Neuron 52 205 220 October 5 2006 2006 Elsevier Inc DOI 10 1016 j neuron 2006 09 019 Brain Controlled Interfaces Movement Restoration with Neural Prosthetics Andrew B Schwartz 1 X Tracy Cui 2 Douglas J Weber 3 and Daniel W Moran4 1 Departments of Neurobiology and Bioengineering Center for the Neural Basis of Cognition McGowan Institute for Regenerative Medicine 2 Department of Bioengineering McGowan Institute for Regenerative Medicine 3 Departments of Physical Medicine and Rehabilitation and Bioengineering The University of Pittsburgh Pittsburgh Pennsylvania 15213 4 Departments of Biomechanical Engineering and Neurobiology Washington University St Louis Missouri 63130 Brain controlled interfaces are devices that capture brain transmissions involved in a subject s intention to act with the potential to restore communication and movement to those who are immobilized Current devices record electrical activity from the scalp on the surface of the brain and within the cerebral cortex These signals are being translated to command signals driving prosthetic limbs and computer displays Somatosensory feedback is being added to this control as generated behaviors become more complex New technology to engineer the tissue electrode interface electrode design and extraction algorithms to transform the recorded signal to movement will help translate exciting laboratory demonstrations to patient practice in the near future From the rapid growth in biotechnology neural engineering has emerged as a new field The merger of systems neurophysiology and engineering has resulted in approaches to link brain activity with man made devices to replace lost sensory and motor function The excitement in this field is based not only on the prospect of helping a wide range of patients with neural disorders but also on the certainty that this new technology will make it possible to gain scientific insight into the way populations of neurons interact in the complex distributed systems that generate behavior This review will address recent progress in cortical motor prosthetics Related reviews are also available Schwartz 2004 Wilson et al 2006 Lebedev and Nicolelis 2006 Leuthardt et al 2006 Neural prosthetics are devices that link machines to the nervous system for the purpose of restoring lost function Two broad approaches are used in this field neurons are stimulated or inhibited by applied current or their activity is recorded to intercept motor intention Stimulation can be used for its therapeutic efficacy as in deep brain stimulation to ameliorate the symptoms of Parkinson s disease or to communicate input to the nervous system for example by transforming sound to Correspondence abs21 pitt edu Review neural input with cochlear prosthetics In contrast recordings are used to decode ongoing activity for use as a command or input signal to an external device Capturing motor intention and executing the desired movement form the basis of brain controlled interfaces BCI a subset of neural prosthetics used to decode intention in order to restore motor ability or communication to impaired individuals Every BCI has four broad components recording of neural activity extraction of the intended action from that activity generation of the desired action with a prosthetic effector and feedback either through intact sensation such as vision or generated and applied by the prosthetic device Figure 1 Recording Technology The first step in the BCI process is to capture signals containing information about the subject s intended movement While researchers have envisioned using methods based on either magnetic Georgopoulos et al 2005 or electromagnetic Weiskopf et al 2004 Yoo et al 2004 signals from the brain these devices are not yet practical for BCI use Currently the four primary recording modalities are electroencephalography EEG electrocorticography ECoG local field potentials LFPs and single neuron action potential recordings single units All of these methods record microvolt level extracellular potentials generated by neurons in the cortical layers The methods are classified by whether the electrodes are placed on the scalp dura cortical surface or in the parenchyma and by the spatial and spectral frequency of their recorded signals Generally there is a tradeoff between these parameters the more invasive the recording technique the higher the spatial spectral frequency content of the recorded signal which in turn depends on the current densities conducted through the volume of the head The primary current sources and sinks i e where current enters the cell and leaves the cell respectively are synapses both excitatory and inhibitory and the voltage sensitive gates underlying neuronal action potentials Because most nonspherical neurons are oriented radially these currents approximate a dipole source which contains both equal and opposite polarities oriented perpendicular to the cortical surface Taken as a whole the cortex can be modeled as a thin convoluted sheet of aligned dipoles whose individual magnitudes vary continuously in time BCI recording aims to sample this dipole sheet and extract the desired control signal From a purely engineering point of view the optimal method of recording this electrical information would be to place a series of small electrodes directly into the dipole sheet to intercept signals from individual neurons single unit BCI designs The ability of a microelectrode to record single unit action potentials depends on many factors such as electrode impedance tip size and shape whether the target cell has an open or closed extracellular field and the size and orientation of the target neuron Layer V cells in the motor cortex have the largest Neuron 206 Figure 1 BCI Schematic Neural activity recorded from the brain is transmitted to a processor that operates an extraction algorithm on the recorded signal The extracted control signal is fed to a robot controller to move the prosthetic arm which generates feedback to close the control loop Electrode inset courtesy of Daryl Kipke University of Michigan cell bodies in the cerebrum 100 mm and generate large electrical fields making them an ideal source for extracellular recording Multisite silicon probes can record distinguishable spikes from layer V neurons in rat sensorimotor cortex located more than 300 mm away in the axial direction Buzsaki and Kandel 1998 although this distance is likely smaller in the tangential direction Henze et al 2000 At this