Statistical evaluation of recurrence quantification analysis applied on single trial evoked potential studies

Abstract 

Objective

We evaluated the potential of recurrence quantification analysis (RQA) to improve the analysis of trial-by-trial-variability in event-related potentials (ERPs) experiments.

Methods

We use an acoustic oddball paradigm to compare the efficiency of RQA with a linear amplitude based analysis of single trial ERPs with regard to the power to distinguish responses to different tone types. We further probed the robustness of both analyses towards structured noise induced by parallel magnetic resonance imaging (MRI).

Results

RQA provided robust discrimination of responses to different tone types, even when EEG data were contaminated by structured noise. Yet, its power to discriminate responses to different tone types was not significantly superior to a linear amplitude analysis. RQA measures were only moderately correlated with EEG amplitudes, suggesting that RQA may extract additional information from single trial responses not detected by amplitude evaluation.

Conclusions

RQA allows quantifying signal characteristics of single trial ERPs measured with and without noise induced by parallel MRI. RQA power to discriminate responses to different tone types was similar to linear amplitude based analysis.

Significance

RQA has the potential to detect differences of signal features in response to a standard oddball paradigm and provide additional trial-by-trial information compared to classical amplitude based analysis.

Keywords: Event related potential, Single trial, RQA, MRI, EEG, Oddball