Realistic models of cortical source analysis in infant participants

Greg D. Reynolds, Department of Psychology, Appalachian State University
John E. Richards, Department of Psychology, University of South Carolina

Brain activity generates electrical potentials that may be recorded on the scalp, the "electroencephalogram" (EEG). Ongoing EEG can be modified by psychophysiological experiments requiring cognitive activity. When events are time-locked to the EEG recording the "event-related potential" (ERP) is measured. The ERP is thought to reflect activity of the brain in areas that control the cognitive processes involved in the psychophysiological tasks. A quantitative technique called "cortical source analysis" can identify the brain areas that are the source of the electrical activity recorded on the scalp, and by inference, the areas of the brain involved in the cognitive processes. Cortical source analysis has used simple models of brain topography and impedance in order to simply calculations for this analysis, e.g., multi-sphere models. However, realistic models of the media inside the head (gray matter, white matter, skull, scalp, CSF, dura) may be used in order to do cortical source analyses. These realistic models give more accurate locations of the sources inside the head, realistic calculation of the topography of the resistances in head media, and whole field electrical calculations of the effect of the cortical sources on the electrical activity on the scalp. We have used these models in a limited way with infant participants doing attention and recognition memory tasks; one of use (JER) has used these models successfully with adult and adolescent participants.

Our current work in this area involves the development of models for realistic cortical source in infant participants. These models demand that anatomical MRIs be done on infant participants. In one scenario an individual infant has the anatomical MRI and then participates in psychophysiological experiments. The MRI then can be used as a realistic model of the media inside the head and cortical source analysis of the resulting ERP data in the psychophysiological experiment is more accurately done. We currently are working on developing recording protocols for infant MRIs in order to pursue this work. A second scenario is to have a library of anatomical MRIs in order to use a infant head that is of similar shape to an infant being tested in a psychophysiological experiment. A third scenario is the development of a stereotaxic atlas (e.g., Talairach atlas; MNI atlas). The latter two scenarios will be examined both with MRIs recorded in our laboratory, and in conjunction with the library of MRIs being recorded in the NIH MRI Study of Brain Development.

Several complications are being examined. First, the resistance values of the media inside infant's heads are not known. Bone density values are much smaller in infant participants, and skin / scalp have fewer dead cells, leading to less resistance in these media in infants in adults. Second, infants have places in the skull where the bones are not yet joined (seams, fontanel). This leads to current leakage to the recording electrodes on the scalp. Third, the overall topography of the brain-skull relation differs in infants and adults, and individual anatomical areas (e.g., Brodmann areas) have a different relation to external skull landmarks in infants and adults. Fourth, the lack of axonal myelination in the first few months and changes in myelination over childhood affect the degree to which cell bodies and axons can be differentiated. So called "gray matter" in adults and adolescents primarily consists of cell bodies and "white matter" are myelinated axons. In the infant however, gray matter consists of both cell bodies and axons. These complications are being evaluated in the development of realistic models of the head for cortical source analysis of infant EEG and ERP.