I hope to use machine learning techniques to extract greater information from electroencephalographic signals gathered in visual psychophysics experiments, and so better inform models of cognition.
In the past, I have published work on neuropsychiatric genetics - Stewart 2009 SchzTGM2. I completed a neuroscience BSc at Edinburgh before being awarded an MSc for research on the use of calcium imaging to monitor the activity of custom wet neural networks. I am now a member of the Neuroinformatics Doctoral Training Centre, where I have worked on mathematical and computational approaches to neuroscience research and have recently completed an MSc on the use of graphics cards to greatly enhance the capability of real-time EEG analysis.
Current work applies this new capability to help tease apart the relationship between visual processing, attention and metrics of consciousness as can be assessed by EEG.
Neuronal gamma-band synchronisation is an intriguing putative neural mechanism (Fries_07_Gamma_Tins). EEG experimentation has the advantage of allowing this to be investigated investigated in awake humans. A recent study (Yuval_09_iGBR_Neuron) has suggested that one kind of the gamma related EEG signal may be artefacts of eye muscle movement. I plan to use powerful real time EEG analysis alongside eye tracking in order to untangle this, and believe this is an informative and accessible novel methodology for studying cognition in humans.
We report results from a simple visual object recognition experiment where independent component analysis (ICA) data processing and machine learning classification were able to correctly distinguish presence of visual stimuli at around 87% (0.70 AUC, p < 0.0001) accuracy within single trials, using data from single ICs.
Seven subjects observed a series of everyday visual object stimuli while EEG was recorded. The task was to indicate whether or not they recognised each object as familiar to them. EEG or IC data from a subset of initial object presentations was used to train support vector machine (SVM) classifiers, which then generated a label for subsequent data. Task-label classifier accuracy gives a proxy measure of task-related information present in the data used to train.
This allows comparison of EEG data processing techniques – here, we found selected single ICs that give higher performance than when classifying from any single scalp EEG channel (0.70 AUC vs 0.65 AUC,p < 0.0001). Most of these single selected ICs were found in occipital regions. Scoring a sliding analysis window moving through the time-points of the trial revealed that peak accuracy is when using data from +75 to +125 ms relative to the object appearing on screen. We discuss the use of such classification and potential cognitive implications of differential accuracy on IC activations.
Organisations: Institute for Adaptive and Neural Computation .
Authors: Stewart, Andrew, Nuthmann, Antje & Sanguinetti, Guido.
Keywords: (EEG, SVM, ICA, classification, single-trial. )
Publication Date: 15 May 2014
Journal: Journal of Neuroscience Methods
Original Language: English