Use of a running signal to noise ratio to identify voluntary event related cortical potentials

Signal-to-noise ratio (SNR) measures have long been used to examine the quality of evoked potential recordings. O/spl uml/zdamar and Delgado (1996) derived an energy-based SNR measure and applied it to auditory brainstem responses. Preliminary evidence is presented that this running SNR estimate can also be used to identify voluntary, motor-related, cortical potentials using a very small number of trials. The running SNR indicator is applied to actual electrocorticogram (ECoG) recordings, and its ability to identify voluntary event-related potential (VERP) estimates is compared with the morphologically-based peak-to-mean (pm) ratio measure developed by Huggins et al. (1995, 1996). The running SNR-based classification is shown to distinguish between supra-threshold pm-ratio VERP estimates yielding good cross-correlation results and those that perform poorly.