Technical Considerations and Polysomnography Equipment

Biological signals recorded during PSG are of very small amplitude (EEG, EOG, and EMG activity is in the microvolt range) and need to be amplified to be displayed and analyzed. These waveforms also need to be filtered to best visualize activity of interest and exclude artifact. PSG equipment is thus basically a series of amplifiers that records and amplifies this activity and then passes it through adjustable filters for display at different sensitivity settings.

High-frequency (or low-pass) filters attenuate all activity at frequencies higher than the value at which they are set, while allowing lower frequency activity to pass. For example, EEG channels are typically set with a high-frequency filter at 70 Hz. This setting would attenuate any activity above 70 Hz but allows lower-frequency activity to pass through unchanged. Low-frequency (or high-pass) filters, on the other hand, attenuate all activity lower than the value at which they are set, while allowing faster-frequency activity to pass. EEG channels, typically set at 0.3 Hz, would therefore attenuate out any activity slower than 0.3 Hz but allow higher-frequency activity to pass through unchanged. These filters can be adjusted to eliminate known sources of artifact. For example, decreasing the high-frequency filter from 70 Hz to 35 Hz may eliminate faster-frequency artifacts like muscle artifact, whereas increasing the low-frequency filter from 0.3 Hz to 1 Hz may attenuate slower-frequency artifacts like sweat artifact. Most PSG amplifiers also have a 60-Hz notch filter, which attenuates the main frequency while attenuating activity of surrounding frequency less extensively. This is particularly useful in studies contaminated by 60-Hz artifact (the frequency of electrical activity in North America; the frequency in other countries is usually 50 Hz), generally seen if the electrode application and impedance are suboptimal. The use of the notch filter, however, is generally discouraged because some important components in the recording, such as muscle activity and epileptiform spikes, may be attenuated. Thus the technician should ideally attempt to eliminate this artifact at the time of recording by reapplying electrodes and attempting to isolate and eliminate the source of the artifact. The differential amplifier is generally sufficient to reject 60-Hz artifacts.

PSG equipment uses differential amplifiers, which amplify the potential difference between the two amplifier inputs. The result of this is that unwanted extraneous environmental noise, which is likely to be seen at the two electrodes, is subtracted out and therefore cannot contaminate the recording. The ability of the amplifier to suppress an extraneous signal such as noise that is simultaneously present in both electrodes is measured by the common mode rejection ratio. Ideally this ratio must exceed 1000 to 1, but most contemporary PSG amplifiers use a ratio in excess of 10,000 to 1.

The amplifiers used consist of both alternating current (AC) and direct current (DC) amplifiers. The AC amplifiers are used to record physiological characteristics showing high frequencies, such as EEG, EOG, EMG, and ECG. The AC amplifier contains both high- and low-frequency filters. DC amplifiers have no low-frequency filters and are typically used to record potentials with slow frequency, such as the output from the oximeter, the output from the pH meter, CPAP titration pressure changes, and intraesophageal pressure readings. AC or DC amplifiers may be used to record respiratory flow and effort.

Sensitivity is expressed in microvolts per millimeter or millivolts per centimeter. Sensitivity switches should be adjusted to obtain sufficient amplitude for interpretation. Sensitivity and filter settings vary according to the physiological characteristics recorded (Table 1.1).

The standard speed for recording traditional PSG is 10 mm/sec so that each monitor screen is a 30-second epoch, making sleep staging easiest. A 30-mm/sec speed is the traditional speed at which EEGs are analyzed, because it allows easy identification of epileptiform activities. While reviewing the PSG at the traditional 10-mm/sec speed, the polysomnographer may pick up EEG abnormalities that can be better analyzed by slowing the recording down to 30 mm/sec. On the other hand, with experience, polysomnographers may choose a 5-mm/sec speed, rendering a 60-second epoch, to better visualize respiratory events.

Electroencephalography

The main purpose of EEG recording performed during PSGs is to distinguish between wakefulness and the various stages of sleep. This is further elaborated on in Chapter 3. According to the 2007 AASM Manual for the Scoring of Sleep and Associated Events, a minimum of three channels (F4-M1, C4-M1, O2-M1) representing the right frontal, central, and occipital electrodes referenced to the contralateral mastoid electrode is recommended, with corresponding backup electrodes over the left hemisphere (F3-M2, C3-M2, O1-M2) referenced to the contralateral mastoid electrode, in case of malfunction of the primary electrodes. Although the montage just described would theoretically be sufficient to detect a posterior dominant rhythm in wakefulness (best seen in occipital leads) and major sleep architecture (vertex waves, sleep spindles, and K complexes, best seen in frontal and central derivations), there are serious limitations to adhering to this minimum prescribed montage. Recording over only one hemisphere may result in inability to score sleep accurately if that hemisphere is lesioned (as in a patient with stroke or tumor) or in missing serious pathological conditions if the contralateral hemisphere is involved. The absence of a temporal lead may result in missing epileptiform activity, which is most common in this region. The American Academy of Sleep Medicine (AASM) also recommends an alternative derivation that includes the midline and a central channel (Fz-Cz, Cz-Oz, C4-M1); however, in addition to the same concerns that arise with the recommended standard derivation, another limitation with this montage is the tendency for midline, centrally predominant activity such as sleep spindles, K complexes, and slow wave sleep to be attenuated and easily missed. This would pose a particular problem in elderly patients while scoring stage N3 sleep, in which the slow wave activity must meet particular amplitude criteria (see Chapter 3). Therefore we recommend a montage that records over both hemispheres and includes the temporal regions (Table 1.2) in addition to electrodes recommended by the AASM for the scoring of sleep. For patients in whom nocturnal seizures are suspected or likely to occur, a full seizure montage with parasagittal and temporal chains is recommended (Table 1.3; see also Fig. 2.1). It is important for the polysomnographer and the polysomnographic technologist to be familiar with major patterns of EEG abnormalities that may be encountered during PSG recording (see Chapter 2).

Electro-oculography

EOG recording is crucial to staging sleep accurately. The two recommended electrodes are labeled E1 (1 cm below the left outer canthus) and E2 (placed 1 cm above the right outer canthus), both referenced to the right mastoid; this allows simultaneous recording of both vertical eye movements (such as blinking) and horizontal eye movements (both slow and rapid). Gold cup or silver–silver chloride electrodes can be used to monitor the EOG.

The underlying concept is that the eye is an electric dipole, with relative positivity at the cornea and a relative negativity at the retina. Any eye movement changes the orientation of the dipole, and it is the movement of the dipole that is recorded as a potential difference between the two electrodes used to record the EOG. In this arrangement, conjugate eye movements produce out-of-phase deflections in the two channels, whereas the EEG slow activities contaminating the eye electrodes are in phase. The sensitivity and filter settings for EOG are similar to those used for EEG (see Table 1.1).

Eye movements are generally characteristic of the sleep stage in which they occur and are an essential part of scoring. Eye blinks, seen only in wakefulness, are conjugate vertical eye movements occurring at 0.5 to 2 Hz with the eyes open or closed. Rapid eye movements (conjugate, irregular, sharp eye movements with an initial deflection of less than half a second) occur in wakefulness along with high chin EMG tone, eye blinks, and a posterior dominant rhythm, but they also occur in REM sleep, especially in phasic REM, where they occur in bursts seen in all directions (horizontal, oblique, and vertical) and are accompanied by low to absent chin tone (interspersed with similar phasic bursting) and a desynchronized, amorphous EEG pattern. In REM sleep, rapid eye movements are frequently preceded by sawtooth waves (see Fig. 3.15), although both may occur independently. The frequency with which bursts of rapid eye movements occur in REM sleep is called REM density. It typically increases in later REM cycles during the course of a normal PSG; this may be reversed in patients with depression.

Slow lateral eye movements are seen in drowsiness and light sleep and are defined as conjugate, sinusoidal, regular eye movements with an initial deflection of greater than half a second (Fig. 1.1). These eye movements are not under voluntary control and cannot be volitionally simulated. In patients who do not generate a posterior dominant rhythm, their appearance heralds stage N1 sleep. Although they may persist into stage N2 during the early part of the night, they generally disappear in stage N3 and REM sleep. However, patients on antidepressants such as selective serotonin reuptake inhibitors (SSRIs) like fluoxetine and paroxetine, as well as serotonin-norepinephrine reuptake inhibitors (SNRIs) such as duloxetine, may have unusual eye movements that appear to be a mixture of rapid and slow eye movements occurring well into stage N3 and often into REM sleep (colloquially referred to among polysomnographers as Prozac eyes); their presence makes sleep staging difficult (Fig. 1.2) and can render scoring of a multiple sleep latency test (MSLT) equally frustrating. The role of EOG in sleep staging is further discussed in Chapter 3.

Electromyography Recording During Standard Polysomnography

EMG channels provide important physiological characteristics that help determine sleep stage, as well as aiding in the diagnosis and classification of a variety of parasomnias. At a minimum, a PSG consists of chin EMG channels recording activity from the mentalis and submental muscles (the mylohyoid and anterior belly of the digastric) and bilateral leg EMG channels recording activity from the tibialis anterior muscles. EMG is recorded using a gold cup or a silver–silver chloride electrode applied to a clean surface using a tape or electrode glue. For chin EMG recordings at least three EMG electrodes are applied so that in the event of a problem with one of the electrodes the additional electrodes can be connected during the recording without disturbing the patient. The electrode impedance should be less than 5000 ohms. The high- and low-frequency filter settings for the EMG recordings are different from those used for EEG and EOG and are listed in Table 1.1. The sensitivity should be at least 20 μV/mm for mental or submental EMG activity.

As recorded during a PSG, the EMG channels represent the surface recording of intracellular changes occurring as a result of muscle depolarization during contraction. Unlike needle EMG performed in patients with suspected neuromuscular disease in the neurophysiology laboratory, analysis of motor unit morphological characteristics and firing pattern is not the focus of these recordings. Rather, the EMG channels provide important information about overall muscle tone. EMG tone is seen to progressively decrease with sleep onset and continue to diminish through NREM sleep to a point where it is at its minimum and almost absent in REM sleep. Phasic bursts in the chin EMG (as well as limb EMG) are seen in phasic REM sleep.

Lower-limb EMGs are generally recorded with electrodes placed over the tibialis anterior muscles 2 to 2.5 cm apart. The main utility of these channels is to record limb movements in patients with periodic limb movements in sleep (PLMS). Although generally seen in up to 80% of patients with restless legs syndrome (RLS, recently renamed Willis-Ekbom disease), PLMS are often seen in normal patients with no daytime complaints, especially those above the age of 65 years. PLMS are also seen in those with a variety of sleep disorders (such as REM sleep behavior disorder [RBD] and narcolepsy, in which they may be abundant in REM sleep) and those on antidepressants such as SSRIs and SNRIs. For this reason a careful sleep history is essential while determining the importance of PLMS. In many cases PLMS occur in association with respiratory events as part of OSA. These are referred to as respiratory-related limb movements and may respond to CPAP treatment. This association cannot be made without PSG that allows simultaneous analysis of respiratory and EMG channels, although kicking leg movements or similar body flailing is often the presenting complaint of the patient or spouse.

Many patients with a history of abnormal movements or behavior in sleep require a more extended EMG montage, known as a multiple muscle montage, which includes extra channels that record from additional cranially innervated muscles (such as the sternocleidomastoideus, masseter, and mentalis), upper limb muscles (e.g., biceps, triceps, extensor digitorum communis), lower limb muscles (e.g., quadriceps, gastrocnemius), and axial muscles (e.g., paraspinals, rectus abdominis, intercostals) (Table 1.4; see also Fig. 9.4). This is of particular utility in patients with suspected RBD, in whom REM without atonia may be missed if an adequate number of muscles is not sampled. Although a standard montage for RBD has not yet been agreed upon, Frauscher et al found that simultaneous recording and quantitative analysis of the mentalis and flexor digitorum superficialis in 3-second miniepochs was 100% specific for RBD, when activity was present in more than 31.9% of miniepochs. The heterogeneity of RBD appears to be expressed in the dissociated EMG findings in muscles innervated by the cranial nerves, as well as the arms and the legs, requiring recording from multiple muscles. Multiple muscle montage recording may also be useful in patients with suspected RLS, because PLMS may also occur in the arm muscles or, rarely, in the axial or cranially innervated muscles (see Fig. 9.3).

Additional EMG channels aid in the analysis of unusual movements in sleep, especially myoclonus, and help assess their propagation and thus their generators. A dystonic muscle burst refers to prolonged EMG activity of 500 to 1000 milliseconds or longer (see Fig. 9.6). Myoclonic muscle bursts are also phasic bursts, which are characteristically noted during REM sleep and may be seen as excessive fragmentary myoclonus also during non-REM (NREM) sleep in many sleep disorders. Myoclonic bursts refer to EMG activity lasting for a brief duration of generally 20 to 250 milliseconds. In patients with tremor, EMG may record rhythmical activity in agonist-antagonist muscle pairs.

It is often helpful to also include intercostal and diaphragmatic EMG channels to record respiratory muscle activity. The intercostal EMG recorded from the seventh to ninth intercostal space with active electrodes on the anterior axillary line and the reference electrodes on the midaxillary line may also include some diaphragmatic muscle activity in addition to the intercostal activity. Diaphragmatic activity can be recorded by placing surface electrodes over the right or left side of the umbilicus or over the anterior costal margin, but these are contaminated by a mixture of intercostal activity and such noninvasive techniques are unreliable for quantitative assessment of diaphragmatic EMG. True diaphragmatic activity is typically recorded by intraesophageal recording. Intercostal and diaphragmatic EMG is particularly useful in the differentiation between central and obstructive apneas, especially when the respiratory channels are unreliable; continued bursts of activity in these channels during such an event would identify it as obstructive, whereas the absence of such bursts would implicate a central event (Fig. 1.3). The 2007 AASM Manual for the Scoring of Sleep and Associated Events recommends the use of the intercostal and diaphragmatic EMG channels for scoring apneas/hypopneas when the airflow channels are unreliable.