Minimum requirements for accurate and efficient real-time on-chip spike sorting

An unsupervised real-time spike sorting system based on optimized OSort

Objective. The OSort algorithm, a pivotal unsupervised spike sorting method, has been implemented in dedicated hardware devices for real-time spike sorting. However, due to the inherent complexity of neural recording environments, OSort still grapples with numerous transient cluster occurrences during the practical sorting process. This leads to substantial memory usage, heavy computational load, and complex hardware architectures, especially in noisy recordings and multi-channel systems.Approach. This study introduces an optimized OSort algorithm (opt-OSort) which utilizes correlation coefficient (CC), instead of Euclidean distance as classification criterion. TheCCmethod not only bolsters the robustness of spike classification amidst the diverse and ever-changing conditions of physiological and recording noise environments, but also can finish the entire sorting procedure within a fixed number of cluster slots, thus preventing a large number of transient clusters. Moreover, the opt-OSort incorporates two configurable validation loops to efficiently reject cluster outliers and track recording variations caused by electrode drifting in real-time.Main results. The opt-OSort significantly reduces transient cluster occurrences by two orders of magnitude and decreases memory usage by 2.5-80 times in the number of pre-allocated transient clusters compared with other hardware implementations of OSort. The opt-OSort maintains an accuracy comparable to offline OSort and other commonly-used algorithms, with a sorting time of 0.68µs as measured by the hardware-implemented system in both simulated datasets and experimental data. The opt-OSort's ability to handle variations in neural activity caused by electrode drifting is also demonstrated.Significance. These results present a rapid, precise, and robust spike sorting solution suitable for integration into low-power, portable, closed-loop neural control systems and brain-computer interfaces.

Calibration-Free and Hardware-Efficient Neural Spike Detection for Brain Machine Interfaces

Recent translational efforts in brain-machine interfaces (BMI) are demonstrating the potential to help people with neurological disorders. The current trend in BMI technology is to increase the number of recording channels to the thousands, resulting in the generation of vast amounts of raw data. This in turn places high bandwidth requirements for data transmission, which increases power consumption and thermal dissipation of implanted systems. On-implant compression and/or feature extraction are therefore becoming essential to limiting this increase in bandwidth, but add further power constraints - the power required for data reduction must remain less than the power saved through bandwidth reduction. Spike detection is a common feature extraction technique used for intracortical BMIs. In this article, we develop a novel firing-rate-based spike detection algorithm that requires no external training and is hardware efficient and therefore ideally suited for real-time applications. Key performance and implementation metrics such as detection accuracy, adaptability in chronic deployment, power consumption, area utilization, and channel scalability are benchmarked against existing methods using various datasets. The algorithm is first validated using a reconfigurable hardware (FPGA) platform and then ported to a digital ASIC implementation in both 65 nm and 0.18 μm CMOS technologies. The 128-channel ASIC design implemented in a 65 nm CMOS technology occupies 0.096 mm2 silicon area and consumes 4.86 μW from a 1.2 V power supply. The adaptive algorithm achieves a 96% spike detection accuracy on a commonly used synthetic dataset, without the need for any prior training.

A Multimodal and Multifunctional CMOS Cellular Interfacing Array for Digital Physiology and Pathology Featuring an Ultra Dense Pixel Array and Reconfigurable Sampling Rate

The article presents a fully integrated multimodal and multifunctional CMOS biosensing/actuating array chip and system for multi-dimensional cellular/tissue characterization. The CMOS chip supports up to 1,568 simultaneous parallel readout channels across 21,952 individually addressable multimodal pixels with 13 μm × 13 μm 2-D pixel pitch along with 1,568 Pt reference electrodes. These features allow the CMOS array chip to perform multimodal physiological measurements on living cell/tissue samples with both high throughput and single-cell resolution. Each pixel supports three sensing and one actuating modalities, each reconfigurable for different functionalities, in the form of full array (FA) or fast scan (FS) voltage recording schemes, bright/dim optical detection, 2-/4-point impedance sensing (ZS), and biphasic current stimulation (BCS) with adjustable stimulation area for single-cell or tissue-level stimulation. Each multi-modal pixel contains an 8.84 μm × 11 μm Pt electrode, 4.16 μm × 7.2 μm photodiode (PD), and in-pixel circuits for PD measurements and pixel selection. The chip is fabricated in a standard 130nm BiCMOS process as a proof of concept. The on-chip electrodes are constructed by unique design and in-house post-CMOS fabrication processes, including a critical Al shorting of all pixels during fabrication and Al etching after fabrication that ensures a high-yield planar electrode array on CMOS with high biocompatibility and long-term measurement reliability. For demonstration, extensive biological testing is performed with human and mouse progenitor cells, in which multidimensional biophysiological data are acquired for comprehensive cellular characterization.

A flexible adhesive surface electrode array capable of cervical electroneurography during a sequential autonomic stress challenge

This study introduces a flexible, adhesive-integrated electrode array that was developed to enable non-invasive monitoring of cervical nerve activity. The device uses silver-silver chloride as the electrode material of choice and combines it with an electrode array consisting of a customized biopotential data acquisition unit and integrated graphical user interface (GUI) for visualization of real-time monitoring. Preliminary testing demonstrated this electrode design can achieve a high signal to noise ratio during cervical neural recordings. To demonstrate the capability of the surface electrodes to detect changes in cervical neuronal activity, the cold-pressor test (CPT) and a timed respiratory challenge were employed as stressors to the autonomic nervous system. This sensor system recording, a new technique, was termed Cervical Electroneurography (CEN). By applying a custom spike sorting algorithm to the electrode measurements, neural activity was classified in two ways: (1) pre-to-post CPT, and (2) during a timed respiratory challenge. Unique to this work: (1) rostral to caudal channel position-specific (cephalad to caudal) firing patterns and (2) cross challenge biotype-specific change in average CEN firing, were observed with both CPT and the timed respiratory challenge. Future work is planned to develop an ambulatory CEN recording device that could provide immediate notification of autonomic nervous system activity changes that might indicate autonomic dysregulation in healthy subjects and clinical disease states.

Selecting an effective amplitude threshold for neural spike detection

This paper assesses and challenges whether commonly used methods for defining amplitude thresholds for spike detection are optimal. This is achieved through empirical testing of single amplitude thresholds across multiple recordings of varying SNR levels. Our results suggest that the most widely used noise-statistics-driven threshold can suffer from parameter deviation in different noise levels. The spike-noise-driven threshold can be an ideal approach to set the threshold for spike detection, which suffers less from the parameter deviation and is robust to sub-optimal settings.

Low-Power Lossless Data Compression for Wireless Brain Electrophysiology

Wireless electrophysiology opens important possibilities for neuroscience, especially for recording brain activity in more natural contexts, where exploration and interaction are not restricted by the usual tethered devices. The limiting factor is transmission power and, by extension, battery life required for acquiring large amounts of neural electrophysiological data. We present a digital compression algorithm capable of reducing electrophysiological data to less than 65.5% of its original size without distorting the signals, which we tested in vivo in experimental animals. The algorithm is based on a combination of delta compression and Huffman codes with optimizations for neural signals, which allow it to run in small, low-power Field-Programmable Gate Arrays (FPGAs), requiring few hardware resources. With this algorithm, a hardware prototype was created for wireless data transmission using commercially available devices. The power required by the algorithm itself was less than 3 mW, negligible compared to the power saved by reducing the transmission bandwidth requirements. The compression algorithm and its implementation were designed to be device-agnostic. These developments can be used to create a variety of wired and wireless neural electrophysiology acquisition systems with low power and space requirements without the need for complex or expensive specialized hardware.

Algorithm and hardware considerations for real-time neural signal on-implant processing

Objective Various on-workstation neural-spike-based brain machine interface(BMI) systems have reached the point of in-human trials, but on-node and on-implant BMI systems are still under exploration. Such systems are constrained by the area and battery. Researchers should consider the algorithm complexity, available resources, power budgets, CMOS technologies, and the choice of platforms when designing BMI systems. However, the effect of these factors is currently still unclear. Approaches. Here we have proposed a novel real-time 128 channel spike detection algorithm and optimised it on Microcontroller(MCU) and Field Programmable Gate Array(FPGA) platforms towards consuming minimal power and memory/resources. It is presented as a use case to explore the different considerations in system design. Main results. The proposed spike detection algorithm achieved over 97% sensitivity and a smaller than 3% false detection rate. The MCU implementation occupies less than 3KB RAM and consumes 31.5μW/ch. The FPGA platform only occupies 299 logic cells and 3KB RAM for 128 channels and consumes 0.04μW/ch. Significance. On the spike detection algorithm front, we have eliminated the processing bottleneck by reducing the dynamic power consumption to lower than the hardware static power, without sacrificing detection performance. More importantly, we have explored the considerations in algorithm and hardware design with respect to scalability, portability, and costs. These findings can facilitate and guide the future development of real-time on-implant neural signal processing platforms.

The impact of reducing signal acquisition specifications on neuronal spike sorting

Measuring electrical potentials in the extracellular space of the brain is a popular technique because it can detect action potentials from putative individual neurons. Electrophysiology is undergoing a transformation where the number of recording channels, and thus number of neurons detected, is growing at a dramatic rate. This rapid scaling is paving the way for both new discoveries and commercial applications; however, as the number of channels increases there will be an increasing need to make these systems more power efficient. One area ripe for optimization are the signal acquisition specifications needed to detect and sort action potentials (i.e., "spikes") to putative single neuron sources. In this work, we take existing recordings collected using Intan hardware and modify them in a way that corresponds to reduced recording performance. The accuracy of these degraded recordings to spike sort using MountainSort4 is evaluated by comparing against expert labels. We show that despite reducing signal specifications by a factor of 2 or more, spike sorting accuracy does not change substantially. Specifically, reducing both sample rate and bit depth from 30 kHz and 16 bits to 12 kHz and 12 bits resulted in a 3% drop in spike sorting accuracy. Our results suggest that current neural acquisition systems are over-specified. These results may inform the design of next generation neural acquisition systems enabling higher channel count systems.

A 32-Channel Time-Multiplexed Artifact-Aware Neural Recording System

This paper presents a low-power, low-noise microsystem for the recording of neural local field potentials or intracranial electroencephalographic signals. It features 32 time-multiplexed channels at the electrode interface and offers the possibility to spatially delta encode data to take advantage of the large correlation of signals captured from nearby channels. The circuit also implements a mixed-signal voltage-triggered auto-ranging algorithm which allows to attenuate large interferers in digital domain while preserving neural information. This effectively increases the system dynamic range and avoids the onset of saturation. A prototype, fabricated in a standard 180 nm CMOS process, has been experimentally verified in-vitro with cellular cultures of primary cortical neurons from mice. The system shows an integrated input-referred noise in the 0.5-200 Hz band of 1.4 μV_rms for a spot noise of about 85 nV /√{Hz}. The system draws 1.5 μW per channel from 1.2 V supply and obtains 71 dB + 26 dB dynamic range when the artifact-aware auto-ranging mechanism is enabled, without penalising other critical specifications such as crosstalk between channels or common-mode and power supply rejection ratios.

HTsort: Enabling Fast and Accurate Spike Sorting on Multi-Electrode Arrays

Spike sorting is used to classify the spikes (action potentials acquired by physiological electrodes), aiming to identify their respective firing units. Now it has been developed to classify the spikes recorded by multi-electrode arrays (MEAs), with the improvement of micro-electrode technology. However, how to improve classification accuracy and maintain low time complexity simultaneously becomes a difficulty. A fast and accurate spike sorting approach named HTsort is proposed for high-density multi-electrode arrays in this paper. Several improvements have been introduced to the traditional pipeline that is composed of threshold detection and clustering method. First, the divide-and-conquer method is employed to utilize electrode spatial information to achieve pre-clustering. Second, the clustering method HDBSCAN (hierarchical density-based spatial clustering of applications with noise) is used to classify spikes and detect overlapping events (multiple spikes firing simultaneously). Third, the template merging method is used to merge redundant exported templates according to the template similarity and the spatial distribution of electrodes. Finally, the template matching method is used to resolve overlapping events. Our approach is validated on simulation data constructed by ourselves and publicly available data and compared to other state-of-the-art spike sorters. We found that the proposed HTsort has a more favorable trade-off between accuracy and time consumption. Compared with MountainSort and SpykingCircus, the time consumption is reduced by at least 40% when the number of electrodes is 64 and below. Compared with HerdingSpikes, the classification accuracy can typically improve by more than 10%. Meanwhile, HTsort exhibits stronger robustness against background noise than other sorters. Our more sophisticated spike sorter would facilitate neurophysiologists to complete spike sorting more quickly and accurately.

Adaptive spike detection and hardware optimization towards autonomous, high-channel-count BMIs

BACKGROUND

The progress in microtechnology has enabled an exponential trend in the number of neurons that can be simultaneously recorded. The data bandwidth requirement is however increasing with channel count. The vast majority of experimental work involving electrophysiology stores the raw data and then processes this offline; to detect the underlying spike events. Emerging applications however require new methods for local, real-time processing.

NEW METHODS

We have developed an adaptive, low complexity spike detection algorithm that combines three novel components for: (1) removing the local field potentials; (2) enhancing the signal-to-noise ratio; and (3) computing an adaptive threshold. The proposed algorithm has been optimised for hardware implementation (i.e. minimising computations, translating to a fixed-point implementation), and demonstrated on low-power embedded targets.

MAIN RESULTS

The algorithm has been validated on both synthetic datasets and real recordings yielding a detection sensitivity of up to 90%. The initial hardware implementation using an off-the-shelf embedded platform demonstrated a memory requirement of less than 0.1 kb ROM and 3 kb program flash, consuming an average power of 130 μW.

COMPARISON WITH EXISTING METHODS

The method presented has the advantages over other approaches, that it allows spike events to be robustly detected in real-time from neural activity in a completely autonomous way, without the need for any calibration, and can be implemented with low hardware resources.

CONCLUSION

The proposed method can detect spikes effectively and adaptively. It alleviates the need for re-calibration, which is critical towards achieving a viable BMI, and more so with future 'high bandwidth' systems' targeting 1000s of channels.

Lossless Compression of Intracortical Extracellular Neural Recordings using Non-Adaptive Huffman Encoding

This paper investigates the effectiveness of four Huffman-based compression schemes for different intracortical neural signals and sample resolutions. The motivation is to find effective lossless, low-complexity data compression schemes for Wireless Intracortical Brain-Machine Interfaces (WI-BMI). The considered schemes include pre-trained Lone 1^st and 2^nd order encoding [1], pre-trained Delta encoding, and pre-trained Linear Neural Network Time (LNNT) encoding [2]. Maximum codeword-length limited versions are also considered to protect against overfit to training data. The considered signals are the Extracellular Action Potential signal, the Entire Spiking Activity signal, and the Local Field Potential signal. Sample resolutions of 5 to 13 bits are considered. The result show that overfit-protection dramatically improves compression, especially at higher sample resolutions. Across signals, 2^nd order encoding generally performed best at lower sample resolutions, and 1^st order, Delta and LNNT encoding performed best at higher sample resolutions. The proposed methods should generalise to other remote sensing applications where the distribution of the sensed data can be estimated a priori.

A framework for on-implant spike sorting based on salient feature selection

On-implant spike sorting methods employ static feature extraction/selection techniques to minimize the hardware cost. Here we propose a novel framework for real-time spike sorting based on dynamic selection of features. We select salient features that maximize the geometric-mean of between-class distances as well as the associated homogeneity index effectively to best discriminate spikes for classification. Wave-shape classification is performed based on a multi-label window discrimination approach. An external module calculates the salient features and discrimination windows through optimizing a replica of the on-implant operation, and then configures the on-implant spike sorter for real-time online operation. Hardware implementation of the on-implant online spike sorter for 512 channels of concurrent extra-cellular neural signals is reported, with an average classification accuracy of ~88%. Compared with other similar methods, our method shows reduction in classification error by a factor of ~2, and also reduction in the required memory space by a factor of ~5.

Low-Impedance 3D PEDOT:PSS Ultramicroelectrodes

The technology for producing microelectrode arrays (MEAs) has been developing since the 1970s and extracellular electrophysiological recordings have become well established in neuroscience, drug screening and cardiology. MEAs allow monitoring of long-term spiking activity of large ensembles of excitable cells noninvasively with high temporal resolution and mapping its spatial features. However, their inability to register subthreshold potentials, such as intrinsic membrane oscillations and synaptic potentials, has inspired a number of laboratories to search for alternatives to bypass the restrictions and/or increase the sensitivity of microelectrodes. In this study, we present the fabrication and in vitro experimental validation of arrays of PEDOT:PSS-coated 3D ultramicroelectrodes, with the best-reported combination of small size and low electrochemical impedance. We observed that this type of microelectrode does not alter neuronal network biological properties, improves the signal quality of extracellular recordings and exhibits higher selectivity toward single unit recordings. With fabrication processes simpler than those reported in the literature for similar electrodes, our technology is a promising tool for study of neuronal networks.

On-Line Recursive Decomposition of Intramuscular EMG Signals Using GPU-Implemented Bayesian Filtering

OBJECTIVE

Real-time intramuscular electromyography (iEMG) decomposition, which is needed in biofeedback studies and interfacing applications, is a complex procedure that involves identifying the motor neuron spike trains from a streaming iEMG recording.

METHODS

We have previously proposed a sequential decomposition algorithm based on a Hidden Markov Model of EMG, which used Bayesian filter to estimate unknown parameters of motor unit (MU) spike trains, as well as their action potentials (MUAPs). Here, we present a modification of this original model in order to achieve a real-time performance of the algorithm as well as a parallel computation implementation of the algorithm on Graphics Processing Unit (GPU). Specifically, the Kalman filter previously used to estimate the MUAPs, is replaced by a least-mean-square filter. Additionally, we introduce a number of heuristics that help to omit the most improbable decomposition scenarios while searching for the best solution. Then, a GPU-implementation of the proposed algorithm is presented.

RESULTS

Simulated iEMG signals containing up to 10 active MUs, as well as five experimental fine-wire iEMG signals acquired from the tibialis anterior muscle, were decomposed in real time. The accuracy of decompositions depended on the level of muscle activation, but in all cases exceeded 85 %.

CONCLUSION

The proposed method and implementation provide an accurate, real-time interface with spinal motor neurons.

SIGNIFICANCE

The presented real time implementation of the decomposition algorithm substantially broadens the domain of its application.

Low-latency single channel real-time neural spike sorting system based on template matching

Recent technical advancements in neural engineering allow for precise recording and control of neural circuits simultaneously, opening up new opportunities for closed-loop neural control. In this work, a rapid spike sorting system was developed based on template matching to rapidly calculate instantaneous firing rates for each neuron in a multi-unit extracellular recording setting. Cluster templates were first generated by a desktop computer using a non-parameter spike sorting algorithm (Super-paramagnetic clustering) and then transferred to a field-programmable gate array digital circuit for rapid sorting through template matching. Two different matching techniques–Euclidean distance (ED) and correlational matching (CM)–were compared for the accuracy of sorting and the performance of calculating firing rates. The performance of the system was first verified using publicly available artificial data and was further confirmed with pre-recorded neural spikes from an anesthetized Mongolian gerbil. Real-time recording and sorting from an awake mouse were also conducted to confirm the system performance in a typical behavioral neuroscience experimental setting. Experimental results indicated that high sorting accuracies were achieved for both template-matching methods, but CM can better handle spikes with non-Gaussian spike distributions, making it more robust for in vivo recording. The technique was also compared to several other off-line spike sorting algorithms and the results indicated that the sorting accuracy is comparable but sorting time is significantly shorter than these other techniques. A low sorting latency of under 2 ms and a maximum spike sorting rate of 941 spikes/second have been achieved with our hybrid hardware/software system. The low sorting latency and fast sorting rate allow future system developments of neural circuit modulation through analyzing neural activities in real-time.

An Improved Update Rate CDR for Interference Robust Broadband Human Body Communication Receiver

Broadband Human Body Communication (HBC) enables energy efficient communication between body area network devices by utilizing the electrical conductivity property of the human body. However, environmental interference remains a primary bottleneck in its implementation. An integrating front-end receiver with resettable integration followed by periodic sampling can be utilized to enable interference robust broadband HBC. However, as required in all broadband communication systems, a Clock Data Recovery (CDR) loop is necessary to correctly sample the received data at the appropriate instant. The CDR is required to be sensitive to the clock-data phase mismatch at the receiver end and take corrective action for reducing it, similar to the CDR of a traditional receiver. In addition to that, the CDR for a broadband HBC receiver also requires to be tolerant to environmental interference. This paper analyzes the traditional Baud Rate CDR for an integrating front-end receiver and proposes a modified integrating CDR architecture with a higher update rate. Simulation results show 2.5X higher clock data frequency offset tolerance of the proposed CDR compared to the traditional Baud Rate CDR, >1.25X higher clock data frequency offset tolerance in presence of interference and >10% interference frequency offset tolerance with respect to the integration clock. The proposed CDR is also implemented in a Xilinx Spartan-3E FPGA board to validate its closed loop functionality in real time.

SpikeDeeptector: a deep-learning based method for detection of neural spiking activity

OBJECTIVE

In electrophysiology, microelectrodes are the primary source for recording neural data (single unit activity). These microelectrodes can be implanted individually or in the form of arrays containing dozens to hundreds of channels. Recordings of some channels contain neural activity, which are often contaminated with noise. Another fraction of channels does not record any neural data, but only noise. By noise, we mean physiological activities unrelated to spiking, including technical artifacts and neural activities of neurons that are too far away from the electrode to be usefully processed. For further analysis, an automatic identification and continuous tracking of channels containing neural data is of great significance for many applications, e.g. automated selection of neural channels during online and offline spike sorting. Automated spike detection and sorting is also critical for online decoding in brain-computer interface (BCI) applications, in which only simple threshold crossing events are often considered for feature extraction. To our knowledge, there is no method that can universally and automatically identify channels containing neural data. In this study, we aim to identify and track channels containing neural data from implanted electrodes, automatically and more importantly universally. By universally, we mean across different recording technologies, different subjects and different brain areas.

APPROACH

We propose a novel algorithm based on a new way of feature vector extraction and a deep learning method, which we call SpikeDeeptector. SpikeDeeptector considers a batch of waveforms to construct a single feature vector and enables contextual learning. The feature vectors are then fed to a deep learning method, which learns contextualized, temporal and spatial patterns, and classifies them as channels containing neural spike data or only noise.

MAIN RESULTS

We trained the model of SpikeDeeptector on data recorded from a single tetraplegic patient with two Utah arrays implanted in different areas of the brain. The trained model was then evaluated on data collected from six epileptic patients implanted with depth electrodes, unseen data from the tetraplegic patient and data from another tetraplegic patient implanted with two Utah arrays. The cumulative evaluation accuracy was 97.20% on 1.56 million hand labeled test inputs.

SIGNIFICANCE

The results demonstrate that SpikeDeeptector generalizes not only to the new data, but also to different brain areas, subjects, and electrode types not used for training.

CLINICAL TRIAL REGISTRATION NUMBER

The clinical trial registration number for patients implanted with the Utah array is NCT01849822. For the epilepsy patients, approval from the local ethics committee at the Ruhr-University Bochum, Germany, was obtained prior to implantation.

Multifractal detrended fluctuation analysis of continuous neural time series in primate visual cortex

BACKGROUND

Local field potential (LFP) recordings have become an important tool to study the activity of populations of neurons. The functional activity of LFPs is usually compared with the activity of neighboring single spike neurons with sampling rates much higher than those of the continuous field potential channel (5 kHz). However, comparison of these signals generated with the lower sampling rate technique is important.

NEW METHOD

In this study, we provide an analysis of extracellular field potential time series using the sophisticated nonlinear multifractal detrended fluctuation analysis (MF-DFA). Using the MF-DFA, we demonstrate that the integral of the singularity spectrum is a powerful new method to measure the response tuning of spikes in the continuous field potential channel.

RESULTS

Results show that the spikes in the continuous channel at frequency ranges above the LFP component signals were consistently tuned similar to those in the spike channel. Our results also show that using a low-pass filter (<250 Hz), which is commonly applied as a preprocessing step to insulate LFPs from spikes, significantly influences the nonlinearity of the multifractal time series.

COMPARISON WITH EXISTING METHODS AND CONCLUSIONS

Our approach for inferring the tuning curve of spiking activity from the continuous channel has some advantages compared to conventional methods such as spike trains. The MF-DFA does not require any preprocessing of the raw signal data and makes no assumptions about the time series characteristics. This method is robust and can be applied to short time series of continuous raw signals.

Seamlessly fused digital-analogue reconfigurable computing using memristors

As the world enters the age of ubiquitous computing, the need for reconfigurable hardware operating close to the fundamental limits of energy consumption becomes increasingly pressing. Simultaneously, scaling-driven performance improvements within the framework of traditional analogue and digital design become progressively more restricted by fundamental physical constraints. Emerging nanoelectronics technologies bring forth new prospects yet a significant rethink of electronics design is required for realising their full potential. Here we lay the foundations of a design approach that fuses analogue and digital thinking by combining digital electronics with analogue memristive devices for achieving charge-based computation; information processing where every dissipated charge counts. This is realised by introducing memristive devices into standard logic gates, thus rendering them reconfigurable and capable of performing analogue computation at a power cost close to digital. The versatility and benefits of our approach are experimentally showcased through a hardware data clusterer and an analogue NAND gate.

A 64-Channel Versatile Neural Recording SoC With Activity-Dependent Data Throughput

Modern microtechnology is enabling the channel count of neural recording integrated circuits to scale exponentially. However, the raw data bandwidth of these systems is increasing proportionately, presenting major challenges in terms of power consumption and data transmission (especially for wireless systems). This paper presents a system that exploits the sparse nature of neural signals to address these challenges and provides a reconfigurable low-bandwidth event-driven output. Specifically, we present a novel 64-channel low-noise (2.1 V), low-power (23  W per analogue channel) neural recording system-on-chip (SoC). This features individually configurable channels, 10-bit analogue-to-digital conversion, digital filtering, spike detection, and an event-driven output. Each channel's gain, bandwidth, and sampling rate settings can be independently configured to extract local field potentials at a low data-rate and/or action potentials (APs) at a higher data rate. The sampled data are streamed through an SRAM buffer that supports additional on-chip processing such as digital filtering and spike detection. Real-time spike detection can achieve 2 orders of magnitude data reduction, by using a dual polarity simple threshold to enable an event driven output for neural spikes (16-sample window). The SoC additionally features a latency-encoded asynchronous output that is critical if used as part of a closed-loop system. This has been specifically developed to complement a separate on-node spike sorting coprocessor to provide a real-time (low latency) output. The system has been implemented in a commercially available 0.35-m CMOS technology occupying a silicon area of 19.1 mm (0.3 mm gross per channel), demonstrating a low-power and efficient architecture that could be further optimized by aggressive technology and supply voltage scaling.

Event-driven processing for hardware-efficient neural spike sorting

OBJECTIVE

The prospect of real-time and on-node spike sorting provides a genuine opportunity to push the envelope of large-scale integrated neural recording systems. In such systems the hardware resources, power requirements and data bandwidth increase linearly with channel count. Event-based (or data-driven) processing can provide here a new efficient means for hardware implementation that is completely activity dependant. In this work, we investigate using continuous-time level-crossing sampling for efficient data representation and subsequent spike processing.

APPROACH

(1) We first compare signals (synthetic neural datasets) encoded with this technique against conventional sampling. (2) We then show how such a representation can be directly exploited by extracting simple time domain features from the bitstream to perform neural spike sorting. (3) The proposed method is implemented in a low power FPGA platform to demonstrate its hardware viability.

MAIN RESULTS

It is observed that considerably lower data rates are achievable when using 7 bits or less to represent the signals, whilst maintaining the signal fidelity. Results obtained using both MATLAB and reconfigurable logic hardware (FPGA) indicate that feature extraction and spike sorting accuracies can be achieved with comparable or better accuracy than reference methods whilst also requiring relatively low hardware resources.

SIGNIFICANCE

By effectively exploiting continuous-time data representation, neural signal processing can be achieved in a completely event-driven manner, reducing both the required resources (memory, complexity) and computations (operations). This will see future large-scale neural systems integrating on-node processing in real-time hardware.

Filter based phase distortions in extracellular spikes

Extracellular recordings are the primary tool for extracting neuronal spike trains in-vivo. One of the crucial pre-processing stages of this signal is the high-pass filtration used to isolate neuronal spiking activity. Filters are characterized by changes in the magnitude and phase of different frequencies. While filters are typically chosen for their effect on magnitudes, little attention has been paid to the impact of these filters on the phase of each frequency. In this study we show that in the case of nonlinear phase shifts generated by most online and offline filters, the signal is severely distorted, resulting in an alteration of the spike waveform. This distortion leads to a shape that deviates from the original waveform as a function of its constituent frequencies, and a dramatic reduction in the SNR of the waveform that disrupts spike detectability. Currently, the vast majority of articles utilizing extracellular data are subject to these distortions since most commercial and academic hardware and software utilize nonlinear phase filters. We show that this severe problem can be avoided by recording wide-band signals followed by zero phase filtering, or alternatively corrected by reversed filtering of a narrow-band filtered, and in some cases even segmented signals. Implementation of either zero phase filtering or phase correction of the nonlinear phase filtering reproduces the original spike waveforms and increases the spike detection rates while reducing the number of false negative and positive errors. This process, in turn, helps eliminate subsequent errors in downstream analyses and misinterpretations of the results.

Decoding Local Field Potentials for Neural Interfaces

The stability and frequency content of local field potentials (LFPs) offer key advantages for long-term, low-power neural interfaces. However, interpreting LFPs may require new signal processing techniques which should be informed by a scientific understanding of how these recordings arise from the coordinated activity of underlying neuronal populations. We review current approaches to decoding LFPs for brain-machine interface (BMI) applications, and suggest several directions for future research. To facilitate an improved understanding of the relationship between LFPs and spike activity, we share a dataset of multielectrode recordings from monkey motor cortex, and describe two unsupervised analysis methods we have explored for extracting a low-dimensional feature space that is amenable to biomimetic decoding and biofeedback training.

Improving data quality in neuronal population recordings

Understanding how the brain operates requires understanding how large sets of neurons function together. Modern recording technology makes it possible to simultaneously record the activity of hundreds of neurons, and technological developments will soon allow recording of thousands or tens of thousands. As with all experimental techniques, these methods are subject to confounds that complicate the interpretation of such recordings, and could lead to erroneous scientific conclusions. Here we discuss methods for assessing and improving the quality of data from these techniques and outline likely future directions in this field.

A real-time spike classification method based on dynamic time warping for extracellular enteric neural recording with large waveform variability

BACKGROUND

Computationally efficient spike recognition methods are required for real-time analysis of extracellular neural recordings. The enteric nervous system (ENS) is important to human health but less well-understood with few appropriate spike recognition algorithms due to large waveform variability.

NEW METHOD

Here we present a method based on dynamic time warping (DTW) with high tolerance to variability in time and magnitude. Adaptive temporal gridding for "fastDTW" in similarity calculation significantly reduces the computational cost. The automated threshold selection allows for real-time classification for extracellular recordings.

RESULTS

Our method is first evaluated on synthesized data at different noise levels, improving both classification accuracy and computational complexity over the conventional cross-correlation based template-matching method (CCTM) and PCA+k-means clustering without time warping. Our method is then applied to analyze the mouse enteric neural recording with mechanical and chemical stimuli. Successful classification of biphasic and monophasic spikes is achieved even when the spike variability is larger than millisecond in width and millivolt in magnitude.

COMPARISON WITH EXISTING METHOD(S)

In comparison with conventional template matching and clustering methods, the fastDTW method is computationally efficient with high tolerance to waveform variability.

CONCLUSIONS

We have developed an adaptive fastDTW algorithm for real-time spike classification of ENS recording with large waveform variability against colony motility, ambient changes and cellular heterogeneity.

Past, present and future of spike sorting techniques

Spike sorting is a crucial step to extract information from extracellular recordings. With new recording opportunities provided by the development of new electrodes that allow monitoring hundreds of neurons simultaneously, the scenario for the new generation of algorithms is both exciting and challenging. However, this will require a new approach to the problem and the development of a common reference framework to quickly assess the performance of new algorithms. In this work, we review the basic concepts of spike sorting, including the requirements for different applications, together with the problems faced by presently available algorithms. We conclude by proposing a roadmap stressing the crucial points to be addressed to support the neuroscientific research of the near future.

The Principle of the Micro-Electronic Neural Bridge and a Prototype System Design

The micro-electronic neural bridge (MENB) aims to rebuild lost motor function of paralyzed humans by routing movement-related signals from the brain, around the damage part in the spinal cord, to the external effectors. This study focused on the prototype system design of the MENB, including the principle of the MENB, the neural signal detecting circuit and the functional electrical stimulation (FES) circuit design, and the spike detecting and sorting algorithm. In this study, we developed a novel improved amplitude threshold spike detecting method based on variable forward difference threshold for both training and bridging phase. The discrete wavelet transform (DWT), a new level feature coefficient selection method based on Lilliefors test, and the k-means clustering method based on Mahalanobis distance were used for spike sorting. A real-time online spike detecting and sorting algorithm based on DWT and Euclidean distance was also implemented for the bridging phase. Tested by the data sets available at Caltech, in the training phase, the average sensitivity, specificity, and clustering accuracies are 99.43%, 97.83%, and 95.45%, respectively. Validated by the three-fold cross-validation method, the average sensitivity, specificity, and classification accuracy are 99.43%, 97.70%, and 96.46%, respectively.