Title: Time-domain Classification of the Brain Reward System: Analysis of Natural- and Drug-Reward Driven Local Field Potential Signals in Hippocampus and Nucleus Accumbens

Abstract: Addiction is a major public health concern characterized by compulsive reward-seeking behavior. The excitatory glutamatergic signals from the hippocampus (HIP) to the Nucleus accumbens (NAc) mediate learned behavior in addiction. Limited comparative studies have investigated the neural pathways activated by natural and unnatural reward sources. This study has evaluated neural activities in HIP and NAc associated with food (natural) and morphine (drug) reward sources using local field potential (LFP). We developed novel approaches to classify LFP signals into the source of reward and recorded regions by considering the time-domain feature of these signals. Proposed methods included a validation step of the LFP signals using autocorrelation, Lyapunov exponent and Hurst exponent to assess the meaningful stability of these signals (lack of chaos). By utilizing the probability density function (PDF) of LFP signals and applying Kullback-Leibler divergence (KLD), data were classified to the source of the reward. Also, HIP and NAc regions were visually separated and classified using the symmetrized dot pattern technique, which can be applied in real-time to ensure the deep brain region of interest is being targeted accurately during LFP recording. We believe our method provides a computationally light and fast, real-time signal analysis approach with real-world implementation.