Biomedical Engineering Reference
In-Depth Information
embraced prematurely. The set of criteria we have proposed can be used in the prag-
matic evaluation of candidate biomarkers.
11.3 Research Applications of EEG to Examine Pathophysiology in
Depression
11.3.1 Resting State or Task-Related Differences Between Depressed and
Healthy Subjects
Coherence is analogous to the squared correlation in the frequency domain between
two EEG signals (time series) measured simultaneously at different scalp locations
[27]. When interpreting differences in EEG coherence between electrodes with dif-
ferent physical distance within hemispheres (e.g., F3-P3 versus F3-O1) or between
them (O1-O2 versus T5-T6), two particularly important factors come into play.
Volume conduction may serve to inflate measures of coherence at short (<10-cm)
interelectrode distances, while an increasing phase difference may reduce coherence
estimates at large distances (
15 cm) [140-142]. Thatcher et al. [28] mapped the
physical distinction between “short” and “large” distances between scalp elec-
trodes onto an anatomical model, by employing Braitenberg's [143] two-compart-
ment model of axonal systems in the cerebral cortex. According to Braitenberg,
compartment A is composed of the basal dendrites that receive input from the axon
collaterals from adjacent pyramidal cells, whereas compartment B consists of the
apical dendrites of pyramidal cells that receive input from remotely originating
corticocortical projections.
Our laboratory and others have used coherence to study circuit function in
patient groups. We examined the use of electrode pairings selected because they
could assess connectivity over known neuroanatomic pathways. For example, to
study connectivity in the superior longitudinal fasciculus, we used an average of
coherence from one anterior pair of channels to three posterior pairings [35]. This
approach yielded useful data in differentiating healthy elders from individuals with
Alzheimer's disease or vascular dementia [35] and in relating structural damage in
white matter tracts to cognitive performance in asymptomatic older adults [32].
We have recently examined a three-dimensional elaboration of the coherence
construct to address a fundamental limitation: Coherence is calculated using two
EEG signals recorded from separate locations, but conventionally this means two
different scalp recording sites. To examine coherence in pathways between cortex
near the scalp electrodes and locations deeper within the skull (remote from scalp
electrodes), we developed a new method referred to as
current source coherence
(CSC) [144]. Whereas determining electrical sources in three dimensions from sur-
face measurements is inherently ambiguous (i.e., the “inverse problem”), a reason-
able estimate can be obtained if some assumptions are made about the distribution
of current sources. Our implementation of CSC employs the LORETA algorithm
[145] as a solution to the inverse problem, but other approaches could also be used.
Instantaneous current vectors can be calculated with LORETA in the 2,394 gray
matter voxels of its solution space, using the Montreal Neurological Institute (MNI)
standard brain model.
>
Search WWH ::
Custom Search