Does a Subject Posture Affect EEG and TES? A Study Using EIT and FEM Head Modeling

MARIANO FERNÁNDEZ CORAZZA; EASWARA MOORTHY ESSAKI ARUMUGAM; SERGEI TUROVETS

Conferencia; 4th Annual Brain Stimulation and Imaging Meeting; 2018

Does Subject Posture affectEEG/TES ? : A Study Using EIT and Finite Element Modeling EaswaraMoorthy Essaki Arumugam1?, Mariano Fernández-Corazza2,3?,Sergei Turovets1,3  1PhilipsNeuro, Eugene, OR, USA, 2LEICI Instituto de Investigaciones enElectrónica, Control y Procesamiento de Señales, Universidad Nacional de LaPlata (UNLP), CONICET, La Plata, Argentina, 3NeuroInformaticsCenter, University of Oregon, Eugene, OR, USA ?These authors contributed equally to thiswork IntroductionMany technologicaladvances have been made recently in studying and manipulating the human brainusing Electroencephalography (EEG) and Transcranial Electrical Stimulation(TES) respectively. Yet some basic things have been ignored or not given much importancein the neuroscience field ? once such thing is the influence of subject postureon EEG and TES. Subject posturehas been mostly studied in the context of baroreceptors, which aremechanoreceptors that affects the neural activity and cognition (Lipnicki,2009; Thibault & Raz, 2016). Rice et al., (2013) showed that the brainshifts in supine or prone position due to negative buoyancy and while lying insupine position, EEG amplitude increased by 80 % since the occipital area ofthe brain is closer to the EEG electrodes. While EEG changes between subjectpostures could be due to actual neurophysiology and/or brain shift, it isdifficult to isolate which one plays the major role. In this study, toeliminate neurophysiology effect with posture, we have used ElectricalImpedance Tomography (EIT) method to see if there are any changes in theimprinted scalp potentials due to the injected current in different subjectpositions. Also we tried to simulate using Finite Element Method (FEM) toverify the experimental EIT data.  MethodsExperimentalProcedure: One Caucasian male subject (37 years old)participated in the study approved by Electrical Geodesic Inc. (EGI)Institutional Review Board. The subject wore a 256-channel HydroCel GeodesicSensor Net (HCGSN-130) with Elefix paste as the electrolyte. The hardware usedwas Geodesic Transcranial Electrical neuromodulation (GTEN-100; Luu et al.,2016) prototype system, in which eight sources and eight sinks (Figure 1.A) wereused to inject AC current (frequency of 27 Hz and 10 mA peak to peak per pair with the total being 80 mA) and the rest 240 passive channels measuring the imprintedpotentials. The data was collected in NetStation 5 software (Philips Neuro,Eugene, OR, USA) and processed in Matlab 2016b (The MathWorks, Inc., Natick,MA, USA). The amplitudes were extracted by averaging the peak to peak value ofthe imprinted potentials after band-pass filtering of 22-32 Hz andaverage-referenced. Bad channels were removed from the analysis. EIT data wascollected three times with the same subject: first with the subject sittingupright, second with lying horizontally in a reclinable TMS chair and the thirdwith sitting upright. Electrical HeadModel: T1-weighted MRI (3T) was taken (supineposture) from 52 year Asian male and 7 different tissues (Scalp, Skull, CSF,Gray Matter, White Matter, Eye balls, internal air) was segmented through ModalImage Pipeline (MIP; Li et al., 2016) software and a detailed electrical headmodel was built using FEM (Fernández-Corazza et al., 2017). The conductivities inS/m assigned were: Scalp: 0.315, Skull: 0.008, CSF: 1.79, GM: 0.33, WM: 0.2, Air:0, Eye: 1.5, Electrode: 500, Electrolyte: 1.5. Supine PositionSimulation: While the regular MRI was taken assupine posture, the MIP software pads CSF with a minimum of 1 voxel between theskull and the Gray Matter (Li et al., 2016). This makes the CSF thickness about2-4 mm around the head, which is considered as upright posture here (Figure 3.A) in this analysis. To simulate supine posture in FEM, wetried three things: (1)   shifting the brain 2 mm posteriorlytouching the skull (Figure 3.D), making it less CSF in the back of the head andmore in the front (similar to Rice et al, 2013).(2)   assuming more blood flow inthe back of the head while in supine posture due to gravity, we made conductivityof the scalp tissue as a linear gradient from 0.30 (in the front) to 0.33 (inthe back) with the average being 0.316 S/m.(3)   we assumed that there couldbe more pressure on the back of the scalp, while lying in supine position withthose corresponding electrodes touching the head rest of the TMS chair. Soalong with (2), we simulated a local ischemia (by reducing the scalpconductivity by 10%) around those electrodes which were resting on the headrest (See Figure 1.B). ResultsThere was cleardifference between upright position before and supine position (RelativeDifference Measure, RDM = 0.13) imprinted potentials as well as upright positionafter and supine position (RDM = 0.16) due to experimental EIT. RDM of uprightbefore and after supine was about 5 times less (RDM=0.03). FEM simulationof the scalp conductivity changes, brain shift or scalp conductivity changes/localischemia did not show much differences in the EIT imprinted potentials (Figure1.D). Figure 2 shows localconductivity changes (experimental EIT data) by taking the linear gradient ofthe differences between the different subject postures. One can see maximallocal conductivity changes near the back of the head. To analyze theimpact of the subject posture on EEG and TES, we used brain shift (by 2 mm) model.EEG scalp topography was simulated due to activation of an occipital dipole pointingbackwards (Figure 3.B and E). By shifting the brain by 2 mm, one can clearlysee the scalp negativity shifted as well as TES reciprocity optimization patternof sink electrodes (Figure 3. C and F).  ConclusionsAlthough our FEMmodel did not show drastic differences in EIT imprinted potentials due tosubject postures, our experimental EIT data showed a significant redistributionof voltage gradient. Mostimportantly, we showed that brain shift can cause considerable impact in EEGand TES by simulation of a realistic head model. We think that more experimentsand simulations are needed to understand subject posture in neuroscience field. Figure 1. (A) Geodesic sensor montage with the injectionelectrodes colored. (B) Simulation of (3) in methods section, where the scalpconductivity is varied from the front to back and also simulating localischemia on the back electrodes. (C) Experimental EIT imprinted potentials ofupright position, supine and then again upright. (D) Differences of the uprightand supine positions in experiment, FEM simulation by shifting brain and FEMsimulation due to Fig. 1.B Figure 2. Local resistivity changes of theexperimental EIT data. First the differences between two postures are computedand then a gradient was taken to calculate the local potential differences,which is proportional to the resistance changes as the total current isconstant. One can see the similarity of the upright before vs supine andupright after vs supine positions. A 3D head model is shown in 2.A inset forreference. Bad channels are interpolated.Figure 3. FEM simulation of brain shift in supine position.(A) and (D) MRI slice showing upright and supine position simulation. (B) and(E) showing the EEG simulation of the scalp topography due to the sameoccipital dipole. (C) and (F) showing TES reciprocity optimization for the sameoccipital dipole in upright and supine positions. One can see a cleardifference of scalp negativity between (B) and (E) as well the sink electrodesselection in (C) and (F). References[1] Corazza, M. F., Turovets, S., Luu, P.,Price, N., Muravchik, C., & Tucker, D. (2017). Skull modeling effects inconductivity estimates using parametric electrical impedance tomography. IEEETransactions on Biomedical Engineering.[2] Li, K., Papademetris, X., & Tucker,D. M. (2016). BrainK for structural image processing: creating electricalmodels of the human head. Computational intelligence and neuroscience, 2016.[3] Lipnicki, D. M. (2009). Baroreceptor activity potentiallyfacilitates cortical inhibition in zero gravity. Neuroimage, 46(1),10-11.[4] Luu, P., Arumugam, E., Moorthy, E.,Anderson, E., Gunn, A., Rech, D., ... & Tucker, D. M. (2016).Slow-Frequency pulsed transcranial electrical stimulation for modulation ofcortical plasticity based on reciprocity targeting with precision electricalhead modeling. Frontiers in human neuroscience, 10,377.[5] Rice, J. K., Rorden, C., Little, J. S.,& Parra, L. C. (2013). Subject position affects EEG magnitudes. NeuroImage, 64,476-484.[6] Thibault, R. T., & Raz, A. (2016).Imaging posture veils neural signals. Frontiers in human neuroscience, 10,520.s