Unravelling the multiscale temporal modulations in seizure features using long-term iEEG recordings

Abstract

Epilepsy poses a significant treatment challenge due to its dynamic nature, characterised by fluctuations in pathological brain activity and symptoms. Seizures can manifest in different ways, resulting in diverse spatial and temporal patterns of features within individual patients. While researchers have extensively studied fluctuations over different timescales in seizure occurrence, the presence of similar fluctuations in other seizure features and the expression of the underlying fluctuations in pathological brain regions remain unclear. To address these gaps, I investigated cyclical patterns of seizure network evolutions and seizure duration (used as proxy for seizure severity) in refractory focal epilepsy patients. Using long-term intracranial EEG (iEEG) recordings, I analysed how seizure features change over time within subjects. Initially, I explored whether temporal fluctuations in iEEG band power over seconds to days could explain variability in seizure evolutions. Within each subject, a combination of ultradian, circadian and some slower fluctuations accounted for most of the diversity in seizure evolution. Then, I explored how seizure severity changes over time using iEEG band power cycles. Combinations of multiple band power cycles explained most of the variability in seizure duration. These findings suggest that cycles over multiple timescales in interictal iEEG properties, such as band power, may modulate seizure features and serve as markers for seizure-modulating processes. I then examined the functional expression of band power cycles in both pathological and healthy brain regions. Interestingly, ultradian and circadian cycles were diminished in brain regions identified as pathological, indicating that brain pathology may alter biological rhythms on similar timescales. In summary, my research contributes to the understanding of temporal changes in seizure features. Disentangling the dynamic nature of the disease and specifically the seizure modulating factors could improve personalised epilepsy treatments for seizure control. Furthermore, by predicting not only seizure occurrences but also their dynamics, evolutions, severity, and symptoms, my work expands the scope of alternative treatments. Fluctuations in iEEG features may serve as biomarkers for monitoring treatment response and enabling on-demand treatment options. Finally, investigati ing the relationship between altered cycles and pathology could unveil the biological rhythm’s role in ictogenesis and epileptogenesis.