David E. Morledge
Garfield Auditory Res. Lab., 3440 N. Broad St., Temple Univ. School of Medicine, Philadelphia, PA 19140
A frequency-domain detection method, based on statistical estimates derived from coherence measures, was used to determine optimal signal processing and stimulus parameter settings. The method was evaluated using simulations of electrophysiologic recording noise and signals in noise, as well as, real physiologic noise and click evoked auditory brain-stem responses. Parameters varied in the study were artifact rejection level, spectral filter shape, and stimulus presentation rate. Signal detection was improved following objective selection of parameter settings by this method. Some systematic estimate biases were found under the noise conditions evaluated. This method should be useful in selecting stimulus and recording parameter settings for improved signal detectability. [Work supported by U.C. San Francisco Graduate Division.]