Clinical Significance
Differentiating normal waveforms from abnormal patterns is essential for EEG to achieve high specificity and diagnostic value. EEG waveforms can be characterized by location, amplitude, frequency, morphology, continuity (rhythmic, intermittent, or continuous), synchrony, symmetry, and reactivity (see Image. Normal Awake Electroencephalogram).
Frequency remains the most commonly used method to classify EEG waveforms, with waves named according to their frequency range using Greek letters. Key waveforms include δ (0.5-4 Hz), θ (4-7 Hz), α (8-12 Hz), σ (12-16 Hz), and β (13-30 Hz).
Normal rhythms are categorized into awake and sleep patterns.
Awake patterns include the α rhythm and its variants, μ rhythm, λ waves, posterior slow waves of youth, hyperventilation-induced slowing, photic driving, and photomyogenic response.
Sleep patterns include positive occipital sharp transients of sleep, vertex waves, spindles, K complexes, sleep-related hypersynchrony, and frontal arousal rhythm.
Breach rhythms can affect both awake and sleep patterns.
Additional waveforms exist outside the conventional clinical EEG bandwidth, but they have gained importance with the advent of digital signal processing. These patterns include infraslow oscillations (<0.5 Hz), γ waves (30-80 Hz), and high-frequency oscillations (>80 Hz).
Several patterns or variants of uncertain clinical significance previously considered abnormal in early EEG history are now classified as normal. These patterns include:
Wicket spikes and wicket rhythms: The most commonly overread normal pattern for epileptiform activity
Small sharp spikes: Also known as benign epileptiform transients of sleep
Rhythmic midtemporal θ of drowsiness: Also called psychomotor variant
Cigánek rhythm (midline θ)
6-Hz phantom spike-wave
14- and 6-Hz positive spikes
Subclinical rhythmic epileptiform discharges of adults
Slow-fused transients
Occipital spikes associated with blindness
Temporal slowing in older adults
[2]
Frequency
Mapping the full frequency bandwidth of brain electrophysiological signals is crucial for understanding both physiological and pathological states. Conventional clinical EEG typically focuses on waveforms ranging from 0.5 to 70 Hz, analyzed using bandpass filtering techniques. However, broader EEG bandwidths have been examined by clinical neurophysiologists and researchers and shown to have clinical relevance in specific contexts.
Simultaneous recording of brain direct current shifts, infraslow oscillations (<0.1 Hz), typical local field potentials (0.1-80 Hz), and higher frequencies (80-600 Hz) from the same site holds promise for preclinical epilepsy research and may offer clinical biomarkers for more precise delineation of seizure onset zones.[3] Excluding infraslow or ultrafast frequency bands from routine EEG omits physiologically and pathologically significant features of brain activity.
Full-bandwidth EEG enables analysis of all physiologically and clinically meaningful waveforms without sacrificing a frequency band for another.[4] Despite its potential, routine clinical use of full-bandwidth EEG remains limited, as capturing extremely high-frequency signals requires specialized equipment capable of higher sampling rates and expanded data storage. Based on full-bandwidth EEG recordings, EEG waveforms can be characterized into several types, discussed below.
Infraslow Oscillations
In preterm neonates, infraslow oscillations dominate the EEG at frequencies below 0.5 Hz, typically ranging from 0.01 to 0.1 Hz. These patterns, also referred to as spontaneous activity transients, reflect endogenously driven, spontaneous activity that plays a critical role in shaping neuronal connectivity during early development when sensory input is minimal or absent.[5] Infraslow oscillations within a broader range (0.02-0.2 Hz) also occur during non-rapid eye movement sleep—a phase that coincides with higher-frequency EEG activity.[6]
Research on low-frequency EEG has largely focused on cognitive states and tasks, including contingent stimulation, such as contingent negative variation; voluntary motor activity, such as Bereitschafts potential; and orienting responses.[7][8] These slow scalp-recorded potentials often persist for several seconds and typically exhibit amplitudes of only a few microvolts. Accurate capture of these signals requires full-bandwidth EEG recordings and electrodes with true direct current properties, along with stable skin-electrode interfaces.
Both invasive and noninvasive EEG recordings in animal models and human studies have demonstrated that seizures are associated with very slow EEG responses and variable low-frequency fluctuations at the seizure focus.[9] Recent noninvasive recordings of ictal direct current activity have demonstrated that focal-onset seizures often coincide with prolonged and relatively high-amplitude direct current shifts.[10] Additional evidence indicates that infraslow activity (<0.2 Hz) may support long-range spatial coupling in the brain. A study demonstrated the potential to examine infraslow activity through its modulation of higher-frequency signals, even without direct current–coupled recordings, and emphasized the clinical significance of infraslow dynamics in evaluating seizure risk.[11]
δ Rhythm
The δ rhythm (0.5-4 Hz) appears physiologically during deep sleep and predominates in the frontocentral regions. Pathological δ activity arises during wakefulness in cases of generalized encephalopathy or focal cerebral dysfunction. Frontal intermittent rhythmic δ activity occurs in adults, whereas occipital intermittent rhythmic δ activity presents in children.[12][13] Temporal intermittent rhythmic δ activity is frequently observed in individuals with temporal lobe epilepsy.[14] Recent studies have identified stepping movements as δ-rhythmic (1-4 Hz), modulated by sensory input. Stepping-related δ-rhythmic neural activity, along with β (10-30 Hz) frequency dynamics, also emerges in sensorimotor circuits.[15]
θ Rhythm
The θ rhythm (4-7 Hz) appears during drowsiness and early sleep stages, such as N1 and N2. This pattern is most prominent in the frontocentral head regions and slowly migrates backward, replacing the α rhythm as drowsiness begins. Heightened emotional states can also enhance frontal rhythmic θ activity in children and young adults.
Focal θ activity during wakefulness suggests focal cerebral dysfunction. A recent study showed that patients with obstructive sleep apnea exhibit increased θ phase and β amplitude in the frontal and occipital regions during N1 sleep and wakefulness. This finding implies that cortical coupling is prevalent and displays sleep-stage–specific patterns in obstructive sleep apnea. θ-β Phase-amplitude coupling during N1 and wakefulness exhibited a positive correlation with hypoxia-related indices, suggesting a potential relationship between these neural oscillations and the severity of obstructive sleep apnea.[16]
Recent studies have shown that oscillatory activity in frontal brain regions in the θ range (4-8 Hz) correlates with cognitive processing and typically appears as increased θ activity, especially in post-response (late) periods. This oscillatory activity may be modulated by neurofeedback and has proven effective in upregulating frontal midline θ. Neural oscillatory activity in the θ band (approximately 3-8 Hz), which is linked to cognitive control in nonlinguistic tasks, serves as a real-time index of cognitive control during real-time language comprehension. Increases in θ activity may reflect greater conflict within a sentence. These effects emerge as early as 300 ms after the onset of the initiating event, indicating the rapid recruitment of cognitive control during sentence processing in response to conflicting representations.[17][18][19]
α Rhythm
The posterior dominant α rhythm (8-12 Hz) is characteristically present in normal awake EEG recordings in the occipital head region. This waveform defines the normal background rhythm of the adult EEG. The posterior rhythm reaches the α range of 8 Hz by age 3 and remains stable even into the 9th decade of life in healthy individuals. Fast variants of background α rhythm occur in the normal population. Slowing of the background α rhythm indicates generalized cerebral dysfunction.[20]
The amplitude of the α rhythm varies among individuals and over time in the same individual. Reactivity of the α rhythm is characteristic and aids in its identification. This pattern is most pronounced when eyes are closed and during mental relaxation, and is typically attenuated by eye opening and mental effort. In diffuse encephalopathy, patients may exhibit generalized α activity that is nonreactive to internal or external stimuli, a pattern known as α coma.
The μ rhythm, another type of α rhythm, appears in the central head regions and displays an arch-like morphology. This pattern disappears with motor activity of the contralateral limbs or with thoughts of initiating motor activity.[21] However, the μ rhythm remains relatively unchanged with eye-opening. This waveform is frequently observed in young adults and is less common in children and older individuals.
Attenuating factors of the μ rhythm include fatigue, somatosensory stimulation, and mental arithmetic. μ rhythms are often asymmetric and asynchronous on both sides. Desynchronization of the μ rhythm (8-13 Hz) over centroparietal areas reliably reflects activation of the mirror neuron system, which is associated with social skills. A recent study reported significant μ suppression during hand movement observation across mirror neuron system-attributed frontoparietal areas following neurofeedback training.[22]
Brain oscillatory activity in the α band has been associated with perception, memory, decision-making, and overall cognitive functioning. Recent literature has linked a higher peak α frequency to better visual temporal processing skills, with α oscillations playing a crucial role in temporal binding within and across different senses. Peak α frequency has been revealed to be an electrophysiological index of visual information processing speed, which is also linked to cognitive ability. These findings suggest that peak α frequency deficits are characteristic of schizophrenia but not of bipolar disorder and that individual differences in peak α frequency are due to abnormalities in visual information processing and cognition in schizophrenia.[23][24][25]
Individual α frequency, conventionally ranging from approximately 7 to 13 Hz, with faster α oscillations, is associated with greater temporal resolution and more refined perceptual experience. Research suggests that interindividual differences in α frequency contribute to performance variability in low-level perceptual tasks. These findings support the hypothesis that individual α frequency underlies a fundamental temporal sampling mechanism that shapes visual objective performance, with higher frequencies promoting enhanced sensory evidence per time unit.
In contrast, the α wave has been associated with behavioral inhibition and has been studied in relation to the default mode network, which is predominantly active during resting states. Studies have shown that individuals with low α activity exhibit greater behavioral inhibition and display distinct functional connectivity patterns within the default mode network, suggesting a significant role for low α oscillations in inhibitory control.[26][27]
σ Waves
σ Waves (12-16 Hz), commonly known as sleep spindles, are physiologically observed in N2 sleep. These waveforms serve as markers of sleep mechanisms, including brain development, sensory information processing, and consolidation, as they reflect thalamocortical circuit activity. A recent study involving 8-month-old infants found local differences in the non-rapid eye movement sleep quality that are mostly attributable to the topical phase of brain maturation associated with psychomotor development. Furthermore, sleep spindles in stage N2 sleep are highly heritable, whereas occipital σ activity during non-rapid eye movement sleep is significantly correlated with performance gains of visual perception learning.[28][29][30]
σ Waves may be slow (12-14 Hz) or fast (14-16 Hz) and are most prominently observed in the frontocentral head regions.[31] The fast spindles oscillate in the high-σ band and are related to sensorimotor processing, which is affected by deprived vision. A recent study revealed that children with blindness or severe visual impairment have no developmental spindles within the central area. In contrast, young ones presented low central high σ and high β (25-30 Hz) event-related spectral perturbations and showed no signs of maturational decrease. High σ and high β activity among children with blindness or severe visual impairment is associated with clinical indices that predict perceptual and motor disorders, suggesting that fast spindles are pivotal biomarkers for identifying early developmental deviations in children with blindness or severe visual impairment.[32]
Other studies also identified σ frequency activity as a hallmark of non-rapid eye movement sleep EEG and a key indicator of neural plasticity, central nervous system development, and cognition. Research has shown that sleep EEG spectral power densities within the σ frequency activity range in infants are more often linked to seasonal variation, likely due to changes in seasonal light exposure. Moreover, σ frequency activity was lower in the last part of the night sleep, suggesting different regulation of spindle activity in infants and toddlers compared to adults.[33] In adolescents, the decrease in σ activity is believed to reflect depressed sleep spindle activity, highlighting the impact of sleep restriction on cognitive performance during wakefulness.[34]
Pathological spindle rhythm, also known as spindle coma, is observed in generalized encephalopathy. Current literature suggests that reduced EEG power density in the σ range may represent early signs of overnight withdrawal from the continuous presence of daytime stimulants. Studies have shown that EEG power density in the σ frequencies (12-16 Hz) during non-rapid eye movement sleep is reduced in both caffeine and withdrawal conditions, indicating that daily caffeine intake in the morning and afternoon does not significantly affect nighttime sleep structure or subjective sleep quality in healthy individuals who regularly consume caffeine.[35]
Other findings have shown lower σ wave power in N2 sleep spindle activity and N3 waves in the last sleep cycle during daytime sleep in both the normal group and patients with Down syndrome. Cortical thickness in the σ band during non-rapid eye movement sleep is observed in individuals with Alzheimer disease.[36][37][38][39]
β Rhythm
The β rhythm (13-30 Hz) is the most frequently observed pattern in normal adults and children. This waveform is most prominent in the frontal and central head regions and attenuates as it goes posteriorly. The amplitude of β activity is typically 10 to 20 µV, rarely exceeding 30 µV.
The β rhythm often rises in amplitude during drowsiness and N1 sleep, and subsequently decreases in N2 and N3 sleep. Most sedative medications, such as barbiturates, chloral hydrate, and benzodiazepines, increase the amplitude and quantity of β activity in individuals who use them.[40] Focal, regional, or hemispheric attenuation of β can occur with a cortical injury, malformations, subdural, epidural, or subgaleal fluid collections. Recent case reports indicate an increase in β-band activity in the sensory thalamus associated with severe facial pain, suggesting the potential for closed-loop deep brain stimulation in facial pain management.[41]
Recent studies have shown that low- and high-frequency β rhythms in the motor cortex are correlated with band-specific roles in both movement control and spatiotemporal attention. Findings show that the low β band (<20 Hz) dominant in the primary motor cortex was linked positively to visual cue onset and negatively to uninstructed hand postural micro-movements. Meanwhile, a high β band (>20 Hz) dominant in the dorsal premotor cortex has been shown to exhibit temporal task prediction, with selective modulations before and during cues, which are enhanced in moments of increased focal attention when the gaze is directed at the work area.[42]
Another study investigated the influence of β-frequency sensory stimulation on stepping and dorsal striatal regulation of stepping. Although stimulation at 10- and 145-Hz frequencies produced locomotion and desynchronized striatal network, only 10-Hz stimulation enhanced the δ rhythmicity of stepping and strengthened the coupling between stepping and striatal local field potential δ and β oscillations. Findings revealed that an increase in the frequency of sensory stimulation can modulate lower-frequency striatal neural dynamics and enhance stepping rhythmicity, highlighting the translational potential of noninvasive β-frequency sensory stimulation for improving gait.
Another study suggested that excessive oscillatory activity across basal ganglia nuclei in the β frequency range (12-30 Hz) may be a hallmark of Parkinson disease. Findings revealed that pathophysiological strengthening of striatal and pallidal synapses following dopamine depletion can result in the emergence of synchronized oscillatory activity in the mid-β range, with spike-phase association between the neuronal populations of the basal ganglia. Furthermore, external pallidum inhibition, contrary to the subthalamic nucleus, eliminates oscillations.[43]
γ Waves
γ waves (30-80 Hz) were initially recorded in the visual cortex of monkeys and are now known to be widespread across various brain regions, including the premotor, parietal, temporal, and frontal cortices. γ oscillations are commonly observed in the cerebral cortex during different stages of the wake-sleep cycle and are associated with sensory processing and cognition.[44] The cortico-basal-ganglia-thalamo-cortical loop appears to communicate using this rhythm and is extensively involved in patients with Parkinson disease undergoing deep brain stimulation. An increase in β-γ phase-amplitude coupling over the sensorimotor cortex is consistently observed in patients with Parkinson disease, reflecting neural entrainment in the basal ganglia.[45]
Recent studies have identified narrowband γ oscillations as potential biomarkers, noting their decline with healthy aging and the onset of Alzheimer disease. Visual gratings induce narrowband γ oscillations accompanied by suppression in the α band (8-12 Hz), whereas auditory ripples elicit strong broadband high γ responses alongside suppression in the β band (14-26 Hz).[46]
In addition to its role in sensory perception and the integration of different brain regions, the γ rhythm is also linked to cognitive functions, including working memory and attention. Current literature describes θ-γ coupling as the modulation of γ oscillations by θ-phasic activity, which plays a key role in signaling information during working memory performance. Mental arithmetic, involving calculations with numbers, serves as an essential tool for evaluating sensory processing and working memory management. A decrease in θ-γ coupling and θ power has been observed in patients with panic disorder.[47]
High-Frequency Oscillations
High-frequency oscillations (>80 Hz) are further classified into ripples (80-200 Hz) and fast ripples (200-500 Hz). The γ rhythm is associated with sensory perception and the integration of different brain areas. Recent studies have analyzed spikes and high-frequency oscillations in epileptic EEG patterns, identifying them as potential biomarkers for surgical outcomes in patients with epilepsy, including pediatric patients with frequent seizures.
Spontaneous fast ripples are always pathological, while ripples may be physiological or pathological. Research shows that morphologically defined putative pathological high-frequency oscillations strongly correlate with expert-defined spikes and predominantly localize within the seizure onset zone. Novel pathological features identified include high power in the γ (30-80 Hz) and ripple (>80 Hz) bands centered on the event.
Differentiating pathological from physiological ripples can improve the prediction of surgical outcomes.[48][49][50][51][52] Furthermore, intracranial depth recordings from epileptic hippocampi in animal and human models report ultrafast frequency bursts (fast ripples), which likely correlate with the local epileptogenicity of brain tissue.[53] Subdural recordings during presurgical epilepsy evaluation also show that activity bursts in a relatively lower frequency range (60-100 Hz) may indicate the location of an epileptic focus.
Ultrafast EEG activity correlates with cognitive states and event-related potentials. The significance of γ rhythms in numerous cognitive functions has been well established.[54][55] Brainstem evoked potentials represent a well-established and routinely measured category of ultrafast EEG signals. Reports indicate that high-frequency oscillations greater than 200 Hz are associated with somatosensory stimulation or motor movements. The sensitivity of these waveforms to vigilance states, motor interference, and pharmacological manipulations, such as anesthetics or sedatives, offers new possibilities for brain monitoring and diagnostics.[56][57] These signals may also aid in the early detection of demyelination and other disorders affecting cortical integrity.
Morphology
EEG transients are isolated waveforms or complexes that stand out from the background activity. Several EEG transients observed in normal individuals are benign and must be distinguished from those that are pathological. Correct identification of these nonepileptic waveforms requires training and experience.
A recent cohort study examined the prevalence and characteristics of normal EEG variants. The distribution of these normal variants was reported as follows:
Sharp transients, including wicket spikes: 19.21%
Rhythmic temporal θ of drowsiness: 6.03%
Temporal slowing in older adults: 2.89%
Slowly fused transients: 2.59%
14- and 6-Hz bursts: 1.83%
Breach rhythm: 1.25%
Small sharp spikes: 1.05%
6-Hz spike-and-slow-wave complexes: 0.69%
Subclinical rhythmic epileptiform discharges of adults: 0.03%
EEG readers must be familiar with these variants to avoid misinterpreting nonepileptiform transients, which could lead to overdiagnosis of epilepsy, unnecessary long-term antiepileptic treatment, and serious medicolegal consequences.[58]
Identification of Nonepileptiform Transients
Nonepileptiform transients are waveforms that resemble epileptiform activity but are not associated with epileptic seizures.[59] These patterns often have a sharp contour and may appear as isolated, arrhythmic bursts. Most commonly, these transients occur during drowsiness or light sleep.[60] Some of the frequently observed nonepileptiform transients are discussed below.
λ Waves: λ Waves are positive sharp transients observed in the occipital region during wakefulness, with maximal amplitude over the bilateral occipital areas and concurrent wicket spikes in the ipsilateral temporal lobe (see Image. λ Waves and μ Rhythms).[61] These transients appear most prominently during visual exploration and typically disappear with eye closure.[62]
Positive occipital sharp transients of sleep: Positive occipital sharp transients of sleep are sharply contoured waveforms resembling λ waves and are observed in approximately 50% to 80% of healthy individuals during non-rapid eye movement sleep. Like λ waves, positive occipital sharp transients of sleep show maximal amplitude over the bilateral occipital regions and may coincide with wicket spikes in the ipsilateral temporal lobe. These waveforms are considered physiological patterns of non-rapid eye movement sleep, most frequently observed in adolescents and young adults, with the highest prevalence during the early stages of drowsiness rather than deeper phases of non-rapid eye movement sleep.[63][64]
6-Hz Spike-and-wave complexes: 6-Hz spike-and-wave complexes, also known as phantom spike-and-wave, are low-amplitude, infrequently encountered, and poorly defined spikes occurring within a repeating spike-and-slow-wave pattern.[65] The frequency of these waveforms typically ranges from 5 to 6 Hz. The complexes have amplitudes below 40 μV and exhibit spike durations shorter than 30 ms.[66] Frontal or occipital predominance may be observed. These complexes are most commonly observed in adolescents and young adults.
14- and 6-Hz positive spikes: 14- and 6-Hz positive spikes, also known as ctenoids, are unilateral, bisynchronous or asynchronous waveforms with regular repetitions and arciform morphology. These waveforms are centered in the posterior temporal region but may have a broad distribution. These spikes are frequently observed during drowsiness and light sleep.[67]
Vertex sharp transients: Vertex sharp transients appear as monophasic, diphasic, or triphasic waves and surface-negative sharp waves with phase reversal at or near the vertex. These transients typically occur during drowsiness and non-rapid eye movement sleep. The duration of vertex sharp transients is typically around 100 ms.
K Complex: K complexes are sharp, well-delineated, high-voltage polyphasic waves lasting longer than 0.5 s. These patterns are less sharply contoured than epileptiform spikes and are often followed by sleep spindles.[68] The K complex is considered the most significant event in a healthy human EEG, characterized by a short positive voltage peak at approximately 200 ms, followed by a large negative complex around 550 ms, and a prolonged positive peak around 900 ms.
K complexes may occur in the following ways:
Spontaneously, referred to as spontaneous K complexes.
In response to internal stimuli, such as respiratory interruptions.
In response to external stimuli, such as a tactile sensation. These K complexes are referred to as evoked K complexes.
These waveforms are most prominent over the frontal and superior frontal cortices. Recent studies indicate that K complexes can be generated across a wide range of cortical areas. Notably, the motor cortex exhibits awake-like EEG activity before the onset of a K complex, followed by microarousals. Although K complexes are a hallmark of N2 sleep, they may also play dual roles in region-specific sleep regulation and arousal responses.[69][70]
Benign epileptiform transients of sleep: Benign epileptiform transients of sleep, also known as small sharp spikes or benign sporadic sleep spikes, are the most common normal EEG variants. These waveforms appear as low-amplitude, sharply contoured monophasic or posteriorly propagating, inferiorly rotating diphasic patterns, most frequent during N1 and N2 sleep.[71] Benign epileptiform transients of sleep typically occur in adults aged 30 to 60. These patterns have amplitudes under 90 µV and durations longer than 90 ms. Benign epileptiform transients of sleep most often appear in the midtemporal region with a broad field extending into the adjacent frontal area.
Recent studies describe benign epileptiform transients of sleep as traveling, rotating hippocampal spikes, where the diphasic waveform propagates back and forth along the anteroposterior axis of the hippocampus. This finding suggests a sleep-related functional role for these hippocampal spikes.[72]
Wicket waves: Wicket waves, also known as wicket rhythms, are common EEG transients characterized by monophasic, sharply contoured waves with symmetric upward and downward phases. These patterns blend into ongoing background activity and do not cause disruption.[73] These physiological rhythms occur during non-rapid eye movement sleep and relaxed wakefulness facilitated by drowsiness. Wicket waves are most commonly observed in middle-to-late adulthood.
Rhythmic midtemporal θ of drowsiness: Previously referred to as the psychomotor variant, rhythmic midtemporal θ of drowsiness consists of trains of θ activity observed during non-rapid eye movement and rapid eye movement sleep, sleep-wake transitions, or triggered by hyperventilation, according to recent case reports.[74] Rhythmic midtemporal θ of drowsiness centers on the midtemporal region but may spread to anterior and posterior temporal areas, as well as the posterior parietal region. Rhythmic midtemporal θ of drowsiness displays a monomorphic pattern with a distinctly sharp or notched contour.[75]
Subclinical rhythmic electroencephalographic discharges of adults: Subclinical rhythmic epileptiform discharges of adults is a pattern of unclear clinical significance but is often misdiagnosed as epileptiform activity. This pattern consists of sharply contoured rhythmic θ activity that is bilateral, synchronous, and symmetric, ranging from slower δ to faster θ frequencies. The pattern typically peaks over the parietal and posterior temporal regions.[76] These paroxysms occur without objective or subjective clinical symptoms, commonly during wakefulness or light sleep, and may be induced by hyperventilation. Duration ranges from 10 s to 5 min, averaging 40 to 80 s, resolving either abruptly or gradually. Although subclinical rhythmic epileptiform discharges of adults is most commonly observed in older adults, recent reports have documented its occurrence in young adults and children as well.[77][78]
