Date of Award

Winter 2-2022

Document Type


Degree Name

Doctor of Philosophy (PhD)


Biomedical Engineering

First Advisor

Leonidas Iasemidis


The gold standard for localization of the epileptogenic zone (EZ) continues to be the visual inspection of electrographic changes during seizure onset by experienced EEG readers. Development of an epileptogenic focus localization tool that can accurately delineate the EZ from the analysis of interictal (seizure-free) period is still an open question of great significance for improved diagnosis (e.g., presurgical evaluation) and treatment (e.g., surgical) outcome. We developed an EZ localization algorithm (LA) based on novel analysis of the power in a narrow frequency band of the recorded signals from the brain, that is, using a simple univariate periodogram-type power measure, a straight-forward statistical ranking technique, and a robust dimensional reduction approach; out of the sampled brain regions, the EZ was most frequently maximally active in this band even in seizure-free (interictal) periods. Ten patients with frontal and temporal lobe epilepsy and matching the inclusion criteria (Engel Class I ≥ 12 months post-surgery) were recruited at the University of Alabama Medical Center. In a multi-folded leave-one-out statistical framework, the accuracy of the LA in identifying EZ from interictal intracranial stereo EEG (sEEG) in these patients at the epilepsy monitoring unit (EMU) was 100% (10/10) during the first hours of the sEEG recording and up to their 1st clinical seizure. To further evaluate the LA, data from two new (Engel Class I) patients were analyzed in a double-blinded study; the LA was successful in localizing the focal channels in both of these patients. These findings support the hypothesis that the EZ is

interictally regulated by inhibitory neurons with resonance behavior in a narrow gamma band (~65-85 Hz). We further show that this characteristic band can be determined by a wide range of spectral power estimation univariate or multivariate techniques as well as directed functional connectivity network measures and can therefore easily be missed or rejected due to the use of 60Hz power noise reduction filters as pre-processing tools of the EEG. In conclusion, our LA achieves reliable epileptogenic focus localization across different focal epilepsy types in humans when the analysis is done within a specific “focal frequency band” (FFB) during the early hours of the interictal period at the EMU.

To assist with the elucidation of the mechanisms of seizure generation (ictogenesis) and seizure prediction, we analyzed the dynamics of intracellular calcium concentration ([Cai++]) signals following in vitro stimulation of primary neuronal cultures by nanomolar (subthreshold) concentrations of glutamate (Glu). Employing novel measures of connectivity for the network formed by the firing neurons, we showed that: a) connectivity between neurons increased by addition of Glu, b) the neurons became hyperexcitable following the depletion of astrocytes from the cultures, and c) neuronal networks with astrocytes were less complex and exhibited more predictable outcomes following a wide range of Glu stimuli. These preliminary results, if confirmed in a more extensive set of cultures, suggest that the developed methodology and network measures may be useful for seizure prediction and as biomarkers in the in vivo monitoring and quantitative analysis of the epileptic brain’s activity en route to seizures based on the hypothesis that increase of the levels of glutamate in hyperexcitable neurons in the epileptogenic focus is not balanced by the function of neighboring pathologically impaired astrocytes.