Investigation of the relationship between 0.5–1200 Hz signal characteristics of cortical high-frequency oscillations and epileptogenicity through multivariate analysis

Abstract

Fast ripples (FRs) (250–500 Hz) on the electroencephalogram (EEG) are closely related to epileptogenicity and are important to determine cortical regions resected in epilepsy surgery. However, FR-related epileptogenicity may be variable, and may depend on information associated with FRs. We enrolled nine epilepsy patients who had undergone intracranial 5 kHz-sampling-rate EEG for surgical treatment and had final Engel class I outcomes. Three electrodes were selected from each epileptogenic area (EA) and the unlikely EA (the region outside the EA) in each patient. Up to 100 candidate FRs were automatically detected from interictal nocturnal EEG at each of the selected electrodes and were visually reviewed independently by two researchers. Multivariate logistic regression analysis was performed using the frequency and log-power value of the corresponding FRs, presence of concurrent spike, ripple, very-high-frequency oscillations (vHFO)1 (500–600 Hz), and vHFO2 (600–1200 Hz), and whether the timing of the spectral peak of corresponding FRs was in the peak–trough or trough–peak transition of each slow activity (0.5–1, 1–2, 2–3, 3–4, and 4–8 Hz) as independent variables. Factors significantly related to epileptogenicity were FR power, the concurrent presence of spike and vHFO2, coupling with 0.5–1 and 1–2 Hz slow waves in the peak–trough transition, and coupling with 3–4 and 4–8 Hz slow waves in the trough–peak transition. Multifactorial analysis of FRs may increase their usefulness, potentially leading to improved treatment outcomes in epilepsy surgery.