SPATIO-TEMPORAL ANALYSIS OF EVOKED POTENTIALS

This research seeks to increase the information obtained from noninvasive evoked potential (EP) studies by developing (1) techniques which separate the contributions of temporally overlapping generators from the surface-recorded data, and (2) improved models for the volume conduction mechanisms which relate generator location to surface topography. These techniques will be applied to analysis of human short-latency auditory EPs and subcortical and cortical somatosensory EPs to median and peroneal nerve stimulation. Human EPs recorded intracranially during neurosurgery will be used for verification and optimization of the models and techniques. Estimated generator waveforms will be defined by principal component analysis of surface topographic mappings with rotation of the factors, and examined in normal subjects and in patients with neurologic disease. In the patient group, the sensitivity of the test and its ability to localize areas of dysfunction will be compared to those of conventionally interpreted EPs. The volume conduction processes by which the surface potentials are derived from activity in distant generators will be examined through a numeric or finite-differences computer model. Such a model takes into account the anatomy of conductance inhomogeneities within the head, and should more accurately represent volume conduction of human EPs than models assuming spherical symmetry. The "forward" volume conduction model defines the surface distribution produced by a hypothetical array of generators; the "inverse" problem will be approached by heurisitically modifying the generator hypotheses to obtain the best fit between the predicted surface topography and the empirical data. The techniques developed in the course of these studies of specific EPs are also applicable to the analysis of surface- recorded event-related potentials produced by various other stimuli and by cognitive and motor tasks. Thus, they will increase both clinical accuracy and experimental utility of many types of event-related potentials.