lecture 7 - EEG/MEG Analysis Flashcards
origin of signals in MEG
MEG - primary (ionic/electrical) currents flowing within dendrite of neuron
intracellular
origin of signal in EEG
secondary, volume currents
measures difference on scalp due to extracellular currents
EEG/MEG similarities
- same neural events can be recorded
- require synchronous current flow across 10-50k neurons
sampling signals
sampling of EEG/MEG signals
no discrete jumps in signal, continuous signal so need to sample ongoing current
sample discrete time intervals & digitalize them to reproduce signal
sampling signals
what is the Nyquist sampling theorem?
calculates the frequency a signal must be sampled at to produce correct sampling rate that is accurately reconstructred
at least 2x expected frequency components in continuous signal
ex: 15Hz signal: sample rate = 30 Hz
sampling signals
what is aliasing
when continous signal is undersampled and higher frequency shows as lower frequency in sampled time series
levels of analysis
levels of EEG/MEG analysis
- sensor/electrode space
- source space
levels of analysis
sensor/electrode space analysis
analyzing raw data directly from sensors, time series for different electrodes
levels of analysis
source space analysis
distinguishes origin of activity, mapping
varieties of activity
what is spontaneous activity
recording activity in absence of a task, endogenously generated
varieties of activity
what is Fourier Transform
method that takes complex signals and decomposes it into discrete frequency components
varieties of activity
how does Fourier transformation work
analog signals recorded as a function of time or space can be represented by a large number of sinusoids - each with a specific amplitude, phase, and frequency
varieties of activity
what is Fourier transform used for
converting between time and frequency domains
domain analysis
time domain analysis
how a signal changes over time
units: time (ms, mins etc)
domain analysis
frequency domain analysis
how much of the signal (energy) lies within each given frequency over a range of frequencies
units: Hz
domain analysis
time-frequency domain analysis
combines time and frequency domains to see how frequency components of a signal change over time; temporal evolution of different frequencies
varieties of activity
why use event-related potentials
assumes stimulus-evoked signal is buried in noise, aligns each trial and averages data across all trials = smoother data than individual trials
varieties of activity
what are evoked fields & how do they work?
MEG
magnetic signals generated by the brain in response in sensory stimulation
primary intracellular current produces orthogonal circular magnetic field > MEG data at each sensory averaged > peak signal found & time course explored in detail
varieties of activity
big assumption of averaging evoked fields
across all trials, there is a specific signal that is the response to the stimulus & everything else is noise
varieties of activity
evoked activity
phased-locked & time-locked to stimulus event, fixed latency
occurs consistently at the same time after stimulus activity
varieties of activity
induced activity
phase-jittered and slightly time-shifted
brain activity is influenced by a stimulus but not phase-locked, timing is relative to the stimulus is variable & not easily averaged out
varieties of activity
what brain processes do evoked vs induced activity measure?
evoked - direct neural response to specific sensory input
induced - changes in brain activity due to cognitive processing related to stimulus
source analysis
inverse problem
“what and where is the source that produces that distribution across the scalp?”
spatial distribution of MEG/EEG across scalp
no certain analytic way to figure this out
source analysis
forward problem
starts with understanding of source, can figure out when, how, speed, and direction that current is flowing
can derive what signal looks like outside of head, can solve this problem (Maxwell’s equation)
