A tunable local field potentials computer simulator to assess minimal requirements for phase–amplitude cross-frequency-coupling estimation. (1st October 2016)

• Record Type: Journal Article
• Authors: Rubega, Maria; Fontana, Roberto; Vassanelli, Stefano; Sparacino, Giovanni
• Is Part Of: Network. Volume 27:Number 4(2016)
• Page Start: 268
• Page End: 288
• Publication Date: 2016-10-01
• Subjects: Cross-frequency coupling -- local field potential -- model -- neuroscience -- simulation; Neural computers -- Periodicals; Neural networks (Computer science) -- Periodicals; 006.32
• DOI: 10.1080/0954898X.2016.1213440
• Languages: English
• ISSNs: 0954-898X

Abstract

ABSTRACT: The quantitative study of cross-frequency coupling (CFC) is a relevant issue in neuroscience. In local field potentials (LFPs), measured either in the cortex or in the hippocampus, how γ-oscillation amplitude is modulated by changes in θ-rhythms-phase is thought to be important in memory formation. Several methods were proposed to quantify CFC, but reported evidence suggests that experimental parameters affect the results. Therefore, a simulation tool to support the determination of minimal requirements for CFC estimation in order to obtain reliable results is particularly useful. An approach to generate computer-simulated signals having CFC intensity, sweep duration, signal-to-noise ratio (SNR), and multiphasic-coupling tunable by the user has been developed. Its utility has been proved by a study evaluating minimal sweep duration and SNR required for reliable θ–γ CFC estimation from signals simulating LFP measured in the mouse hippocampus. A MATLAB® software was made available to facilitate methodology reproducibility. The analysis of the synthetic LFPs created by the simulator shows how the minimal sweep duration for achieving accurate θ–γ CFC estimates increases as SNR decreases and the number of CFC levels to discriminate increases. In particular, a sufficient reliability in discriminating five different predetermined CFC levels is reached with 35-s sweep with SNR = 20, while SNR = 5 requires at least 140-s sweep.