Pattern analysis of sleep-deprived human EEG

Progress during the past decade in non-linear dynamics and instability theo ry has provided useful tools for understanding spatio-temporal pattern form ation. Procedures which apply principle component analysis (using the Karhu nen-Loeve decomposition technique) to the multichannel electroencephalograp h (EEG) time series have been developed. This technique shows localized cha nges of cortical functioning; it identifies increases and decreases of the activity of localized cortical regions over time while the subject performs a simple task or test. It can be used to demonstrate the change in cortica l dynamics in response to a continuous challenge. Using 16 EEG electrodes, the technique provides spatio-temporal information not obtained with power spectrum analysis, and includes the weighted information given with omega c omplexity. As an application, we performed a pattern analysis of sleep-depr ived human EEG data in 20 healthy young men. Electroencephalograph recordin gs were performed on subjects for < 2 min, with eyes closed after normal sl eep and after 24 h of experimentally-induced sleep deprivation. The signifi cant changes in the eigenvector components indicated the relative changes o f local activity in the brain with progressive sleep deprivation. A sleep d eprivation effect was observed, which was hemispherically correlated but wi th opposite directional dynamics. These changes were seen in the temporo-pa rietal regions bilaterally. The application of the technique showed that th e simple test task was performed with a limited unilateral hemispheric invo lvement at baseline, but needed a much larger cortical participation with d ecreased frontal activity and increased coherence and bilateral hemispheric involvement. The calculations performed demonstrated that the same weighte d changes as those obtained with omega complexity were shown, but the techn ique had the added advantage of showing the localized directional changes o f the principle eigenvector at each studied electrode, pointing out the cor tical localized region affected by the sleep deprivation and toward which d irection the environmental challenge induced the spatial change. This metho dology may allow the evaluation of changes in local dynamics in brain activ ity in normal and pathological conditions.