Abstract Obstructive Sleep Apnea Syndrome (OSAS) is a very common sleep disorder with potential severe implications in essential aspects and the patient's health; This study addresses the evaluation of the EEG energy-based brain mapping in before , during and after the OSAS episode using continuous wavelet spectral analyses. This is the second study applying topographic EEG brain mapping to assess spectral power changes during OSA. The first one remotes to 1993 and had a very poor resolution ( Walsleben, et al. (1993)); The results shows that there is a significant reduction in the energy of delta frequency band but an increase in beta band, in overall brain. theta and alpha have an increased energy in the parietal and

occipital cortex, but decreased in the other areas, mainly after the OSAS episode.

RecPad2010 - 16th edition of the Portuguese Conference on Pattern Recognition, UTAD University, Vila Real city, October 29th

Problem Formulation

(i) Delta waves SD during apnea episode

(j) Theta waves SD during apnea episode (k) Alpha waves SD during apnea episode

l) Beta waves SD during apnea episode

(m) Delta waves SD after apnea episode

(n) Theta waves SD after apnea episode (o) Alpha waves SD after apnea episode

p) Beta waves SD after apnea episode

Experimental Results

Standard Deviation (SD) of Relative Energy Mapping

(a) Delta waves during apnea episode

(b) Theta waves during apnea episode (c) Alpha waves during apnea episode

d) Beta waves during apnea episode

(e) Delta waves after apnea episode

(f) Theta waves after apnea episode (g) Alpha waves after apnea episode

h) Beta waves after apnea episode

Mean of Relative Energy MappingEEG extraction Signals were recorded on a computerized polysomnography system during the patient sleep time (8h);

EEG electrodes were positioned according to the International 10-20 System with 21 electrodes and 2 references;

The EEG signals were acquired at a sampling frequency of 100 Hz.

The nasal airflow was recorded as well.

EEG

Air Flow

Episode selection The cessation of the air flow during more than 10 s in NREM-2 sleep stage and more than 60 s isolation were the criteria of selecting an apnea episode;

Each episode was divided into three segments according to the episode's phase: before, during, and after an apnea event;

The resulting dataset includes 10773 epochs (171 episodes x 3 phases x 21 channels).

Energy Mapping

After processing the EEG signal (χ), the mean energy for each channel (c) and frequency band (fb) was calculated by the following equation:

where Ep represents the energy for each apnea segment (p={pre, dur, post}), i represents the episode (i=1,...,M) and n the sample (n=1,...,N);

Finally, the next equation gives the relative energy of each channel:

The brain mapping was based on the approximation of the head to a semi-sphere and interpolating the channel energy values using the Matlab toolbox EEGLAB with the function topoplot().

A second map was made by using a 2-sample t-test where a p-value<0.05 were considered significant. Only the significant points are shown.

Wavelet Processing

 

 

Conclusions Considering that each patient has significant amount of episodes, and many of them aren't isolated, it is possible to

conclude that these patients don't have a full replenishment of the brain energy during the night; Not all areas are affected in the same way during an apnea episode; It is plausible to say that this patients don't rest while sleeping how they should. It is also curious to establish the relationship that the zones with a higher standard deviation for all frequency bands

are in areas which represent the pre-motor, motor and visual areas of the brain.

References( Walsleben, et al. (1993)) - (J. Walsleben, E. O’Malley, K. Bonnet, R. Norman, and D. Rapoport, “The utility of topographic eeg mapping in obstructive sleep apnea syndrome,” Sleep, vol. 16, pp. 76–78, 1993)

Statistical tested (T-test) mappings with p<0.05

The lengend is the same as the Mean of Relative Energy Mapping

D. Belo1, A. Coito1, T. Paiva2, J.M. Sanches1

1Institute for Systems and Robotics / Instituto Superior Técnico, Lisboa, Portugal2Centro de Electroencefalografia e Neurofisiologia Clínica / Faculdade de Medicina da Universidade de Lisboa

Lisbon, Portugal

Brain Mapping of a Wavelet Analyses of Subjects with Obstructive Sleep Apnea