A 'Total Unique Variation Analysis' for Brain-Machine Interfaces.

When designing a fully implantable brain-machine interface (BMI), the primary aim is to detect as much neural information as possible with as few channels as possible. In this paper, we present a total unique variance analysis (TUVA) for evaluating the signal unique to each channel that cannot be predicted by linear combination of signals on other channels. TUVA is a statistical method for determining the total unique variance in multidimensional data, ordering channels from most to least informative, to aid in the design of maximally-efficacious BMIs. We demonstrate how this method can be applied to the design of BMIs by comparing TUVA values computed for simulated lead-field maps for high-channel-count electrocorticography (ECoG) with values computed for recordings in the interictal period in the context of surgery planning for epileptic resection.Clinical Relevance- This paper introduces a new statistical method for comparison of neural interface designs, focused on quantifying recording efficiency by minimizing channel crosstalk, which may help improve the risk-benefit profile of invasive neural recording.

Selective recording of physiologically evoked neural activity in a mixed autonomic nerve using a minimally invasive array.

Real-time closed-loop control of neuromodulation devices requires long-term monitoring of neural activity in the peripheral nervous system. Although many signal extraction methods exist, few are both clinically viable and designed for extracting small signals from fragile peripheral visceral nerves. Here, we report that our minimally invasive recording and analysis technology extracts low to negative signal to noise ratio (SNR) neural activity from a visceral nerve with a high degree of specificity for fiber type and class. Complex activity was recorded from the rat pelvic nerve that was physiologically evoked during controlled bladder filling and voiding, in an extensively characterized in vivo model that provided an excellent test bed to validate our technology. Urethane-anesthetized male rats (n = 12) were implanted with a four-electrode planar array and the bladder instrumented for continuous-flow cystometry, which measures urodynamic function by recording bladder pressure changes during constant infusion of saline. We demonstrated that differential bipolar recordings and cross-correlation analyses extracts afferent and efferent activity, and discriminated between subpopulations of fibers based on conduction velocity. Integrated Aδ afferent fiber activity correlated with bladder pressure during voiding (r2: 0.66 ± 0.06) and was not affected by activating nociceptive afferents with intravesical capsaicin (r2: 0.59 ± 0.14, P = 0.54, and n = 3). Collectively, these results demonstrate our minimally invasive recording and analysis technology is selective in extracting mixed neural activity with low/negative SNR. Furthermore, integrated afferent activity reliably correlates with bladder pressure and is a promising first step in developing closed-loop technology for bladder control.

Fractal Microelectrodes for More Energy-Efficient Cervical Vagus Nerve Stimulation.

Vagus nerve stimulation (VNS) has the potential to treat various peripheral dysfunctions, but the traditional cuff electrodes for VNS are susceptible to off-target effects. Microelectrodes may enable highly selective VNS that can mitigate off-target effects, but they suffer from the increased impedance. Recent studies on microelectrodes with non-Euclidean geometries have reported higher energy efficiency in neural stimulation applications. These previous studies use electrodes with mm/cm-scale dimensions, mostly targeted for myelinated fibers. This study evaluates fractal microelectrodes for VNS in a rodent model (N = 3). A thin-film device with fractal and circle microelectrodes is fabricated to compare their neural stimulation performance on the same radial coordinate of the nerve. The results show that fractal microelectrodes can activate C-fibers with up to 52% less energy (p = 0.012) compared to circle microelectrodes. To the best of the knowledge, this work is the first to demonstrate a geometric advantage of fractal microelectrodes for VNS in vivo.

Computational modelling of nerve stimulation and recording with peripheral visceral neural interfaces.

Objective.Neuromodulation of visceral nerves is being intensively studied for treating a wide range of conditions, but effective translation requires increasing the efficacy and predictability of neural interface performance. Here we use computational models of rat visceral nerve to predict how neuroanatomical variability could affect both electrical stimulation and recording with an experimental planar neural interface.Approach.We developed a hybrid computational pipeline,VisceralNerveEnsembleRecording andStimulation (ViNERS), to couple finite-element modelling of extracellular electrical fields with biophysical simulations of individual axons. Anatomical properties of fascicles and axons in rat pelvic and vagus nerves were measured or obtained from public datasets. To validate ViNERS, we simulated pelvic nerve stimulation and recording with an experimental four-electrode planar array.Main results.Axon diameters measured from pelvic nerve were used to model a population of myelinated and unmyelinated axons and simulate recordings of electrically evoked single-unit field potentials (SUFPs). Across visceral nerve fascicles of increasing size, our simulations predicted an increase in stimulation threshold and a decrease in SUFP amplitude. Simulated threshold changes were dominated by changes in perineurium thickness, which correlates with fascicle diameter. We also demonstrated that ViNERS could simulate recordings of electrically-evoked compound action potentials (ECAPs) that were qualitatively similar to pelvic nerve recording made with the array used for simulation.Significance.We introduce ViNERS as a new open-source computational tool for modelling large-scale stimulation and recording from visceral nerves. ViNERS predicts how neuroanatomical variation in rat pelvic nerve affects stimulation and recording with an experimental planar electrode array. We show ViNERS can simulate ECAPS that capture features of our recordings, but our results suggest the underlying NEURON models need to be further refined and specifically adapted to accurately simulate visceral nerve axons.

Rapid Analysis of Visual Receptive Fields by Iterative Tomography.

Many receptive fields in the early visual system show standard (center-surround) structure and can be analyzed using simple drifting patterns and a difference-of-Gaussians (DoG) model, which treats the receptive field as a linear filter of the visual image. But many other receptive fields show nonlinear properties such as selectivity for direction of movement. Such receptive fields are typically studied using discrete stimuli (moving or flashed bars and edges) and are modelled according to the features of the visual image to which they are most sensitive. Here, we harness recent advances in tomographic image analysis to characterize rapidly and simultaneously both the linear and nonlinear components of visual receptive fields. Spiking and intracellular voltage potential responses to briefly flashed bars are analyzed using non-negative matrix factorization (NNMF) and iterative reconstruction tomography (IRT). The method yields high-resolution receptive field maps of individual neurons and neuron ensembles in primate (marmoset, both sexes) lateral geniculate and rodent (mouse, male) retina. We show that the first two IRT components correspond to DoG-equivalent center and surround of standard [magnocellular (M) and parvocellular (P)] receptive fields in primate geniculate. The first two IRT components also reveal the spatiotemporal receptive field structure of nonstandard (on/off-rectifying) receptive fields. In rodent retina we combine NNMF-IRT with patch-clamp recording and dye injection to directly map spatial receptive fields to the underlying anatomy of retinal output neurons. We conclude that NNMF-IRT provides a rapid and flexible framework for study of receptive fields in the early visual system.

Preliminary Minimum Reporting Requirements for In-Vivo Neural Interface Research: I. Implantable Neural Interfaces.

The pace of research and development in neuroscience, neurotechnology, and neurorehabilitation is rapidly accelerating, with the number of publications doubling every 4.2 years. Maintaining this progress requires technological standards and scientific reporting guidelines to provide frameworks for communication and interoperability. The present lack of such neurotechnology standards limits the transparency, repro-ducibility, and meta-analysis of this growing body of literature, posing an ongoing barrier to research, clinical, and commercial objectives. Continued neurotechnological innovation requires the development of some minimal standards to promote integration between this broad spectrum of technologies and therapies. To preserve design freedom and accelerate the translation of research into safe and effective technologies with maximal user benefit, such standards must be collaboratively co-developed by the full range of neuroscience and neurotechnology stakeholders. This paper summarizes the preliminary recommendations of IEEE P2794 Standards Working Group, developing a Reporting Standard for in-vivo Neural Interface Research (RSNIR).

Simulating bidirectional peripheral neural interfaces in EIDORS.

Bioelectronic neural interfaces that deliver adaptive therapeutic stimulation in an intelligent manner must be able to sense and stimulate activity within the same nerve. Existing minimally-invasive peripheral neural interfaces can provide a read-out of the aggregate level of activity via electrical recordings of nerve activity, but these recordings are limited in terms of their specificity. Computational simulations can provide fine-grained insight into the contributions of different neural populations to the extracellular recording, but integration of the signals from individual nerve fibers requires knowledge of spread of current in the complex (heterogenous, anisotropic) extracellular space. We have developed a model which uses the open-source EIDORS package for extracellular stimulation and recording in the pelvic nerve. The pelvic nerve is the primary source of autonomic innervation to the pelvic organs, and a prime target for electrical stimulation to treat a variety of voiding disorders. We simulated recordings of spontaneous and electrically-evoked activity using biophysical models for myelinated and unmyelinated axons. As expected, stimulus thresholds depended strongly on both fibre type and electrode-fibre distance. In conclusion, EIDORS can be used to accurately simulate extracellular recording in complex, heterogenous neural geometries.

Recording of Electrically Evoked Neural Activity and Bladder Pressure Responses in Awake Rats Chronically Implanted With a Pelvic Nerve Array.

Bioelectronic medical devices are well established and widely used in the treatment of urological dysfunction. Approved targets include the sacral S3 spinal root and posterior tibial nerve, but an alternate target is the group of pelvic splanchnic nerves, as these contain sacral visceral sensory and autonomic motor pathways that coordinate storage and voiding functions of the bladder. Here, we developed a device suitable for long-term use in an awake rat model to study electrical neuromodulation of the pelvic nerve (homolog of the human pelvic splanchnic nerves). In male Sprague-Dawley rats, custom planar four-electrode arrays were implanted over the distal end of the pelvic nerve, close to the major pelvic ganglion. Electrically evoked compound action potentials (ECAPs) were reliably detected under anesthesia and in chronically implanted, awake rats up to 8 weeks post-surgery. ECAP waveforms showed three peaks, with latencies that suggested electrical stimulation activated several subpopulations of myelinated A-fiber and unmyelinated C-fiber axons. Chronic implantation of the array did not impact on voiding evoked in awake rats by continuous cystometry, where void parameters were comparable to those published in naïve rats. Electrical stimulation with chronically implanted arrays also induced two classes of bladder pressure responses detected by continuous flow cystometry in awake rats: voiding contractions and non-voiding contractions. No evidence of tissue pathology produced by chronically implanted arrays was detected by immunohistochemical visualization of markers for neuronal injury or noxious spinal cord activation. These results demonstrate a rat pelvic nerve electrode array that can be used for preclinical development of closed loop neuromodulation devices targeting the pelvic nerve as a therapy for neuro-urological dysfunction.

Binocular summation in marmoset lateral geniculate nucleus.

In primates and carnivores, the main laminae of the dorsal lateral geniculate nucleus (LGN) receive monocular excitatory input in an eye-alternating fashion. There is also evidence that nondominant eye stimulation can reduce responses to dominant eye stimulation and that a subset of LGN cells in the koniocellular (K) layers receives convergent binocular excitatory input from both eyes. What is not known is how the two eye inputs summate in the K layers of LGN. Here, we aimed to answer this question by making extracellular array electrode recordings targeted to K layers in the marmoset (Callithrix jacchus) LGN, as visual stimuli (flashed 200 ms temporal square-wave pulses or drifting gratings) were presented to each eye independently or to both eyes simultaneously. We found that when the flashed stimulus was presented to both eyes, compared to the dominant eye, the peak firing rate of most cells (61%, 14/23) was reduced. The remainder showed response facilitation (17%) or partial summation (22%). A greater degree of facilitation was seen when the total number of spikes across the stimulus time window (200 ms) rather than peak firing rates was measured. A similar pattern of results was seen for contrast-varying gratings and for small numbers of parvocellular (n = 12) and magnocellular (n = 3) cells recorded. Our findings show that binocular summation in the marmoset LGN is weak and predominantly sublinear in nature.

Receptive Field Properties of Koniocellular On/Off Neurons in the Lateral Geniculate Nucleus of Marmoset Monkeys.

The koniocellular (K) layers of the primate dorsal lateral geniculate nucleus house a variety of visual receptive field types, not all of which have been fully characterized. Here we made single-cell recordings targeted to the K layers of diurnal New World monkeys (marmosets). A subset of recorded cells was excited by both increments and decrements of light intensity (on/off-cells). Histological reconstruction of the location of these cells confirmed that they are segregated to K layers; we therefore refer to these cells as K-on/off cells. The K-on/off cells show high contrast sensitivity, strong bandpass spatial frequency tuning, and their response magnitude is strongly reduced by stimuli larger than the excitatory receptive field (silent suppressive surrounds). Stationary counterphase gratings evoke unmodulated spike rate increases or frequency-doubled responses in K-on/off cells; such responses are largely independent of grating spatial phase. The K-on/off cells are not orientation or direction selective. Some (but not all) properties of K-on/off cells are consistent with those of local-edge-detector/impressed-by-contrast cells reported in studies of cat retina and geniculate, and broad-thorny ganglion cells recorded in macaque monkey retina. The receptive field properties of K-on/off cells and their preferential location in the ventral K layers (K1 and K2) make them good candidates for the direct projection from geniculate to extrastriate cortical area MT/V5. If so, they could contribute to visual information processing in the dorsal ("where" or "action") visual stream.SIGNIFICANCE STATEMENT We characterize cells in an evolutionary ancient part of the visual pathway in primates. The cells are located in the lateral geniculate nucleus (the main visual afferent relay nucleus), in regions called koniocellular layers that are known to project to extrastriate visual areas as well as primary visual cortex. The cells show high contrast sensitivity and rapid, transient responses to light onset and offset. Their properties suggest they could contribute to visual processing in the dorsal ("where" or "action") visual stream.

Multipolar Field Shaping in a Suprachoroidal Visual Prosthesis.

Visual prostheses are now an available mobility aid for patients blinded by degenerative retinal diseases. However, the spatial resolution of existing devices is still insufficient to deliver normal levels of mobility vision without stimulation strategies, which enable existing devices to deliver several different percepts per stimulation site. A stimulation strategy, in which field shaping is achieved by incorporating multipolar (bipolar and tripolar) stimulation could convey additional information to a user of a visual prosthesis, as compared with monopolar stimulation, is investigated. Electrical stimulus response thresholds were simulated using morphologically and physiologically accurate cable models of human retinal ganglion cell (RGC) axons. From the population response patterns which could be evoked in simulation, multipolar field-shaping stimulation from one location could convey as much information as monopolar array stimulation. This result is confirmed in vitro by applying a Bayesian classification analysis to multielectrode array recordings of RGC population responses to extracellular stimulation. In vitro recorded population responses to individual stimuli in vitro could be used to train a Baysian classifier, which could correctly identify individual stimuli as predicted by the simulated population responses. In both simulation and in vitro experiments, monopolar thresholds were not significantly different to multipolar thresholds.

A spectral element method with adaptive segmentation for accurately simulating extracellular electrical stimulation of neurons.

The capacity to quickly and accurately simulate extracellular stimulation of neurons is essential to the design of next-generation neural prostheses. Existing platforms for simulating neurons are largely based on finite-difference techniques; due to the complex geometries involved, the more powerful spectral or differential quadrature techniques cannot be applied directly. This paper presents a mathematical basis for the application of a spectral element method to the problem of simulating the extracellular stimulation of retinal neurons, which is readily extensible to neural fibers of any kind. The activating function formalism is extended to arbitrary neuron geometries, and a segmentation method to guarantee an appropriate choice of collocation points is presented. Differential quadrature may then be applied to efficiently solve the resulting cable equations. The capacity for this model to simulate action potentials propagating through branching structures and to predict minimum extracellular stimulation thresholds for individual neurons is demonstrated. The presented model is validated against published values for extracellular stimulation threshold and conduction velocity for realistic physiological parameter values. This model suggests that convoluted axon geometries are more readily activated by extracellular stimulation than linear axon geometries, which may have ramifications for the design of neural prostheses.

The Effect of Electric Cross-Talk in Retinal Neurostimulation.

PURPOSE: To investigate the efficacy of electric field shaping in modulating the extent and activation threshold in retinal neurostimulation. This study aims to quantify the interference of neighboring stimulation sites by assessing the shift in the activation threshold produced by a concomitant interfering stimulus. METHODS: Electrical stimuli were applied to healthy retinae in a feline model (n = 4) using a 24-channel electrode array surgically implanted in the suprachoroidal space. A 96-channel penetrating electrode array was used for recording cortical responses to a number of stimulation paradigms. Data were analyzed offline. Concurrent monopolar and hexapolar stimuli were delivered at primary and interfering sites separated by up to 2.19 mm to evaluate electric cross-talk. The spike rate was fit to a sigmoidal curve to estimate the P50 threshold. The slope of the linear regression of the P50 value versus interfering current level was considered as a measure of cross-talk. RESULTS: Concurrent monopolar stimulation produced a proportional drop in the P50 of approximately 20% of the interfering current level in presence of a primary monopolar and hexapolar stimulus. On the other hand, hexapolar interference did not alter activation thresholds at the primary site. CONCLUSIONS: Hexapolar stimulation reduces electric cross-talk between neighboring sites and represents a technique to reduce interference between individual stimulation sites. In contrast, concurrent monopolar stimulation produces a reduction of the activation threshold of stimuli delivered nearby. Thus, a single source of subthreshold monopolar charge injection can provide benefit in the form of significant threshold reduction simultaneously at multiple stimulation sites.

Electrically evoked potentials in an ovine model for the evaluation of visual prosthesis efficacy.

Visual prostheses are becoming a reality as a therapy to restore functional vision to the blind. New stimulation strategies and novel electrode designs are contributing to accelerate the development of such devices triggering the interest of scientists, clinicians and the blind community worldwide. In this scenario, there is a need for large animal models that are suitable for preclinical testing of retinal neuroprostheses. This study presents an electrophysiology assessment of an ovine model for single and simultaneous electrode stimulation from the suprachoroidal space, using symmetric biphasic current pulses with a monopolar return configuration. Visually and electrically evoked potentials were recorded using supradural surface electrodes, showing charge thresholds comparable to those in humans. This model represents an alternative to feline or canine models with analogous activation levels and an eye anatomy similar to that of humans.

A cortical integrate-and-fire neural network model for blind decoding of visual prosthetic stimulation.

We present a computational model of the optic pathway which has been adapted to simulate cortical responses to visual-prosthetic stimulation. This model reproduces the statistically observed distributions of spikes for cortical recordings of sham and maximum-intensity stimuli, while simultaneously generating cellular receptive fields consistent with those observed using traditional visual neuroscience methods. By inverting this model to generate candidate phosphenes which could generate the responses observed to novel stimulation strategies, we hope to aid the development of said strategies in-vivo before being deployed in clinical settings.

Attaining higher resolution visual prosthetics: a review of the factors and limitations.

Visual prosthetics is an expanding subfield of functional electrical stimulation which has gained increased interest recently in light of new advances in treatments and technology. These treatments and technology represent a major improvement over prior art, but are still subject to a host of limitations which are dependent on the manner in which one approaches the topic of visual prosthetics. These limitations pose new research challenges whose solutions are directly applicable to the well-being of blind individuals everywhere. In this review, we will outline and critically compare major current approaches to visual prosthetics, and in particular retinal prosthetics. Then, we will engage in an in-depth discussion of the limitations imposed by current technology, physics, and the underlying biology of the retina to highlight several of the challenges currently facing researchers.