Paper Submission & Registration
8th Dutch Bio-Medical Engineering Conference
12:30   Brain - I
Chair: Ying Wang
15 mins
Diagnostic value of stimulation-induced epileptic seizures: concordance with spontaneous seizures and association with clinical outcome: a stereo-electroencephalography study
Daniel Kreiter, Raf van Hoof, Karolina Gasztych, Balu Krishnan, Olaf Schijns, Louis Wagner, Albert Colon, Simon Tousseyn
Abstract: RATIONALE Epileptic seizures may be induced by cortical electrical stimulation during stereo-electroencephalography (SEEG). These seizures may aid in epileptogenic zone (EZ) identification. Currently, the exact concordance between spontaneous and stimulation-induced seizures is unknown. Aim of our study was to quantitatively determine electrical concordance between spontaneous and stimulation-induced seizures to improve understanding of the diagnostic value of stimulation-induced seizures. METHODS 39 consecutive SEEG registrations from 36 patients (June 2015 – May 2020) with at least one spontaneous and one stimulation-induced electroclinical seizure were retrospectively included. Spectral decomposition was performed on SEEG signals for the complete seizure period. Ictal changes of SEEG power compared to a baseline period were calculated. Pearson correlation coefficients were obtained between spontaneous seizures mutually and between spontaneous and stimulation-induced seizures. We tested whether stimulation in- or outside the clinically defined seizure-onset zone (SOZ) yields seizures with higher correlation to spontaneous seizures and if concordance was associated with postsurgical (radiofrequency thermocoagulation and/or resective surgery) outcome (ILAE class 1-2 versus >2). RESULTS Spontaneous seizures (n=721) showed good concordance per registration (median: 0.68) with similar concordance between outcome groups. Stimulation-induced seizures (n=120) with high electrical concordance (ρ>0.7) to spontaneous seizures could be elicited in 67% of the study population. Concordance for stimulation in- or outside the spontaneous SOZ was similar. Maximal concordance per SEEG was not significantly different in the good (n=13) compared to the poor (n=17) outcome group for stimulation inside the spontaneous SOZ (median ρ 0.66 vs. 0.88, p=0.16), but was significantly lower for stimulation outside the spontaneous SOZ (0.30 vs. 0.88, p<0.01). In receiver operating characteristic (ROC) analysis, highly correlated seizures stimulated outside spontaneous SOZ proved to be predictive for poor outcome (area under curve: 0.86, 95%CI: 0.71-0.97). CONCLUSIONS Stimulation-induced seizures with high electrical concordance to spontaneous seizures can be elicited. Highly concordant seizures resulting from stimulation outside spontaneous SOZ were associated with poor outcome. Results suggest that seizure-inducing contacts of highly concordant stimulation-induced seizures may be part of EZ.
15 mins
Validation of multipin, dry EEG electrodes
Janne Heijs, Jurjan Havelaar, Patrique Fiedler, Richard van Wezel, Ciska Heida
Abstract: Current developments towards dry electrodes in the field of electroencephalography (EEG) are promising for research in non-laboratory environments. In contrast to conventional gel electrodes, multipin dry-EEG electrodes do not require the application of conductive gel, which is time consuming, requires trained staff and, therefore, confines the use of gel-EEG systems to the laboratory environment. Multipin, dry electrodes consist of multiple pins that pass through the hair layer to establish a physical connection with the scalp. Fiedler et al. (2016, 2018) optimized the number of pins per electrode, the pin length and pin flexibility, to improve signal quality and wearing comfort. The aim of the current study is to validate these multipin, dry-EEG electrodes by comparing the performance with conventional gel-EEG electrodes. Healthy volunteers participated in two experimental sessions with either the 32-channel gel-EEG system, or the 32-channel multipin, dry-EEG system. Three tasks were performed to evaluate signal quality in time and frequency domain: 1) the 40-Hz Auditory Steady-State Response (ASSR) to evaluate power spectrum, signal-to-noise ratio (SNR) and the global field power in time domain (GFPt) of the auditory evoked potential; 2) the Checkerboard pattern reversal paradigm to evaluate GFPt of the visual evoked potential; and 3) the Eyes-Open/Eyes-Closed paradigm to evaluate power spectra in the alpha-band. Moreover, application time and experienced wearing comfort were reported. Preliminary results of four subjects, showed a reduced wearing comfort for the dry-EEG system compared to the gel-EEG system. Moderate to high Spearman’s rank correlation coefficients were found for GFPt of the auditory evoked potential (ρ=0.36, p<0.001) and visual evoked potential (ρ=0.69, p<0.001). No differences were found between the two systems for average alpha-band power (p=0.524; Eyes Open/Closed) and 40-Hz peak amplitude (p=0.250; ASSR). SNR was not significantly (p=0.307), decreased for the dry-EEG system (6.57 ± 7.38) compared to the gel-EEG system (7.48 ± 8.23). Preliminary results suggest that similar task-related signals are observed for the dry-EEG system and gel-EEG system, in terms of amplitude, time, spectral power and signal-to-noise ratios.
15 mins
EEG-in-a-dish towards precision medicine for epilepsy
Areti Sfakianou, Eline van Hugte, Nael Nadif Kasri, Michel J.A.M. van Putten, Monica Frega
Abstract: Dravet Syndrome (DS) and Generalized Epilepsy with Febrile Seizures Plus (GEFS+) are epileptic, infantile disorders presenting de novo mutations in gene SCN1A [1]. Human induced Pluripotent Stem Cell (hiPSC) models of epilepsy are a revolutionary platform for mechanistic studies and drug discoveries [2]. Furthermore, neuronal activity can be measured with the use of micro-electrode arrays (MEA). MEA data analysis focuses on high frequency spike data (100-3500 Hz). However, little is known whether MEAs can detect local field potentials (LFP, 0.5-50 Hz) originating from synaptic currents, observed on EEG. In this project, we developed an epilepsy human model to detect both spikes and LFPs and study patient-specific phenotypes. HiPSCs generated from blood of healthy individuals and patients with DS or GEFS+ were differentiated into excitatory neurons via the forced expression of Ngn2 [3] and were grown on 24-well MEAs (12 electrodes/well). We performed 10 min recordings on a commercial MEA System in a broad frequency band (0.5-3500 Hz) to extract both spikes and LFPs. In addition, we established an analysis tool for LFP detection in MATLAB [4]. Features of interest were firing rate and synchronization of spikes, LFPs’ frequency and their waveforms’ characteristics, as well as power spectral density (PSD). We found a lower firing and synchronized activity in neuronal networks derived from patients as compared to control (100-3500 Hz). In addition, we achieved to detect LFPs which were significantly lower in neuronal networks derived from patients than control. Although LFPs’ duration appeared the same in all cell lines (3-4 seconds), their peak-to-peak amplitude fluctuated up to 1.8 times higher in control compared to diseased lines. Moreover, we found a lower PSD on diseased cells compared to control at frequencies below 10 Hz confirming the low LFPs’ frequency of diseased cells. Overall, we achieved to detect neuronal activity below 50 Hz. Activity exhibited by DS and GEFS+ neuronal networks was significantly lower in all frequency band compared to control. The extraction of high frequency spike data in parallel with novel LFP parameters allow us to better describe patient-specific phenotypes and improve the characterization and evaluation of treatment strategies.
15 mins
A method to experimentally estimate the conductivity of chronic stroke lesions
Joris van der Cruijsen, Maria Carla Piastra, Ruud Selles, Thom Oostendorp
Abstract: The anatomical properties of stroke lesions, such as volume and location, are acknowledged sources of inter-subject variability that are taken into account in the design of transcranial electric stimulation aimed to maximally stimulate specific brain areas. However, visual inspection of MRI images of stroke subjects indicates that the content of the lesion, and therefore the conductive properties, are an additional source of inter-subject variability that is currently overlooked. The goal of our study was to show through simulations that the conductivity of stroke lesions can be estimated by applying transcranial stimulation and simultaneously record the related scalp potentials. We started with an MRI-based three-layer (scalp, skull, brain) boundary element head model of a healthy subject, from which we created two additional models with a 180 ml or a 35 ml central cortical lesion. The assigned conductivities were 0.225, 0.015, 0.225, and 0.45 S/m for scalp, skull, brain, and lesion, respectively. 128 electrodes were modelled on the scalp, of which 20 were considered as stimulation electrodes. We identified the electrode pair whose scalp potentials were most affected by the presence of the lesion. Then, for this electrode pair, we simulated scalp potentials in a model with a fixed lesion conductivity of 0.45 S/m and in a second model, in which a random lesion conductivity was set. We varied the conductivity in the second model to minimize the error between the models. We performed this for 8, 16, 32, 64 and 128 recording electrodes and added displacements on the electrode positions to reflect experimental conditions. We found that, in absence of electrode position displacements, the optimisation algorithm converged to the correct lesion conductivity for all electrode numbers. For mean displacement errors below 1 mm, a relative estimation error below 5% was reached. These findings show that our proposed method can be used to estimate stroke lesion conductivity. The method can easily be implemented in experimental procedures that already combine transcranial electric stimulation and EEG. Adding this method may improve the accuracy of volume conductor models of stroke patients and lead to more effective stimulation configurations and treatment for this population.

end %-->