Nonlinear Analysis and Biomarkers in Neurological Diseases (Temporal Lobe Epilepsy and Alzheimer's Disease).

Abstract

In our study, we used scalp electroencephalography (EEG) to examine brain activities for both temporal lobe epilepsy (TLE) and Alzheimer’s disease (AD). We applied a nonlinear dynamic metric, phase coherence, to measure the synchronization of cellular activities. The linear mixed models were implemented to correlate the changes in repeated phase coherence measures controlling for brain wave frequency, state of consciousness, epileptic state (interictal and preictal) and anatomical position of EEG electrodes.

For patients with temporal lobe epilepsy, our results suggested a lack of information transfer of the brain in epilepsy patients during interictal periods, but not during preictal periods. In particular, we found increased phase coherence in the anterior region compared to the posterior region in the preictal period, in agreement with experimental observation that seizures mainly propagate through the anterior region of the brain. For Alzheimer’s disease, our results indicated significant increases in both phase coherence and cross correlation in posterior local pairs of AD patients compared to normal subjects during intermittent photic stimulation (IPS). Anterior pairs do not show such a marked difference. Using mean phase coherence in the band for the T5-O1 pair as a biomarker for AD resulted in a 75% true positive rate together with a false positive rate that was only 0 to 16.7%.

Our detailed study of phase coherence and other linear and nonlinear metrics in scalp EEG data had three important implications. First, it improves our understanding of seizure generation and propagation in temporal lobe epilepsy. Second, our studies may lay the groundwork for the development of a non-invasive, reliable method of seizure prediction. If one is able to predict seizures reliably, one may be able to develop acute therapies that can be used to abort the seizures. This will be a major step forward in the treatment of epilepsy, particularly for those patients whose conditions are medically intractable. Third, our observation of statistically significant increases in phase coherence served well as a potential biomarker for AD. It is likely that the sensitivity can be further improved with additional studies using larger sample size.