World Sleep Industry ENCYCLOPEDIA

Polysomnography Systems

2.2.3

Definition, System Boundary, and Intended Use

A polysomnography system is a medical-device system that synchronously acquires, displays, stores, and supports analysis of multiple physiologic signals during sleep. The defining feature of a complete polysomnogram is not merely a large channel count. Electroencephalography (EEG), electrooculography (EOG), and chin electromyography (EMG) establish a sleep-wake timeline, against which respiratory, oxygenation, cardiac, and movement events can be interpreted in their actual stage and time context. The output is a record for scoring by trained personnel and interpretation by a physician, not a diagnosis that stands independently of clinical information.

A typical system includes patient electrodes and sensors, a headbox, isolated amplifiers, analog-to-digital acquisition, a technologist workstation, synchronized audio and video, two-way communication, data storage, and analysis and reporting software. Testing may take place in an attended sleep laboratory or, when supported by the intended use, with a portable system elsewhere. Location, attendance, and channel count cannot by themselves classify the examination; the recorded signals, ability to stage sleep, operating model, and device labeling all matter.

This entry does not equate a home sleep apnea test that records respiratory variables but cannot reliably stage sleep, a stand-alone pulse oximeter, consumer wristband, single-purpose EEG recorder, viewing workstation, or independent automated scoring application with a complete PSG system. The device, sleep laboratory, testing service, and final medical report are also separate layers. Authorization of one hardware component does not demonstrate that every facility, workflow, or report using it is acceptable.

Physiologic Channels, Signal Chain, and Common Time Base

Routine respiratory PSG commonly combines EEG, EOG, chin EMG, ECG or heart rate, oronasal airflow, thoracic and abdominal respiratory effort, and pulse oximetry. Bilateral anterior tibialis EMG, body position, snoring, and synchronized video often add movement and behavioral context. Suspected sleep-related hypoventilation may require transcutaneous or end-tidal carbon dioxide; parasomnia or nocturnal spell evaluation may require fuller video and an expanded EEG montage; positive airway pressure titration also records pressure, flow, leak, and treatment mode. Each channel answers a different question, and oxygen saturation, snoring, or position cannot substitute unconditionally for airflow, effort, or sleep staging.

The path from sensor to report includes the patient contact, wired or wireless link, input protection, amplification and signal conditioning, anti-aliasing, analog-to-digital conversion, buffering, clock, file format, display filtering, and analysis software. Sampling, digital resolution, gain, and filtering should suit the amplitude and frequency content of each signal while preserving traceable raw waveforms. Filtering merely to make a trace appear cleaner can remove slow activity, muscle activity, sharp transients, or genuine respiratory morphology. A normal-looking display does not show that the source data avoided saturation, clipping, or packet loss.

All physiologic channels, event markers, lights-off and lights-on times, technologist notes, PAP changes, audio, and video need a verifiable common time base. Wireless reconnection, workstation sleep, network delay, and clock correction must not create an undisclosed offset. The system should also let users distinguish a disconnected sensor, low perfusion, displacement, held value, or algorithmic interpolation from a true measurement. Estimated or duplicated intervals must not silently appear as acquired waveforms.

Setup, Calibration, Artifact Recognition, and Overnight Quality Control

Before a study, the team selects a montage and sensor sizes for the clinical question, verifies identity and the order, records hardware and software versions, and checks physical condition, power, available storage, time synchronization, and accessories. Electrode placement, contact impedance, and leads should meet the governing technical specification. Biocalibrations using eye opening and closure, eye movement, jaw contraction, leg movement, breathing, breath holding, or position change can confirm channel direction and response, while equipment calibration checks amplification against reference signals. Post-study calibration can expose drift or failure that became evident only after the night.

Common artifacts include electrode popping, slow sweat drift, mains interference, ECG contamination of EEG, muscle contamination, movement saturation, twisted effort belts, compression or condensation in a nasal-pressure tube, displacement of a thermal airflow sensor, a compressed oximeter probe or low perfusion, and an obstructed camera view. An artifact can contaminate several derived measures at once and should not be reduced to a generic note that signal quality was poor. During an attended study, the technologist continuously observes waveforms and the patient, restores critical channels without unnecessary awakening when feasible, and documents the cause, intervention time, and irrecoverable interval.

Comfort and signal integrity require joint management. Skin preparation, conductive materials, adhesives, and fixation should suit the contact duration and population, and leads need strain relief and routing that reduces neck entanglement, pulling, and falls. Children, people with fragile skin or marked movement, and those using oxygen or PAP require appropriate sizes and layouts. More abrasion, adhesive, or binding is not a universal answer to high impedance, and an unverified sensor must not be mixed into an interface merely because its connector fits; contact injury, electrical safety, and measurement error can change together.

Study Design, Professional Operation, and Appropriate Pathway

The clinical question determines the study, rather than the device's available inputs determining the order. A full-night diagnostic PSG, a split-night diagnostic and PAP titration study, a full-night PAP titration, an expanded EEG-video examination, a hypoventilation study, and a PSG serving as the night before a multiple sleep latency test differ in preparation, montage, interventions, and reporting. The record should also capture sleep schedule, medications and substances, existing treatment, symptoms, comorbidities, and ways in which the test night differed from usual sleep.

In attended PSG, a trained technologist continuously monitors technical validity and relevant behavior, records posture, events, interventions, and patient reports, and manages detached sensors, PAP changes, or safety concerns within the authorized protocol. Remote viewing can extend visibility but does not automatically replace on-site capability required by a particular protocol. Many PSG amplifiers and recorders display SpO2, ECG, and respiration yet are not vital-sign alarm monitors. Unless a specific model has that intended use and supporting validation, it must not be the sole system responsible for critical alarms.

For adults with suspected obstructive sleep apnea, laboratory PSG is the standard diagnostic method, while a technically adequate HSAT may be appropriate for selected adults at increased risk who lack complicating conditions. A negative, inconclusive, or technically inadequate HSAT should not be treated as exclusion when clinical concern remains. Significant cardiorespiratory disease, possible respiratory-muscle weakness, awake or suspected sleep-related hypoventilation, chronic opioid use, prior stroke, severe insomnia, or a need to observe stage or behavior will often favor PSG under the applicable guideline.

Scoring, Clinical Interpretation, and Limits of Automated Analysis

Raw data should be scored for sleep stage, arousal, respiration, cardiac rhythm, and limb movement according to the current rules declared for that examination, with the history of human edits retained. A report should at least make total recording time, total sleep time, sleep efficiency, stage proportions and latencies, arousals, respiratory events, oxygenation, position, and movement results identifiable. The apnea-hypopnea index uses sleep time as its denominator and is not equivalent to an index based on recording or monitoring time when sleep cannot be reliably staged. Summary values must remain traceable to waveforms, valid intervals, and event definitions.

The airflow sensor used for hypopnea, the desaturation threshold, and whether arousal association is accepted can change the event count. Age, PAP status, carbon-dioxide method, and movement rules can also call for different criteria. The report should name the manual or protocol version, primary sensors, scoring rule, technical limitations, valid time, substituted signals, and major deviations, so values produced under different rules are not compared uncritically across time or facilities. A low AHI also does not automatically explain other nocturnal behavior, hypoxemia, hypoventilation, or daytime sleepiness.

Automated staging and event recognition can provide preliminary labels, prioritize review, or improve workflow, but they require validation for the intended population, montage, signal quality, and target rule set. Human scoring itself has inter-scorer variability; that fact does not justify unsupervised replacement of professional judgment. Performance in typical adults cannot establish performance in children, neurologic disease, arrhythmia, PAP use, severe artifact, or new hardware. A change in algorithm, threshold, or software version requires reassessment of performance, report consistency, and audit records and must not silently rewrite prior results.

Basic Safety, Data Governance, and Maintenance

The patient-connected amplifier, sensors, computers, video equipment, and attached PAP or carbon-dioxide modules together form a medical electrical system. Verification should cover patient isolation, leakage current, temperature rise, power failure, mechanical strength, fluid ingress, batteries and charging, electromagnetic compatibility, and essential performance. In the intended environment, interference should not cause unrecognizable waveform distortion, restart, or data loss. An unassessed power supply, USB peripheral, network converter, or third-party accessory can alter the isolation boundary and the compliance state of the assembled system.

Operational safety also encompasses skin irritation or injury, cleaning and disinfection of reusable accessories, disposal of single-use items, cable entanglement, falls, leaving bed at night, and combined use with PAP, oxygen, or other equipment. Video, audio, and bedroom behavior are highly sensitive data. Before testing, the service should explain the acquisition field, who can view it, retention, and access purposes and provide appropriate changing and privacy arrangements. Chest pain, major breathing difficulty, severe hypoxemia, or acute neurologic or behavioral abnormality calls for the facility's emergency procedure, not merely completion of the recording.

Raw waveforms, configuration, calibration, annotations, scoring version, report, audio-video, and modification history should be linked and retained according to responsibility, protected by authentication, least privilege, encryption, backup, integrity checks, and audit logs. A connected system needs an asset inventory, assessment of patches and security updates, vulnerability response, end-of-support plan, and verifiable export. Network failure behavior should define continued local acquisition, buffer limits, and recovery merging. Retaining only a PDF after deleting raw waveforms, or preventing export in a readable form, weakens review, migration, and quality tracing.

Acceptance Metrics, Failure Response, and Development Direction

System acceptance should exercise a representative end-to-end study, not merely power each port. Measures include valid-channel fraction and noise, gain and frequency response, recovery from saturation, respiratory and oximetry response, cross-channel and audio-video synchronization, event markers, continuous overnight writing, recovery from network and power interruption, file integrity, report traceability, and re-reading of exported data. After deployment, scheduled maintenance, functional checks, accessory life management, fault trending, and a scoring-consistency program should verify that performance has not drifted over time.

The consequence of a failed channel depends on the task. Loss of principal EEG weakens staging, absent airflow or effort impairs respiratory-event classification, prolonged invalid oximetry limits oxygenation assessment, and audio-video misalignment compromises review of behavioral events. The system should promptly identify actionable faults, and the technologist should document repair or use of an alternate signal. If the remaining data cannot answer the ordered question, a professional decides whether to reconnect, extend, repeat, or use another method rather than interpolating every gap and issuing an unchanged definitive conclusion.

Development directions include lighter wireless sensors, dry electrodes, wearable Type II PSG, automatic quality detection, longitudinal multinight recording, and explainable assisted scoring. Convenience becomes valuable only after channel meaning, time synchronization, valid-signal fraction, data security, and clinical agreement are demonstrated. Removing wires or moving a study home changes the risks of dropout, incorrect placement, and absence of intervention. Related entries include sleep EEG monitoring systems, sleep respiratory monitors, home sleep screening devices, sleep oximeters, sleep carbon-dioxide monitors, and sleep-test data analysis software.