Data-driven estimation of mutual information using frequency domain and its application to epilepsy
Rakesh Malladi, Don H. Johnson, Giridhar P. Kalamangalam, Nitin Tandon, Behnaam Aazhang · 2017
We consider the problem of estimating mutual information between dependent data, an important problem in many science and engineering applications. We propose a data-driven estimator of mutual information in this paper. The main novelty of our solution lies in transforming the data to frequency domain to make the problem tractable. We define a novel metric-mutual information in frequency (Ml-in-frequency)-to detect and quantify the dependence between two random processes across frequency using Cramer's spectral representation. Our solution calculates mutual information as a function of frequency to estimate the mutual information between the dependent data over time and validate its performance on linear and nonlinear models. We then use our MI-in-frequency metric to infer the cross-frequency coupling during epileptic seizures, by analyzing electrocorticographic recordings from a total of eleven seizures in four medial temporal lobe epilepsy patients.