A Spatial-Temporal Graph Attention Network for Automated Detection and Width Estimation of Cortical Spreading Depression Using Scalp EEG.

We present an end-to-end Spatial-Temporal Graph Attention Network (STGAT) for non-invasive detection and width estimation of Cortical Spreading Depressions (CSDs) on scalp electroencephalography (EEG). Our algorithm, that we refer to as CSD Spatial-temporal graph attention network or CSD-STGAT, is trained and tested on simulated CSDs with varying width and speed ranges. Using high-density EEG, CSD-STGAT achieves less than 10.96% normalized width estimation error for narrow CSDs, with an average normalized error of 6.35%±3.08% across all widths, enabling non-invasive and automated estimation of the width of CSDs for the first time. In addition, CSD-STGAT learns the temporal and spatial features of CSDs simultaneously, which improves the "spatio-temporal tracking accuracy" (i.e., the defined detection performance metric at each electrode) of the narrow CSDs by up to 14%, compared to the state-of-the-art CSD-SpArC algorithm, with only one-tenth of the network size. CSD-STGAT achieves the best spatio-temporal tracking accuracy of 86.27%±0.53% for wide CSDs using high-density EEG, which is comparable to the performance of CSD-SpArC with less than 0.38% performance reduction. We further stitch the detections across all electrodes and over time to evaluate the "temporal accuracy". Our algorithm achieves less than 0.7% false positive rate in the simulated dataset with inter-CSD intervals ranging from 5 to 60 minutes. The lightweight architecture of CSD-STGAT paves the way towards real-time detection and parameter estimation of these waves in the brain, with significant clinical impact.

Scaling of Algorithmic Bias in Pulse Oximetry with Signal-to-Noise Ratio.

Recent work has noted a skin-color bias in existing pulse oximetry systems in their estimation of arterial oxygen saturation. Frequently, the algorithm used by these systems estimate a "ratio-of-ratios", called the "R-value", on their way to estimating the oxygen saturation. In this work, we focus on an "SNR-related" bias that is due to noise in measurements. We derive expressions for the SNR-related bias in R-value estimation, and observe how it scales with the signal-to-noise ratio (SNR). We show that the bias can arise at two steps of R-value estimation: in estimating the max and min of a pulsatile signal, and, additionally in taking ratios to estimate the R-value. We assess the bias resulting from the combination of the two steps, but also separate out contributions of each step. By doing so, we deduce that the bias induced in max and min estimation is likely to dominate. Because the SNR tends to get worse with higher melanin concentration, our result provides a sense of scaling of this bias with melanin concentration.

EEG Source Imaging of Infarct Core and Penumbra for Ischemic Stroke Patients.

A quantitative method of analyzing EEG signals after stroke onset can help monitor disease progression and tailor treatments. In this work, we present an EEG-based imaging algorithm to estimate the location and size of the stroke infarct core and penumbra tissues. Building on recent advancements in localizing neural silences, we develop an algorithm that utilizes known spectral properties of the infarct core and penumbra to separately localize them. Our algorithm uses these properties to estimate source contributions to the scalp EEG recordings in different frequency bands. Subsequently, it utilizes optimization techniques to search for the affected brain sources iteratively. We test our algorithm on simulated datasets using a realistic MRI head model, achieving center-of-mass error of 12.80mm and 17.24mm, and size estimation error of 21.78% and 36.62% for infarct core and penumbra respectively.

Effect of electric field direction on neuronal activity: an ex-vivo study.

The effect of electrical stimulation on neurons depends on the spatiotemporal properties of the applied electric field as well as on the biophysical properties of the neural tissue, which includes geometric and electrical characteristics of the cells, and the neural circuit dynamics. In this work, we characterize the effect of electric field direction on neural response in cortical layers. This can, for instance, enable more efficient (e.g., with reduced currents) and/or more selective stimulation. We stimulated mice brain slices using a recently developed brain slice platform to study transcranial currents in an ex-vivo model, where electrodes are separated from the brain slice to inject electric fields at a distance. By rotating the electrode array with respect to the slice, we changed the direction of electric field with respect to the cortical column. Our results demonstrate that in somatosensory cortex, the maximum local field potential (LFP) response is attained when the electric field is oriented parallel to the cortical column. For the same field intensity, when the field is oriented perpendicular to the cortical column, the LFP response is absent. This confirms that electric field direction is an important quantity to determine the effect of neuronal stimulation.

Noninvasive and reliable automated detection of spreading depolarization in severe traumatic brain injury using scalp EEG.

BACKGROUND: Spreading depolarizations (SDs) are a biomarker and a potentially treatable mechanism of worsening brain injury after traumatic brain injury (TBI). Noninvasive detection of SDs could transform critical care for brain injury patients but has remained elusive. Current methods to detect SDs are based on invasive intracranial recordings with limited spatial coverage. In this study, we establish the feasibility of automated SD detection through noninvasive scalp electroencephalography (EEG) for patients with severe TBI. METHODS: Building on our recent WAVEFRONT algorithm, we designed an automated SD detection method. This algorithm, with learnable parameters and improved velocity estimation, extracts and tracks propagating power depressions using low-density EEG. The dataset for testing our algorithm contains 700 total SDs in 12 severe TBI patients who underwent decompressive hemicraniectomy (DHC), labeled using ground-truth intracranial EEG recordings. We utilize simultaneously recorded, continuous, low-density (19 electrodes) scalp EEG signals, to quantify the detection accuracy of WAVEFRONT in terms of true positive rate (TPR), false positive rate (FPR), as well as the accuracy of estimating SD frequency. RESULTS: WAVEFRONT achieves the best average validation accuracy using Delta band EEG: 74% TPR with less than 1.5% FPR. Further, preliminary evidence suggests WAVEFRONT can estimate how frequently SDs may occur. CONCLUSIONS: We establish the feasibility, and quantify the performance, of noninvasive SD detection after severe TBI using an automated algorithm. The algorithm, WAVEFRONT, can also potentially be used for diagnosis, monitoring, and tailoring treatments for worsening brain injury. Extension of these results to patients with intact skulls requires further study.

Physical injury to the brain, for example due to head trauma, may worsen over time, resulting in long-term disability or death. A spreading depolarization is a slowly spreading wave in the brain, which, if detected, can be used to predict worsening brain injuries. Current methods to detect spreading depolarizations require surgeries, which are risky and unlikely to be recommended to patients with mild brain injuries. In this work, we develop an automated monitoring technique for non-surgical, non-invasive detection of spreading depolarizations, called WAVEFRONT. We validated the performance of WAVEFRONT in 12 patients with severe brain injury. Our results demonstrate the feasibility of non-invasive detection of spreading depolarizations. Our approach can potentially help clinicians predict outcomes of brain injury patients, and tailor treatments accordingly.

A model of neurovascular coupling and its application to cortical spreading depolarization.

Cortical spreading depolarization (CSD) is a neuropathological condition involving propagating waves of neuronal silence, and is related to multiple diseases, such as migraine aura, traumatic brain injury (TBI), stroke, and cardiac arrest, as well as poor outcome of patients. While CSDs of different severity share similar roots on the ion exchange level, they can lead to different vascular responses (namely spreading hyperemia and spreading ischemia). In this paper, we propose a mathematical model relating neuronal activities to predict vascular changes as measured with near-infrared spectroscopy (NIRS) and fMRI recordings, and apply it to the extreme case of CSD, where sustained near-complete neuronal depolarization is seen. We utilize three serially connected models (namely, ion exchange, neurovascular coupling, and hemodynamic model) which are described by differential equations. Propagating waves of ion concentrations, as well as the associated vasodynamics and hemodynamics, are simulated by solving these equations. Our proposed model predicts vasodynamics and hemodynamics that agree both qualitatively and quantitatively with experimental literature. Mathematical modeling and simulation offer a powerful tool to help understand the underlying mechanisms of CSD and help interpret the data. In addition, it helps develop novel monitoring techniques prior to data collection. Our simulated results strongly suggest that fMRI is unable to reliably distinguish between spreading hyperemia and spreading ischemia, while NIRS signals are substantially distinct in the two cases.

Diffuse optical tomography spatial prior for EEG source localization in human visual cortex.

Electroencephalography (EEG) and diffuse optical tomography (DOT) are imaging methods which are widely used for neuroimaging. While the temporal resolution of EEG is high, the spatial resolution is typically limited. DOT, on the other hand, has high spatial resolution, but the temporal resolution is inherently limited by the slow hemodynamics it measures. In our previous work, we showed using computer simulations that when using the results of DOT reconstruction as the spatial prior for EEG source reconstruction, high spatio-temporal resolution could be achieved. In this work, we experimentally validate the algorithm by alternatingly flashing two visual stimuli at a speed that is faster than the temporal resolution of DOT. We show that the joint reconstruction using both EEG and DOT clearly resolves the two stimuli temporally, and the spatial confinement is drastically improved in comparison to reconstruction using EEG alone.

PATHFINDER: Designing Stimulus for Neuromodulation Through Data-Driven Inverse Estimation of Non-Linear Functions.

OBJECTIVE: Using data-driven methods to design stimuli (e.g., electrical currents) which evoke desired neural responses in different neuron-types for applications in treating neural disorders. METHODS: The problem of stimulus design is formulated as estimating the inverse of a many-to-one non-linear "forward" mapping, which takes as input the parameters of waveform and outputs the corresponding neural response, directly from the data. A novel optimization framework "PATHFINDER" is proposed in order to estimate the previously mentioned inverse mapping. A comparison with existing data-driven methods, namely conditional density estimation methods and numerical inversion of an estimated forward mapping is performed with different dataset sizes in toy examples and in detailed computational models of biological neurons. RESULTS: Using data from toy examples, as well as computational models of biological neurons, we show that PATHFINDER can outperform existing methods when the number of samples is low (i.e., a few hundred). SIGNIFICANCE: Traditionally, the design of such stimuli has been model-driven and/or uses simplistic intuition, often aided by trial-and-error. Due to the inherent challenges in accurately modeling neural responses, as well as the sophistication of stimuli's effect on neural membrane potentials, data-driven approaches offer an attractive alternative. Our results suggest that PATHFINDER can be applied for optimizing stimulation parameters in experiments and treatments of neural disorders due to it requiring low number of data points.

Focused Epicranial Brain Stimulation by Spatial Sculpting of Pulsed Electric Fields Using High Density Electrode Arrays.

Transcranial electrical neuromodulation of the central nervous system is used as a non-invasive method to induce neural and behavioral responses, yet targeted non-invasive electrical stimulation of the brain with high spatial resolution remains elusive. This work demonstrates a focused, steerable, high-density epicranial current stimulation (HD-ECS) approach to evoke neural activity. Custom-designed high-density (HD) flexible surface electrode arrays are employed to apply high-resolution pulsed electric currents through skull to achieve localized stimulation of the intact mouse brain. The stimulation pattern is steered in real time without physical movement of the electrodes. Steerability and focality are validated at the behavioral, physiological, and cellular levels using motor evoked potentials (MEPs), intracortical recording, and c-fos immunostaining. Whisker movement is also demonstrated to further corroborate the selectivity and steerability. Safety characterization confirmed no significant tissue damage following repetitive stimulation. This method can be used to design novel therapeutics and implement next-generation brain interfaces.

Building a culture of responsible neurotech: Neuroethics as socio-technical challenges.

Scientists around the globe are joining the race to achieve engineering feats to read, write, modulate, and interface with the human brain in a broadening continuum of invasive to non-invasive ways. The expansive implications of neurotechnology for our conception of health, mind, decision-making, and behavior has raised social and ethical considerations that are inextricable from neurotechnological progress. We propose "socio-technical" challenges as a framing to integrate neuroethics into the engineering process. Intentionally aligning societal and engineering goals within this framework offers a way to maximize the positive impact of next-generation neurotechnologies on society.

HingePlace: Focused transcranial electrical current stimulation that allows subthreshold fields outside the stimulation target.

Transcranial Electrical Stimulation (TES) is a promising tool for treating many neurological disorders, but it classically results in diffused stimulation. Many optimization algorithms have been proposed for focusing TES, commonly by creating multi-electrode arrangements and choosing current amplitudes such that the resulting current fields in the brain are focused in the target region, and are as small as possible outside the target region. Consequently, it is likely that such optimization does not harness the non-linear nature of neural dynamics, particularly their thresholding phenomenon, i.e., the observation that neurons fire only when the stimulating currents are above a certain threshold. In this work, we propose HingePlace which explicitly harnesses this thresholding phenomenon by designing multi-electrode arrangements which allow the electric fields outside the target region to be non-zero but still below the stimulation threshold. In idealized simulated models, we compare HingePlace with existing algorithms and find that HingePlace performs strictly better, in some cases providing ~20% reduction in stimulated area for a specified limit on maximum injected current.

Projections and the Potential Societal Impact of the Future of Neurotechnologies.

Traditionally, recording from and stimulating the brain with high spatial and temporal resolution required invasive means. However, recently, the technical capabilities of less invasive and non-invasive neuro-interfacing technology have been dramatically improving, and laboratories and funders aim to further improve these capabilities. These technologies can facilitate functions such as multi-person communication, mood regulation and memory recall. We consider a potential future where the less invasive technology is in high demand. Will this demand match that the current-day demand for a smartphone? Here, we draw upon existing research to project which particular neuroethics issues may arise in this potential future and what preparatory steps may be taken to address these issues.

Abnormalities in cortical pattern of coherence in migraine detected using ultra high-density EEG.

Individuals with migraine generally experience photophobia and/or phonophobia during and between migraine attacks. Many different mechanisms have been postulated to explain these migraine phenomena including abnormal patterns of connectivity across the cortex. The results, however, remain contradictory and there is no clear consensus on the nature of the cortical abnormalities in migraine. Here, we uncover alterations in cortical patterns of coherence (connectivity) in interictal migraineurs during the presentation of visual and auditory stimuli and during rest. We used a high-density EEG system, with 128 customized electrode locations, to compare inter- and intra-hemispheric coherence in the interictal period from 17 individuals with migraine (12 female) and 18 age- and gender-matched healthy control subjects. During presentations of visual (vertical grating pattern) and auditory (modulated tone) stimulation which varied in temporal frequency (4 and 6 Hz), and during rest, participants performed a colour detection task at fixation. Analyses included characterizing the inter- and intra-hemisphere coherence between the scalp EEG channels over 2-s time intervals and over different frequency bands at different spatial distances and spatial clusters. Pearson's correlation coefficients were estimated at zero-lag. Repeated measures analyses-of-variance revealed that, relative to controls, migraineurs exhibited significantly (i) faster colour detection performance, (ii) lower spatial coherence of alpha-band activity, for both inter- and intra-hemisphere connections, and (iii) the reduced coherence occurred predominantly in frontal clusters during both sensory conditions, regardless of the stimulation frequency, as well as during the resting-state. The abnormal patterns of EEG coherence in interictal migraineurs during visual and auditory stimuli, as well as at rest (eyes open), may be associated with the cortical hyper-responsivity that is characteristic of abnormal sensory processing in migraineurs.

Neural silences can be localized rapidly using noninvasive scalp EEG.

A rapid and cost-effective noninvasive tool to detect and characterize neural silences can be of important benefit in diagnosing and treating many disorders. We propose an algorithm, SilenceMap, for uncovering the absence of electrophysiological signals, or neural silences, using noninvasive scalp electroencephalography (EEG) signals. By accounting for the contributions of different sources to the power of the recorded signals, and using a hemispheric baseline approach and a convex spectral clustering framework, SilenceMap permits rapid detection and localization of regions of silence in the brain using a relatively small amount of EEG data. SilenceMap substantially outperformed existing source localization algorithms in estimating the center-of-mass of the silence for three pediatric cortical resection patients, using fewer than 3 minutes of EEG recordings (13, 2, and 11mm vs. 25, 62, and 53 mm), as well for 100 different simulated regions of silence based on a real human head model (12 ± 0.7 mm vs. 54 ± 2.2 mm). SilenceMap paves the way towards accessible early diagnosis and continuous monitoring of altered physiological properties of human cortical function.

Quantifying a frequency modulation response biomarker in responsive neurostimulation.

Objective.Responsive neurostimulation (RNS) is an effective treatment for controlling seizures in patients with drug-resistant focal epilepsy who are not suitable candidates for resection surgery. A lack of tools for detecting and characterizing potential response biomarkers, however, contributes to a limited understanding of mechanisms by which RNS improves seizure control. We developed a method to quantify ictal frequency modulation, previously identified as a biomarker of clinical responsiveness to RNS.Approach.Frequency modulation is characterized by shifts in power across spectral bands during ictal events, over several months of neurostimulation. This effect was quantified by partitioning each seizure pattern into segments with distinct spectral content and measuring the extent of change from the baseline distribution of spectral content using the squared earth mover's distance.Main results.We analyzed intracranial electroencephalography data from 13 patients who received RNS therapy, six of whom exhibited frequency modulation on expert evaluation. Patients in the frequency modulation group had, on average, significantly larger and more sustained changes in their squared earth mover's distances (mean = 13.97 × 10-3± 1.197 × 10-3). In contrast, those patients without expert-identified frequency modulation exhibited statistically insignificant or negligible distances (mean = 4.994 × 10-3± 0.732 × 10-3).Significance.This method is the first step towards a quantitative, feedback-driven system for systematically optimizing RNS stimulation parameters, with an ultimate goal of truly personalized closed-loop therapy for epilepsy.

Effect of skull thickness and conductivity on current propagation for noninvasively injected currents.

Objective.When currents are injected into the scalp, e.g. during transcranial current stimulation, the resulting currents generated in the brain are substantially affected by the changes in conductivity and geometry of intermediate tissue. In this work, we introduce the concept of 'skull-transparent' currents, for which the changing conductivity does not significantly alter the field while propagating through the head.Approach.We establish transfer functions relating scalp currents to head potentials in accepted simplified models of the head, and find approximations for which skull-transparency holds. The current fields resulting from specified current patterns are calculated in multiple head models, including MRI heads and compared with homogeneous heads to characterize the transparency. Experimental validation is performed by measuring the current field in head phantoms.Main results.The main theoretical result is derived from observing that at high spatial frequencies, in the transfer function relating currents injected into the scalp to potential generated inside the head, the conductivity terms form a multiplicative factor and do not otherwise influence the transfer function. This observation is utilized to design injected current waveforms that maintain nearly identical focusing patterns independently of the changes in skull conductivity and thickness for a wide range of conductivity and thickness values in an idealized spherical head model as well as in a realistic MRI-based head model. Experimental measurements of the current field in an agar-based head phantom confirm the transparency of these patterns.Significance.Our results suggest the possibility that well-chosen patterns of current injection result in precise focusing inside the brain even withouta prioriknowledge of exact conductivities of intermediate layers.

Enhanced spatiotemporal resolution imaging of neuronal activity using joint electroencephalography and diffuse optical tomography.

Significance: Electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) are both commonly used methodologies for neuronal source reconstruction. While EEG has high temporal resolution (millisecond-scale), its spatial resolution is on the order of centimeters. On the other hand, in comparison to EEG, fNIRS, or diffuse optical tomography (DOT), when used for source reconstruction, can achieve relatively high spatial resolution (millimeter-scale), but its temporal resolution is poor because the hemodynamics that it measures evolve on the order of several seconds. This has important neuroscientific implications: e.g., if two spatially close neuronal sources are activated sequentially with only a small temporal separation, single-modal measurements using either EEG or DOT alone would fail to resolve them correctly. Aim: We attempt to address this issue by performing joint EEG and DOT neuronal source reconstruction. Approach: We propose an algorithm that utilizes DOT reconstruction as the spatial prior of EEG reconstruction, and demonstrate the improvements using simulations based on the ICBM152 brain atlas. Results: We show that neuronal sources can be reconstructed with higher spatiotemporal resolution using our algorithm than using either modality individually. Further, we study how the performance of the proposed algorithm can be affected by the locations of the neuronal sources, and how the performance can be enhanced by improving the placement of EEG electrodes and DOT optodes. Conclusions: We demonstrate using simulations that two sources separated by 2.3-3.3 cm and 50 ms can be recovered accurately using the proposed algorithm by suitably combining EEG and DOT, but not by either in isolation. We also show that the performance can be enhanced by optimizing the electrode and optode placement according to the locations of the neuronal sources.

Current Conceptual Understanding of the Epileptogenic Network From Stereoelectroencephalography-Based Connectivity Inferences.

Localization of the epileptogenic zone (EZ) is crucial in the surgical treatment of focal epilepsy. Recently, EEG studies have revealed that the EZ exhibits abnormal connectivity, which has led investigators to now consider connectivity as a biomarker to localize the EZ. Further, abnormal connectivity of the EZ may provide an explanation for the impact of focal epilepsy on more widespread brain networks involved in typical cognition and development. Stereo-electroencephalography (sEEG) is a well-established method for localizing the EZ that has recently been applied to examine altered brain connectivity in epilepsy. In this manuscript, we review recent computational methods for identifying the EZ using sEEG connectivity. Findings from previous sEEG studies indicate that during interictal periods, the EZ is prone to seizure generation but concurrently receives inward connectivity preventing seizures. At seizure onset, this control is lost, allowing seizure activity to spread from the EZ. Regulatory areas within the EZ may be important for subsequently ending the seizure. After the seizure, the EZ appears to regain its influence on the network, which may be how it is able to regenerate epileptiform activity. However, more research is needed on the dynamic connectivity of the EZ in order to build a biomarker for EZ localization. Such a biomarker would allow for patients undergoing sEEG to have electrode implantation, localization of the EZ, and resection in a fraction of the time currently needed, preventing patients from having to endure long hospital stays and induced seizures.

Hydrophilic Conductive Sponge Sensors for Fast Setup, Low Impedance Bio-potential Measurements.

Low electrode-skin impedance can be achieved if the interface has an electrolytic medium that allows the movement of ions across the interface. Maintaining good physical contact of the sensor with the skin is imperative. We propose a novel hydrophilic conductive sponge interface that encapsulates both of these fundamental concepts into an effective physical realization. Our implementation uses a hydrophilic polyurethane prepolymer doped with conductive carbon nanofibers and cured to form a flexible sponge material that conforms to uneven surfaces, for instance, on parts of the scalp with hair. Our results show that our sponges are able to stay in a hydrated state with a low electrode-skin impedance of around 5kΩ for more than 20 hours. The novelty in our conductive sponges also lies in their versatility: the carbon nanofibers make the electrode effective even when the electrode dries up. The sensors remain conductive with a skin impedance on the order of 20kΩ when dry, which is substantially lower than typical impedance of dry electrodes, and are able to extract alpha wave EEG activity in both wet and dry conditions.

Novel Electrodes for Reliable EEG Recordings on Coarse and Curly Hair.

EEG is a powerful and affordable brain sensing and imaging tool used extensively for the diagnosis of neurological disorders (e.g. epilepsy), brain computer interfacing, and basic neuroscience. Unfortunately, most EEG electrodes and systems are not designed to accommodate coarse and curly hair common in individuals of African descent. In neuroscience studies, this can lead to poor quality data that might be discarded in scientific studies after recording from a broader population set. In clinical diagnoses, it may lead to an uncomfortable and/or emotionally taxing experience, and, in the worst cases, misdiagnosis. Our prior work demonstrated that braiding hair in cornrows to expose the scalp at target locations leads to reduced electrode-skin impedance for existing electrodes. In this work, we design and implement novel electrodes that harness braided hair, and demonstrate that, across time, our electrodes, in conjunction with braiding, lower the impedance further, attaining 10x lower impedance than existing systems.