Surgery – Diagnostic testing – Detecting muscle electrical signal
Reexamination Certificate
2002-01-29
2004-08-31
Jones, Mary Beth (Department: 3736)
Surgery
Diagnostic testing
Detecting muscle electrical signal
Reexamination Certificate
active
06785574
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a myoelectric feature-pattern classification method and apparatus in a method of interfacing muscle action potential (myoelectric pattern).
2. Description of the Prior Art
FIG. 2
is a drawing used to illustrate a prior-art apparatus that manipulates a control target by extracting a feature pattern from a myoelectric pattern and classifying the extracted feature pattern. In the drawing, {circle around (1)} denotes myoelectric patterns, {circle around (2)} a surface electrode group, {circle around (3)} amplification and smoothing apparatuses, {circle around (4)} a feature-pattern extraction apparatus, {circle around (5)} a pattern classifier, and {circle around (6)} control target such as a motor, a robot, a device for the disabled, a rehabilitation device, a myoelectric arm prosthesis, a game, and so forth.
As shown in the drawing, a myoelectric pattern {circle around (1)} that is an action potential generated by the coordinated action of a plurality of muscles, is measured by a set or a plurality of sets of surface electrode groups {circle around (2)} on a skin surface. What is measured is the sum of the action potentials generated by the plurality of muscles.
Next, the measured potential is subjected to amplification and smoothing by the amplification and smoothing apparatuses {circle around (3)} The feature-pattern extraction apparatus {circle around (4)} extracts a feature pattern from the amplified, smoothed signal. The pattern classifier {circle around (5)} classifies feature patterns from the extracted signal, and generates control signals to control the control target {circle around (6)}. A real value filter, such as a neural network or the like, or a logical value filter, such as a logic circuit or the like, can be used for the pattern classifier.
This type of prior-art apparatus requires, as the feature-pattern extraction apparatus, a high-level arithmetic processing device (a high-specification CPU or DSP, or a special LSI) for FFT calculations and the like and for solving inverse mapping problems, which has been a problem standing in the way of reducing the size and cost of the apparatus. This has prevented the apparatus coming into widespread use as a myoelectric pattern interface.
Sometimes a bias in myoelectric pattern distributions can make it difficult to achieve classification with good accuracy. It is possible that the bias in the distribution is caused by the distance relationship between the measurement electrodes and contracted muscle. That is, because a myoelectric pattern at the contraction of a muscle that is far from the electrodes is attenuated as it passes through the living body tissue, it distributes in low-value regions. In contrast to this, in the case of the contraction of a muscle that is close to the electrodes, there is little attenuation from the propagation in the body, so it distributes in high-value regions.
The present invention is proposed to resolve the above problems, and has as an object to provide a myoelectric feature-pattern classification method and apparatus that is smaller and cheaper, promoting the wider use of a myoelectric interface apparatus.
Another object of the present invention is to provide a myoelectric feature-pattern classification method and apparatus that, in cases where the myoelectric pattern distribution is biased, can reduce the distribution bias and increase the pattern classification accuracy.
SUMMARY OF THE INVENTION
The myoelectric pattern classification method of the present invention comprises using logarithmic transformation processing to extract a feature pattern from a myoelectric pattern that is a muscle action potential, classifying the extracted feature pattern and generating an output control signal.
Also, the myoelectric pattern classification apparatus of the present invention comprises a feature-pattern extraction apparatus that uses a logarithmic transformation apparatus to extract a feature pattern from a myoelectric pattern that is a muscle action potential, and a pattern classifier that classifies the extracted feature pattern and generates an output control signal.
The distance relationship between a contracted muscle and the electrodes can give rise to distributions that are biased towards low-value regions and high-value regions. In this case, it is possible to decrease the distribution bias and raise pattern classification accuracy by using logarithmic transformation to transform the patterns to increase the low-value region resolution and decrease the high-value region resolution.
The above logarithmic transformation can be realized by means of an analogue filter, or by software on a CPU or microprocessor, or by a lookup-table or the like. This makes it possible to reduce the size and cost, thereby promoting the wider use of a myoelectric interface apparatus.
Other objects and features of the invention will be more apparent from the following detailed description of the invention based on the accompanying drawings.
REFERENCES:
patent: 5376128 (1994-12-01), Bozeman, Jr.
patent: 6272479 (2001-08-01), Farry et al.
patent: 2002/0136328 (2002-09-01), Shimizu
I. Kajitani, et al., Technical Report of IEICE, PRMU98-86, vol. 98, No. 275, pp. 9-16, “Developing Adaptable EMG Controlled Prostheses to Disabled People”, Sep. 18, 1998 (with English abstract).
Higuchi Tetsuya
Kajitani Isamu
Jones Mary Beth
National Institute of Advanced Industrial Science and Technology
Szmal Brian
LandOfFree
Myoelectric feature-pattern classification method and apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Myoelectric feature-pattern classification method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Myoelectric feature-pattern classification method and apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3280848