Comparison of Phase Synchronization Measures for Identifying Stimulus-Induced Functional Connectivity in Human Magnetoencephalographic and Simulated Data
Kenji Yoshinaga, Masao Matsuhashi, Tatsuya Mima, Hidenao Fukuyama, Ryōsuke Takahashi, Takashi Hanakawa, Akio Ikeda
Abstract
Although several methods are available for estimating phase synchronization of brain oscillations, it remains unclear which method is to be chosen to detect event-related functional connectivity. Using somatosensory-evoked magnetic fields recorded from healthy participants, we compared performances of four phase synchronization measures (imaginary part of phase-locking value, imaginary part of coherence, phase lag index and weighted phase lag index) for detecting stimulus-induced functional connectivity between the contralateral primary and ipsilateral secondary somatosensory cortices. The analyses converged to reveal that the imaginary part of coherence exhibited the best performance for detecting stimulus-induced functional connectivity, followed by the weighted phase lag index. Because these two measures depend in part on amplitudes of signals, we investigated the effects of amplitudes on computation of phase synchronization measures. We found that amplitude weighting effectively attenuated the influence of noise contamination, without any substantial influences from event-related amplitude coherence. A simulation study modeling noise-contaminated periodograms replicated these findings. In this simulation, the imaginary part of coherence showed slightly better performance in simulating multi-pixel periodograms, while this phenomenon was not observed at the single-pixel level. Overall the present results suggest that the amplitude-dependent measures, the imaginary part of coherence in particular, may have several properties suitable for detecting event-related functional connectivity.