I will talk about a novel and robust method based on making use of higher criticism and large scale inference for detecting signals and sorting peaks in electrophysiological measurements of neuronal activities, which are accompanied with considerable noise. The detection method relies solely on the intrinsic statistical properties of the data and avoids any preprocessing, which prevents the loss of any invaluable information. Some discussions of time frequency analysis based on the Hilbert-Huang transform and classification using geometric methods will also be presented. This is joint work with E. Mitricheva, R. Kimura, N. K. Logothetis and H. R. Noori.