Research Article

Noise-Assisted Instantaneous Coherence Analysis of Brain Connectivity

Figure 1

Schematic representation of the proposed noise-assisted instantaneous coherence (NAIC). A trivariate data [X Y Z] is used as an example. The first step (A) consists of transforming each time series to the corresponding analytic matrix by virtue of the MEMD and Hilbert transform. A random noise complex matrix is then added to the analytic matrix of data (B) to facilitate the calculation of coherence (C). Two random noise complex matrices are independently generated to compute their coherence. The process is repeated for (e.g., 1000) times (D) to obtain a null distribution of the maximum coherence (E). Here, we set the value as 0.01, thus the threshold “ ” corresponds to the 10th value from the maximum of the null distribution (F). Finally, we use the “ ” to threshold the coherence from (C) to be considered as statistically significant from noise (G). The output of NAIC (H) provides high-resolution time-frequency coherence spectrum.
275073.fig.001