Brain Imaging Workshop 12/11/02
An Introduction to EEG
References:
Nunez, PL. Electric fields of the brain. 1st ed. New York, Oxford University Press, 1981. (New edition due out in March, 2003)
Jasper, HH. Appendix to report to Committee on Clinical Examination in EEG: the
ten-twenty electrode system of the International
Federation. Electroencephalography Clinical Neurophysiology, 1958, 10:371-375.
American Electroencephalographic Society Guidelines for Standard Electrode
Position Nomenclature,
Journal of Clinical Neurophysiology, 1991, 8(2):200-202.
Davidson, RJ, Jackson, DC,
Larson, CL. Human electroencephalography. In: Cacioppo, JT, Tassinary, LG,
Bernston, GG, editors.
Handbook of Psychophysiology. Cambridge: Cambridge University Press, 2000:27-56.
Tomarken, AJ, Davidson, RJ,
Wheeler, R, Kinney, L. Psychometric properties of resting anterior EEG
asymmetry:
Temporal stability and internal consistency. Psychophysiology, 1992, 29:576-592.
For ERPs see me about using BESA or Netstation
Single pyramidal cell may have more than 104 synapses distributed over its soma and dendritic surface.
Perhaps 105 neurons under each square millimeter of cortical surface. Scalp EEG measures space-averaged activity of 107 or more neurons implying a source area of at least a square centimeter. Scalp EEG potentials can be estimated as if generated by a dipole source (i.e. no need for quadrupole or higher order terms). Notice that electrical negativity is directed towards the outer cortex
EEG potentials are measured as the difference between two points, one on the scalp where EEG effects are strong and one (the reference electrode) hopefully isolated from these effects. Some commonly used reference sites are Cz, earlobes, mastoids, tip of nose and average reference ("reference free"). Earlobes or mastoids are generally linked either physically or mathematically in order to maintain symmetry. The average reference uses the constraint that the sum of the potentials over a spherical surface is zero and requires fairly high density recording (~128 channels) and can be improved by estimating potentials for the inferior spherical area. Spherical spline interpolation is sometimes used for these estimates.
Pitfalls of Cz reference (or any reference still containing EEG signal):
F3 actual
F4 actual
Cz1 actual
Measured F3-Cz1
Measured F4-Cz1
At later time actual F3 and F4 have not changed but Cz has changed phase
by 180 degrees
Cz2
Measured F3-Cz2
Measured F4-Cz2
The actual signal at F3 has twice the amplitude as F4's. Referenced to Cz and time 1 the measured (F3-Cz1) has four times the measured amplitude at (F4-Cz1) while at time 2 the measured signal (F3-Cz2) is flatlined since actual F3 equals actual Cz2. This is an extreme example but emphasizes the importance of a reference free from EEG signal.
Standard Electrode positions (10-20)
Traditional anatomical terms designate electrode positions: Frontral pole (Fp), Frontal (F), Central (C), Parietal (P), and Occipital (O). Central represents cortex in the vicinity of the Central Sulcus and is sometimes called the sensori-motor area. Numbering starts at zero for the centerline (z for zero) increasing with distance from centerline with even number on right and odd on left. In addition pharyngeal electrodes are designated by Pg1 and Pg2.
EGI GSN128:
EGI's technical note "ElectrodePositions.pdf" in N:\EGImanuals lists the electrodes on the GSN128 which are closest to the standard 10-20 positions. Built with plastic electrode pedestals connected by monofilament lines having overall geodesic structure (below) so that tensions on electrodes result in force towards the center of approximate head sphere. Flare at the bottom of each electrode pedestal is designed to keep electrolyte in sponge from being wicked away by hair. Thus the base of electrode needs to be below overhanging hair.
Example of mean electrode power values in the alpha band (8-13Hz) during one minute eyes open resting EEG recording plotted versus electrode number. The pluses and zeros represent two different GSN128s with color corresponding to recording run. Notice the decrease in power as electrodes become closer to reference at Cz.
Artifacts:
Blinks and eye movement:
These data are from an eye movement test. Starting at the left the large EEG deflections correspond to up, down, right, left eye movements followed by a blink and repeating the sequence.
Muscle artifact (EMG) shown here near C4:
Heart artifact (EKG) shown here near O1:
Perhaps now is a good time to outline the Lab's procedure for computing asymmetry scores (see Tomarken et al. listed above for details):
Collect eight one-minute resting EEG recordings, four each of eyes open (EO)
and eyes closed (EC)
Remove clipped data with flagclip
Remove 60 Hz line noise with filt60
Visually score out artifacts with eegscore
Compute power spectral for selected bands and references with dopower
Perform statistics in SAS: separate EC and EO average powers weighted by number
of good seconds of data
Final average power from equal weighting of EO and EC averages.
Have encountered electrolyte bridging problems with EGI nets:
(b)
LD 
C3 C4
HD 
C3 C4
Forward modelling with Patrick Berg's Dipole Simulator (available at www.besa.de)
Use default head model and tangential dipole beneath Cz:
Now lower the skull conductivity to 0.0010 mho/m or about 1/4 of default value. This will demonstrate "blurring" by low conductivity skull layer and also why comparing power values between subjects mainly reflects variations in skull conductivity.
This is the reason asymmetry scores are computed as the difference in log powers. F3L = log(F4power) - log(F3power) = log(F4power/F3power) so that the asymmetry score is a ratio independent of between subject power differences.
Now let's assume that we put a GSN128 or GSN256 on the head with lots of electrode sponges soaked in electrolyte with a conductivity several times that of the scalp. This should increase the effective conductivity of the scalp.
Default head model (scalp conductivity 0.33 mho/m):
Head model with net (scalp conductivity 0.50 mho/m):