A window into the brain networks: magnetoencephalography (MEG) and simultaneous Transcranial Current Stimulation (tCS).

Based on already published large evidence, non-invasive brain stimulation (NIBS) techniques like tdCS represent very important approach for the improvement of abnormal brain functions in various conditions (psychiatric and neurological). NIBS can induce temporary changes of neural oscillations and performance on various functional tasks. One of the key-points in understanding a mechanism of NIBS is the knowledge about the brains response to current stimulation and underlying brain network dynamics changes. Until recently, concurrent observation of the effect of NIBS on multiple brain networks interactions and most importantly, how current stimulation modifies these networks remained unknown because of difficulties in simultaneous recording and current stimulation. Recently, in Neuroelectrics wireless hybrid EEG/tCS 8-channel neurostimulator system has been developed that allows simultaneous EEG recording and current stimulation. Now, a relatively new imaging technique called magnetoencephalography (MEG) has emerged as a procedure that can bring new inside into brain dynamics. In this context, our group conducted a successfully proof of concept test to ensure the feasibility of concurrent MEG recording and current stimulation using Starstim and a set of non-ferrous electrodes (Figure 1). But first of all, what actually is MEG? Magnetoencephalography (MEG) is a noninvasive recording method of the magnetic flux from the head surface. Magnetic flux is associated with intracranial electrical currents produced by neural activity (the neural currents are caused by a flow of ions through postsynaptic dendritic membranes). From Maxwell equations, magnetic fields are found whenever there is a current flow, whether in a wire or a neuronal element. Hence, MEG detects these magnetic fields generated by spontaneous or evoked brain activity.

The magnetic fields generated by neural activity are extremely small. These fields are about one billion times weaker than the ambient magnetic field of the earth, on the order of femtoTesla (10-15Tesla) to picoTesla (10-12Tesla). For a measurement of these tiny fields a magnetic detector (a conductive wire loop) that is sensitive to magnetic flux passing through it is used. For such very weak magnetic fields to induce current in the loop it must have practically no electrical resistance (must be superconducting) and this can be achieved by reducing the temperature of the wires to close to absolute zero. Hence, the (loop) magnetometer wires are housed in a thermally insulated reservoir filled with liquid helium (the coldest cryogenic liquid), which keeps them at a temperature of about 4.2 Kelvin (-269°C). In MEG systems as amplifiers superconductive quantum interference devices (SQUIDS) are used (loops containing Josephson junction). These devices convert the feeble induced currents to high-amplitude voltages. MEG systems are equipped with a head-shaped array of more than hundred SQUIDS sensors (Figure 2). As I mentioned already, brain fields are about one billion times weaker than the ambient magnetic field of the earth. Cars moving past the building where MEG system is located generate a larger magnetic field than that from the brain, as does the nearby lift, and even people walking up the metal staircase in the building generates a measurable noise signal. The magnetic activity of the brain is substantially smaller than ambient noise and this is the main reason why MEG recordings are performed in a magnetically shielded room to isolate them from external magnetic fields. Reduction of environmental noise is achieved by placing the MEG system inside those magnetically shielded room (MSR). The MSR provides passive shielding of high frequency noise achieved by the use of a layer of highly conductive metal (typically aluminium). Low frequencies (e.g., < 100 Hz), shielded rooms also have layers of high magnetic permeability material (Mu-metal), which, depending on the number of layers can provide attenuation ranging from a modest factor of 30 for very low frequencies up to 100 or more at higher frequencies. Recently, MEG has become an important tool in neurological signal processing and functional neuroimaging. During the last decade, an increasing number of studies of language and cognitive functions and brain connectivity have been carried out. The main applications of MEG are clinical investigations (epilepsy, language, somatosensory, auditory, motor and visual area mapping), and cognitive neuroscience research (you can find more information in Hamalainen et al. 1993; Papanicolaou 2009 or on a webpage htttp://megcommunity.org) (Figure 3 and 4).

Its high temporal resolution permits assessment of fractions of milliseconds. MEG also has excellent spatial resolution; sources can be localized with a millimetre precision. Due to the fact that magnetic fields are not volume conducted over the scalp as in the case of electrical potentials (the influence of skull, scalp, cerebrospinal fluid and brain tissue on magnetic fields is very weak), detailed topographical patterns of magnetic field activity overlying the scalp can be used to infer the location of activity and enables an almost undistorted view on brain activity. Therefore, MEG source analysis, after being overlaid onto subject’s individual MRI yields a realistic image of the location of underlying tissue causing neurophysiological activity. A major attraction of MEG is that its temporal resolution is limited only by the sampling rate of the electronics, which in MEG systems typically far exceeds the largest bandwidth of interest for brain signals. Using MEG we can measure both DC shifts related to slow polarization of the cortex, as well as high-frequency oscillations and transient spikes. This is in contrast to functional brain imaging technique such as PET or fMRI, which are based on metabolic/hemodynamic phenomena and have temporal resolution on the order of minutes or seconds, thus are incapable of imaging rapidly changing patterns of brain activity.


Regarding our proof of concept test with MEG-tCS simultaneously recording. We used a whole-head magnetoencephalography system (148 sensors – magnetometers, 4D Neuroimaging Magnes 2500WH, San Diego, CA,) to record magnetic fields at 678 Hz sampling rate using the phantom head included with the MEG system. The stimulation test (tDCS) has been performed using MRI Sponstim electordes (Neuroelectrics® sponge electrodes for MRI compatible stimulation). Stimulation electrodes have been placed on the phantoms right center (C4) and right posterior (O2) positions. We performed the quantification of noise induced by tDCS relative to sensor noise. As expected our Starstim device stimulators generated significant electromagnetic noise resulting in FFT power changes up to 5 dB (relative to sensor noise) on average across the MEG sensor array. Noise was highest at lower frequencies (0-2Hz) than (higher frequencies (2-200Hz). Noise levels were highest around the stimulating tDCS electrodes and detectable in areas remote from the stimulation electrode with varying amplitude. We can compare levels of the noise produced by tCS with a noise produced by metallic interference that come from inside the head in clinical studies, such as implanted intracranial electrodes and dental ferromagnetic prosthesis and brackets, or from outside, such as pacemakers and vagal stimulators. It has been already shown already that algorithms based on signal space separation (SSS) and blind source separation (BSS) techniques to remove metallic artifacts from MEG signals can be successfully applied to any MEG dataset affected by artifacts, allowing further analysis of unusable or even poor quality data. For instance, the goal of BSS algorithms is to estimate the different original source signals or components from the observation signals assuming a linear mixture model. As has been shown in our recent study (Migliorelli et al. 2015), this can be done because, although original source signals and the mixing signals are unknown, a certain statistical independence between sources is assumed. Both (SSS and BSS) algorithms can increase the SNR by approximately 100%. Noise values from our test with tCS are comparable to those with vagal stimulator or ferromagnetic prosthesis. We believe that the same methods that effectively remove metallic interference from MEG signals can be successfully applied to tCS unwanted signal components reduction or removal (as has been shown already in Soekadar et al. 2103, Garcia-Cossio et al. 2015 and Marshall et al. 2015). Our next step will be to perform in-vivo study with real subjects to open a window into the brain networks with MEG and simultaneous transcranial current stimulation. This will help in the identification of affected networks and thus help in optimal NIBS stimulation approaches.

1.Hamalainen, M., Hari, R., Ilmoniemi, R.J., Knuutila, J., and Lounasmaa, O.V. (1993). Magnetoencephalography – Theory, Instrumentation, and Applications to Noninvasive Studies of the Working Human Brain. Rev Mod Phys 65, 413-497. DOI: 10.1103/RevModPhys.65.413

2.Andrew C. Papanicolaou (editor), Clinical Magnetoencephalography and Magnetic Source Imaging, Cambridge University Press; 1 edition (14 September, 2009); 220 pages; ISBN-13: 978-0521873758

3.Carolina Migliorelli, Joan F Alonso, Sergio Romero, Miguel A Mañanas, Rafał Nowak and Antonio Russi.

Automatic BSS-based filtering of metallic interference in MEG recordings: definition and validation using simulated signals, J Neural Eng. 2015 May 27;12(4):046001.

4.Soekadar SR, Witkowski M, García Cossio E, Birbaumer N, Robinson SE, Cohen LG. In vivo assessment of human brain oscillations during application of transcranial electric currents. Nature Communications 2013;4.

5.Garcia-Cossio, E., Witkowski, M., Robinson, S.E., Cohen, L.G., Birbaumer, N.,

Soekadar, S.R. Simultaneous transcranial direct current stimulation (tDCS) and whole-head magnetoencephalography (MEG): assessing the impact of tDCS on slow cortical magnetic fields. Neuroimage. 2015; doi: 10.1016/j.neuroimage.2015.09.068 [in press].

6.Marshall T, Esterer S, Herring JD, Bergmann TO & Jensen O (2015). On the relationship between cortical excitability and visual oscillatory responses – a concurrent tDCS-MEG study. NeuroImage, doi:10.1016/j.neuroimage.2015.09.069.

Leave a Reply

Your email address will not be published. Required fields are marked *