P.A. Parker

Improving myoelectric signal classification using wavelet packets and principal components analysis

An accurate and computationally efficient means of classifying surface myoelectric signals has been the subject of considerable research effort in recent years. This work demonstrates how this may be achieved, using a wavelet packet based feature set in conjunction with principal components analysis.

A dynamic feedforward neural network for subset classification of myoelectric signal patterns

Many biological signals are transient in nature, and the myoelectric signal (MES) is no exception. This is problematic for pattern classifiers that fail to incorporate the structure present in the temporal dimension of these signals. Standard feedforward neural network classifiers have difficulty processing temporal signals-time cannot be implicitly represented by the network architecture. A dynamic feedforward neural network architecture is described here that more effectively integrates the temporal information in transient signals, The internal representation of time also allows the dynamic network to classify subsets of the full temporal record. This reduces the time needed to obtain a classification result-an obvious benefit to real-time identification applications, such as the control of prosthetic devices.

Myoelectric signal classification using a finite impulse response neural network

Recent work by Hudgins (1993) has proposed a neural network-based approach to classifying the myoelectric signal (MES) elicited at the onset of movement of the upper limb. A standard feedforward artificial network was trained (using the backpropagation algorithm) to discriminate amongst four classes of upper-limb movements from the MES, acquired from the biceps and triceps muscles. The approach has demonstrated a powerful means of classifying limb function intent from the MES during natural muscular contraction, but the static nature of the network architecture fails to fully characterize the dynamic structure inherent in the MES. It has been demonstrated previously that a finite-impulse response (FIR) network has the ability to incorporate the temporal structure of a signal, representing the relationships between events in time and providing translation invariance of the relevant feature set. The application of this network architecture to limb function discrimination from the MES is described here.

A novel approach to localized muscle fatigue assessment

A method for generating a function which maps a set of surface myoelectric parameters to a fatigue index is proposed in this work. This forms the basis of a novel approach to assessing localized muscle fatigue with the myoelectric signal. An artificial neural network with a multilayer perceptron architecture was utilized to tune the function to emphasize trends in input parameters which are due to fatigue. The concept was tested empirically under static, cyclic, and random conditions. Results indicate improved performance when compared to fatigue assessment performance of mean frequency estimates.

A comparison between force and position control strategies in myoelectric prostheses

This work studies the simultaneous and proportional myoelectric force and position estimation of multiple degrees of freedom (DOFs) for unilateral transradial amputees. Two experiments were conducted to compare force and position control paradigms. In the first, a force experiment, subjects performed isometric contractions, while the force applied by the limb and EMG were recorded. In the second, a position experiment, dynamic contractions were permitted during which position of the limb and EMG were measured. Artificial neural networks (ANNs) were trained to estimate force/position from EMG of the contralateral limb during mirrored bilateral contractions. This study involved contractions with combined activations of three DOFs including wrist: flexion/extension, radial/ulnar deviation and forearm supination/pronation. For the given data set, while force estimation demonstrated high accuracy (R2=0.84±0.02), position estimation performance was relatively poor (R2=0.57±0.05). Two healthy subjects participated in this work.

The effects of force and joint angle on muscle conduction velocity estimation

Conduction velocity estimated from the surface myoelectric signal has been proposed as a potential index of fatigue for dynamic muscle contractions, in which joint angle and/or muscle force may be changing unpredictably. To be more useful as an index than power spectral parameters such as mean frequency, the conduction velocity estimate would have to be more resilient to changes in joint angle and/or muscle force. Results from this study using myoelectric signals collected from the biceps brachii, indicate that conduction velocity may indeed be more resilient to dynamic factors but also revealed that measurement techniques must be refined before reliable estimates can be obtained at joint angles in which extreme shortening of the muscle occurs.

Examining the mean frequency of myoelectric signals produced by dynamic muscle contractions

The mean frequency of the power spectrum of a myoelectric signal is an accepted index for monitoring fatigue in static contractions. There is, however, indication that it may be a useful index even in dynamic contractions. The purpose of this investigation was to examine this possibility. Results obtained while comparing mean frequency data produced from static and dynamic contractions reveal that it may be feasible to use mean frequency as an index for monitoring fatigue in dynamic contractions.

Adaptive stimulus artifact and ECG reduction in somatosensory evoked potential studies

Somatosensory Evoked Potentials (SEPs) are an important class of bioelectric signals which contain clinically valuable information. However, surface measurements of these signals are often contaminated by the stimulus artifact which, depending on the stimulating and recording measurement characteristics, may obscure some of the information contained in the SEPs. In addition, the SEP recordings on the spinal cord are also influenced by the more powerful ECG interference. The purpose of this paper is two fold-firstly, the authors apply a nonlinear adaptive filter based on the second order Volterra series to iteratively minimize the stimulus artifact. Secondly, a two stage adaptive filter structure is proposed to simultaneously reduce the ECG and stimulus artifact components for spinal cord SEPs. Preliminary experimental results showing the effectiveness of the proposed filter structures are included.

On the experimental results of a noninvasive estimation technique of muscle conduction velocity distribution

This paper is intended to present an assessment of the experimental results of a previously proposed muscle conduction velocity distribution (CVD) estimator. The performance of the proposed technique was seriously deteriorated when applied to experimental data. The causes for this decline were evaluated by introducing real-world errors in the model parameters and looking at how sensitive the estimator is to these. The simulation results show the high sensitivity of the estimator to parameter errors. The similitude found between the simulation results and the CVD estimates obtained on experimental data helped confirm this observation as well. Given the sensitivity observed the proposed technique is not practical for muscle dimensions such as those found in the biceps brachii.

A pattern based continuous multifunction myoelectric control strategy

A new technique to extract more control information from the myoelectric signal (MES) is introduced. The technique is based on the correlation of the MES obtained from a linear array of surface electrodes. The goal is to develop a continuous classifier of the MES to be used for myoelectric control.