Detecting epileptic seizures from EEG recordings

Neurologists and epileptologists spend a significant amount of their time sifting through EEG recordings that span anywhere between 1 to 9 hours of raw data, by manually inspecting the data in 10 second increments, looking for markers of a malfunctioning brain. These markers, however, are often small, transient changes in a complex and chaotic signal.

I’m interested in developing a method to automatically flag neural events (such as seizures, eye-blinks or sleep) from EEG data. My approach is focussed on overcoming the complex, non-stationary nature of the EEG signal, which poses a difficult challenge for everything from LDA to deep learning.

Specifically, I’m interested in feature mining and feature engineering, to enable a drastic reduction in the size of the feature space of the model. I’ve implemented a technique that I’ve termed ‘spectral contrasting’, that significantly reduces the feature space of the input dataset by attempting to overcome the non-stationarity in the data. This enables us to classify seizure events at a much better accuracy than by using the original timeseries data alone!

A jupyter notebook documenting the method and the analysis is here. Spectral contrasting is already enabling subject-independent classification of seizures from single trials, something which is considered incredibly difficult on EEG data.

I’ve developed a python toolkit for implementing spectral contrasting. The code can found here.

The dataset used is a subset of the Temple University EEG Corpus dataset (a freely available dataset), which contains labelled time-series data for normal and seizure activity.