Year: 2012
Type: dissertation
Abstract: Epilepsy is one the most common neurological disorders. For patients suffering from epilepsy who are medically intractable, in certain cases surgical resection of pathologic brain tissue is one remaining possibility. Prior to surgery, intracranial Electroencephalography (IEEG) study is conducted to localize seizure-generating zones. IEEG typically consists of studying epileptiform spikes and seizure discharges; however some researchers have observed short bursts of high frequency oscillations (HFOs), mostly in the range of 100-500 Hz in the seizure generating areas. A retrospective correlation analysis between the post-surgical outcome and HFO generating tissue supports the idea that HFO events may play a fundamental role in epilepsy and epileptogenesis. Recently, HFO events have been scored visually by clinicians. Typically, it is necessary to record IEEG for several days or weeks to collect sufficient epileptiform activities for precise evaluation. Needless to say, manual review of the data makes visual scoring an extremely tiresome process in which subjectivity is inevitable. Due to the recent explosion of HFO research, the development of algorithms for automatic detection of HFO events poses a great benefit to researchers and clinicians. In the literature, two methods have been widely used for automatic detection of HFO events based on the energy of the signal in the 100-500 Hz frequency band. In this thesis, we present three new methods for automatic detection of HFO events based on the sharpness property of the IEEG signals. By using simulated and real-data signals, the performance of the proposed methods are compared to the existing energy-based approaches using sensitivity and specificity metrics. Additionally, we present the clinical implication of the HFO event detections for four epileptic patients. The results indicate that the performance of the proposed detectors are robust and stable and do not deteriorate in the presence of the noise and artifacts.
Authors: Roshanak Yazdanpour-Naeini, Ry
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W615369775