Statistically reliable and fast direct estimation of phase-amplitude cross-frequency coupling.

Abstract

There is growing interest in phase-amplitude cross-frequency coupling (PAC), which is widely observed in human and animal brain recordings. The choice of the estimation method is vital while extracting accurate PAC parameters from data. Two desired properties of PAC estimators are reliability and computational efficiency. This study offers a methodology called normalized direct PAC (ndPAC) for the rapid and statistically reliable estimation of PAC strength. A plain confidence limit formula, depending solely on data length and confidence level, is derived. Confidence level derivation is validated numerically. It is shown through simulations that ndPAC exhibits high specificity and sensitivity performances. The suggested methodology is also demonstrated on monkey electrocorticogram recorded during a visual task.