Richards, J.E. (2007). Realistic head models for cortical source analysis in infant participants. Society for Research in Child Development, Boston. (PDF)
Cortical source analysis can
identify cortical areas that are active during infant cognitive processing. This
analysis uses high-density EEG recording and quantitative models that identify
dipole sources inside the head to account for the EEG data. These sources can be
related to the EEG activity in the time domain, to the experimental procedures,
and to the cognitive processes occurring during the task. Cortical source
analysis with infant participants has used adult models for the electrical and
spatial characteristics of the head. The current poster will overview a method
for using infant MRIs to develop realistic head models for infant participants
for cortical source analysis.
The method consists of
several steps: 1) the MRI of a representative infant or the infant in the
psychophysiological experiment must be obtained; 2) the head must be “segmented”
with computer programs that analyze the MRI recording; 3) values for impedance
for skull, scalp, CSF, and brain for infant participants must be estimated; 4)
sufficient number of electrodes must be recorded; 5) realistic “forward” models
using the spatial topography of the MRI must be computed.
These steps allow the use of “realistic models” in computer programs that
do cortical source analysis (e.g., “equivalent current dipole” analysis in BESA,
or Source-Signal EMSE programs).
This technique has been
applied to cortical sources involve in infant spatial cueing, infant visual
recognition memory, the response of infants to mother’s and stranger’s face, and
EEG during object disappearance and appearance. The result of using realistic
head models is illustrated in four findings.
First, infant skull and scalp impedances (conductivity; resistance) are
much lower than adults, so that using appropriate infant values results in
cortical sources closer to the surface.
Second, the number of recording channels affects cortical source
analysis, with less than ~50 electrodes resulting in models that fit poorly and
that mislocate the cortical source. Third, using realistic head models for areas
of current leakage in the skull (sutures / seams between skull bones, partially
closed fontenal) or for varying head characteristics (varying skull thickness,
placement of brain with respect to skull landmarks) result in better fits and
more appropriate locations of cortical sources. Finally, this is illustrated
with a “representative MRI” approach in which a single MRI is used for all
participants, and with an “individual MRI” approach in which the infant has both
an anatomical MRI and participates in the psychophysiological experiment.