Tracking of informative gamma frequency range in local field potentials of anesthetized rat olfactory bulb for odor discrimination

Abstract

The study of responses in the local field potentials (LFP) of the olfactory bulb (OB) induced by odorants’ presentation is one of the intensively developing trends in the olfactory perception research. These studies are often aimed at solving the problem of recognizing and classifying specific responses to various odorants including the usage of machine learning methods. Carrying out studies on anesthetized animals, on the one hand, facilitates the registration of LFP, but, on the other hand, it requires considering the possible non-stationarity of LFP characteristics. We propose a method that makes it possible to track the non-stationarity of the spectral characteristics of LFP in a narrow (10 Hz) frequency range corresponding to gamma oscillations. Predictor-corrector methods, such as α-β and Kalman filter, for tracking the informative gamma frequency range are considered. Tracking the informative frequency range makes it possible to significantly increase the accuracy of recognizing rat OB activity patterns specific to the presented odorants. The results of binary classification for the task of LFP air/tobacco patterns discrimination are presented.