The GUIDE: Brain Interfaces (Part 2 of Bionics and Brain-Machine Interfaces)

http://teddysratlab.blogspot.com [Full link to blog for email clients.]

[BTW, I'm still dictating these blogs using Dragon NaturallySpeaking. I'm trying to be alert for typographical errors caused by the software, but if you are still slipping by. I will have about a week between the time I wrote this blog, and the due date for the next blog. Hopefully medication and therapy will allow me to get back to typing by that time.]

Brain machine interfaces are important not just for "bionics" but also for interfaces that would allow quadriplegics and those with neurodegenerative diseases to interface with the outside world. One of the key questions in developing a brain machine interface is: What is the nature of the information that is to be interfaced? With sensory information such as been discussed in the two previous blogs, we need to know two things: (1) how much information do we want to input to the brain, and (2) what pattern does the brain expect to see? With the visual and auditory prosthetic discussed in Monday's blog, researchers and doctors utilize residual neural function in the eye and ear, respectively, and simply provide a pattern of electrical stimulation that is "mapped" in a manner similar to the normal input to retina or cochlea. Tactile and proprioceptive sensation, as required for an effective bionic limb, is a little bit more complex, since the inputs will likely need to be transmitted directly to the appropriate area of the brain that normally receives input from limbs. Likewise advanced improvements in visual and auditory prosthetics – the degree of sophistication required to truly replace sight and sound – will also need to provide inputs directly to the visual and auditory processing regions of the brain.

This problem is already being faced by the researchers working on the Revolutionizing Prosthetics program by DARPA, and that direct brain control of motor movements also necessitates interfacing with the part of the brain that normally controls muscles. Both of these problems require a means of either transmitting any electrical stimulus directly to the brain or reading the brains electrical activity and translating it to an external device. in many ways this is a separate field of neuroscience, termed Brain Machine Interfacing.

Brain Machine Interfacing, or BMI, is a field with two major goals: (1) understanding the normal brain "code" that corresponds to sensation, movement, memory, or any other cognitive function; (2) providing an appropriate "interface" between neurons and electronics. In many ways part one of BMI is no different from any of the studies of neuroscience, in which we try to better understand the functions of the brain. However, the particular goal of BMI with understanding the brain is to be able to point to specific patterns of neural activity and say "this represents an input" or "this represents a memory."

"Utah array" http://www.sci.utah.edu/~gk/abstracts/bisti03/

The second half, or second challenge, of BMI is, in many ways, much harder even though it would seem to be the simpler problem. The type of electric is commonly used for recording neural activity is simply a wire – insulated along its length, with only a small area exposed at the tip. Sharpened tungsten electrodes may have an exposed surface of less than a micron, while stainless steel, nichrome, or platinum iridium wire, typically have surface areas of about 20 to 50 square microns. Since the diameter of a single neuron is itself about 20 square microns, one would think that these electrode sizes would be adequate for recording single neurons. In fact, the larger surface area of non-sharpened electrodes is more appropriate for recording multiple neurons, requiring software that can separate the activity of more than one neuron. While smaller diameter electrodes would seem to be more precise, and more accurate, they do not remain effective as long as non-sharpened electrodes. The major problem with this type of electrode is that either the wire or the brain tissue itself can move reducing the effectiveness of the electrode. Another problem is that the glial cells which provide metabolic support to neurons will tend to encapsulate and insulate foreign objects in the brain. When this happens to electrode its ability to record neural activity is decreased and eventually eliminated.

"Michigan arrays" by NeuroNexus

A much more recent approach is to develop recording electrodes that resemble printed circuits, with recording surfaces, and structural materials, that are inert. Silicon and ceramic substrates, with printed platinum "wiring" provide many advantages for BMI. In the first place, silica and ceramic and platinum are much less likely to cause inflammation and gliosis. In the second place, the "printed circuit" process allows manufacture of patterns of electrode-like recording sites arranged in grids that can stimulate many neurons in many different areas and and provides many orders of magnitude more information to be either recorded or stimulated. The "Utah Array" (http://www.sci.utah.edu/~gk/abstracts/bisti03/) shown above is one such array, while the "Michigan Arrays" (http://www.neuronexustech.com/) shown at left represent a more classic printed circuit approach. It is still the case that implanted electrodes have a finite duration of function. We still don't know entirely how long electrodes of this type will remain functional when implanted into the human brain. Even the patient's who will be working with the "bionic" arm prosthesis, will only have electrodes implanted into the brain for a few years. Research is ongoing to determine what is the "best" type of electrode that can be implanted once and remain functional for the lifetime of the patient.

The field of neural prosthetics requires a number of developments before we can reach the goal of true "bionic" prosthetic organs and limbs. In addition, creation of "cyborgs", or cybernetic organisms, is even further in the future, requiring great advances in interfacing between the brain and external devices. In the next blog, I will finish this series on advanced prosthetics, bionics, and brain machine interfaces with the discussion of techniques for interfacing brains and computers by means of "reading" brain activity without electrodes implanted in the brain.

Until next time, take care of your brain. It's better than a poke in the eye with a sharp… electrode!