Riemannian classification of single-trial surface EEG and sources during checkerboard and navigational images in humans

Objective Different visual stimuli are classically used for triggering visual evoked potentials comprising well-defined components linked to the content of the displayed image. These evoked components result from the average of ongoing EEG signals in which additive and oscillatory mechanisms contribute to the component morphology. The evoked related potentials often resulted from a mixed situation (power variation and phase-locking) making basic and clinical interpretations difficult. Besides, the grand average methodology produced artificial constructs that do not reflect individual peculiarities. This motivated new approaches based on single-trial analysis as recently used in the brain-computer interface field. Approach We hypothesize that EEG signals may include specific information about the visual features of the displayed image and that such distinctive traits can be identified by state-of-the-art classification algorithms based on Riemannian geometry. The same classification algorithms are also applied to the dipole sources estimated by sLORETA. Main results and significance We show that our classification pipeline can effectively discriminate between the display of different visual items (Checkerboard versus 3D navigational image) in single EEG trials throughout multiple subjects. The present methodology reaches a single-trial classification accuracy of about 84% and 93% for inter-subject and intra-subject classification respectively using surface EEG. Interestingly, we note that the classification algorithms trained on sLORETA sources estimation fail to generalize among multiple subjects (63%), which may be due to either the average head model used by sLORETA or the subsequent spatial filtering failing to extract discriminative information, but reach an intra-subject classification accuracy of 82%.

Detailed measurements and simulations of electric field distribution of two TMS coils cleared for obsessive compulsive disorder in the brain and in specific regions associated with OCD

The FDA cleared deep transcranial magnetic stimulation (Deep TMS) with the H7 coil for obsessive-compulsive disorder (OCD) treatment, following a double-blinded placebo-controlled multicenter trial. Two years later the FDA cleared TMS with the D-B80 coil on the basis of substantial equivalence. In order to investigate the induced electric field characteristics of the two coils, these were placed at the treatment position for OCD over the prefrontal cortex of a head phantom, and the field distribution was measured. Additionally, numerical simulations were performed in eight Population Head Model repository models with two sets of conductivity values and three Virtual Population anatomical head models and their homogeneous versions. The H7 was found to induce significantly higher maximal electric fields (p<0.0001, t=11.08) and to stimulate two to five times larger volumes in the brain (p<0.0001, t=6.71). The rate of decay of electric field with distance is significantly slower for the H7 coil (p < 0.0001, Wilcoxon matched-pairs test). The field at the scalp is 306% of the field at a 3 cm depth with the D-B80, and 155% with the H7 coil. The H7 induces significantly higher intensities in broader volumes within the brain and in specific brain regions known to be implicated in OCD (dorsal anterior cingulate cortex (dACC), dorsolateral prefrontal cortex (dlPFC), inferior frontal gyrus (IFG), orbitofrontal cortex (OFC) and pre-supplementary motor area (pre-SMA)) compared to the D-B80. Significant field ≥ 80 V/m is induced by the H7 (D-B80) in 15% (1%) of the dACC, 78% (29%) of the pre-SMA, 50% (20%) of the dlPFC, 30% (12%) of the OFC and 15% (1%) of the IFG. Considering the substantial differences between the two coils, the clinical efficacy in OCD should be tested and verified separately for each coil.

EEG Dipole Source Localization using Deep Neural Network

Electroencephalography (EEG) inverse problem is a typical inverse problem, in which the electrical activity within the brain is reconstructed based on EEG data collected from the scalp electrodes. In this paper, the four-layer concentric head model is used for simulation firstly, four deep neural network models including a multilayer perceptron (MLP) model and three convolutional neural networks (CNNs) are adopted to solve EEG inverse problem based on equal current dipole (ECD) model. In the simulations, 100,000 samples are generated randomly, of which 60% are used for network training and 20% are used for cross-validation. Eventually, the generalization performance of the model using the optimal function is measured by the errors in the rest 20% testing set. The experimental results show that the absolute error, relative error, mean positioning error and standard deviation of the four models are extremely low. The CNN with 6 convolutional layers and 3 pooling layers (CNN-3) is the best model. Its absolute error is about 0.015, its relative error is about 0.005, and its dipole position error is 0.040±0.029 cm. Furthermore, we use CNN-3 for source localization of the real EEG data in Working Memory. The results are in accord with physiological experience. The deep neural network method in our study needs fewer calculation parameters, takes less time, and has better positioning results.

Fast and numerically stable Mie solution of EM near field and absorption for stratified spheres

Purpose The purpose of this paper is the development of an analytic computational model for electromagnetic (EM) wave scattering from spherical objects. The main application field is the modeling of electrically large objects, where the standard numerical techniques require huge computational resources. An example is full-wave modeling of the human head in the millimeter-wave regime. Hence, an approximate model or analytical approach is used. Design/methodology/approach The Mie–Debye theorem is used for calculating the EM scattering from a layered dielectric sphere. The evaluation of the analytical expressions involved in the infinite sum has several numerical instabilities, which makes the precise calculation a challenge. The model is validated through an application example with comparing results to numerical calculations (finite element method). The human head model is used with the approximation of a two-layer sphere, where the brain tissues and the cranial bones are represented by homogeneous materials. Findings A significant improvement is introduced for the stable calculation of the Mie coefficients of a core–shell stratified sphere illuminated by a linearly polarized EM plane wave. Using this technique, a semi-analytical expression is derived for the power loss in the sphere resulting in quick and accurate calculations. Originality/value Two methods are introduced in this work with the main objective of estimating the final precision of the results. This is an important aspect for potentially unstable calculations, and the existing implementations have not included this feature so far.

A hybrid boundary element-finite element method to solve the EEG forward problem

This article presents a hybrid boundary element-finite element (BE–FE) method to solve the EEG forward problem and take advantages of both the boundary element method (BEM) and finite element method (FEM). Although realistic EEG forward problems with heterogeneous and anisotropic regions can be solved by FEM accurately, the FEM modeling of the brain with dipolar sources may lead to singularity. In contrast, the BEM can solve EEG forward problems with isotropic tissue regions and dipolar sources using a suitable integral formulation. This work utilizes both FEM and BEM strengths attained by dividing the regions into some homogeneous BE regions with sources and some heterogeneous and anisotropic FE regions. Furthermore, the BEM is applied for modeling the brain, including dipole sources and the FEM for other head layers. To validate the proposed method, inhomogeneous isotropic/anisotropic three– and four–layer spherical head models are studied. Moreover, a four&-layer realistic head model is investigated. Results for six different dipole eccentricities and two different dipole orientations are computed using the BEM, FEM, and hybrid BE–FE method together with statistical analysis and the related error criteria are compared. The proposed method is a promising new approach for solving realistic EEG forward problems.

The Impact of Head Model Choice on the In Vitro Evaluation of Aerosol Drug Delivery

There are variations in the values reported for aerosol drug delivery across in vitro experiments throughout the published literature, and often with the same devices or similar experimental setups. Factors contributing to this variability include, but are not limited to device type, equipment settings, drug type and quantification methods. This study assessed the impact of head model choice on aerosol drug delivery using six different adults and three different paediatric head models in combination with a facemask, mouthpiece, and high-flow nasal cannula. Under controlled test conditions, the quantity of drug collected varied depending on the choice of head model. Head models vary depending on a combination of structural design differences, facial features (size and structure), internal volume measurements and airway geometries and these variations result in the differences in aerosol delivery. Of the widely available head models used in this study, only three were seen to closely predict in vivo aerosol delivery performance in adults compared with published scintigraphy data. Further, this testing identified the limited utility of some head models under certain test conditions, for example, the range reported across head models was aerosol drug delivery of 2.62 ± 2.86% to 37.79 ± 1.55% when used with a facemask. For the first time, this study highlights the impact of head model choice on reported aerosol drug delivery within a laboratory setting and contributes to explaining the differences in values reported within the literature.

Modeling implanted metals in electrical stimulation applications

BackgroundMetal implants impact the dosimetry assessment in electrical stimulation techniques. Therefore, they need to be included in numerical models. While currents in the body are ionic, metals only allow electron transport. In fact, charge transfer between tissues and metals requires electric fields to drive the electrochemical reactions at the interface. Thus, metal implants may act as insulators or as conductors depending on the scenario.Objective/HypothesisThe aim of this paper is to provide a theoretical argument that guides the choice of the correct representation of metal implants using purely electrical models but considering the electrochemical nature of the problem in the technology of interest.MethodsWe built a simple model of a metal implant exposed to a homogeneous electric field of various magnitudes to represent both weak (e.g., tDCS), medium (TMS) or strong field stimulation. The same geometry was solved using two different models: a purely electric one (with different conductivities for the implant), and an electrochemical one. As an example of application, we also modeled a transcranial electrical stimulation (tES) treatment in a realistic head model with a skull plate using a high and low conductivity value for the plate.ResultsMetal implants generally act as electric insulators when exposed to electric fields up to around 100 V/m (tES and TMS range) and they only resemble a perfect conductor for fields in the order of 1000 V/m and above. The results are independent of the implant’s metal, but they depend on its geometry.Conclusion(s)Metal implants can be accurately represented by a simple electrical model of constant conductivity, but an incorrect model choice can lead to large errors in the dosimetry assessment. In particular, tES modeling with implants incorrectly treated as conductors can lead to errors of 50% in induced fields or more. Our results can be used as a guide to select the correct model in each scenario.

Small Triple-Band Meandered PIFA for Brain-Implantable Biotelemetric Systems: Development and Testing in a Liquid Phantom

We present a meandered triple-band planar-inverted-F antenna (PIFA) for integration into brain-implantable biotelemetric systems. The target applications are wireless data communication, far-field wireless power transfer, and switching control between sleep/wake-up mode at the Medical Device Radiocommunication Service (MedRadio) band (401–406 MHz) and Industrial, Scientific and Medical (ISM) bands (902–928 MHz and 2400–2483.5 MHz), respectively. By embedding meandered slots into the radiator and shorting it to the ground, we downsized the antenna to the volume of 11 × 20.5 × 1.8 mm3. We optimized the antenna using a 7-layer numerical human head model using full-wave electromagnetic field simulation. In the simulation, we placed the implant in the cerebrospinal fluid (CSF) at a depth of 13.25 mm from the body surface, which is deeper than in most works on implantable antennas. We manufactured and tested the antenna in a liquid phantom which we replicated in the simulator for further comparison. The measured gain of the antenna reached the state-of-the-art values of −43.6 dBi, −25.8 dBi, and −20.1 dBi at 402 MHz, 902 MHz, and 2400 MHz, respectively.