Date of Completion
Epilepsy, Biomedical Engineering, High Frequency Oscillation
Field of Study
Doctor of Philosophy
There is a need for novel biomarkers to aid in the clinical treatment of epilepsy and improve the understanding of seizure generation and the seizure onset zone (SOZ). High frequency oscillations (HFO) are a promising biomarker with the potential to fill this role, but early efforts to apply them have fallen short of clinical quality tools and validity and application of their association with seizure have not been fully explored. This study will advance the understanding and application of HFO to the clinical setting in three ways. First, the relevance of HFO to SOZ and their utility for SOZ localization will be determined practically by implementation of an automated method for SOZ localization using intracranial EEG. A novel feature of HFO will be used to train a machine learning system that can accurately identify the SOZ in a patient and patient state independent way with as little 30 minutes of recording. Second, the temporal evolution of HFO occurrence will be characterized using point process modeling and differences in motif manifestation of HFO will be investigated in areas of seizure generation and across epilepsy etiologies to open the door to epilepsy subtype studies using HFO. Finally, to expand the applicability of HFO in both clinical and research settings, a system for the automated detection of HFO from scalp EEG, rather than intracranial, will be developed, using methods of feature and waveform clustering to overcome challenges in noninvasive identification of these events. Together, these approaches will result in the creation of a tool for clinical application of HFO in SOZ localization, provide insights into HFO occurrence and their link to different etiologies of epilepsy, and lay a foundation for new applications of HFO in the previously unutilized area of noninvasive HFO.
Sumsky, Stefan, "Application of Machine Learning to Enhance the Diagnostic Utility of Interictal High Frequency Oscillations in Drug-resistant Epilepsy" (2019). Doctoral Dissertations. 2390.