Dear Users

Today, David Halliday from the Department of Electronics will give a talk. David's talk will be followed by a YNiC project presentation by Philip Quinlan and Nikos Andreadis titled "Switching tasks and anticipating switching tasks: what is the brain doing?".

These events will take place at YNiC from 4-5 pm.

The title of David's talk is "The Noisy Brain – An introduction to time and frequency domain signal processing techniques for neurophysiological data".

http://www.elec.york.ac.uk/staff/dh20.html

Abstract: A common feature of neurophysiological signals is the presence of noise, i.e. the signals have the appearance of random processes. In addition, records taken under similar conditions do not have a similar appearance. Analysis of such data therefore requires the use of some form of statistical analysis. For example sample records of EEG recorded from different scalp locations may not look alike, but a statistical analysis may demonstrate that the records share a common underlying rhythm. The field of time-series analysis provides a conceptual and mathematical framework within which random signals, including neurophysiological signals, can be analysed. Two approaches are traditionally used in the analysis of time series – these are the time and frequency domain approaches. The frequency domain approach is generally based on Fourier methods. In the case of two or more simultaneously recorded signals, an important concept is that of correlation, the objective being to assess to what extent the activity in one signal is correlated with that in a second signal. A key frequency domain parameter in this context is the coherence function, which provides a normative linear measure of association between two time series, as a function of frequency. The talk will consider time and frequency domain analyses of neurophysiological signals, including coherence functions. These will be developed within a unified framework where time and frequency domain techniques sit naturally alongside each other, and can be used as complementary forms of analysis. A number of extensions will be discussed, which look at alternative measures of correlation (phase synchronization, causal measures); deal with time dependency (Wavelets, Kalman filtering) and non linearities (higher order spectra).

Everyone is welcome to attend.

Best wishes Rebecca