Synchronization Tomography: A Method for Three-Dimensional Localization of Phase Synchronized Neuronal Populations in the Human Brain using Magnetoencephalography

We present a noninvasive technique which allows the anatomical localization of phase synchronized neuronal populations in the human brain with magnetoencephalography. We study phase synchronization between the reconstructed current source density (CSD) of different brain areas as well as between the CSD and muscular activity. We asked four subjects to tap their fingers in synchrony with a rhythmic tone, and to continue tapping at the same rate after the tone was switched off. The phase synchronization behavior of brain areas relevant for movement coordination, inner voice, and time estimation changes drastically when the transition to internal pacing occurs, while their averaged amplitudes remain unchanged. Information of this kind cannot be derived with standard neuroimaging techniques like functional magnetic resonance imaging or positron emission tomography.

Figure 1

Time-dependent power spectral density in dB (a) of an original MEG signal from a channel over the left sensorimotor cortex (SMC), (b) of the strongest component of the cerebral current source density $\mathbf{j}$ from SMC, (c) of the preprocessed EMG signal of the right flexor digitorum muscle after standard preprocessing, i.e., high-pass filtering ($>30\mathrm{H}\mathrm{z}$) followed by rectification (where $x\to |x|$), applied to extract the burst activity, and (d) of the strongest component of $\mathbf{j}$ from the secondary auditory cortex. Yule-Walker algorithm [13] with window length $=12\mathrm{s}$, step size $=1.3\mathrm{s}$.

Figure 2

(a) Simulation: A simulated current with time course ${\xi }_1$ and fixed orientation and location ${\mathbf{x}}_0$ (“$\to$”) in the left sensorimotor cortex (SMC) of a real brain generates a magnetic field. ${\mathbf{x}}_0$ is reliably detected with the $1:1$ synchronization index (1) between ${\xi }_1$ and all voxels [15]. (b)–(d) Experiment: Synchronization indices ${\rho }_{1,2}$ [16] between the EMG of the right flexor digitorum muscle (RFM) (2-Hz band) and brain currents (4-Hz band) [(b),(c) during first minute], and ${\rho }_{1,1}$ between RFM (2-Hz band) and brain currents (2-Hz band) [(d), during second minute]; window length $T=60\mathrm{s}$. Areas with local maxima of ${\rho }_{1,2}$ and ${\rho }_{1,1}$ are shown in corresponding coronal and transversal brain sections: left secondary auditory cortex (AC), left cerebellum (LC), left premotor cortex (PMC), and left SMC [14].

Figure 3

(a)–(c): Averaged amplitudes $\overline{A}$ of the current source density from (3) in arbitrary units of the 2-Hz band (1–3 Hz, bold line) and the 4-Hz band (3–5 Hz, thin line) of SMC (a), of LC (b), and of AC (c). $\overline{A}$ of PMC is similar to $\overline{A}$ of AC. Brain/muscle synchronization (CMS) in (d),(e): CMS between RFM and SMC (d), RFM and LC (e). Synchronization index ${\rho }_{1,1}$ [16] for the 2-Hz band of both muscle activity and brain currents (bold) and ${\rho }_{2,1}$ for 2-Hz band of muscle activity and 4-Hz band of brain currents (thin). $1:1$ brain/brain synchronization (CCS) between SMC and PMC (f): ${\rho }_{1,1}$ for the 2-Hz band (bold) and for the 4-Hz band (thin) of the currents of both areas. No $2:1$ CCS within or between brain areas was found. (a)–(f) Window length $T=20\mathrm{s}$ [14].