LAUSR

This paper discusses the various methods of identifying the seizure onset zone (SOZ) from the intracranial electroencephalography (iEEG) data. Epilepsy, also known as seizure disorder, is a neurological condition caused due to disruption in the regular electrical communication within the neuron network. With almost a third of epileptic conditions being drug-resistant and several cases with no known cause, there is a need to resort to alternative treatment methods such as neurostimulation or surgical resection. Both these methods require the identification of regions within the brain that need to be stimulated or resected. For most of the patients, this corresponds to the part that initiates the seizure. These are called seizure onset zone (SOZ) or epileptogenic zone (EZ). Epileptologists locate the SOZ by analyzing the iEEG data of patients suffering from seizures. This, however, is time-consuming and can be prone to human error. Thus, there has been significant research on the automatic detection of SOZ. High-frequency oscillations (HFOs), characterized by iEEG oscillations above 80 Hz, are one of the most promising candidates for identifying SOZ. Functional connectivity and graph theory measures have also distinguished SOZ and non-SOZ regions using different features. Newer works on phase-amplitude coupling have also shown promising results. With the increased data availability, it has also become possible to build supervised learning algorithms to improve the predictive power of anomaly detection algorithms used to localize SOZ.