Basic neuroscienceA novel approach to identify time-frequency oscillatory features in electrocortical signals

- A novel approach was proposed to optimally identify oscillatory features.
- Time-frequency oscillatory features were identified based on their spatial distributions.
- The novel approach was validated using both simulated and real electrocortical datasets.
- The novel approach can help explore the precise functions of oscillatory activities.

BackgroundSensory, motor, and cognitive events could not only evoke phase-locked event-related potentials in ongoing electrocortical signals, but also induce non-phase-locked changes of oscillatory activities. These oscillatory activities, whose functional significances differ greatly according to their temporal, spectral, and spatial characteristics, are commonly detected when single-trial signals are transformed into time-frequency distributions (TFDs). Parameters characterizing oscillatory activities are normally measured from multi-channel TFDs within a time-frequency region-of-interest (TF-ROI), pre-defined using a hypothesis-driven or data-driven approach. However, both approaches could ignore the possibility that the pre-defined TF-ROI contains several spatially/functionally distinct oscillatory activities.New methodWe proposed a novel approach based on topographic segmentation analysis to optimally and automatically identify detailed time-frequency features. This approach, which could effectively exploit the spatial information of oscillatory activities, has been validated in both simulation and real electrocortical studies.ResultsSimulation study showed that the proposed approach could successfully identify noise-contaminated time-frequency features if their signal-to-noise ratio was relatively high. Real electrocortical study demonstrated that several time-frequency features with distinct scalp distributions and evident neurophysiological functions were identified when the same analysis was applied on stimulus-elicited TFDs.Comparison with existing methodsUnlike traditional approaches, the proposed approach could provide an optimal identification of detailed time-frequency features by making use of their distinct spatial distributions.ConclusionsOur findings illustrated the validity and usefulness of the presented approach in isolating detailed time-frequency features, thus having wide applications in cognitive neuroscience to provide a precise assessment of the functional significance of oscillatory activities