Neurochip based on Light-Addressable Potentiometric Sensor(LAPS), whose sensing elements are excitable cells, can monitor electrophysiological properties of cultured neuron networks with cellular signals well analyzed. Here we are highlighting a kind of neurochip with rat pheochromocytoma (PC12) cells hybrid with LAPS and a method of de-noising signals based on wavelet transform. Cells were cultured on LAPS for several days to form networks, and we then used LAPS system to detect the extracellular potentials with signals de-noised according to decomposition in the time-frequency space. The signal was decomposed into various scales, and coefficients were processed based on the properties of each layer. At last, signal was reconstructed based on the new coefficients. The results show that after de-noising, baseline drift is removed and signal-to-noise ratio is increased. It suggests that the neurochip of PC12 cells coupled to LAPS is stable and suitable for long-term and non-invasive measurement of cell electrophysiological properties with wavelet transform, taking advantage of its time-frequency localization analysis to reduce noise.
Until recently, neurobiologists have used computers for simulation, data collection, and data analysis, but not to interact directly with nerve tissue in live, behaving animals. Although digital computers and nerve tissue both use voltage waveforms to transmit and process information, engineers and neurobiologists have yet to cohesively link the electronic signalling of digital computers with the electronic signalling of nerve tissue in freely behaving animals.
Recent advances in micro-electro-mechanical systems (MEMS), CMOS electronics, and embedded computer systems will finally let us link computer circuitry to neural cells in live animals and, in particular, to re-identifiable cells with specific, known neural functions. The key components of such a brain-computer system include neural probes, analog electronics, and a miniature microcomputer. Researchers developing neural probes such as sub- micron MEMS probes, microclamps, microprobe arrays, and similar structures can now penetrate and make electrical contact with nerve cells with out causing significant or long-term damage to probes or cells.
Researchers developing analog electronics such as low-power amplifiers and analog-to-digital converters can now integrate these devices with micro- controllers on a single low-power CMOS die. Further, researchers developing embedded computer systems can now incorporate all the core circuitry of a modern computer on a single silicon chip that can run on miniscule power from a tiny watch battery. In short, engineers have all the pieces they need to build truly autonomous implantable computer systems.
Until now, high signal-to-noise recording as well as digital processing of real-time neuronal signals has been possible only in constrained laboratory experiments. By combining MEMS probes with analog electronics and modern CMOS computing into self-contained, implantable microsystems, implantable computers will free neuroscientists from the lab bench.
INTEGRATING SILICON and NEUROBIOLOGY
Neurons and neuronal networks decide, remember, modulate, and control an animals every sensation, thought, movement, and act. The intimate details of this network, including the dynamic properties of individual neurons and neuron populations, give a nervous system the power to control a wide array of behavioural functions.
The goal of understanding these details motivates many workers in modern neurobiology. To make significant progress, these neurobiologists need methods for recording the activity of single neurons or neuron assemblies, for long timescales, at high fidelity, in animals that can interact freely with their sensory world and express normal behavioral responses.
Neurobiologists examine the activities of brain cells tied to sensory inputs, integrative processes, and motor outputs to understand the neural basis of animal behavior and intelligence. They also probe the components of neuronal control circuitry to understand the plasticity and dynamics of control. They want to know more about neuronal dynamics and networks, about synaptic interactions between neurons, and about the inextricable links between environmental stimuli and neuronal signalling, behavior, and control.
To explore the details of this biological circuitry, neurobiologists use two classes of electrodes to record and stimulate electrical signals in tissue
- intracellular micropipettes to impale or patch- clamp single cells for interrogation of the cells internal workings, and
- Extracellular wires or micro machined probes for interrogating multisite patterns of extra- cellular neural signalling or electrical activity in muscles.
Neurobiologists use amplifiers and signal generators to stimulate and record to and from neurons through these electrodes, and signal-processing systems to analyse the results. They have used these techniques for decades to accumulate a wealth of understanding about the nervous system. Unfortunately, to date, most of these experiments have been performed on slices of brain tissue or on restrained and immobilized animals, primarily because the electronic instruments required to run the experiments occupy the better part of a lab bench.
This situation leaves neurobiologists with a nagging question: Are they measuring the animals nor mal brain signals or something far different? Further, neurobiologists want to understand how animal brains respond and react to environmental stimuli. The only way to truly answer these questions is to measure a brains neural signalling while the animal roams freely in its natural environment.
The solution to these problems lies in making the test equipment so small that a scientist can implant it into or onto the animal, using materials and implantation techniques that hurt neither computer nor animal. Recent developments in MEMS, semi-conductor electronics, embedded systems, bio compatible materials, and electronic packaging finally allow neuroscientists and engineers to begin packaging entire neurobiology experiments into hardware and firmware that occupy less space than a human fingernail.
Researchers call these bioembedded systems neurochips. Scientists from the University of Washing-ton, Caltech, and Case Western Reserve University have teamed to build these miniaturized implantable experimental setups to explore the neural basis of behavior.
This research effort has developed or is in the process of developing the following:
Miniaturized silicon MEMS probes for recording from the insides of nerve cells;
- biocompatible coatings that protect these probes from protein fouling;
- a stand-alone implantable microcomputer that records from and stimulates neurons, sensory pathways, or motor control pathways in an intact animal, using intracellular probes, extra- cellular probes, or wire electrodes;
- neurophysiological preparations and techniques for implanting microchips and wire electrodes or MEMS probes into or onto animals in a way that does not damage the probes or tissue;
- firmware that performs real-time biology experiments with implanted computers, using analytical models of the underlying biology; and
- software to study and interpret the experimental results, eventually leading to reverse- engineered studies of animal behavior.
As the Neuroscience Application Examples sidebar shows, the first neurochip experiments use sea slugs and moths in artificial environments, but broad interest has already arisen for using implantable computers in many other animals.
Like their bench top experimental counterparts, neurochips use amplifiers to boost low-voltage biological signals, analog-to-digital converters (ADCs) to digitize these signals, microcomputers to process the signals, on-board memory to store the signals, digital-to-analog converters (DACs) to stimulate nerves, and software to control the overall experiment.
solid lines show required components, dashed lines show some optional components
Figure 1 shows a neurochips basic elements. The key requirements are that the neurochip be small and lightweight enough to fit inside or onto the animal, have adequate signal fidelity for interacting with the millivolt-level signals characteristic of nerve tissue, and have sufficient processing power to perform experiments of real scientific value.
The basic components of a neurochip are commercially available today. They include instrumentation amplifiers, ADCs/DACs, reconfigurable microcomputers, and high-density memory. For example, a Programmable System-on-a-Chip from Cypress Microsystems integrates a microprocessor, variable-gain amplifiers, an ADC, a memory controller, and a DAC into a single integrated circuit.
First-generation neurochips integrate one or more ICs, passive elements such as capacitors, batteries, and 110 pads on small micro-PCBs.
Building the probes that let a neurochip eaves drop on the electrical signaling in a nerve bundle, group of neurons, or single neuron presents a daunting task. Benchtop experiments on con strained animals typically use metallic needles often made of stainless steel or tungstento communicate with nerve bundles, micromachined silicon probes to record from groups of neurons, or glass capillaries filled with a conductive ionic solution to penetrate and record from the inside of individual neurons. In unconstrained animals, flexible metallic needles, attached to the animal with surgical superglue, and micromachined silicon probes still work. However, replicating the performance of glass capillaries in flying, swimming, wiggling animals is a different story entirely.
Several centimeters long and quite fragile, the glass capillaries that neurobiologists use to probe the insides of nerve cells typically have tip diameters smaller than 0.3 microns. They impale neurons even more fragile than the probes themselves.Neuro biologists use micromanipulators to painstakingly and precisely drive single probes into single neurons. Fortunately, MEMS technology offers a possible alternative to these glass capillaries.
Researchers seek to implant both probes and neurochips inside an animals brain. Unfortunately, an animals immune system rapidly and indiscriminately encapsulates all foreign bodies with proteins, without regard for the research value of implanted probes and neurochips. The adsorbed proteins not only attenuate the recorded electrical signals, but can also jeopardize the animals survival by causing abnormal tissue growth.
Researchers at the University of Washingtons Center for Engineered Biomaterials have developed plasma-deposited ether-terminated oligoethylene glycol coatings that inhibit protein fouling. Preliminary research indicates that these glyme coatings can reduce the protein fouling of probes and neurochips to levels acceptable for week-long experiments.
Neurochips can derive power from on-board batteries, external radiofrequency sources, a wire tether, or the nerve tissue itself, The ultimate decision on the power source depends on the nature of the experiments and the animals environment. Batteries are attractive because they avoid the antennas and charge pumps required to capture RF energy, operate in all environments, do not restrict the animals movement the way a tether does, and provide much more power than tapping nerve cells for energy.
Batteries have a weight disadvantage, but thin- film technologies using LiCoO2/LiPON/Li and Ni/KOHIZn promise flexible rechargeable batteries with peak current densities greater than 12 mA per square centimeter for short-duration experiments, and lifetimes measured in days or longer at low-current densities.
Batteries are ideal for the two sample preparations shown in the Neuroscience Application Examples sidebar. The typical hawkmoth flight time is less than 60 seconds. The 12 mA provided by a 200 mg, one-square-centimeter battery easily powers a neurochip for this experiments duration. The sea slug trolling methodically along the seafloor lies at the opposite end of the spectrum, needing only a few milliamps of current to power a neurochip for a week. The slug can easily accommodate a large battery in its visceral cavity, allowing extended untethered experiments.
Once implanted, an embedded neurochip must read its experimental procedure from memory, run the experiment, acquire the neural spike trains, then store the results in memory. As with all computer systems, memory size is an issue for neurochips. Fortunately, the electrical spike trains generated by nerve tissue have a stereotyped shape, suggesting that neurochips should com press the neural waveforms before storing them in memory.
Compressing the signals has two advantages. First, it effectively increases the on-board storage capacity. Second, it decreases the frequency of memory writes, reducing power consumption. Even simple compression algorithms such as run- length encoding can achieve better than 10 to 1 compression ratios on neural signals.
Custom algorithms that apply vector quantization, run-length encoding, and Huffman encoding to different parts of the neural waveform can achieve up to 1,000 to 1 compression ratios. Given the limited computing power of an implantable microcomputer, simpler is better when it comes to compression, but even simple RLE offers huge power and memory-size benefits.
A stimulating world
Passive neurochips that do nothing more than record will provide neurobiologists with a wealth of data. But even now, with the first neurochips barely in production, neurobiologists are already calling for designs that stimulate nerve tissue as well as record from it. Active neurochips will allow stimulus-response experiments that test models of how nervous systems control behavior, such as how sensory inputs inform motor-circuit loops and the logic or model behind the response.
Indeed, the neurochip projects long-term goal is to develop a hardware and software environment in which a neurobiologist conceives a stimulus-response experiment, encodes that experiment in software, downloads the experiment to an implanted neurochip, and recovers the data when the experiment concludes.
With advances in integrated circuit processing we can expect ever more capable and power-efficient embedded computers. The simple neurochips of today will become the complex embedded systems of tomorrow, when embedding in this ultimate sense will mean computer electronics embedded in nerve tissue.
Enabling neuroscientists to better understand the neural basis of behavior is reason enough to develop such devices. The long-term promise is much greater, however, perhaps leading one day to neural prosthetics, hardware-based human-computer interfaces, and artificial systems that incorporate principles of biological intelligence.