Authors :
Presenting Author: Lorenzo Fabbri, BS – The University of Texas at Arlington
Cecilia Liberati, BS – Universitá Campus Bio-medico di Roma; Emily Brock, BS – Cook Children's Health Care System; Calandra Jones, BS – Cook Children's Health Care System; Steven Stufflebeam, MD – Massachusetts General Hospital; Phillip Pearl, MD – Boston Children's Hospital; Scott Perry, MD – Cook Children's Health Care System; Eleonora Tamilia, PhD – Boston Children's Hospital; Christos Papadelis, PhD – Cook Children's Health Care System
Rationale:
High-frequency oscillations (HFOs) are interictal biomarkers of the epileptogenic zone (EZ) in drug resistant epilepsy (DRE). Yet, their clinical value is unsure since the HFO-generating area is often large and not specific to the EZ, so its complete resection is often unnecessary for seizure freedom. This is due to the presence of physiological HFOs in non-epileptogenic areas. Previous intracranial EEG (iEEG) studies divided physiological from pathological HFOs but were limited by the inability of iEEG to obtain whole-head coverage and record data from healthy controls. Here, we use noninvasive techniques, such as magnetoencephalography (MEG) and high-density EEG (HD-EEG), to detect HFOs from children with DRE and healthy controls and to identify features that differentiate pathological from physiological HFOs.Methods:
We analyzed MEG (306 sensors) and HD-EEG (256 channels) data from 28 healthy controls (10.9 ± 3.6 y, 12 males) and 34 children with DRE (13.0 ± 3.2 y, 18 males) who were separated between those having focal and generalized/diffuse DRE. In focal DRE, the epileptogenic hemisphere (EH) was defined based on presurgical evaluation. We identified HFOs (ripples: 80-160 Hz) on MEG and HD-EEG (automated detection with visual review) and localized their cortical generators via source imaging (Figure 1A). For each HFO, we extracted a set of temporal, spatial, morphological, and spectral features (Figure 1B). HFOs were grouped in four classes (Figure 2C) based on whether they were generated by: (1) the controls’ healthy brain (HFOs-controls); (2) the EH in focal DRE (HFOs-EH); (3) the non-EH in focal DRE (HFOs-nonEH); and (4) the generalized/diffuse DRE brain (HFOs-Gen). HFO features were compared between groups with Kruskal-Wallis test.
Results:
We found more HFOs on HD-EEG than MEG in both controls (0.95 vs 0.13 HFOs/min, p < 0.001) and DRE (0.62 vs 0.26 HFOs/min, p < 0.001). We generated the cortical distribution maps of physiological HFOs (HFOs-controls) for HD-EEG and MEG, which show high rates in somatosensory (0.3 HFOs/min) and temporal areas (0.03 HFOs/min) (Figure 2A). HFO-controls did not differ between dominant and non-dominant hemispheres; their amplitude, frequency and duration on HD-EEG correlated with age (R=0.46, R=0.45, R=0.40). For HD-EEG, various HFO features differed between four classes: HFO-controls showed lower and less variable amplitude and duration than HFOs-EH, and lower frequency, longer propagation latency and smaller spatial extent (Figure 2B). Looking at the HFO groups from DRE, HFOs-nonEH and HFOs-Gen often differed from HFOs-EH, with the latter being the most different from controls. For MEG, similar differences were found in HFO frequency, duration, latency, and spatial extent (Figure 2C).