Neural Engineering

Summary

Neural engineering is an emerging discipline that translates research discoveries into neurotechnologies. These technologies provide new tools for neuroscience research, while leading to enhanced care for patients with nervous-system disorders. Neural engineers aim to understand, represent, repair, replace, and augment nervous-system function. They accomplish this by incorporating principles and solutions derived from neuroscience, computer science, electrochemistry, materials science, robotics, and other fields. Much of the work focuses on the delicate interface between living neural tissue and nonliving constructs. Efforts focus on elucidating the coding and processing of information in the sensory and motor systems, understanding disease states, and manipulating neural function through interactions with artificial devices such as brain-computer interfaces and neuroprosthetics.

Definition and Basic Principles

Neural engineering (or neuroengineering, NE) is an emerging interdisciplinary research area within biomedical engineering that employs neuroscientific and engineering methods to elucidate neuronal function and design solutions for neurological dysfunction. Restoring sensory, motor, and cognitive function in the nervous system is a priority. The strong emphasis on engineering and quantitative methods separates NE from the “traditional” fields of neuroscience and neurophysiology. The strong neuroscientific approach distinguishes NE from other engineering disciplines such as artificial neural networks. Despite being a distinct discipline, NE draws heavily from basic neuroscience and neurology and brings together engineers, physicians, biologists, psychologists, physicists, and mathematicians.

Neural engineering can be viewed as the driving technology behind several overlapping fields: functional electrical stimulation, stereotactic and functional neurosurgery, neuroprosthetics and neuromodulation. The broad scope of NE also encompasses neurodiagnostics, neuroimaging, neural tissue regeneration, and computational approaches. By using mathematical models of neural function (computational neuroscience), researchers can perform robust testing of therapeutic strategies before they are used on patients.

The human brain, arguably the most complex system known to humankind, contains about 10¹¹ neurons and several times more glial cells. Understanding the functional neuroanatomy of this exquisite device is a sine qua non for anyone aiming to manipulate and repair it. The “neuron doctrine,” pioneered by Spanish neuroscientist Santiago Ramón y Cajal, considers the neuron to be a distinct anatomical and functional unit. The extension introduced by American neuroscientist Warren S. McCullogh and American logician Walter Pitts asserts that the neuron is the basic information-processing unit of the brain. For neuroengineers, this means that a particular goal can be reached just by manipulating a cell or group of cells. One argument in favor of this view is that stimulating groups of neurons produces a regular effect. Motor activity, for example, can be induced by stimulating the motor cortex with electrodes. In addition, lesions to specific brain areas due to neurodegenerative disorders or stroke lead to more or less predictable clinical manifestation patterns.

Background and History

Electricity (in the form of electric fish) was used by ancient Egyptians and Romans for therapeutic purposes. In the eighteenth century, the work of Swiss anatomist Albrecht von Haller, Italian physician Luigi Galvani, and American scientist Benjamin Franklin set the stage for the use of electrical stimulation to restore movement to paralyzed limbs. The basis of modern NE is early neuroscience research demonstrating that neural function can be recorded, manipulated, and mathematically modeled. In the mid-twentieth century, electrical recordings became popular as a window into neuronal function. Metal wire electrodes recorded extracellularly, while glass pipettes probed individual cells. Functional electrical stimulation (FES) emerged with a distinct engineering orientation and the aim to use controlled electrical stimulation to restore function. Modern neuromodulation has developed since the 1970s, driven mainly by clinical professionals. The first peripheral nerve, then spinal cord and deep brain stimulators were introduced in the 1960s. In 1997, the Food and Drug Administration (FDA) approved deep brain stimulation (DBS) for the treatment of Parkinson's disease. An FES-based device that restored grasp was approved the same year.

In the 1970s, researchers developed primitive systems controlled by electrical activity recorded from the head. The U.S. Pentagons Advanced Research Projects Agency (ARPA) supported research aimed at developing bionic systems for soldiers. Scientists demonstrated that recorded brain signals can communicate a user's intent in a reliable manner and found cells in the motor cortex the firing rates of which correlate with hand movements in two-dimensional space.

Since the 1960s, engineers, neuroscientists, and physicists have constructed mathematical models of the retina that describe various aspects of its function, including light-stimulus processing and transduction. In addition, scientists have made attempts to treat blindness using engineering solutions, such as nonbiological “visual prostheses.” In 1975, the first multichannel cochlear implant (CI) was developed and implanted two years later. The cochlear implant is an example of a neurostimulator (NS or neuroprosthesis). Twenty-first-century developments include other neurostimulators such as deep brain stimulators and brain-computer interfaces (BCIs). BCIs involve mapping and recording neural signals that correspond to an intention such as wanting to hold an object, then using a computer and a device such as a robotic arm to perform the task.

How It Works

Neuromodulation and Neuroaugmentation. Neural engineering applications have two broad (and sometimes overlapping) goals: neuromodulation and neuroaugmentation. Neuromodulation (altering nervous system function) employs stimulators and infusion devices, among other techniques. It can be applied at multiple levels: cortical, subcortical, spinal, or peripheral. Neural augmentation aims to amplify neural function and uses sensory (auditory, visual) and motor prostheses.

Neuromuscular Stimulation. Based on a method that has remained unchanged for decades, electrodes are placed within the excitable tissue that provide current to activate certain pathways. This supplements or replaces lost motor or autonomic functions in patients with paralysis. An example is application of electrical pulses to peripheral motor nerves in patients with spinal cord injuries. These pulses lead to action potentials that propagate across neuromuscular junctions and lead to muscle contraction. Coordinating the elicited muscle contractions ultimately reconstitutes function.

Neural Prosthetics. Neural prostheses (NP) aim to restore sensory or motor function—lost because of disease or trauma—by linking machines to the nervous system. By artificially manipulating the biological system using external electrical currents, neuroengineers try to mimic normal sensorimotor function. Electrodes act as transducers that excite neurons through electrical stimulation, or record (read) neural signals. In the first approach, stimulation is used for its therapeutic efficacy, for example, to alleviate the symptoms of Parkinson's disease, or to provide input to the nervous system, such as converting sound to neural input with a cochlear implant. The second paradigm uses recordings of neural activity to detect motor intention and provide input signal to an external device. This forms the basis of a subset of neural prosthetics called brain-controlled interfaces (BCI).

Microsystems. Miniaturization is a crucial part of designing instruments that interface efficiently with neural tissue and provide adequate resolution with minimal invasiveness. Microsystems technology integrates devices and systems at the microscopic and submicroscopic levels. It is derived from microelectronic batch-processing fabrication techniques. A “neural microsystem” is a hybrid system consisting of a microsystem and its interfacing neurons (be they cultured, part of brain slices, or in the intact nervous system). Technologies such as microelectrodes, microdialysis probes, fiber optic, and advanced magnetic materials are used. The properties of these systems render them suitable for simultaneous measurements of neuronal signals in different locations (to analyze neural network properties) as well as for implantation within the body.

Applications and Products

Some of the most common applications of NE methods are described below.

Cochlear Implants. Cochlear implants (CI), by far the most successful sensory neural prostheses to date, have penetrated the mainstream therapeutic arsenal. Their popularity is rivaled only by the cardiac pacemakers and deep brain stimulation (DBS) systems. Implanted in patients with sensorineural deafness, these devices process sounds electronically and transmit stimuli to the cochlea. A CI includes several components: a microphone, a small speech processor that transforms sounds into a signal suitable for auditory neurons, a transmitter to relay the signal to the cochlea, a receiver that picks up the transmitted signal, and an electrode array implanted in the cochlea. Individual results vary, but achieving a high degree of accuracy in speech perception is possible, as is the development of language skills.

Retinal Bioengineering. Retinal photoreceptor cells contain visual pigment, which absorbs light and initiates the process of transducing it into electrical signals. They synapse onto other types of cells, which in turn carry the signals forward, eventually through the optic nerve and into the brain, where they are interpreted. Every neuron in the visual system has a “receptive field,” a particular portion of the visual space within which light will influence that neuron's behavior. This is directly related to (and represented by) a specific region of the retina. Inherited retinal degenerations such as retinitis pigmentosa (RP) or age-related macular degeneration (AMD) are responsible for the compromised or nonexistent vision of millions of people. In these disorders, the retinal photoreceptor cells lose function and die, but the secondary neurons are spared.

Using an electronic prosthetic device, a signal is sent to these secondary neurons that ultimately causes an external visual image. A miniature video camera is mounted on the patient's eyeglasses that captures images and feeds them to a microprocessor, which converts them to an electronic signal. Then the signal is sent to an array of electrodes located on the retina's surface. The electrodes transmit the signal to the viable secondary neurons. The neurons process the signal and pass it down the optic nerve to the brain to establish the visual image.

Several different versions of this device exist and are implanted either into the retina or brain. Cortical visual prostheses could entirely bypass the retina, especially when this structure is damaged from diseases such as diabetes or glaucoma. Retinal prostheses, or artificial retinas (AR), could take advantage of any remaining functional cells and would target photoreceptor disorders such as RP. Two distinct retinal placements are used for AR. The first type slides under the retina (subretinal implant) and consists of small silicon-based disks bearing microphotodiodes. The second type would be an epiretinal system, which involves placing the camera or sensor outside the eye, sending signals to an intraocular receiver. In addition to challenges related to miniaturization and power supply, developing these systems faces obstacles pertaining to biocompatibility, such as retinal health and implant damage, and vascularization.

Functional Electrical Stimulation (FES). Some FES devices are commercialized, and others belong to clinical research settings. A typical unit includes an electronic stimulator, a feedback or control unit, leads, and electrodes. Electrical stimulators bear one or multiple channels (outputs) that are activated simultaneously or in sequence to produce the desired movement. Applications of FES include standing, ambulation, cycling, grasping, bowel and bladder control, male sexual assistance, and respiratory control. Although not curative, the method has numerous benefits, such as improved cardiovascular health, muscle-mass retention, and enhanced psychological well-being through increased functionality and independence.

Brain-Controlled Interfaces. A two-electrode device was implanted into a 1998 stroke victim who could communicate only by blinking his eyes. The device read from only a few neurons and allowed him to select letters and icons with his brain. A team of researchers helped a young patient with a spinal cord injury by implanting electrodes into his motor cortex that were connected to an interface. The patient was able to use the system to control a computer cursor and move objects using a robotic arm.

Brain-controlled interfaces (BCIs), a subset of NP, represent a new method of communication based on brain-generated neural activity. They offer hope to patients with severe motor dysfunction. Practical uses that have already been developed include provision of feedback used for neurofeedback training and cognitive function. Devices that track the level of a user's engagement are on the market. Independent of peripheral nerves and muscles, BCI have the ability to restore communication and movement. This exciting technological advance is not only poised to help patients, but it also provides insight into the way neurons interact.

Every BCI has four main components: recording of electrical activity, extraction of the planned action from this activity, execution of the desired action using the prosthetic effector (actuator), and delivery of feedback (via sensation or prosthetic device). Brain-controlled interfaces rely on four main recording modalities: electroencephalography, electrocorticography, local field potentials, and singe-neuron action potentials. The methods are noninvasive, semi-invasive, or invasive, depending on where the transducer is placed: scalp, brain surface, or cortical tissue.

The field is still in its infancy. However, several basic principles have emerged from these and other early experiments. A crucial requirement in BCI function, for example, is for the reading device to obtain sufficient information for a particular task. Another observation refers to the “transparency of action” in brain-machine interface (BMI) systems: Upon reaching proficiency, the action follows the thought, with no awareness of intermediate neural events.

Deep Brain Stimulation (DBS) and Other Modulation Methods. Deep brain stimulation of thalamic nuclei decreases tremors in patients with Parkinson's disease. It may alleviate depression, epilepsy, and other brain disorders. One or more thin electrodes, about 1 millimeter in diameter, are placed in the brain. An external signal generator with a power supply is also implanted somewhere in the body, typically in the chest cavity. An external remote control sends signals to the generator, varying the parameters of the stimulation, including the amount and frequency of the current and the duration and frequency of the pulses. The exact mechanism by which this method works is still unclear. It appears to exert its effect on axons and act in an inhibitory manner, by inducing an effect akin to ablation of target area, much like early Parkinson's treatment. One major advantage of DBS over other previously employed methods is its reversibility and absence of structural damage. Another valuable neuromodulatory approach, the electrical stimulation of the vagus nerve, can reduce seizure frequency in patients with epilepsy and alleviate treatment-resistant depression. Transcutaneous electrical nerve stimulation (TENS) represents the most common form of electrotherapy and is still in use for pain relief. Cranial electrotherapy stimulation involves passing small currents across the skull. The approach shows good results in depression, anxiety, and sleep disorders.

Transcranial magnetic stimulation uses the magnetic field produced by a current passing through a coil and can be applied for diagnostic (multiple sclerosis, stroke), therapeutic (depression), or research purposes.

Careers and Course Work

Most careers in NE require a bachelor's degree in engineering, biology, neuroscience, physics, or computer science. Neuroengineers often undergo formal training in mechanical or electronics engineering, combined with biomedical training. Educational programs in bioengineering, including NE, are growing rapidly in the United States. Typical course work integrates engineering and life sciences studies. Students may receive instruction in neuroengineering fundamentals; chemistry; fluid mechanics; engineering electrophysiology; diagnostic imaging physics; and drug development. In addition to core courses, students take electives related to their ultimate career goals. Subsequently, many pursue a master's or doctoral degree, sometimes followed by one or more years of postdoctoral training. As in other biomedical engineering fields, some graduates go on to obtain a medical, law, or business degree.

Neuroengineering researchers are employed by universities, medical institutions, industry (medical devices, pharmaceutical, biotechnology), and governmental agencies. They work as physicians, clinical engineers, product engineers, researchers, managers, or teachers.

Social Context and Future Prospects

Bioelectrodes for neural recording and neurostimulation are an essential part of neuroprosthetic devices. Designing an optimal, stable electrode that records long-term and interacts adequately with neural tissue remains a priority for neural engineers. The implementation of microsystem technology opens new perspectives in the field.

More than 1.5 billion people around the world had hearing loss in 2022, according to the World Health Organization, mainly because sensory hair cells in the cochlea have degenerated. The only efficient therapy for patients with profound hearing loss is the CI. Improvements in CI performance have increased the average sentence recognition with multichannel devices. An exciting new development, auditory brainstem implants, show improved performance in patients with impaired cochlear nerves.

Millions of Americans have vision loss. The need for a reliable prosthetic retina is significant, and rivals the one for CI. Technological progress makes it quite likely that a functioning implant with a more sophisticated design and higher number of electrodes will be on the market soon. The epiretinal approach is promising, but providing interpretable visual information to the brain represents a challenge. In addition, even if they prove to be successful, retinal prostheses under development address only a limited number of visual disorders. Much is left to be discovered and tested in this field.

The coming years will also see rapid gains in the area of BCI. Whether they achieve widespread use will depend on several factors, including performance, safety, cost, and improved quality of life.

The advent of gene therapy, stem cell therapy, and other regenerative approaches offers new hope for patients and may complement prosthetic devices. However, many ethical and scientific issues still have to be solved.

Implanted devices are changing the way neurological disorders are treated. An unprecedented transition of NE discoveries from the research to the commercial realm is taking place. At the same time, new discoveries constantly challenge the basic tenets of neuroscience and may alter the face of NE in the coming decades. People's understanding of the nervous system, especially of the brain, changes, and so do the strategies designed to enhance and restore its function.

Bibliography

Blume, Stuart. The Artificial Ear: Cochlear Implants and the Culture of Deafness. New Brunswick, N.J.: Rutgers University Press, 2010.

DiLorenzo, Daniel J., and Joseph D. Bronzino, eds. Neuroengineering. Boca Raton, Fla.: CRC Press, 2008.

Durand, Dominique M. “What Is Neural Engineering?” Journal of Neural Engineering 4, no. 4 (September, 2006).

Gonfalonieri, Alexandre. "What Brain-Computer Interfaces Could Mean for the Future of Work." Harvard Business Review, 6 Oct. 2020, hbr.org/2020/10/what-brain-computer-interfaces-could-mean-for-the-future-of-work. Accessed 8 June 2022.

He, Bin, ed. Neural Engineering. New York: Kluwer Academic/Plenum Publishers, 2005.

Hu, Xiaoling. "Advances in Neural Engineering for Rehabilitation." Behavioural Neurology, vol. 2017, 2017, doi: 10.1155/2017/9240921. Accessed 28 Feb. 2018.

Katz, Bruce F. Neuroengineering the Future: Virtual Minds and the Creation of Immortality. Hingham, Mass.: Infinity Science Press, 2008.

Liu, Xilin, and Andrew G. Richardson. "Edge Deep Learning for Neural Implants: A Case Study of Seizure Detection and Prediction." Journal of Neural Engineering, vol. 18, no. 4, 26 Apr. 2021, iopscience.iop.org/article/10.1088/1741-2552/abf473/meta. Accessed 7 June 2022.

Montaigne, Fen. Medicine By Design: The Practice and Promise of Biomedical Engineering. Baltimore: The Johns Hopkins University Press, 2006.