Combining epileptogenic zone localization algorithms and machine learning methods to improve surgery outcomes for drug-resistant epilepsy.

A third of people with epilepsy have drug-resistant epilepsy. Surgical resection of the epileptogenic zone is currently the most effective method of seizure freedom in this population but requires extensive data analysis. Automated epileptogenic zone detection algorithms have shown promise in assisting experienced clinicians with localization of the EZ, however lack of standardization and difficult to reproduce complex algorithmic techniques hinder broad acceptance. This project aims to address these challenges by developing free comprehensive and tested toolboxes to improve epileptogenic zone localization through validated algorithms.

Audiovisual multisensory integration

Comprehension of noisy speech improves when a listener is allowed to view a speaker’s face. However, the neural mechanism for audiovisual integration is poorly understood. We use direct recordings from the human brain to develop models to better understand how visual and auditory speech signals are represented and combined. Specifically, populations of neurons are known to spontaneously fluctuate between excitable and less excitable states. These fluctuations are known as low frequency oscillations and are recorded as part of the local field potential. The phase of these oscillations is of particular interest because neuronal excitation varies based on the phase of the oscillation. Phase resetting is when an external stimulus abruptly interrupts the phase of an oscillation, changing the timing of maximum and minimum excitability to a new time cycle. Considerable debate has taken place regarding the function of low frequency oscillations and phase resetting. Some evidence supports the assertion that the phase of low frequency oscillations is an important form of data encoding, while other evidence suggests phase resetting is merely an epiphenomenon from evoked responses or time-frequency decomposition methodologies. Our overarching hypothesis is that phase resetting is the mechanism for fast, long-distance communication from visual to auditory cortex underpinning audiovisual multisensory integration.

Software Development

We are conducting our research and development with the RAVE platform (R Analysis and Visualization of intracranial EEG), an open-source, free, NIH-supported software for consistently replicable statistical analysis and visualization of iEEG data. RAVE (https://openwetware.org/wiki/RAVE) can be used through a web browser, making it accessible even to individuals with no prior programming experience.