked by a steep positive upslope and slower downslope confined to a single electrode.
Don't confuse electrode pop for lateral eye movements, which involve the respective contralateral electrode(s). This tracing also has excess background beta activity.
In ICU patients, ventilators can cause a plethora of artifact. Here we see intermittent, periodic bursts of sharp activity, most prominent in the left anterior temporal region that is caused by the motion of the ventilator tubes, along with the condensed water inside them.
For drug resistant epilepsies, RNS (responsive nerve stimulator) can be implanted at the site of seizure foci if eloquent cortex at that spot precludes surgical resection. RNS stim artifact appears as high amplitude, very fast bursts over the region implanted. Note the slowing and breach in the right hemisphere here, discussed in the nonepileptifrom abnormalities section.
ECG artifact appears as spike-like activity that is time locked to the QRS complex on the ECG tracing, and is more prominent on the left side because the heart is on the left.