Neuroelectric computational devices and networks

Details Neuroelectric computational devices and networks Neuroelectric computational devices and networks

2001-08-09

2003-11-25

Starks, Jr., Wilbert L. (Department: 2121)

Data processing: artificial intelligence

Neural network

C600S544000, C607S046000, C607S057000

Reexamination Certificate

active

06654729

ABSTRACT:

FIELD OF THE INVENTION
The present invention is related to hybrid to logic devices comprising biological and electrical components. The invention particularly relates to interfacing biological cells, such as neurons, with electrical circuitry.
BACKGROUND OF THE INVENTION
The concept of biological computing currently can be divided into at least four general approaches. The first of these are algorithms for neural networks based on schemes like backpropagation [Thimm et al., 1996; Erb, 1993]. These have seen wide use but are beginning to reach their limitations. The second approach involves promising new algorithms that utilize specific synaptic learning rules derived from cellular neurobiology and are now being developed for engineering applications [Granger et al. (1991); Ambrose-Ingerson et al. (1990)]. This olfactory-based algorithm substantially outperforms conventional Artificial Intelligence (AI) algorithms and neural networks in simulated space telemetry data having superimposed noise [Kowtha et al. (1995)] and in actual Specific Emitter Identification (SEI) using rigorous blind-test data in field conditions [Barrows et al. (1996)]. A third approach involves computation using DNA in a test tube [Adleman, L. (1994)] or possibly the manipulation of DNA in bacteria or other cells.
The limitation of these three approaches is that the system has order imposed on it from a predetermined pattern. However, if one could examine a reproducible system that controls the connections between living neurons, new paradigms in computing could be realized. Thus, a fourth approach involves the use of experiments on live animals, slice preparations, and in cultured networks [Gross, G. W. (1994)]. Many complicated and complex techniques have been developed to study neuronal networks in a living system using techniques such as MRI and PET, or in culture using dual patch-clamp and imaging systems. One of the difficulties arising from cultured networks is that the random spatial distributions and overlapping of dendrites and axons on homogeneous substrates in culture have historically made geometrically-dependent studies of synaptic function virtually impossible. The problem then becomes sorting out all of the complex connections and arrangements and making them reproducible. Consequently, a reductionist approach to this fourth method is desired which uses a minimum number of neurons to construct simple reproducible circuits and connect them to silicon devices. These new hybrid neuroelectric devices can then be connected in a multitude of configurations much like current computer chips without a predetermined hierarchial system.
SUMMARY OF THE INVENTION
The present invention is for neuroelectric components and logic devices which comprise one or more neurons having a predefined orientation on a substrate. One or more neural stimulating means, e.g., stimulator pads, electrodes, magnetic induction coils, and the like, are provided adjacent the neurons(s) and are capable of establishing a signal, e.g., an electrical signal therein. A transducer is provided adjacent at least one of the neurons and is capable of detecting the signal produced in the neuron. Typically, the neuronal cell(s) is/are hippocampal in origin.
It is preferred that the predefined orientation(s) of the neuron(s) on the substrate is/are set by the provision of a self-assembled monolayer (SAM) on the substrate in a predefined pattern, e.g., by providing a cell-repulsive surface at the periphery of the SAM. A particularly preferred self-assembled monolayer is composed of trimethoxysilylpropyl diethylene tetraamine (DETA), and a preferred cell-repulsive surface is provided by polyethylene glycol (molecular weight 550, i.e., PEG
550
). Methods for providing a SAM on the surface of a substrate are described by the present inventors in U.S. Ser. No. 08/689,970, the disclosure of which is incorporated herein by reference.
A neuroelectric device of the invention preferably comprises a gigaohm seal provided between the neuron and the substrate, which facilitates detection of a signal, e.g., an electrical signal, in the neuron with the transducer. It is preferred that the substrate on which the neuron rests comprises a layer of silicon dioxide and/or a layer of silicon nitride.
The neural stimulation means of the invention is typically an electrode, although other means are contemplated as long as they can produce a detectable signal in the neuron. Thus, chemical agents, such as ion channel blockers, magnetic fields, and the like, which can generate and/or sustain an electrical signal in the neuron are contemplated. The signal produced in the neuron and detected by the transducer can be a membrane current, membrane voltage, action potential, and the like. The stimulator can be formed integrally with the substrate, e.g., as a stimulator pad integrated with the substrate, or can be provided separately, e.g., by directly contacting the neuron from a side opposite the substrate.
A transducer of the invention is capable of detecting a signal, e.g., an electrical signal, propagating in a neuron of an instant neuroelectric device. Exemplary of such transducer is a field effect transistor (FET) or microelectrode array. The transducer can be provided either integrally with the substrate, or can be contacted directly with the neuron independent of the substrate. The neuroelectric device can further comprise a patch clamp attached to the neuron for measuring transmembrane potentials thereof.
Also contemplated is a neuroelectric logic device that comprises more than one of the above-mentioned neuroelectric devices. The neurons can be excitatory or inhibitory in nature. In one device, two neurons are provided on a substrate, and stimulators are provided adjacent the neurons. The stimulators are each capable of establishing a signal in the neurons. A transducer is provided adjacent one of the neurons and is capable of detecting, e.g. capacitatively, the signal propagating in the neuron. Thus, the neurons are in synaptic relationship, i.e., a synapse is defined between the neurons so that a signal established in one neuron can be attenuated upon stimulation of the other neuron.


REFERENCES:
patent: 4232680 (1980-11-01), Hudleson et al.
patent: RE31031 (1982-09-01), Kissiah, Jr.
patent: 4450530 (1984-05-01), Llinas et al.
patent: 4603703 (1986-08-01), McGill et al.
patent: 6011991 (2000-01-01), Mardirossian
Sehmi, A.S.; Jones , N.B.; Wang, S.Q.; Loudon, G.H., Knowledge-based systems for neuroelectric signal processing, Science, Measurement and Technology, IEE Proceedings- , vol. 141 Issue 3 , May 1994, pp. 215-223.*
Schapiro, D., A neurointelligence: multisignificance images and neurocomputing, Neural Networks, 1991., IJCNN-91-Seattle International Joint Conference on, vol.: ii, Jul. 8-14, 1991, pp. 948 vol. 2.

Affiliated with

Also associated with

Rating

Neuroelectric computational devices and networks does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Neuroelectric computational devices and networks, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Neuroelectric computational devices and networks will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3136615