Definition, signal source, and product boundary
Eye movement electrodes are surface electrodes that record the electrooculogram from intact periorbital skin during polysomnography. The corneal side of the eye is relatively positive and the retinal side relatively negative, so rotation of this standing dipole changes the voltage sensed near each eye. The left and right channels can therefore display blinks, slow eye movements, and rapid eye movements. The electrode is only the patient-contact element; the complete recording chain also includes conductive medium, leads, reference and ground electrodes, a patient interface, differential amplifiers, sampling, and display software.
In this entry, an eye movement electrode is a passive biopotential-recording component. It does not deliver therapeutic current to the eye or skin. It is not the same as the ophthalmic electro-oculographic examination used to evaluate retinal function, an electroretinogram, video eye tracking, an infrared camera, an eyelid-motion sensor, or a scalp EEG electrode. Similar materials may be used in some of these products, but their anatomical sites, derivation definitions, signal targets, and validation requirements are different.
The EOG is a mixture of changes in ocular dipole orientation, nearby cerebral and muscular activity, and environmental noise; it is not a direct measurement of eye angle, gaze point, or visual acuity. Amplitude depends on electrode position, reference, skin interface, individual ocular physiology, amplifier, and filtering. Without a method-specific calibration and a validated algorithm, microvolts cannot be converted directly into gaze angle, and absolute amplitudes cannot be compared reliably across devices, nights, or patients.
Two-channel placement, reference, and phase logic
Routine adult PSG uses at least two EOG channels. A common montage places E1 lateral to and slightly below the left outer canthus and E2 lateral to and slightly above the right outer canthus, with a typical offset of about one centimetre. The offsets relative to the two canthi should be equal, and both channels are commonly referenced to the opposite mastoid M2. Exact labels, distances, and alternative montages must follow the current scoring rules, laboratory protocol, and system instructions rather than convenience during application.
During conjugate horizontal eye movement, the corneal-retinal dipole produces opposite-direction deflections at the two vertically offset electrodes, whereas frontal EEG activity reaching both sites is often more nearly in phase. The phase, morphology, and timing relationship between the two channels is therefore central to distinguishing ocular from cerebral activity. Swapped labels, reversed polarity, left-right placement errors, or an incorrect reference on one channel may still produce a visible waveform while destroying that discrimination.
If the M2 reference fails, current technical guidance permits both E1 and E2 to be re-referenced to M1, but the time, reason, and new derivation names must be recorded. Changing the reference on only one channel introduces an asymmetric background that may mimic dysconjugate movement, and amplitude or phase before and after any reference change should not be merged without annotation. Extra periorbital derivations can be added when a specific distinction between vertical and horizontal movement is required; this does not justify casually relocating the two standard electrodes.
Skin interface, component compatibility, and electrical safety
Before application, staff should confirm identity, explain work near the eyes, and gently clean and dry the skin as specified for the electrode. The moving lid margin, globe, broken or infected skin, marked inflammation, and unhealed incisions should be avoided. Periorbital skin is thin, so preparation aims for stable and reasonably balanced contact, not the lowest attainable impedance. Aggressive abrasion, sharp tools, or overflowing paste can injure the skin and may create a conductive bridge to a neighbouring electrode or channel.
Cup, disc, pre-gelled hydrogel, and other disposable electrodes differ in contact area, polarization, adhesive, drying behaviour, connector, and lead strain. The user must confirm that a model is cleared or specified for the intended site and wear duration and that its connection, touch protection, and signal range are compatible with the PSG patient interface. A plug that fits or a waveform that appears does not establish material suitability, electrical performance, or system safety.
Only electrodes, leads, and adapters declared compatible with the recording equipment should be connected to the patient. Exposed metal, damaged insulation, fluid ingress, or improvised laboratory leads require removal from service. When positive airway pressure, oxygen therapy, video, stimulators, or other medical electrical equipment is also present, leakage current, grounding, and the electromagnetic environment must be evaluated as a complete system. An electrode must not be used in MRI, radiofrequency, or another special environment unless the specific model and its entire connection chain are explicitly labelled as compatible.
Biocalibration, overnight quality control, and lead recovery
At the beginning of a study, biocalibration should be performed while the patient is awake and able to cooperate, using a consistent sequence of eyes open, eyes closed, blinks, and gaze left, right, up, and down. The technologist should view both EOG channels with the event markers and verify direction, relative amplitude, channel labels, reference, and timing against each instruction. A large deflection on just one channel is not proof that the montage is correct.
Overnight quality control should look for baseline drift, flattening, clipping, abrupt jumps, persistent mains interference, identical traces in both channels, sudden imbalance, and lead sway coupled to body movement. Tears, sweat, cosmetics or skin products, side-sleeping pressure, pillow friction, mask edges, leak blowing toward the eye, cannula tape, and spreading paste can all alter the interface. Troubleshooting should use raw traces, video, EEG, chin EMG, and neighbouring time segments; software smoothing should not be used to conceal failure.
If one electrode detaches, the remaining channel may still show some ocular activity, but left-right phase discrimination, separation from frontal EEG, and confidence in staging are reduced. When a backup electrode or reference change is used under a predefined protocol, the raw signal and the times of failure, intervention, and recovery should be retained. Copying the other channel, relabelling frontal EEG as EOG, or treating algorithmic interpolation as measurement is unacceptable. Prolonged loss of both channels during important periods must be disclosed, and qualified personnel must decide whether the study remains sufficient.
Role and limits in sleep staging
Awake blinks and intentional saccades, the slow rolling eye movements often seen during the transition into sleep, and the rapid eye movements of REM sleep have different temporal appearances in EOG. The signal is therefore important for distinguishing wake, N1, and stage R and for understanding transitions. Each waveform must nevertheless be interpreted within surrounding epochs and the complete PSG context; one sharp transient or one eye movement cannot label an entire night.
Staging requires EOG to be assessed with EEG background and defining events, chin muscle tone, and relevant technical annotations. A rapid deflection can also be an awake saccade, blink, facial muscle contraction, electrode pop, or head movement. Conversely, some intervals already in stage R do not continuously contain conspicuous rapid eye movements. The presence of a rapid eye movement does not establish REM by itself, and its absence does not by itself exclude REM.
Eye movement electrodes are not independent detectors of respiratory events, arousals, epilepsy, bruxism, or REM sleep behaviour disorder. Ocular activity projects into frontal EEG, which can provide a cross-check but can also distort cerebral morphology. Apparent dysconjugate movement or unilateral absence may reflect a lead error, prosthetic eye, ocular muscle condition, or neurological disorder. Video, history, and an appropriate specialist examination are needed; PSG EOG alone must not be used to make an ophthalmic diagnosis.
Artifact recognition, algorithms, and data comparability
Common physiological contamination includes frontal EEG, orbicularis and temporalis muscle activity, jaw clenching, swallowing, ECG, and pulse. Technical artifacts include electrode pops, unstable contact, cable motion, alternating-current interference, amplifier saturation, and accidental channel bridging. True conjugate movement usually has a spatially coherent pattern across both EOG channels and nearby EEG, whereas a square transient in one electrode, an instantaneous baseline step, or equal-amplitude noise everywhere should prompt inspection of hardware and references first.
Filters, display sensitivity, and sampling alter the slope of slow eye movements and the sharpness of rapid ones. An excessive high-pass setting can attenuate slow rolling activity; excessive low-pass filtering or smoothing can blunt rapid components. Automated ocular-artifact removal may also delete information needed for staging. Settings should follow the equipment and current scoring protocol, and unaltered raw data, processing settings, and software versions should be retained rather than storing only algorithmically modified traces.
Automated staging can use EOG features, but performance depends on montage, reference, channel integrity, training population, comorbidity, and artifact handling. Models derived from standard two-channel PSG, a single frontal derivation, or wearable dry electrodes are not automatically interchangeable. System updates, derivation changes, or a different patient population require renewed validation. Confidence values and human edits should be auditable, and poor-quality intervals should be marked invalid or uncertain rather than forced into an apparently complete hypnogram.
Selection, infection control, and post-use management
Selection should consider recording duration, periorbital space, diameter and edge profile, adhesive strength, conductive medium, all skin-contact materials, lead weight, colour coding, connector, strain relief, and shelf life. Electrodes should be trial-fitted with spectacles, eye masks, positive airway pressure masks, cannulas, and securing tape; closed eyes and side sleeping should not produce sustained lid pressure or skin traction. Material evaluation covers the electrode, gel, adhesive, backing, and cleaning residues, not just the name of the metal.
Disposable or single-patient electrodes must be used and discarded according to their label; an intact appearance does not justify cleaning and reuse between patients. Reusable cup electrodes, leads, and adapters may be reprocessed only by the validated steps for that model, including cleaning, disinfection, rinsing, drying, inspection, and storage. Incompatible disinfectants, soaking, or heat processes must not be applied to adhesive, plating, solder joints, or insulation. Contamination, corrosion, cracks, intermittent conductors, loose connectors, or recurrent quality-control failures require quarantine and retirement.
During removal, the periorbital skin should be supported and the adhesive peeled slowly to avoid pulling lashes or introducing gel into the eye. Persistent erythema, blistering, broken skin, pain, or ocular discomfort warrants discontinuation and clinical follow-up. Receiving inspection and routine quality control should trace model, lot, expiry, storage conditions, reuse count, and failure trends. Eye movement electrodes provide an entry point to sleep physiology; their validity ultimately depends on standardized placement, an intact recording chain, and qualified integrated interpretation.