Systems and methods for decoding intended motor commands from recorded neural signals for the control of external devices or to interact in virtual environments

Patent metadata

Field	Value
Publication number	US-9717440-B2
Application number	US-201414270016-A
Country	US
Kind code	B2
Filing date	May 5, 2014
Priority date	May 3, 2013
Publication date	Aug 1, 2017
Grant date	Aug 1, 2017

Abstract

Systems and methods for decoding neural and/or electromyographic signals are provided. A system can include at least one single channel decoder. Optionally, the system can include a demixer in operable communication with the single channel decoders. Each single channel decoder can include a filter to attenuate noise and sharpen spikes in the neural and/or electromyographic signals, a detection function to identify spikes, and a demodulator to get a real-time estimate of motor intent.

First claim

What is claimed is:

1. A system comprising:
one or more electrodes recording intended motor actions from input signals;
at least one single channel decoder (SCD) in operable communication with the one or more electrodes and configured to decode the input signals; and
a demixer in operable communication with the at least one SCD and identifying, sorting, and segregating the decoded input signals of the at least one SCD as motor intent signals corresponding to a particular class of motor actions through use of a learning stage, the demixer being further configured to generate a set of control signals and to output the control signals to an external device, the control signals being based on the motor intent signals, and the control signals controlling the external device,
the input signals including at least one of neural signals and muscle signals,
the at least one SCD down-sampling and compressing the input signals, and
the at least one SCD being configured to calculate an estimate of false positive rate by processing background noise after setting a threshold parameter and to use the estimate of false positive rate to reduce false positives during recording of the input signals and decoding of the input signals.

2. The system according to claim 1 , wherein each SCD comprises:
a filter configured to attenuate noise of the input signals and to sharpen spikes of the input signals corresponding to nerve firing;
a detector configured to identify spikes of the input signals corresponding to the nerve firing; and
a demodulator configured to generate an estimate of the motor intent signals.

3. The system according to claim 2 , wherein the filter utilizes a nonlinear reshaping function to attenuate noise of the input signals and to sharpen spikes of the input signals corresponding to the nerve firing.

4. The system according to claim 2 , wherein the demodulator is a single filter cascade of filters or a parallel set of filters, and wherein the demodulator is tunable such that it has a tunable time constant.

5. The system according to claim 1 , wherein the at least one SCD outputs the decoded signals to the demixer, and wherein the demixer is configured to identify the motor intent signals as corresponding to a degree of movement of the intended motor actions.

6. The system according to claim 5 , wherein the external device is a powered prosthesis, a television, a computer, a portable electronic device, a single degree-of-freedom robot, or a multiple degree-of-freedom robot.

7. The system according to claim 1 , further comprising a decoder configuration unit configured to train the demixer to identify the decoded signals received from the at least one SCD as motor intent signals corresponding to a degree of movement of the intended motor actions.

8. The system according to claim 1 , wherein the at least one SCD comprises a bank of bandpass filters configured to attenuate noise of the input signals and to sharpen spikes of the input signals corresponding to nerve firing.

9. A method comprising:
receiving input signals by at least one single channel decoder (SCD) configured to decode the input signals;
down-sampling and compressing the input signals by the at least one SCD;
calculating, by the at least one SCD, an estimate of false positive rate by processing background noise after setting a threshold parameter;
using, by the at least one SCD, the estimate of false positive rate to reduce false positives during recording of the input signals and decoding of the input signals;
generating an estimate of the intended motor actions based on the decoded input signals;
generating motor intent signals based on or comprising the generated estimate of the intended motor actions;
outputting the motor intent signals to a demixer in operable communication with the at least one SCD;
identifying, sorting, and segregating, by the demixer, the motor intent signals as corresponding to a particular class of motor actions through use of a learning stage, the input signals including at least one of neural signals and muscle signals; and
generating, by the demixer, a set of control signals and outputting the control signals to an external device, the control signals being based on the motor intent signals, and the control signals controlling the external device.

10. The method according to claim 9 , wherein each SCD comprises:
a filter attenuating noise of the input signals and sharpening spikes of the input signals corresponding to nerve firing;
a detector identifying spikes of the input signals corresponding to the nerve firing: and
a demodulator generating the estimate of the motor intent signals.

11. The method according to claim 10 , wherein the filter utilizes a nonlinear reshaping function to attenuate noise of the input signals and to sharpen spikes of the input signals corresponding to the nerve firing.

12. The method according to claim 10 , wherein the demodulator is a single filter cascade of filters or a parallel set of filters, and wherein the demodulator is tunable such that it has a tunable time constant.

13. The method according to claim 9 , wherein the demixer identifies the motor intent signals as corresponding to a degree of movement of the intended motor actions.

14. The method according to claim 13 , wherein the external device is a powered prosthesis, a television, a computer, a portable electronic device, a single degree-of-freedom robot, or a multiple degree-of-freedom robot.

15. The method according to claim 9 , further comprising training the demixer to identify the motor intent signals as corresponding to a degree of movement of the intended motor actions, wherein the demixer is trained by a decoder configuration unit.

16. The method according to claim 9 , further comprising:
recording the input signals by one or more electrodes in operable communication with the at least one SCD; and
transmitting the input signals from the one or more electrodes to the at least one SCD.

17. A system comprising:
one or more electrodes recording intended motor actions from input signals;
at least one single channel decoder (SCD) in operable communication with the one or more electrodes and configured to decode the input signals;
a demixer in operable communication with the at least one SCD and identifying, sorting, and segregating the decoded input signals of the at least one SCD as motor intent signals corresponding to a particular class of motor actions through use of a learning stage, the demixer being further configured to generate a set of control signals and to output the control signals to an external device, the control signals being based on the motor intent signals, and the control signals controlling the external device; and
a decoder configuration unit providing an initialization parameter to the at least one SCD and training the demixer during a training phase of the demixer, the input signals including at least one of neural signals and muscle signals, each SCD comprising:
a filter configured to attenuate noise of the input signals and to sharpen spikes of the input signals corresponding to nerve firing;
a detector configured to identify spikes of the input signals corresponding to the nerve firing; and
a demodulator configured to generate an estimate of the motor intent signals,

the at least one SCD outputting the estimated motor intent signals to the demixer, and the demixer being configured to identify the estimated motor intent signals as corresponding to a degree of movement of the intended motor actions,
each SCD down-sampling and compressing

Assignees

• Univ Florida Int

Inventors

• Abdelghani Mohamed
• Jung Ranu
• Abbas James J
• Horch Kenneth

Classifications

• A61F2/54

  Artificial arms or hands or parts thereof · CPC title

• A61B5/4851

  Prosthesis assessment or monitoring · CPC title

• A61B5/7203

  for noise prevention, reduction or removal · CPC title

• A61F2/72 Primary

  Bioelectric control, e.g. myoelectric · CPC title

• A61B5/0488

  Human Necessities · mapped topic