“…Then, the FD-MNE solver was executed, which loads the Fourier transformed signals B(f i ) Re and B(f i ) Im , i = 1, 2, …, n, where Re and Im represent the real and imaginary parts of the Fourier transformed signals, respectively, as well as the pre-saved inverse operator W. The real part q j (f i ) Re and imaginary part q j (f i ) Im of the current source vector at the jth cortical vertex with respect to the frequency of interest f i can then be evaluated by multiplying the corresponding rows (3j -2, 3j -1, and 3jth rows) in W with the Fourier transformed signals B(f i ) Re and B(f i ) Im . We used the FD-MNE method instead of time-domain MNE method to estimate the current source vectors because under the current computing environment maximally 20-30 source images could be calculated per every second due to the computational time required for the inverse process [22]. Then, the instantaneous source power changes for a frequency f i at the jth cortical vertex can be readily estimated by inverse Fourier transforming each directional component (x, y, z-directional components) of q j (f i ) Re and q j (f i ) Im into time-domain series (q x,j (t), q y,j (t), q z,j (t)) and calculating the power of the source vector Q j (t k ) as Q j (t k ) = q x,j (t k ) 2 ?…”