Sleep Staging Classification

1
Sleep Staging Classification

• Aristotle University of Thessaloniki
• School of Medicine
• Laboratory of Medical Informatics

2
Sleep Cycle Stages

• 1. Waking
• Referred to as relaxed wakefulness
• The body prepares for sleep
• Muscles begin to relax
• Eye movement slows to a roll

3
Sleep Cycle Stages

• 2. Stage 1 (Non-REM Phase)
• 50 reduction in activity observed between
  wakefulness and stage 1 sleep
• The eyes are closed
• If aroused from it, a person may feel as if he or
  she has not slept
• Stage 1 may last for 5 to10 minutes

4
Sleep Cycle Stages

• 3. Stage 2 (Non-REM Phase)
• Period of light sleep
• Polysomnographic readings show intermittent peaks
  and valleys
• Spontaneous periods of muscle tone mixed with
  periods of muscle relaxation
• The heart rate slows
• Body temperature decreases

5
Sleep Cycle Stages

• 4. Stages 3 and 4 (Non-REM Phase)
• Deep sleep stages
• Stage 4 being more intensive than Stage 3
• Electromyogram records slow waves of high
• amplitude

6
Sleep Cycle Stages

• 5. Stage 5 (REM Phase)
• Rapid Eye Movements
• Heart rate and respiration rate speed up and
  become erratic
• Intense dreaming occurs
• Heightened cerebral activity
• Muscular immobility
• REM typically lasts 10 minutes

7
Sleep classification process

• Data acquisition
• Pattern Identification
• Stage-classification

8
Polysomnography

• EEG (Electroencephalogram)
• EOG (Electrooculogram)
• EMG (Electromyogram)
• ECG (Electrocardiogram)
• BM (Body movements)
• Abdominal movements
• BT (Body Temperature)
• Pulse Oxymetry
• Airflow

9
Sleep stages classification

• Sleep Scoring Criteria According to
  Rechtschaffen-Kales Criteria

10
Automated sleep staging

• Classification problem
• Preprocessing and Enhancement
• Feature Extraction
• Classifier
• Stage

11
Feature extraction

• Feature sets extracted from Electrophysiological
  signals
• 1. Time Domain
• Period Analysis (Zero-Crossing of Peak Analysis)
• Bicoherence Index
• Correlation Dimension
• 2. Frequency Domain
• FFT-based spectral analysis
• ICA in Frequency Domain
• AR Spectral Distance

12
Sleep stages and ECG

• Autonomic Nervous activities change with sleep
  stages
• Frequency analysis of oscillation in R-R
  intervals has been applied to assess HRV
• HRV (Heart Rate Variability) has been used as a
  marker of autonomic nervous activity
• Low Frequency (0.04-0.15 Hz) reflects both
  sympathetic and parasympathetic activities
• High Frequency (gt0.15 Hz) reflects
  parasympathetic nervous activities

13
Sleep stages and ECG

• Autonomic activities (as assessed by HRV) show a
  circadian rhythm
• Sleep is characterized by rapid fluctuations in
  autonomic activities

14
Sleep stages and ECG

• In Non-REM Stages
• Parasympathetic nervous system is activated
• HRV has been reported to present an increase
  in HF components
• HRV has been reported to present a decrease in
  LF components

• In REM Stages
• Sympathetic Nervous system is activated
• HRV has been reported to present a decrease in HF
  components
• HRV has been reported to present an increase in
  LF components

15
Sleep stages and ECG

• Recent studies have investigated the association
  between EEG spectra and HRV indices
• Delta waves seems to be correlated with LF and
  LF/HF
• Sympathetic nervous activities seems to decrease
  with sleep deepening and increased with sleep
  lightning

16
Sleep stages and ECG

• The continuous evaluation of HRV and EEG seems to
  provide an attractive approach for new research
  on autonomic nerve activities in sleep disorders.