Fast activities in the healthy and epileptic brain: High frequency oscillations (HFOs) and high frequency broadband (HFB) response
To accurately delineate the epileptogenic region in patients with uncontrollable seizures, iEEG is often recorded in the epilepsy monitoring unit (EMU) over days where multiple stereotypical clinical seizures are recorded to provide information about the seizure onset zone (SOZ). The prolonged monitoring adds to the risk of complications, increases the cost and places a high demand on the clinical service. Extraction of predictive patterns in iEEG can open the door of efficient diagnosis of SOZ, and may eliminate the need to wait for spontaneous epileptic seizures to occur over days in the hospital setting.
Recent research has introduced surrogate neuro-biomarkers for epilepsy, such neuro-biomarkers in milliseconds time scale, termed high frequency oscillations (HFOs), are transients with frequency spanning from 80 to 500 Hz, far beyond the traditional EEG bandwidth. HFOs are believed to be clinically significant, and thus could be used for seizure localization. However, due to the massive size of modern iEEG datasets and the short duration, low amplitude of the transient, visually marking the events in long-term recordings is not feasible, which stressed the necessity to introduce a robust, reliable auto-detector that is applicable for quantitative data analysis in large-scale modern iEEG database.
On the other hand, in the healthy brain tissue the functional response can be evaluated by high-frequency broadband (HFB, 70 – 180 Hz) activation, known as an indicator for local engagement of the cortical tissue during behavioral tasks (Crone et al., 1998; Pesaran et al. 2002; Gail et al. 2004; Wilke et al. 2006; Manning et al., 2009; Flinker et al., 2011; Foster et al., 2012; Mesgarani et al., 2014).
The topographical patterns of HFB responses are more discrete and somatotopically specific. In particular, studies have shown that task-induced HFB response appeared to be more focally localized with respect to the expected anatomical structure and timing of activation.
One thing remains to be answered in the current clinical practice is a means by which clinicians can distinguish pathological from physiological high frequency activity, and estimate the extent of healthy functions in the pathological epileptic tissue at the individual patient level, to see if the cortical tissue can be removed without significant functional side effects.