2021/02/01 by Sathish Ande, Srinivas Avasarala, Ande, Sathish +13
Neuroscience · #FOS: Electrical engineering #Neural dynamics and brain function #Neuroscience and Neural Engineering #Neuroscience and Neuropharmacology Research #Signal Processing (eess.SP) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2102.00723
openalex publication_date 2021/02/01 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Quantification of information content and its temporal variation in\nintracellular calcium spike trains in neurons helps one understand functions\nsuch as memory, learning, and cognition. Such quantification could also reveal\npathological signaling perturbation that potentially leads to devastating\nneurodegenerative conditions including Parkinson's, Alzheimer's, and\nHuntington's diseases. Accordingly, estimation of entropy rate, an\ninformation-theoretic measure of information content, assumes primary\nsignificance. However, such estimation in the present context is challenging\nbecause, while entropy rate is traditionally defined asymptotically for long\nblocks under the assumption of stationarity, neurons are known to encode\ninformation in short intervals and the associated spike trains often exhibit\nnonstationarity. Against this backdrop, we propose an entropy rate estimator\nbased on empirical probabilities that operates within windows, short enough to\nensure approximate stationarity. Specifically, our estimator, parameterized by\nthe length of encoding contexts, attempts to model the underlying memory\nstructures in neuronal spike trains. In an example Markov process, we compared\nthe performance of the proposed method with that of versions of the Lempel-Ziv\nalgorithm as well as with that of a certain stationary distribution method and\nfound the former to exhibit higher accuracy levels and faster convergence.\nAlso, in experimentally recorded calcium responses of four hippocampal neurons,\nthe proposed method showed faster convergence. Significantly, our technique\ndetected structural heterogeneity in the underlying process memory in the\nresponses of the aforementioned neurons. We believe that the proposed method\nfacilitates large-scale studies of such heterogeneity, which could in turn\nidentify signatures of various diseases in terms of entropy rate estimates.\n