A novel high electrode count spike recording array using an 81,920 pixel transimpedance amplifier-based imaging chip

Extracellular Recording of Entire Neural Networks Using a Dual-Mode Microelectrode Array With 19584 Electrodes and High SNR

Electrophysiological research on neural networks and their activity focuses on the recording and analysis of large data sets that include information of thousands of neurons. CMOS microelectrode arrays (MEAs) feature thousands of electrodes at a spatial resolution on the scale of single cells and are, therefore, ideal tools to support neural-network research. Moreover, they offer high spatio-temporal resolution and signal to-noise ratio (SNR) to capture all features and subcellular resolution details of neuronal signaling. Here, we present a dual-mode (DM) MEA, which enables simultaneous: 1) full-frame readout from all electrodes and 2) high-SNR readout from an arbitrarily selectable subset of electrodes. The DM-MEA includes 19584 electrodes, 19584 full-frame recording channels with noise levels of 10.4 μV_rms in the action potential (AP) frequency band (300 Hz-5 kHz), 246 low-noise recording channels with noise levels of 3.0 μV_rms in the AP band and eight stimulation units. The capacity to simultaneously perform full-frame and high-SNR recordings endows the presented DM-MEA with great flexibility for various applications in neuroscience and pharmacology.

Mapping of primary somatosensory cortex of the hand area using a high-density electrocorticography grid for closed-loop brain computer interface

The ideal modality for generating sensation in sensorimotor brain computer interfaces (BCI) has not been determined. Here we report the feasibility of using a high-density "mini"-electrocorticography (mECoG) grid in a somatosensory BCI system. Thirteen subjects with intractable epilepsy underwent standard clinical implantation of subdural electrodes for the purpose of seizure localization. An additional high-density mECoG grid was placed (Adtech, 8 by 8, 1.2-mm exposed, 3-mm center-to-center spacing) over the hand area of primary somatosensory cortex. Following implantation, cortical mapping was performed with stimulation parameters of frequency: 50 Hz, pulse-width: 250 µs, pulse duration: 4 s, polarity: alternating, and current that ranged from 0.5 mA to 12 mA at the discretion of the epileptologist. Location of the evoked sensory percepts was recorded along with a description of the sensation. The hand was partitioned into 48 distinct boxes. A box was included if sensation was felt anywhere within the box. The percentage of the hand covered was 63.9% (± 34.4%) (mean ± s.d.). Mean redundancy, measured as electrode pairs stimulating the same box, was 1.9 (± 2.2) electrodes per box; and mean resolution, measured as boxes included per electrode pair stimulation, was 11.4 (± 13.7) boxes with 8.1 (± 10.7) boxes in the digits and 3.4 (± 6.0) boxes in the palm. Functional utility of the system was assessed by quantifying usable percepts. Under the strictest classification, "dermatomally exclusive" percepts, the mean was 2.8 usable percepts per grid. Allowing "perceptually unique" percepts at the same anatomical location, the mean was 5.5 usable percepts per grid. Compared to the small area of coverage and redundancy of a microelectrode system, or the poor resolution of a standard ECoG grid, a mECoG is likely the best modality for a somatosensory BCI system with good coverage of the hand and minimal redundancy.

Technical considerations for generating somatosensation via cortical stimulation in a closed-loop sensory/motor brain-computer interface system in humans

Somatosensory feedback is the next step in brain computer interface (BCI). Here, we compare three cortical stimulating array modalities for generating somatosensory percepts in BCI. We compared human subjects with either a 64-channel "mini"-electrocorticography grid (mECoG; 1.2-mm diameter exposed contacts with 3-mm spacing, N = 1) over the hand area of primary somatosensory cortex (S1), or a standard grid (sECoG; 1.5-mm diameter exposed contacts with 1-cm spacing, N = 1), to generate artificial somatosensation through direct electrical cortical stimulation. Finally, we reference data in the literature from a patient implanted with microelectrode arrays (MEA) placed in the S1 hand area. We compare stimulation results to assess coverage and specificity of the artificial percepts in the hand. Using the mECoG array, hand mapping revealed coverage of 41.7% of the hand area versus 100% for the sECoG array, and 18.8% for the MEA. On average, stimulation of a single electrode corresponded to sensation reported in 4.42 boxes (range 1-11 boxes) for the mECoG array, 19.11 boxes (range 4-48 boxes) for the sECoG grid, and 2.3 boxes (range 1-5 boxes) for the MEA. Sensation in any box, on average, corresponded to stimulation from 2.65 electrodes (range 1-5 electrodes) for the mECoG grid, 3.58 electrodes for the sECoG grid (range 2-4 electrodes), and 11.22 electrodes (range 2-17 electrodes) for the MEA. Based on these findings, we conclude that mECoG grids provide an excellent balance between spatial cortical coverage of the hand area of S1 and high-density resolution.

Automatic spike sorting for high-density microelectrode arrays

High-density microelectrode arrays can be used to record extracellular action potentials from hundreds to thousands of neurons simultaneously. Efficient spike sorters must be developed to cope with such large data volumes. Most existing spike sorting methods for single electrodes or small multielectrodes, however, suffer from the "curse of dimensionality" and cannot be directly applied to recordings with hundreds of electrodes. This holds particularly true for the standard reference spike sorting algorithm, principal component analysis-based feature extraction, followed by k-means or expectation maximization clustering, against which most spike sorters are evaluated. We present a spike sorting algorithm that circumvents the dimensionality problem by sorting local groups of electrodes independently with classical spike sorting approaches. It is scalable to any number of recording electrodes and well suited for parallel computing. The combination of data prewhitening before the principal component analysis-based extraction and a parameter-free clustering algorithm obviated the need for parameter adjustments. We evaluated its performance using surrogate data in which we systematically varied spike amplitudes and spike rates and that were generated by inserting template spikes into the voltage traces of real recordings. In a direct comparison, our algorithm could compete with existing state-of-the-art spike sorters in terms of sensitivity and precision, while parameter adjustment or manual cluster curation was not required. NEW & NOTEWORTHY We present an automatic spike sorting algorithm that combines three strategies to scale classical spike sorting techniques for high-density microelectrode arrays: 1) splitting the recording electrodes into small groups and sorting them independently; 2) clustering a subset of spikes and classifying the rest to limit computation time; and 3) prewhitening the spike waveforms to enable the use of parameter-free clustering. Finally, we combined these strategies into an automatic spike sorter that is competitive with state-of-the-art spike sorters.

Neural Parallel Engine: A toolbox for massively parallel neural signal processing

BACKGROUND

Large-scale neural recordings provide detailed information on neuronal activities and can help elicit the underlying neural mechanisms of the brain. However, the computational burden is also formidable when we try to process the huge data stream generated by such recordings.

NEW METHOD

In this study, we report the development of Neural Parallel Engine (NPE), a toolbox for massively parallel neural signal processing on graphical processing units (GPUs). It offers a selection of the most commonly used routines in neural signal processing such as spike detection and spike sorting, including advanced algorithms such as exponential-component-power-component (EC-PC) spike detection and binary pursuit spike sorting. We also propose a new method for detecting peaks in parallel through a parallel compact operation.

RESULTS

Our toolbox is able to offer a 5× to 110× speedup compared with its CPU counterparts depending on the algorithms. A user-friendly MATLAB interface is provided to allow easy integration of the toolbox into existing workflows.

COMPARISON WITH EXISTING METHODS

Previous efforts on GPU neural signal processing only focus on a few rudimentary algorithms, are not well-optimized and often do not provide a user-friendly programming interface to fit into existing workflows. There is a strong need for a comprehensive toolbox for massively parallel neural signal processing.

CONCLUSIONS

A new toolbox for massively parallel neural signal processing has been created. It can offer significant speedup in processing signals from large-scale recordings up to thousands of channels.

Statistically Reconstructed Multiplexing for Very Dense, High-Channel-Count Acquisition Systems

Multiplexing is an important strategy in multichannel acquisition systems. The per-channel antialiasing filters needed in the traditional multiplexing architecture limit its scalability for applications requiring high channel density, high channel count, and low noise. A particularly challenging example is multielectrode arrays for recording from neural systems. We show that conventional approaches must tradeoff recording density and noise performance, at a scale far from the ideal goal of one-to-one mapping between neurons and sensors. We present a multiplexing architecture without per-channel antialiasing filters. The sparsely sampled data are recovered through a compressed sensing strategy, involving statistical reconstruction and removal of the undersampled thermal noise. In doing so, we replace large analog components with digital signal processing blocks, which are much more amenable to scaled CMOS implementation. The resulting statistically reconstructed multiplexing architecture recovers input signals at significantly improved signal-to-noise ratios when compared to conventional multiplexing with antialiasing filters at the same per-channel area. We implement the new architecture in a 65 536-channel neural recording system and show that it is able to recover signals with performance comparable to conventional high-performance, single-channel systems, despite a more than four-orders-of-magnitude increase in channel density.

A very large-scale microelectrode array for cellular-resolution electrophysiology

In traditional electrophysiology, spatially inefficient electronics and the need for tissue-to-electrode proximity defy non-invasive interfaces at scales of more than a thousand low noise, simultaneously recording channels. Using compressed sensing concepts and silicon complementary metal-oxide-semiconductors (CMOS), we demonstrate a platform with 65,536 simultaneously recording and stimulating electrodes in which the per-electrode electronics consume an area of 25.5 μm by 25.5 μm. Application of this platform to mouse retinal studies is achieved with a high-performance processing pipeline with a 1 GB/s data rate. The platform records from 65,536 electrodes concurrently with a ~10 µV r.m.s. noise; senses spikes from more than 34,000 electrodes when recording across the entire retina; automatically sorts and classifies greater than 1700 neurons following visual stimulation; and stimulates individual neurons using any number of the 65,536 electrodes while observing spikes over the entire retina. The approaches developed here are applicable to other electrophysiological systems and electrode configurations.

Large electronics limit low-noise, non-invasive electrophysiological measurements to a thousand simultaneously recording channels. Here the authors build an array of 65k simultaneously recording and stimulating electrodes and use it to sort and classify single neurons across the entire mouse retina.

Inflammatory cytokine-induced changes in neural network activity measured by waveform analysis of high-content calcium imaging in murine cortical neurons

During acute neuroinflammation, increased levels of cytokines within the brain may contribute to synaptic reorganization that results in long-term changes in network hyperexcitability. Indeed, inflammatory cytokines are implicated in synaptic dysfunction in epilepsy and in an array of degenerative and autoimmune diseases of the central nervous system. Current tools for studying the impact of inflammatory factors on neural networks are either insufficiently fast and sensitive or require complicated and costly experimental rigs. Calcium imaging offers a reasonable surrogate for direct measurement of neuronal network activity, but traditional imaging paradigms are confounded by cellular heterogeneity and cannot readily distinguish between glial and neuronal calcium transients. While the establishment of pure neuron cultures is possible, the removal of glial cells ignores physiologically relevant cell-cell interactions that may be critical for circuit level disruptions induced by inflammatory factors. To overcome these issues, we provide techniques and algorithms for image processing and waveform feature extraction using automated analysis of spontaneous and evoked calcium transients in primary murine cortical neuron cultures transduced with an adeno-associated viral vector driving the GCaMP6f reporter behind a synapsin promoter. Using this system, we provide evidence of network perturbations induced by the inflammatory cytokines TNFα, IL1β, and IFNγ.

Spatiotemporal norepinephrine mapping using a high-density CMOS microelectrode array

A high-density amperometric electrode array containing 8192 individually addressable platinum working electrodes with an integrated potentiostat fabricated using Complementary Metal Oxide Semiconductor (CMOS) processes is reported. The array was designed to enable electrochemical imaging of chemical gradients with high spatiotemporal resolution. Electrodes are arranged over a 2 mm × 2 mm surface area into 64 subarrays consisting of 128 individual Pt working electrodes as well as Pt pseudo-reference and auxiliary electrodes. Amperometric measurements of norepinephrine in tissue culture media were used to demonstrate the ability of the array to measure concentration gradients in complex media. Poly(dimethylsiloxane) microfluidics were incorporated to control the chemical concentrations in time and space, and the electrochemical response at each electrode was monitored to generate electrochemical heat maps, demonstrating the array's imaging capabilities. A temporal resolution of 10 ms can be achieved by simultaneously monitoring a single subarray of 128 electrodes. The entire 2 mm × 2 mm area can be electrochemically imaged in 64 seconds by cycling through all subarrays at a rate of 1 Hz per subarray. Monitoring diffusional transport of norepinephrine is used to demonstrate the spatiotemporal resolution capabilities of the system.

Nanotechnology and neurophysiology

Neuroscience would be revolutionized by a technique to measure intracellular electrical potentials that would not disrupt cellular physiology and could be massively parallelized. Though such a technology does not yet exist, the technical hurdles for fabricating minimally disruptive, solid-state electrical probes have arguably been overcome in the field of nanotechnology. Nanoscale devices can be patterned with features on the same length scale as biological components, and several groups have demonstrated that nanoscale electrical probes can measure the transmembrane potential of electrogenic cells. Developing these nascent technologies into robust intracellular recording tools will now require a better understanding of device-cell interactions, especially the membrane-inorganic interface. Here we review the state-of-the art in nanobioelectronics, emphasizing the characterization and design of stable interfaces between nanoscale devices and cells.

Revealing neuronal function through microelectrode array recordings

Microelectrode arrays and microprobes have been widely utilized to measure neuronal activity, both in vitro and in vivo. The key advantage is the capability to record and stimulate neurons at multiple sites simultaneously. However, unlike the single-cell or single-channel resolution of intracellular recording, microelectrodes detect signals from all possible sources around every sensor. Here, we review the current understanding of microelectrode signals and the techniques for analyzing them. We introduce the ongoing advancements in microelectrode technology, with focus on achieving higher resolution and quality of recordings by means of monolithic integration with on-chip circuitry. We show how recent advanced microelectrode array measurement methods facilitate the understanding of single neurons as well as network function.