EEG phase synchronization during absence seizures

Abstract

Absence seizures-generalized rhythmic spike-and-wave discharges (SWDs) are the defining property of childhood (CAE) and juvenile (JAE) absence epilepsies. Such seizures are the most compelling examples of pathological neuronal hypersynchrony. All the absence detection algorithms proposed so far have been derived from the properties of individual SWDs. In this work, we investigate EEG phase synchronization in patients with CAE/JAE and healthy subjects to explore the possibility of using the wavelet phase synchronization index to detect seizures and quantify their disorganization (fragmentation). The overlap of the ictal and interictal probability density functions was high enough to preclude effective seizure detection based solely on changes in EEG synchronization. We used a machine learning classifier with the phase synchronization index (calculated for 1 s data segments with 0.5 s overlap) and the normalized amplitude as features to detect generalized SWDs. Using 19 channels (10-20 setup), we identified 99.2% of absences. However, the overlap of the segments classified as ictal with seizures was only 83%. The analysis showed that seizures were disorganized in approximately half of the 65 subjects. On average, generalized SWDs lasted about 80% of the duration of abnormal EEG activity. The disruption of the ictal rhythm can manifest itself as the disappearance of epileptic spikes (with high-amplitude delta waves persisting), transient cessation of epileptic discharges, or loss of global synchronization. The detector can analyze a real-time data stream. Its performance is good for a six-channel setup (Fp1, Fp2, F7, F8, O1, O2), which can be implemented as an unobtrusive EEG headband. False detections are rare for controls and young adults (0.03% and 0.02%, respectively). In patients, they are more frequent (0.5%), but in approximately 82% cases, classification errors are caused by short epileptiform discharges. Most importantly, the proposed detector can be applied to parts of EEG with abnormal EEG activity to quantitatively determine seizure fragmentation. This property is important because a previous study reported that the probability of disorganized discharges is eight times higher in JAE than in CAE. Future research must establish whether seizure properties (frequency, length, fragmentation, etc.) and clinical characteristics can help distinguish CAE and JAE.

Keywords: absence seizure; childhood absence epilepsy; epilepsy; juvenile absence epilepsy; seizure detection; seizure fragmentation; synchronization; wavelets.

Figure 1

(A) Archetype of high-amplitude, continuous generalized spike and wave discharges with prominent epileptic spikes. The absence seizure in (B) is briefly interrupted two times. Panels (C, D) show the EEG from Cz channel (top panel) as well as the corresponding time series of the wavelet power (middle panel) and the global synchronization index (bottom panel). The wavelet power and synchronization index were calculated using f_c = 1 Hz and f_a = 12 Hz. These case studies demonstrate that global synchronization decreases when epileptic activity subsides.

Figure 2

Synchronization matrices for the regular (A) and disorganized (B) seizure. These seizures are shown in Figure 1.

Figure 3

EEG synchronization during absence seizures. (A) Topographic map of channel synchronization (cohort average) for all interictal segments (0), the first window that partially overlaps seizures (1), the second partially overlapping (2), the first fully embedded in the seizure (3), all embedded without the first and last (4), the last embedded (5), the last but one overlapping (6) and last overlapping (7). (B) Global synchronization boxplots for data segments 0–7 (segments from all patients were used). (C, D) Probability density function (PDF) of the average synchronization index for the 19 channels (S₁₉) and the four-channel subset S₄ (Fp1, Fp2, T5, T6), respectively. One can see that global synchronization is high during seizures and that there is a strong overlap of the interictal and ictal distributions of the synchronization index.

Figure 4

Example of building a k-NN seizure detector with the leave-one-out cross-validation (LOOCV) for patient P1. We used the global synchronization index and mean normalized EEG amplitude as the features. The learning set comprised randomly chosen interictal and segments fully embedded in absences with average synchronization greater than the cut-off value. We used 3:1 ratio of interictal to ictal windows. Panels (A, C) show the spread of the data generated for all 19 channels of 10-20 EEG setup (S₁₉) and the subset S₄ (channels Fp1, Fp2, T5, and T6), respectively. The confusion matrices in (B, D) show the results of 10-fold cross-validation. The classifiers were applied to the segmented EEG of patient P1 (1 s windows with 0.5 s overlap). Panels (E, F) show P1's confusion matrices for S₁₉ and S₄, respectively.

Figure 5

(A) Seizure from Figure 1B is shown with the leading and trailing interictal segments. (B) The output of the seizure detector. The detection function equals 1 for the segments classified as ictal and 0 otherwise. The red vertical lines in both subplots delineate the abnormal EEG activity marked by a neurologist. Note that SWDs do not simultaneously emerge from the background EEG in all channels. At the end of the seizure, ictal activity gradually subsides: epileptic spikes disappear, the amplitude of the EEG decreases, and global synchronization is lost. The initial and final transients are shorter than 0.5 s. The detector correctly identified the two interruptions in the ictal rhythm, marked in (A) in red. Two EEG intervals were marked blue to draw attention to the limitations of fragmentation analysis. Seizure disorganization shorter than about 0.5 s are, in most cases, undetected. Due to the finite size of the data window used to calculate phase synchronization, the fragmentation can be underestimated.

Figure 6

Histograms of (A) average fragmentation of the patient's seizures and (B) fragmentation of individual seizures. Seizure fragmentation is defined as the duration of segments classified as non-ictal embedded in the abnormal EEG activity interval divided by the length of such an interval.